{"step": 1, "loss": 10.539107322692871, "lr": 4.0000000000000003e-07, "elapsed_sec": 9.733261108398438, "step_time_sec": 9.675120422034524, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 12185815552}
{"step": 2, "loss": 10.53579330444336, "lr": 8.000000000000001e-07, "elapsed_sec": 17.95948886871338, "step_time_sec": 8.2260572189698, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3, "loss": 10.503414154052734, "lr": 1.2000000000000002e-06, "elapsed_sec": 26.174293518066406, "step_time_sec": 8.214545798953623, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4, "loss": 10.49658203125, "lr": 1.6000000000000001e-06, "elapsed_sec": 34.38943266868591, "step_time_sec": 8.21497500303667, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5, "loss": 10.501608848571777, "lr": 2e-06, "elapsed_sec": 42.60410141944885, "step_time_sec": 8.214507819036953, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6, "loss": 10.476670265197754, "lr": 2.4000000000000003e-06, "elapsed_sec": 50.81978511810303, "step_time_sec": 8.215544196078554, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7, "loss": 10.437216758728027, "lr": 2.8e-06, "elapsed_sec": 59.03629016876221, "step_time_sec": 8.216276907012798, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8, "loss": 10.430347442626953, "lr": 3.2000000000000003e-06, "elapsed_sec": 67.26639342308044, "step_time_sec": 8.229974735993892, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9, "loss": 10.39728832244873, "lr": 3.6000000000000003e-06, "elapsed_sec": 75.49767327308655, "step_time_sec": 8.231107785948552, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10, "loss": 10.373255729675293, "lr": 4e-06, "elapsed_sec": 83.72742033004761, "step_time_sec": 8.229578123078682, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 11, "loss": 10.332854270935059, "lr": 4.4e-06, "elapsed_sec": 91.95833969116211, "step_time_sec": 8.23074381693732, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 12, "loss": 10.282586097717285, "lr": 4.800000000000001e-06, "elapsed_sec": 100.18812370300293, "step_time_sec": 8.229597944067791, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 13, "loss": 10.279982566833496, "lr": 5.200000000000001e-06, "elapsed_sec": 108.419442653656, "step_time_sec": 8.231148642022163, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 14, "loss": 10.248394966125488, "lr": 5.6e-06, "elapsed_sec": 116.65028119087219, "step_time_sec": 8.230643838993274, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 15, "loss": 10.221309661865234, "lr": 6e-06, "elapsed_sec": 124.88065600395203, "step_time_sec": 8.230219306075014, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 16, "loss": 10.148496627807617, "lr": 6.4000000000000006e-06, "elapsed_sec": 133.11024284362793, "step_time_sec": 8.229417245020159, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 17, "loss": 10.154120445251465, "lr": 6.8e-06, "elapsed_sec": 141.3420970439911, "step_time_sec": 8.231696284958161, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 18, "loss": 10.0968599319458, "lr": 7.2000000000000005e-06, "elapsed_sec": 149.5726170539856, "step_time_sec": 8.230394055950455, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 19, "loss": 10.075146675109863, "lr": 7.6e-06, "elapsed_sec": 157.8025369644165, "step_time_sec": 8.229711522930302, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 20, "loss": 10.0713472366333, "lr": 8e-06, "elapsed_sec": 166.03104877471924, "step_time_sec": 8.228349619079381, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 21, "loss": 10.061881065368652, "lr": 8.400000000000001e-06, "elapsed_sec": 174.260267496109, "step_time_sec": 8.229061660007574, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 22, "loss": 9.98206615447998, "lr": 8.8e-06, "elapsed_sec": 182.48937249183655, "step_time_sec": 8.228944443981163, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 23, "loss": 9.94698715209961, "lr": 9.2e-06, "elapsed_sec": 190.71938180923462, "step_time_sec": 8.229847621987574, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 24, "loss": 9.949137687683105, "lr": 9.600000000000001e-06, "elapsed_sec": 198.9499855041504, "step_time_sec": 8.230416336096823, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 25, "loss": 9.931636810302734, "lr": 1e-05, "elapsed_sec": 207.17992782592773, "step_time_sec": 8.229762475006282, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 26, "loss": 9.870135307312012, "lr": 1.0400000000000002e-05, "elapsed_sec": 215.41092658042908, "step_time_sec": 8.230830056942068, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 27, "loss": 9.855440139770508, "lr": 1.08e-05, "elapsed_sec": 223.64143443107605, "step_time_sec": 8.230326906894334, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 28, "loss": 9.793110847473145, "lr": 1.12e-05, "elapsed_sec": 231.8719174861908, "step_time_sec": 8.230320382979698, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 29, "loss": 9.782800674438477, "lr": 1.16e-05, "elapsed_sec": 240.10306763648987, "step_time_sec": 8.230973205994815, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 30, "loss": 9.751052856445312, "lr": 1.2e-05, "elapsed_sec": 248.33366084098816, "step_time_sec": 8.230440537910908, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 31, "loss": 9.73832893371582, "lr": 1.2400000000000002e-05, "elapsed_sec": 256.5643434524536, "step_time_sec": 8.23048845701851, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 32, "loss": 9.699809074401855, "lr": 1.2800000000000001e-05, "elapsed_sec": 264.7950587272644, "step_time_sec": 8.2305478049675, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 33, "loss": 9.663117408752441, "lr": 1.32e-05, "elapsed_sec": 273.0230255126953, "step_time_sec": 8.22774191503413, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 34, "loss": 9.710238456726074, "lr": 1.36e-05, "elapsed_sec": 281.2507584095001, "step_time_sec": 8.22761196794454, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 35, "loss": 9.625821113586426, "lr": 1.4e-05, "elapsed_sec": 289.4792630672455, "step_time_sec": 8.228330043959431, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 36, "loss": 9.671537399291992, "lr": 1.4400000000000001e-05, "elapsed_sec": 297.7068099975586, "step_time_sec": 8.227390852989629, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 37, "loss": 9.53088092803955, "lr": 1.48e-05, "elapsed_sec": 305.93474435806274, "step_time_sec": 8.227777522057295, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 38, "loss": 9.591455459594727, "lr": 1.52e-05, "elapsed_sec": 314.16451382637024, "step_time_sec": 8.2295902390033, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 39, "loss": 9.477327346801758, "lr": 1.56e-05, "elapsed_sec": 322.39474058151245, "step_time_sec": 8.230050965910777, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 40, "loss": 9.506929397583008, "lr": 1.6e-05, "elapsed_sec": 330.62468814849854, "step_time_sec": 8.229801374021918, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 41, "loss": 9.473328590393066, "lr": 1.6400000000000002e-05, "elapsed_sec": 338.8532729148865, "step_time_sec": 8.228409347007982, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 42, "loss": 9.465479850769043, "lr": 1.6800000000000002e-05, "elapsed_sec": 347.08255076408386, "step_time_sec": 8.229142318014055, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 43, "loss": 9.565993309020996, "lr": 1.72e-05, "elapsed_sec": 355.312087059021, "step_time_sec": 8.229380047065206, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 44, "loss": 9.463728904724121, "lr": 1.76e-05, "elapsed_sec": 363.5418472290039, "step_time_sec": 8.229606995009817, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 45, "loss": 9.451974868774414, "lr": 1.8e-05, "elapsed_sec": 371.772127866745, "step_time_sec": 8.230087361996993, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 46, "loss": 9.36453914642334, "lr": 1.84e-05, "elapsed_sec": 380.0006721019745, "step_time_sec": 8.228444991982542, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 47, "loss": 9.370508193969727, "lr": 1.88e-05, "elapsed_sec": 388.2293474674225, "step_time_sec": 8.228458266938105, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 48, "loss": 9.407164573669434, "lr": 1.9200000000000003e-05, "elapsed_sec": 396.45833706855774, "step_time_sec": 8.228825489990413, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 49, "loss": 9.341413497924805, "lr": 1.96e-05, "elapsed_sec": 404.6904845237732, "step_time_sec": 8.23204569297377, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 50, "loss": 9.311105728149414, "lr": 2e-05, "elapsed_sec": 412.9187092781067, "step_time_sec": 8.227984802913852, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 51, "loss": 9.307572364807129, "lr": 2.04e-05, "elapsed_sec": 421.1488537788391, "step_time_sec": 8.230014849919826, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 52, "loss": 9.234882354736328, "lr": 2.0800000000000004e-05, "elapsed_sec": 429.3783769607544, "step_time_sec": 8.229315200005658, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 53, "loss": 9.281133651733398, "lr": 2.12e-05, "elapsed_sec": 437.60800647735596, "step_time_sec": 8.229510735953227, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 54, "loss": 9.354886054992676, "lr": 2.16e-05, "elapsed_sec": 445.83788228034973, "step_time_sec": 8.229729963932186, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 55, "loss": 9.473189353942871, "lr": 2.2000000000000003e-05, "elapsed_sec": 454.06711053848267, "step_time_sec": 8.229051279951818, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 56, "loss": 9.259405136108398, "lr": 2.24e-05, "elapsed_sec": 462.2966203689575, "step_time_sec": 8.229342162958346, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 57, "loss": 9.25098705291748, "lr": 2.2800000000000002e-05, "elapsed_sec": 470.52609372138977, "step_time_sec": 8.229345889994875, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 58, "loss": 9.155299186706543, "lr": 2.32e-05, "elapsed_sec": 478.75545597076416, "step_time_sec": 8.229201446985826, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 59, "loss": 9.223447799682617, "lr": 2.36e-05, "elapsed_sec": 486.9836838245392, "step_time_sec": 8.228102714987472, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 60, "loss": 9.254584312438965, "lr": 2.4e-05, "elapsed_sec": 495.2145903110504, "step_time_sec": 8.230714526027441, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 61, "loss": 9.146523475646973, "lr": 2.44e-05, "elapsed_sec": 503.4436786174774, "step_time_sec": 8.228921138099395, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 62, "loss": 9.25544548034668, "lr": 2.4800000000000003e-05, "elapsed_sec": 511.6732704639435, "step_time_sec": 8.229447748046368, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 63, "loss": 9.1367826461792, "lr": 2.52e-05, "elapsed_sec": 519.9027576446533, "step_time_sec": 8.22929126303643, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 64, "loss": 9.154520034790039, "lr": 2.5600000000000002e-05, "elapsed_sec": 528.13343334198, "step_time_sec": 8.230509752058424, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 65, "loss": 9.043805122375488, "lr": 2.6000000000000002e-05, "elapsed_sec": 536.3646492958069, "step_time_sec": 8.2310202709632, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 66, "loss": 9.071311950683594, "lr": 2.64e-05, "elapsed_sec": 544.594272851944, "step_time_sec": 8.229529200005345, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 67, "loss": 9.113080024719238, "lr": 2.68e-05, "elapsed_sec": 552.8243463039398, "step_time_sec": 8.229818204999901, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 68, "loss": 9.12352466583252, "lr": 2.72e-05, "elapsed_sec": 561.0535356998444, "step_time_sec": 8.229058580007404, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 69, "loss": 9.100510597229004, "lr": 2.7600000000000003e-05, "elapsed_sec": 569.2829012870789, "step_time_sec": 8.229179583024234, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 70, "loss": 9.050704956054688, "lr": 2.8e-05, "elapsed_sec": 577.5117712020874, "step_time_sec": 8.228754539974034, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 71, "loss": 8.965359687805176, "lr": 2.8400000000000003e-05, "elapsed_sec": 585.7441942691803, "step_time_sec": 8.232217984041199, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 72, "loss": 8.971807479858398, "lr": 2.8800000000000002e-05, "elapsed_sec": 593.9737112522125, "step_time_sec": 8.229407239938155, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 73, "loss": 9.027331352233887, "lr": 2.92e-05, "elapsed_sec": 602.203403711319, "step_time_sec": 8.229516891995445, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 74, "loss": 8.99355411529541, "lr": 2.96e-05, "elapsed_sec": 610.4341912269592, "step_time_sec": 8.230604627984576, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 75, "loss": 8.917706489562988, "lr": 3e-05, "elapsed_sec": 618.6683769226074, "step_time_sec": 8.23402290197555, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 76, "loss": 8.879059791564941, "lr": 3.04e-05, "elapsed_sec": 626.8967745304108, "step_time_sec": 8.228234504000284, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 77, "loss": 8.914286613464355, "lr": 3.08e-05, "elapsed_sec": 635.1274762153625, "step_time_sec": 8.230532860965468, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 78, "loss": 8.928058624267578, "lr": 3.12e-05, "elapsed_sec": 643.3593125343323, "step_time_sec": 8.231600785977207, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 79, "loss": 8.876300811767578, "lr": 3.16e-05, "elapsed_sec": 651.5885684490204, "step_time_sec": 8.229097898001783, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 80, "loss": 8.801525115966797, "lr": 3.2e-05, "elapsed_sec": 659.818521976471, "step_time_sec": 8.229776383028366, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 81, "loss": 8.835502624511719, "lr": 3.24e-05, "elapsed_sec": 668.0483062267303, "step_time_sec": 8.22963797999546, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 82, "loss": 8.833276748657227, "lr": 3.2800000000000004e-05, "elapsed_sec": 676.2781851291656, "step_time_sec": 8.229703977005556, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 83, "loss": 8.783249855041504, "lr": 3.32e-05, "elapsed_sec": 684.508050441742, "step_time_sec": 8.229713910957798, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 84, "loss": 8.785623550415039, "lr": 3.3600000000000004e-05, "elapsed_sec": 692.7411215305328, "step_time_sec": 8.23293456598185, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 85, "loss": 8.789057731628418, "lr": 3.4e-05, "elapsed_sec": 700.9759430885315, "step_time_sec": 8.234626766992733, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 86, "loss": 8.792498588562012, "lr": 3.44e-05, "elapsed_sec": 709.2083656787872, "step_time_sec": 8.232266106992029, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 87, "loss": 8.783129692077637, "lr": 3.4800000000000006e-05, "elapsed_sec": 717.4379556179047, "step_time_sec": 8.229439369984902, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 88, "loss": 8.684282302856445, "lr": 3.52e-05, "elapsed_sec": 725.6672866344452, "step_time_sec": 8.229169860947877, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 89, "loss": 8.713153839111328, "lr": 3.56e-05, "elapsed_sec": 733.8987336158752, "step_time_sec": 8.23131933703553, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 90, "loss": 8.684257507324219, "lr": 3.6e-05, "elapsed_sec": 742.1293592453003, "step_time_sec": 8.230470719980076, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 91, "loss": 8.659427642822266, "lr": 3.6400000000000004e-05, "elapsed_sec": 750.360137462616, "step_time_sec": 8.230533196940087, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 92, "loss": 8.604059219360352, "lr": 3.68e-05, "elapsed_sec": 758.5888338088989, "step_time_sec": 8.22851410903968, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 93, "loss": 8.64678955078125, "lr": 3.72e-05, "elapsed_sec": 766.8177559375763, "step_time_sec": 8.228798049967736, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 94, "loss": 8.564542770385742, "lr": 3.76e-05, "elapsed_sec": 775.0485396385193, "step_time_sec": 8.230623183073476, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 95, "loss": 8.588293075561523, "lr": 3.8e-05, "elapsed_sec": 783.2812831401825, "step_time_sec": 8.23252775601577, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 96, "loss": 8.488482475280762, "lr": 3.8400000000000005e-05, "elapsed_sec": 791.5151660442352, "step_time_sec": 8.233728392980993, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 97, "loss": 8.601215362548828, "lr": 3.88e-05, "elapsed_sec": 799.7452034950256, "step_time_sec": 8.229862420004793, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 98, "loss": 8.52197265625, "lr": 3.92e-05, "elapsed_sec": 807.9751467704773, "step_time_sec": 8.229785592062399, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 99, "loss": 8.550505638122559, "lr": 3.96e-05, "elapsed_sec": 816.2068917751312, "step_time_sec": 8.231562001048587, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 100, "loss": 8.4901704788208, "lr": 4e-05, "elapsed_sec": 824.4401209354401, "step_time_sec": 8.233086313935928, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 101, "loss": 8.435052871704102, "lr": 4.0400000000000006e-05, "elapsed_sec": 832.6740429401398, "step_time_sec": 8.233760240022093, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 102, "loss": 8.484026908874512, "lr": 4.08e-05, "elapsed_sec": 840.9076328277588, "step_time_sec": 8.233416056958959, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 103, "loss": 8.457459449768066, "lr": 4.12e-05, "elapsed_sec": 849.1377732753754, "step_time_sec": 8.23000082501676, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 104, "loss": 8.409107208251953, "lr": 4.160000000000001e-05, "elapsed_sec": 857.3688702583313, "step_time_sec": 8.23091804899741, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 105, "loss": 8.418124198913574, "lr": 4.2000000000000004e-05, "elapsed_sec": 865.5982096195221, "step_time_sec": 8.229201197973453, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 106, "loss": 8.382238388061523, "lr": 4.24e-05, "elapsed_sec": 873.8278901576996, "step_time_sec": 8.229530054959469, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 107, "loss": 8.491365432739258, "lr": 4.2800000000000004e-05, "elapsed_sec": 882.0576648712158, "step_time_sec": 8.229582852916792, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 108, "loss": 8.33547306060791, "lr": 4.32e-05, "elapsed_sec": 890.2871353626251, "step_time_sec": 8.229298896971159, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 109, "loss": 8.357933044433594, "lr": 4.36e-05, "elapsed_sec": 898.5174012184143, "step_time_sec": 8.230097998050041, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 110, "loss": 8.28494644165039, "lr": 4.4000000000000006e-05, "elapsed_sec": 906.7468786239624, "step_time_sec": 8.229352668044157, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 111, "loss": 8.328022956848145, "lr": 4.44e-05, "elapsed_sec": 914.982168674469, "step_time_sec": 8.235123711987399, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 112, "loss": 8.325965881347656, "lr": 4.48e-05, "elapsed_sec": 923.2175524234772, "step_time_sec": 8.23516470391769, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 113, "loss": 8.277388572692871, "lr": 4.52e-05, "elapsed_sec": 931.4521148204803, "step_time_sec": 8.234390683006495, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 114, "loss": 8.364906311035156, "lr": 4.5600000000000004e-05, "elapsed_sec": 939.6820123195648, "step_time_sec": 8.22974300396163, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 115, "loss": 8.313392639160156, "lr": 4.6e-05, "elapsed_sec": 947.9115715026855, "step_time_sec": 8.22940070705954, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 116, "loss": 8.344983100891113, "lr": 4.64e-05, "elapsed_sec": 956.1467735767365, "step_time_sec": 8.23503228300251, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 117, "loss": 8.173270225524902, "lr": 4.68e-05, "elapsed_sec": 964.3818304538727, "step_time_sec": 8.234921452938579, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 118, "loss": 8.262301445007324, "lr": 4.72e-05, "elapsed_sec": 972.6166172027588, "step_time_sec": 8.23463368800003, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 119, "loss": 8.261430740356445, "lr": 4.7600000000000005e-05, "elapsed_sec": 980.8465731143951, "step_time_sec": 8.229800538043492, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 120, "loss": 8.281596183776855, "lr": 4.8e-05, "elapsed_sec": 989.0762722492218, "step_time_sec": 8.229543361929245, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 121, "loss": 8.194727897644043, "lr": 4.8400000000000004e-05, "elapsed_sec": 997.3057708740234, "step_time_sec": 8.229356602067128, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 122, "loss": 8.307713508605957, "lr": 4.88e-05, "elapsed_sec": 1005.5402727127075, "step_time_sec": 8.234327338985167, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 123, "loss": 8.231022834777832, "lr": 4.92e-05, "elapsed_sec": 1013.7757561206818, "step_time_sec": 8.235309337032959, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 124, "loss": 8.104748725891113, "lr": 4.9600000000000006e-05, "elapsed_sec": 1022.0097997188568, "step_time_sec": 8.233869516989216, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 125, "loss": 8.12317943572998, "lr": 5e-05, "elapsed_sec": 1030.2452392578125, "step_time_sec": 8.23526058101561, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 126, "loss": 8.140400886535645, "lr": 5.04e-05, "elapsed_sec": 1038.4818933010101, "step_time_sec": 8.236557079013437, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 127, "loss": 8.187704086303711, "lr": 5.08e-05, "elapsed_sec": 1046.7181956768036, "step_time_sec": 8.2361207959475, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 128, "loss": 8.131332397460938, "lr": 5.1200000000000004e-05, "elapsed_sec": 1054.9529240131378, "step_time_sec": 8.234588107094169, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 129, "loss": 8.154043197631836, "lr": 5.16e-05, "elapsed_sec": 1063.1877129077911, "step_time_sec": 8.234576747985557, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 130, "loss": 8.03116226196289, "lr": 5.2000000000000004e-05, "elapsed_sec": 1071.4228959083557, "step_time_sec": 8.235025044996291, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 131, "loss": 8.079461097717285, "lr": 5.24e-05, "elapsed_sec": 1079.6570122241974, "step_time_sec": 8.233928972040303, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 132, "loss": 8.283520698547363, "lr": 5.28e-05, "elapsed_sec": 1087.8870165348053, "step_time_sec": 8.229836205020547, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 133, "loss": 8.1010160446167, "lr": 5.3200000000000006e-05, "elapsed_sec": 1096.1172924041748, "step_time_sec": 8.230129418079741, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 134, "loss": 8.105408668518066, "lr": 5.36e-05, "elapsed_sec": 1104.3536758422852, "step_time_sec": 8.23624179593753, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 135, "loss": 8.023012161254883, "lr": 5.4e-05, "elapsed_sec": 1112.5882108211517, "step_time_sec": 8.234344340977259, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 136, "loss": 8.04969310760498, "lr": 5.44e-05, "elapsed_sec": 1120.824298620224, "step_time_sec": 8.235944997053593, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 137, "loss": 8.074850082397461, "lr": 5.4800000000000004e-05, "elapsed_sec": 1129.0596704483032, "step_time_sec": 8.235232900013216, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 138, "loss": 8.065653800964355, "lr": 5.520000000000001e-05, "elapsed_sec": 1137.295574426651, "step_time_sec": 8.23566925607156, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 139, "loss": 8.009021759033203, "lr": 5.56e-05, "elapsed_sec": 1145.5316848754883, "step_time_sec": 8.235971308895387, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 140, "loss": 7.9645538330078125, "lr": 5.6e-05, "elapsed_sec": 1153.7682557106018, "step_time_sec": 8.236450005904771, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 141, "loss": 8.006223678588867, "lr": 5.640000000000001e-05, "elapsed_sec": 1162.0048015117645, "step_time_sec": 8.236298494972289, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 142, "loss": 7.9837822914123535, "lr": 5.6800000000000005e-05, "elapsed_sec": 1170.2402641773224, "step_time_sec": 8.235311568016186, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 143, "loss": 7.9925689697265625, "lr": 5.72e-05, "elapsed_sec": 1178.475553035736, "step_time_sec": 8.235149272019044, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 144, "loss": 7.960838317871094, "lr": 5.7600000000000004e-05, "elapsed_sec": 1186.710820198059, "step_time_sec": 8.235111679998226, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 145, "loss": 7.957767486572266, "lr": 5.8e-05, "elapsed_sec": 1194.9469511508942, "step_time_sec": 8.235999328899197, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 146, "loss": 7.8682475090026855, "lr": 5.84e-05, "elapsed_sec": 1203.1823110580444, "step_time_sec": 8.235142320976593, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 147, "loss": 7.928798675537109, "lr": 5.8800000000000006e-05, "elapsed_sec": 1211.4171316623688, "step_time_sec": 8.234640310052782, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 148, "loss": 7.978001594543457, "lr": 5.92e-05, "elapsed_sec": 1219.6531929969788, "step_time_sec": 8.235943146981299, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 149, "loss": 7.915317058563232, "lr": 5.96e-05, "elapsed_sec": 1227.8893146514893, "step_time_sec": 8.235957624972798, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 150, "loss": 8.006235122680664, "lr": 6e-05, "elapsed_sec": 1236.1247653961182, "step_time_sec": 8.235273530008271, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 151, "loss": 7.9694504737854, "lr": 6.0400000000000004e-05, "elapsed_sec": 1244.36079454422, "step_time_sec": 8.235901211039163, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 152, "loss": 7.917299747467041, "lr": 6.08e-05, "elapsed_sec": 1252.5957777500153, "step_time_sec": 8.234771713032387, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 153, "loss": 7.944338321685791, "lr": 6.120000000000001e-05, "elapsed_sec": 1260.8314146995544, "step_time_sec": 8.23551842093002, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 154, "loss": 7.90086555480957, "lr": 6.16e-05, "elapsed_sec": 1269.067151069641, "step_time_sec": 8.235592443030328, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 155, "loss": 7.925395488739014, "lr": 6.2e-05, "elapsed_sec": 1277.302642583847, "step_time_sec": 8.235342507949099, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 156, "loss": 7.946417331695557, "lr": 6.24e-05, "elapsed_sec": 1285.5333275794983, "step_time_sec": 8.230496954987757, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 157, "loss": 7.92129373550415, "lr": 6.280000000000001e-05, "elapsed_sec": 1293.7685854434967, "step_time_sec": 8.235121039906517, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 158, "loss": 7.970491409301758, "lr": 6.32e-05, "elapsed_sec": 1302.0036311149597, "step_time_sec": 8.234857951058075, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 159, "loss": 7.870279312133789, "lr": 6.36e-05, "elapsed_sec": 1310.2394478321075, "step_time_sec": 8.23559129203204, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 160, "loss": 7.875793933868408, "lr": 6.4e-05, "elapsed_sec": 1318.4754366874695, "step_time_sec": 8.235856438055634, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 161, "loss": 7.878444194793701, "lr": 6.44e-05, "elapsed_sec": 1326.7120745182037, "step_time_sec": 8.236468132934533, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 162, "loss": 7.95163631439209, "lr": 6.48e-05, "elapsed_sec": 1334.9472527503967, "step_time_sec": 8.235042824991979, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 163, "loss": 7.8591179847717285, "lr": 6.520000000000001e-05, "elapsed_sec": 1343.183177471161, "step_time_sec": 8.235698878997937, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 164, "loss": 7.8445634841918945, "lr": 6.560000000000001e-05, "elapsed_sec": 1351.4195597171783, "step_time_sec": 8.236200354993343, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 165, "loss": 7.865013599395752, "lr": 6.6e-05, "elapsed_sec": 1359.6560807228088, "step_time_sec": 8.236399946035817, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 166, "loss": 7.78811502456665, "lr": 6.64e-05, "elapsed_sec": 1367.8917820453644, "step_time_sec": 8.235500517999753, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 167, "loss": 7.757575988769531, "lr": 6.68e-05, "elapsed_sec": 1376.1263604164124, "step_time_sec": 8.234397879918106, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 168, "loss": 7.774377346038818, "lr": 6.720000000000001e-05, "elapsed_sec": 1384.3626067638397, "step_time_sec": 8.236134266015142, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 169, "loss": 7.743170261383057, "lr": 6.76e-05, "elapsed_sec": 1392.5981056690216, "step_time_sec": 8.235277237021364, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 170, "loss": 7.802235126495361, "lr": 6.8e-05, "elapsed_sec": 1400.8344130516052, "step_time_sec": 8.236155141959898, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 171, "loss": 7.757760524749756, "lr": 6.840000000000001e-05, "elapsed_sec": 1409.0701267719269, "step_time_sec": 8.235482597956434, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 172, "loss": 7.915954113006592, "lr": 6.88e-05, "elapsed_sec": 1417.3054497241974, "step_time_sec": 8.23515282291919, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 173, "loss": 7.767069339752197, "lr": 6.92e-05, "elapsed_sec": 1425.5417540073395, "step_time_sec": 8.236177698941901, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 174, "loss": 7.724915504455566, "lr": 6.960000000000001e-05, "elapsed_sec": 1433.7787339687347, "step_time_sec": 8.236838362063281, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 175, "loss": 7.792701244354248, "lr": 7.000000000000001e-05, "elapsed_sec": 1442.0142221450806, "step_time_sec": 8.235275772982277, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 176, "loss": 7.7491021156311035, "lr": 7.04e-05, "elapsed_sec": 1450.250904083252, "step_time_sec": 8.236523771076463, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 177, "loss": 7.788399696350098, "lr": 7.08e-05, "elapsed_sec": 1458.4884521961212, "step_time_sec": 8.237450203043409, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 178, "loss": 7.821609020233154, "lr": 7.12e-05, "elapsed_sec": 1466.7250320911407, "step_time_sec": 8.23639708396513, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 179, "loss": 7.771154403686523, "lr": 7.159999999999999e-05, "elapsed_sec": 1474.9615166187286, "step_time_sec": 8.236321287928149, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 180, "loss": 7.694756984710693, "lr": 7.2e-05, "elapsed_sec": 1483.197170972824, "step_time_sec": 8.235443238983862, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 181, "loss": 7.753084659576416, "lr": 7.240000000000001e-05, "elapsed_sec": 1491.433522939682, "step_time_sec": 8.236173350014724, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 182, "loss": 7.707655429840088, "lr": 7.280000000000001e-05, "elapsed_sec": 1499.6700477600098, "step_time_sec": 8.236355229979381, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 183, "loss": 7.655891418457031, "lr": 7.32e-05, "elapsed_sec": 1507.9004290103912, "step_time_sec": 8.23022257303819, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 184, "loss": 7.691425323486328, "lr": 7.36e-05, "elapsed_sec": 1516.1314179897308, "step_time_sec": 8.23085939604789, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 185, "loss": 7.677495956420898, "lr": 7.400000000000001e-05, "elapsed_sec": 1524.3669633865356, "step_time_sec": 8.235369349946268, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 186, "loss": 7.6411895751953125, "lr": 7.44e-05, "elapsed_sec": 1532.6015167236328, "step_time_sec": 8.234372657956555, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 187, "loss": 7.703181266784668, "lr": 7.48e-05, "elapsed_sec": 1540.8376824855804, "step_time_sec": 8.236024169018492, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 188, "loss": 7.72816276550293, "lr": 7.52e-05, "elapsed_sec": 1549.0730385780334, "step_time_sec": 8.23514459899161, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 189, "loss": 7.558027267456055, "lr": 7.56e-05, "elapsed_sec": 1557.3033237457275, "step_time_sec": 8.230124864028767, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 190, "loss": 7.628225326538086, "lr": 7.6e-05, "elapsed_sec": 1565.5330300331116, "step_time_sec": 8.229590683011338, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 191, "loss": 7.716914653778076, "lr": 7.640000000000001e-05, "elapsed_sec": 1573.7694499492645, "step_time_sec": 8.236162742017768, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 192, "loss": 7.632558345794678, "lr": 7.680000000000001e-05, "elapsed_sec": 1582.0058522224426, "step_time_sec": 8.236207133973949, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 193, "loss": 7.637203693389893, "lr": 7.72e-05, "elapsed_sec": 1590.2427036762238, "step_time_sec": 8.236750563024543, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 194, "loss": 7.669851303100586, "lr": 7.76e-05, "elapsed_sec": 1598.479726076126, "step_time_sec": 8.23677743691951, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 195, "loss": 7.555126667022705, "lr": 7.8e-05, "elapsed_sec": 1606.7162334918976, "step_time_sec": 8.236403286922723, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 196, "loss": 7.637351989746094, "lr": 7.84e-05, "elapsed_sec": 1614.952497959137, "step_time_sec": 8.23610580607783, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 197, "loss": 7.562748908996582, "lr": 7.88e-05, "elapsed_sec": 1623.189150094986, "step_time_sec": 8.236416495055892, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 198, "loss": 7.611955165863037, "lr": 7.92e-05, "elapsed_sec": 1631.426012992859, "step_time_sec": 8.236688711098395, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 199, "loss": 7.608701705932617, "lr": 7.960000000000001e-05, "elapsed_sec": 1639.6621043682098, "step_time_sec": 8.235933989984915, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 200, "loss": 7.529603481292725, "lr": 8e-05, "elapsed_sec": 1647.8980906009674, "step_time_sec": 8.235826782067306, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 201, "loss": 7.666306018829346, "lr": 8.04e-05, "elapsed_sec": 1656.1284744739532, "step_time_sec": 8.230243992991745, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 202, "loss": 7.555224895477295, "lr": 8.080000000000001e-05, "elapsed_sec": 1664.364725112915, "step_time_sec": 8.23609655990731, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 203, "loss": 7.529684066772461, "lr": 8.120000000000001e-05, "elapsed_sec": 1672.6006462574005, "step_time_sec": 8.235697382013313, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 204, "loss": 7.538083553314209, "lr": 8.16e-05, "elapsed_sec": 1680.8310403823853, "step_time_sec": 8.230244845035486, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 205, "loss": 7.486595630645752, "lr": 8.2e-05, "elapsed_sec": 1689.0665485858917, "step_time_sec": 8.23531981301494, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 206, "loss": 7.483986854553223, "lr": 8.24e-05, "elapsed_sec": 1697.3028898239136, "step_time_sec": 8.236198050086387, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 207, "loss": 7.465989589691162, "lr": 8.28e-05, "elapsed_sec": 1705.5332963466644, "step_time_sec": 8.230214801966213, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 208, "loss": 7.517803192138672, "lr": 8.320000000000002e-05, "elapsed_sec": 1713.7642395496368, "step_time_sec": 8.230792606016621, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 209, "loss": 7.504088401794434, "lr": 8.360000000000001e-05, "elapsed_sec": 1721.9952664375305, "step_time_sec": 8.2308700310532, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 210, "loss": 7.444090843200684, "lr": 8.400000000000001e-05, "elapsed_sec": 1730.2253108024597, "step_time_sec": 8.229858058970422, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 211, "loss": 7.627027988433838, "lr": 8.44e-05, "elapsed_sec": 1738.4561054706573, "step_time_sec": 8.230624223011546, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 212, "loss": 7.365017414093018, "lr": 8.48e-05, "elapsed_sec": 1746.6863253116608, "step_time_sec": 8.230106014059857, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 213, "loss": 7.423298358917236, "lr": 8.52e-05, "elapsed_sec": 1754.9231443405151, "step_time_sec": 8.236640508053824, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 214, "loss": 7.493337154388428, "lr": 8.560000000000001e-05, "elapsed_sec": 1763.1601448059082, "step_time_sec": 8.236793036921881, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 215, "loss": 7.3725175857543945, "lr": 8.6e-05, "elapsed_sec": 1771.3973970413208, "step_time_sec": 8.237100411090069, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 216, "loss": 7.358531951904297, "lr": 8.64e-05, "elapsed_sec": 1779.634266614914, "step_time_sec": 8.236702412948944, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 217, "loss": 7.436245441436768, "lr": 8.68e-05, "elapsed_sec": 1787.8714442253113, "step_time_sec": 8.236958406982012, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 218, "loss": 7.559871673583984, "lr": 8.72e-05, "elapsed_sec": 1796.108137845993, "step_time_sec": 8.236552693066187, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 219, "loss": 7.629457473754883, "lr": 8.76e-05, "elapsed_sec": 1804.338922739029, "step_time_sec": 8.230620488990098, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 220, "loss": 7.493618488311768, "lr": 8.800000000000001e-05, "elapsed_sec": 1812.5694687366486, "step_time_sec": 8.230370983015746, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 221, "loss": 7.412513256072998, "lr": 8.840000000000001e-05, "elapsed_sec": 1820.8028500080109, "step_time_sec": 8.233251990983263, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 222, "loss": 7.380938529968262, "lr": 8.88e-05, "elapsed_sec": 1829.0389766693115, "step_time_sec": 8.235927468980663, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 223, "loss": 7.357438087463379, "lr": 8.92e-05, "elapsed_sec": 1837.2761273384094, "step_time_sec": 8.236967997974716, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 224, "loss": 7.444799900054932, "lr": 8.96e-05, "elapsed_sec": 1845.5129170417786, "step_time_sec": 8.23664697795175, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 225, "loss": 7.475526809692383, "lr": 9e-05, "elapsed_sec": 1853.743658542633, "step_time_sec": 8.230581362033263, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 226, "loss": 7.351705074310303, "lr": 9.04e-05, "elapsed_sec": 1861.9740617275238, "step_time_sec": 8.230262703960761, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 227, "loss": 7.358314514160156, "lr": 9.080000000000001e-05, "elapsed_sec": 1870.2053763866425, "step_time_sec": 8.231183664058335, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 228, "loss": 7.329686164855957, "lr": 9.120000000000001e-05, "elapsed_sec": 1878.4416525363922, "step_time_sec": 8.236100874026306, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 229, "loss": 7.388152122497559, "lr": 9.16e-05, "elapsed_sec": 1886.678303718567, "step_time_sec": 8.236483121989295, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 230, "loss": 7.364672660827637, "lr": 9.2e-05, "elapsed_sec": 1894.915326833725, "step_time_sec": 8.236879442934878, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 231, "loss": 7.368950843811035, "lr": 9.240000000000001e-05, "elapsed_sec": 1903.1517012119293, "step_time_sec": 8.23619395494461, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 232, "loss": 7.323755741119385, "lr": 9.28e-05, "elapsed_sec": 1911.390845298767, "step_time_sec": 8.239002043032087, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 233, "loss": 7.305913925170898, "lr": 9.32e-05, "elapsed_sec": 1919.6286346912384, "step_time_sec": 8.237631493015215, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 234, "loss": 7.389130592346191, "lr": 9.36e-05, "elapsed_sec": 1927.866225719452, "step_time_sec": 8.237422210979275, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 235, "loss": 7.341746807098389, "lr": 9.4e-05, "elapsed_sec": 1936.1036567687988, "step_time_sec": 8.237257082946599, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 236, "loss": 7.308542251586914, "lr": 9.44e-05, "elapsed_sec": 1944.3411285877228, "step_time_sec": 8.237332516931929, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 237, "loss": 7.237793445587158, "lr": 9.480000000000001e-05, "elapsed_sec": 1952.5789618492126, "step_time_sec": 8.237621932057664, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 238, "loss": 7.248047351837158, "lr": 9.520000000000001e-05, "elapsed_sec": 1960.816400527954, "step_time_sec": 8.237284419010393, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 239, "loss": 7.334654331207275, "lr": 9.56e-05, "elapsed_sec": 1969.0599658489227, "step_time_sec": 8.24341259605717, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 240, "loss": 7.24605655670166, "lr": 9.6e-05, "elapsed_sec": 1977.2986845970154, "step_time_sec": 8.238563438993879, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 241, "loss": 7.236326694488525, "lr": 9.64e-05, "elapsed_sec": 1985.5367147922516, "step_time_sec": 8.237836209009402, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 242, "loss": 7.416637897491455, "lr": 9.680000000000001e-05, "elapsed_sec": 1993.769014596939, "step_time_sec": 8.232118629966862, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 243, "loss": 7.271644115447998, "lr": 9.72e-05, "elapsed_sec": 2002.0101363658905, "step_time_sec": 8.240971375023946, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 244, "loss": 7.292209625244141, "lr": 9.76e-05, "elapsed_sec": 2010.2415931224823, "step_time_sec": 8.231218529981561, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 245, "loss": 7.368719577789307, "lr": 9.800000000000001e-05, "elapsed_sec": 2018.4775195121765, "step_time_sec": 8.235758492955938, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 246, "loss": 7.210270404815674, "lr": 9.84e-05, "elapsed_sec": 2026.71541929245, "step_time_sec": 8.237725791987032, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 247, "loss": 7.240999698638916, "lr": 9.88e-05, "elapsed_sec": 2034.951789855957, "step_time_sec": 8.236220916034654, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 248, "loss": 7.272731304168701, "lr": 9.920000000000001e-05, "elapsed_sec": 2043.18705868721, "step_time_sec": 8.235061137005687, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 249, "loss": 7.159765243530273, "lr": 9.960000000000001e-05, "elapsed_sec": 2051.4233469963074, "step_time_sec": 8.236167161958292, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 250, "loss": 7.173710823059082, "lr": 0.0001, "elapsed_sec": 2059.6601610183716, "step_time_sec": 8.236635274952278, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 251, "loss": 7.252226829528809, "lr": 0.0001004, "elapsed_sec": 2067.896965265274, "step_time_sec": 8.23662374401465, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 252, "loss": 7.235782623291016, "lr": 0.0001008, "elapsed_sec": 2076.1333813667297, "step_time_sec": 8.236284082988277, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 253, "loss": 7.237431526184082, "lr": 0.0001012, "elapsed_sec": 2084.370173931122, "step_time_sec": 8.236598285031505, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 254, "loss": 7.200732231140137, "lr": 0.0001016, "elapsed_sec": 2092.606856107712, "step_time_sec": 8.236546745989472, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 255, "loss": 7.227088928222656, "lr": 0.00010200000000000001, "elapsed_sec": 2100.843894958496, "step_time_sec": 8.236841722973622, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 256, "loss": 7.1914215087890625, "lr": 0.00010240000000000001, "elapsed_sec": 2109.079578638077, "step_time_sec": 8.235487664001994, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 257, "loss": 7.227362632751465, "lr": 0.0001028, "elapsed_sec": 2117.3172373771667, "step_time_sec": 8.237527185003273, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 258, "loss": 7.198492527008057, "lr": 0.0001032, "elapsed_sec": 2125.5529839992523, "step_time_sec": 8.235568108968437, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 259, "loss": 7.157867431640625, "lr": 0.00010360000000000001, "elapsed_sec": 2133.7899100780487, "step_time_sec": 8.236813758034259, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 260, "loss": 7.105377674102783, "lr": 0.00010400000000000001, "elapsed_sec": 2142.027000427246, "step_time_sec": 8.23689663503319, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 261, "loss": 7.2495927810668945, "lr": 0.0001044, "elapsed_sec": 2150.2640738487244, "step_time_sec": 8.236923518008552, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 262, "loss": 7.224955081939697, "lr": 0.0001048, "elapsed_sec": 2158.5010364055634, "step_time_sec": 8.236821350990795, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 263, "loss": 7.223755359649658, "lr": 0.0001052, "elapsed_sec": 2166.736885070801, "step_time_sec": 8.235714065958746, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 264, "loss": 7.2578864097595215, "lr": 0.0001056, "elapsed_sec": 2174.974158525467, "step_time_sec": 8.237108892062679, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 265, "loss": 7.246995449066162, "lr": 0.00010600000000000002, "elapsed_sec": 2183.21014046669, "step_time_sec": 8.235785232041962, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 266, "loss": 7.2641425132751465, "lr": 0.00010640000000000001, "elapsed_sec": 2191.4418540000916, "step_time_sec": 8.23157730710227, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 267, "loss": 7.115641117095947, "lr": 0.00010680000000000001, "elapsed_sec": 2199.6729288101196, "step_time_sec": 8.230897141969763, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 268, "loss": 7.204181671142578, "lr": 0.0001072, "elapsed_sec": 2207.9102873802185, "step_time_sec": 8.237189489998855, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 269, "loss": 7.160394191741943, "lr": 0.0001076, "elapsed_sec": 2216.146405696869, "step_time_sec": 8.236025557969697, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 270, "loss": 7.287374496459961, "lr": 0.000108, "elapsed_sec": 2224.383688688278, "step_time_sec": 8.237073213909753, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 271, "loss": 7.28823184967041, "lr": 0.0001084, "elapsed_sec": 2232.6207287311554, "step_time_sec": 8.236920805997215, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 272, "loss": 7.138877868652344, "lr": 0.0001088, "elapsed_sec": 2240.857522010803, "step_time_sec": 8.23659317707643, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 273, "loss": 7.20466947555542, "lr": 0.00010920000000000001, "elapsed_sec": 2249.095503091812, "step_time_sec": 8.237790085026063, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 274, "loss": 7.068554401397705, "lr": 0.00010960000000000001, "elapsed_sec": 2257.333078145981, "step_time_sec": 8.237427112064324, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 275, "loss": 7.125292778015137, "lr": 0.00011, "elapsed_sec": 2265.570314884186, "step_time_sec": 8.237087560934015, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 276, "loss": 7.143900394439697, "lr": 0.00011040000000000001, "elapsed_sec": 2273.8070158958435, "step_time_sec": 8.23653303005267, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 277, "loss": 7.114806175231934, "lr": 0.00011080000000000001, "elapsed_sec": 2282.044256925583, "step_time_sec": 8.237108998931944, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 278, "loss": 7.146083354949951, "lr": 0.0001112, "elapsed_sec": 2290.281383037567, "step_time_sec": 8.236957787070423, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 279, "loss": 7.087667465209961, "lr": 0.0001116, "elapsed_sec": 2298.5186398029327, "step_time_sec": 8.237103231949732, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 280, "loss": 7.057172775268555, "lr": 0.000112, "elapsed_sec": 2306.7555515766144, "step_time_sec": 8.236765749985352, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 281, "loss": 7.087527275085449, "lr": 0.0001124, "elapsed_sec": 2314.991517305374, "step_time_sec": 8.235772575018927, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 282, "loss": 7.117376327514648, "lr": 0.00011280000000000002, "elapsed_sec": 2323.228574991226, "step_time_sec": 8.236856975010596, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 283, "loss": 7.0000810623168945, "lr": 0.00011320000000000001, "elapsed_sec": 2331.4647319316864, "step_time_sec": 8.236029071966186, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 284, "loss": 7.0375752449035645, "lr": 0.00011360000000000001, "elapsed_sec": 2339.70245218277, "step_time_sec": 8.237523226998746, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 285, "loss": 7.2242431640625, "lr": 0.000114, "elapsed_sec": 2347.9395191669464, "step_time_sec": 8.23687198094558, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 286, "loss": 7.0428900718688965, "lr": 0.0001144, "elapsed_sec": 2356.177042245865, "step_time_sec": 8.237421759986319, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 287, "loss": 7.173214912414551, "lr": 0.0001148, "elapsed_sec": 2364.4132146835327, "step_time_sec": 8.235960630932823, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 288, "loss": 7.083488464355469, "lr": 0.00011520000000000001, "elapsed_sec": 2372.6497790813446, "step_time_sec": 8.236407749936916, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 289, "loss": 7.075937747955322, "lr": 0.0001156, "elapsed_sec": 2380.8872623443604, "step_time_sec": 8.237357721896842, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 290, "loss": 7.073318958282471, "lr": 0.000116, "elapsed_sec": 2389.1239652633667, "step_time_sec": 8.236514632008038, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 291, "loss": 6.99983024597168, "lr": 0.0001164, "elapsed_sec": 2397.361657142639, "step_time_sec": 8.237561082001776, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 292, "loss": 6.977555274963379, "lr": 0.0001168, "elapsed_sec": 2405.599137067795, "step_time_sec": 8.237327820970677, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 293, "loss": 7.061923503875732, "lr": 0.00011720000000000002, "elapsed_sec": 2413.8375160694122, "step_time_sec": 8.238273515948094, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 294, "loss": 7.043135166168213, "lr": 0.00011760000000000001, "elapsed_sec": 2422.0748529434204, "step_time_sec": 8.237132743000984, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 295, "loss": 6.9927897453308105, "lr": 0.00011800000000000001, "elapsed_sec": 2430.3127756118774, "step_time_sec": 8.237804475007579, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 296, "loss": 7.199154853820801, "lr": 0.0001184, "elapsed_sec": 2438.55096244812, "step_time_sec": 8.238034346024506, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 297, "loss": 7.040242671966553, "lr": 0.0001188, "elapsed_sec": 2446.788311958313, "step_time_sec": 8.237157522933558, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 298, "loss": 6.968643665313721, "lr": 0.0001192, "elapsed_sec": 2455.0256974697113, "step_time_sec": 8.237235298962332, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 299, "loss": 7.040628910064697, "lr": 0.00011960000000000001, "elapsed_sec": 2463.263387441635, "step_time_sec": 8.237507028039545, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 300, "loss": 7.061665058135986, "lr": 0.00012, "elapsed_sec": 2471.5012118816376, "step_time_sec": 8.237656341982074, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 301, "loss": 7.038957595825195, "lr": 0.00012040000000000001, "elapsed_sec": 2479.7394285202026, "step_time_sec": 8.238075105007738, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 302, "loss": 7.169844627380371, "lr": 0.00012080000000000001, "elapsed_sec": 2487.977447986603, "step_time_sec": 8.237869509030133, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 303, "loss": 6.92617130279541, "lr": 0.0001212, "elapsed_sec": 2496.214976787567, "step_time_sec": 8.237402145983651, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 304, "loss": 6.978201866149902, "lr": 0.0001216, "elapsed_sec": 2504.452135324478, "step_time_sec": 8.236944233067334, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 305, "loss": 7.033401966094971, "lr": 0.00012200000000000001, "elapsed_sec": 2512.689721107483, "step_time_sec": 8.237470485968515, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 306, "loss": 6.883700370788574, "lr": 0.00012240000000000002, "elapsed_sec": 2520.927220582962, "step_time_sec": 8.237286882009357, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 307, "loss": 6.939644813537598, "lr": 0.0001228, "elapsed_sec": 2529.164583683014, "step_time_sec": 8.237188277067617, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 308, "loss": 6.989223003387451, "lr": 0.0001232, "elapsed_sec": 2537.4046778678894, "step_time_sec": 8.239946427987888, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 309, "loss": 7.021200656890869, "lr": 0.0001236, "elapsed_sec": 2545.642708301544, "step_time_sec": 8.237884926027618, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 310, "loss": 6.967186450958252, "lr": 0.000124, "elapsed_sec": 2553.8795087337494, "step_time_sec": 8.236628390965052, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 311, "loss": 6.995144844055176, "lr": 0.00012440000000000002, "elapsed_sec": 2562.117650747299, "step_time_sec": 8.237961129052565, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 312, "loss": 7.027499675750732, "lr": 0.0001248, "elapsed_sec": 2570.355971097946, "step_time_sec": 8.23818097205367, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 313, "loss": 7.154074192047119, "lr": 0.0001252, "elapsed_sec": 2578.593569278717, "step_time_sec": 8.23742341005709, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 314, "loss": 6.932854175567627, "lr": 0.00012560000000000002, "elapsed_sec": 2586.830732345581, "step_time_sec": 8.237011575023644, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 315, "loss": 7.090041160583496, "lr": 0.000126, "elapsed_sec": 2595.069032430649, "step_time_sec": 8.238155759987421, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 316, "loss": 6.936645030975342, "lr": 0.0001264, "elapsed_sec": 2603.3062584400177, "step_time_sec": 8.237028684001416, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 317, "loss": 7.000851154327393, "lr": 0.0001268, "elapsed_sec": 2611.542401790619, "step_time_sec": 8.235982705024071, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 318, "loss": 6.908951759338379, "lr": 0.0001272, "elapsed_sec": 2619.7793889045715, "step_time_sec": 8.236867806990631, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 319, "loss": 7.011984348297119, "lr": 0.0001276, "elapsed_sec": 2628.017997741699, "step_time_sec": 8.238405750016682, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 320, "loss": 6.928540229797363, "lr": 0.000128, "elapsed_sec": 2636.2562913894653, "step_time_sec": 8.238185219001025, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 321, "loss": 6.931913375854492, "lr": 0.0001284, "elapsed_sec": 2644.492279291153, "step_time_sec": 8.235781451920047, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 322, "loss": 6.9498291015625, "lr": 0.0001288, "elapsed_sec": 2652.730464696884, "step_time_sec": 8.238060088944621, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 323, "loss": 7.036216735839844, "lr": 0.0001292, "elapsed_sec": 2660.96680188179, "step_time_sec": 8.236146120005287, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 324, "loss": 6.960773944854736, "lr": 0.0001296, "elapsed_sec": 2669.2047040462494, "step_time_sec": 8.237777328002267, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 325, "loss": 7.112456798553467, "lr": 0.00013000000000000002, "elapsed_sec": 2677.440399646759, "step_time_sec": 8.23555243294686, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 326, "loss": 6.872456073760986, "lr": 0.00013040000000000003, "elapsed_sec": 2685.677226781845, "step_time_sec": 8.23662401502952, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 327, "loss": 6.952887535095215, "lr": 0.0001308, "elapsed_sec": 2693.914521932602, "step_time_sec": 8.237143768928945, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 328, "loss": 6.824867248535156, "lr": 0.00013120000000000002, "elapsed_sec": 2702.1514298915863, "step_time_sec": 8.236750329029746, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 329, "loss": 6.84409761428833, "lr": 0.0001316, "elapsed_sec": 2710.3904349803925, "step_time_sec": 8.238831738010049, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 330, "loss": 7.051061153411865, "lr": 0.000132, "elapsed_sec": 2718.6283917427063, "step_time_sec": 8.237801786977798, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 331, "loss": 7.055737495422363, "lr": 0.00013240000000000002, "elapsed_sec": 2726.865723133087, "step_time_sec": 8.237157777999528, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 332, "loss": 6.907621383666992, "lr": 0.0001328, "elapsed_sec": 2735.103311777115, "step_time_sec": 8.237416451098397, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 333, "loss": 6.91504430770874, "lr": 0.0001332, "elapsed_sec": 2743.341529607773, "step_time_sec": 8.238095209002495, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 334, "loss": 6.878179550170898, "lr": 0.0001336, "elapsed_sec": 2751.580011367798, "step_time_sec": 8.238259966019541, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 335, "loss": 6.880058288574219, "lr": 0.000134, "elapsed_sec": 2759.817216396332, "step_time_sec": 8.237053257063963, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 336, "loss": 6.880241870880127, "lr": 0.00013440000000000001, "elapsed_sec": 2768.056448698044, "step_time_sec": 8.239075970021077, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 337, "loss": 6.78761100769043, "lr": 0.0001348, "elapsed_sec": 2776.293669939041, "step_time_sec": 8.237117476994172, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 338, "loss": 7.177883625030518, "lr": 0.0001352, "elapsed_sec": 2784.5314245224, "step_time_sec": 8.237540858099237, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 339, "loss": 6.906768798828125, "lr": 0.0001356, "elapsed_sec": 2792.7680699825287, "step_time_sec": 8.236483395914547, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 340, "loss": 6.8012919425964355, "lr": 0.000136, "elapsed_sec": 2801.0047476291656, "step_time_sec": 8.236505364999175, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 341, "loss": 6.846848964691162, "lr": 0.00013639999999999998, "elapsed_sec": 2809.241312265396, "step_time_sec": 8.236417555017397, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 342, "loss": 7.212678909301758, "lr": 0.00013680000000000002, "elapsed_sec": 2817.4778320789337, "step_time_sec": 8.236356924986467, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 343, "loss": 6.9733710289001465, "lr": 0.00013720000000000003, "elapsed_sec": 2825.714985370636, "step_time_sec": 8.236955779022537, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 344, "loss": 6.968395233154297, "lr": 0.0001376, "elapsed_sec": 2833.9526088237762, "step_time_sec": 8.237513886066154, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 345, "loss": 6.926653861999512, "lr": 0.00013800000000000002, "elapsed_sec": 2842.1897983551025, "step_time_sec": 8.23702121002134, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 346, "loss": 6.764826774597168, "lr": 0.0001384, "elapsed_sec": 2850.427537202835, "step_time_sec": 8.237594656995498, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 347, "loss": 6.8832573890686035, "lr": 0.0001388, "elapsed_sec": 2858.6652994155884, "step_time_sec": 8.237646732013673, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 348, "loss": 6.831094264984131, "lr": 0.00013920000000000002, "elapsed_sec": 2866.902891397476, "step_time_sec": 8.237376480945386, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 349, "loss": 6.810600280761719, "lr": 0.0001396, "elapsed_sec": 2875.139073610306, "step_time_sec": 8.23601647594478, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 350, "loss": 6.928903102874756, "lr": 0.00014000000000000001, "elapsed_sec": 2883.37694978714, "step_time_sec": 8.237727488973178, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 351, "loss": 6.86454439163208, "lr": 0.0001404, "elapsed_sec": 2891.61434173584, "step_time_sec": 8.237192229018547, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 352, "loss": 6.886309623718262, "lr": 0.0001408, "elapsed_sec": 2899.8529176712036, "step_time_sec": 8.238473775913008, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 353, "loss": 6.858246326446533, "lr": 0.00014120000000000002, "elapsed_sec": 2908.089900970459, "step_time_sec": 8.236780661973171, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 354, "loss": 6.8368659019470215, "lr": 0.0001416, "elapsed_sec": 2916.3274507522583, "step_time_sec": 8.237433210015297, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 355, "loss": 6.818289756774902, "lr": 0.000142, "elapsed_sec": 2924.565118074417, "step_time_sec": 8.237454132991843, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 356, "loss": 6.847574234008789, "lr": 0.0001424, "elapsed_sec": 2932.8013064861298, "step_time_sec": 8.236041245982051, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 357, "loss": 6.815258026123047, "lr": 0.0001428, "elapsed_sec": 2941.0386526584625, "step_time_sec": 8.237178246024996, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 358, "loss": 6.938912868499756, "lr": 0.00014319999999999998, "elapsed_sec": 2949.2749531269073, "step_time_sec": 8.236129248980433, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 359, "loss": 6.865554332733154, "lr": 0.0001436, "elapsed_sec": 2957.5125918388367, "step_time_sec": 8.237464934005402, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 360, "loss": 6.854211807250977, "lr": 0.000144, "elapsed_sec": 2965.7498359680176, "step_time_sec": 8.237130169989541, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 361, "loss": 6.912259578704834, "lr": 0.0001444, "elapsed_sec": 2973.986878633499, "step_time_sec": 8.236890551983379, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 362, "loss": 6.75367546081543, "lr": 0.00014480000000000002, "elapsed_sec": 2982.2252295017242, "step_time_sec": 8.238170246011578, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 363, "loss": 6.743656158447266, "lr": 0.0001452, "elapsed_sec": 2990.461812019348, "step_time_sec": 8.23642872099299, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 364, "loss": 6.937467575073242, "lr": 0.00014560000000000002, "elapsed_sec": 2998.699556350708, "step_time_sec": 8.237594480975531, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 365, "loss": 6.891429424285889, "lr": 0.00014600000000000003, "elapsed_sec": 3006.936234474182, "step_time_sec": 8.236486479989253, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 366, "loss": 6.882867813110352, "lr": 0.0001464, "elapsed_sec": 3015.173914909363, "step_time_sec": 8.237504578079097, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 367, "loss": 6.789327144622803, "lr": 0.00014680000000000002, "elapsed_sec": 3023.411562681198, "step_time_sec": 8.237489591003396, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 368, "loss": 6.823227405548096, "lr": 0.0001472, "elapsed_sec": 3031.650829076767, "step_time_sec": 8.239115549018607, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 369, "loss": 6.920307159423828, "lr": 0.0001476, "elapsed_sec": 3039.889258623123, "step_time_sec": 8.238253071904182, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 370, "loss": 6.819150924682617, "lr": 0.00014800000000000002, "elapsed_sec": 3048.1261496543884, "step_time_sec": 8.236729111988097, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 371, "loss": 6.8138933181762695, "lr": 0.0001484, "elapsed_sec": 3056.364159345627, "step_time_sec": 8.237876413040794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 372, "loss": 6.753347396850586, "lr": 0.0001488, "elapsed_sec": 3064.6017587184906, "step_time_sec": 8.23738168599084, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 373, "loss": 6.747961521148682, "lr": 0.0001492, "elapsed_sec": 3072.8394107818604, "step_time_sec": 8.237494996981695, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 374, "loss": 6.781036376953125, "lr": 0.0001496, "elapsed_sec": 3081.0770637989044, "step_time_sec": 8.237476373906247, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 375, "loss": 6.836464881896973, "lr": 0.00015, "elapsed_sec": 3089.3140726089478, "step_time_sec": 8.2368681919761, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 376, "loss": 6.804327964782715, "lr": 0.0001504, "elapsed_sec": 3097.551511287689, "step_time_sec": 8.237269740086049, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 377, "loss": 6.848986625671387, "lr": 0.0001508, "elapsed_sec": 3105.786679506302, "step_time_sec": 8.235029730014503, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 378, "loss": 6.7710041999816895, "lr": 0.0001512, "elapsed_sec": 3114.0245995521545, "step_time_sec": 8.237725957063958, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 379, "loss": 6.773014545440674, "lr": 0.00015160000000000003, "elapsed_sec": 3122.2614505290985, "step_time_sec": 8.236708721029572, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 380, "loss": 6.857678413391113, "lr": 0.000152, "elapsed_sec": 3130.4982805252075, "step_time_sec": 8.236682628979906, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 381, "loss": 6.727198123931885, "lr": 0.00015240000000000002, "elapsed_sec": 3138.738124370575, "step_time_sec": 8.239692176110111, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 382, "loss": 6.722187519073486, "lr": 0.00015280000000000003, "elapsed_sec": 3146.97541308403, "step_time_sec": 8.237074585049413, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 383, "loss": 6.728661060333252, "lr": 0.0001532, "elapsed_sec": 3155.2117834091187, "step_time_sec": 8.236210584058426, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 384, "loss": 6.756375789642334, "lr": 0.00015360000000000002, "elapsed_sec": 3163.448601961136, "step_time_sec": 8.236661633010954, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 385, "loss": 6.77174186706543, "lr": 0.000154, "elapsed_sec": 3171.686551809311, "step_time_sec": 8.237770728999749, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 386, "loss": 6.768818378448486, "lr": 0.0001544, "elapsed_sec": 3179.9239695072174, "step_time_sec": 8.23725615197327, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 387, "loss": 6.717320442199707, "lr": 0.00015480000000000002, "elapsed_sec": 3188.162241458893, "step_time_sec": 8.23812788608484, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 388, "loss": 6.866880893707275, "lr": 0.0001552, "elapsed_sec": 3196.3978900909424, "step_time_sec": 8.235536424908787, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 389, "loss": 6.846357822418213, "lr": 0.00015560000000000001, "elapsed_sec": 3204.634852409363, "step_time_sec": 8.236759769031778, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 390, "loss": 6.84161901473999, "lr": 0.000156, "elapsed_sec": 3212.8718736171722, "step_time_sec": 8.236805923981592, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 391, "loss": 6.654271602630615, "lr": 0.0001564, "elapsed_sec": 3221.108498096466, "step_time_sec": 8.23647114704363, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 392, "loss": 6.774636745452881, "lr": 0.0001568, "elapsed_sec": 3229.3465344905853, "step_time_sec": 8.237884634057991, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 393, "loss": 6.765727996826172, "lr": 0.0001572, "elapsed_sec": 3237.5837166309357, "step_time_sec": 8.237031402997673, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 394, "loss": 6.811947822570801, "lr": 0.0001576, "elapsed_sec": 3245.8213982582092, "step_time_sec": 8.237560600973666, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 395, "loss": 6.7840142250061035, "lr": 0.000158, "elapsed_sec": 3254.0590991973877, "step_time_sec": 8.237466482096352, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 396, "loss": 6.589489459991455, "lr": 0.0001584, "elapsed_sec": 3262.2958636283875, "step_time_sec": 8.236647146986797, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 397, "loss": 6.7545623779296875, "lr": 0.00015879999999999998, "elapsed_sec": 3270.5335750579834, "step_time_sec": 8.237546195043251, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 398, "loss": 6.802884578704834, "lr": 0.00015920000000000002, "elapsed_sec": 3278.770346403122, "step_time_sec": 8.236582839977928, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 399, "loss": 6.7181854248046875, "lr": 0.00015960000000000003, "elapsed_sec": 3287.0082585811615, "step_time_sec": 8.23780355101917, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 400, "loss": 6.8305864334106445, "lr": 0.00016, "elapsed_sec": 3295.2445290088654, "step_time_sec": 8.236055254004896, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 401, "loss": 6.844895362854004, "lr": 0.00016040000000000002, "elapsed_sec": 3303.482799768448, "step_time_sec": 8.238153896993026, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 402, "loss": 6.740564823150635, "lr": 0.0001608, "elapsed_sec": 3311.72030878067, "step_time_sec": 8.237312876968645, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 403, "loss": 6.619419574737549, "lr": 0.00016120000000000002, "elapsed_sec": 3319.959615468979, "step_time_sec": 8.239150024950504, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 404, "loss": 6.710301876068115, "lr": 0.00016160000000000002, "elapsed_sec": 3328.1974024772644, "step_time_sec": 8.237627964001149, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 405, "loss": 6.832428932189941, "lr": 0.000162, "elapsed_sec": 3336.433645963669, "step_time_sec": 8.236134330974892, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 406, "loss": 6.639134883880615, "lr": 0.00016240000000000002, "elapsed_sec": 3344.6707816123962, "step_time_sec": 8.236953970976174, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 407, "loss": 6.743284702301025, "lr": 0.0001628, "elapsed_sec": 3352.906905412674, "step_time_sec": 8.235961133032106, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 408, "loss": 6.811192035675049, "lr": 0.0001632, "elapsed_sec": 3361.1449661254883, "step_time_sec": 8.237946617999114, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 409, "loss": 6.7699174880981445, "lr": 0.0001636, "elapsed_sec": 3369.3826410770416, "step_time_sec": 8.23749545996543, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 410, "loss": 6.73295259475708, "lr": 0.000164, "elapsed_sec": 3377.6205158233643, "step_time_sec": 8.237730132066645, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 411, "loss": 6.711637496948242, "lr": 0.0001644, "elapsed_sec": 3385.857996940613, "step_time_sec": 8.237355873920023, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 412, "loss": 6.747705936431885, "lr": 0.0001648, "elapsed_sec": 3394.0950713157654, "step_time_sec": 8.23683907801751, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 413, "loss": 6.61796760559082, "lr": 0.0001652, "elapsed_sec": 3402.3331530094147, "step_time_sec": 8.237947055022232, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 414, "loss": 6.697819709777832, "lr": 0.0001656, "elapsed_sec": 3410.5695509910583, "step_time_sec": 8.236217946046963, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 415, "loss": 6.7362236976623535, "lr": 0.000166, "elapsed_sec": 3418.8073828220367, "step_time_sec": 8.237698357086629, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 416, "loss": 6.690871238708496, "lr": 0.00016640000000000003, "elapsed_sec": 3427.044469356537, "step_time_sec": 8.236853751935996, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 417, "loss": 6.685896873474121, "lr": 0.00016680000000000002, "elapsed_sec": 3435.282183408737, "step_time_sec": 8.237595777027309, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 418, "loss": 6.641459941864014, "lr": 0.00016720000000000003, "elapsed_sec": 3443.519200563431, "step_time_sec": 8.236810547066852, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 419, "loss": 6.824759006500244, "lr": 0.0001676, "elapsed_sec": 3451.755031347275, "step_time_sec": 8.235669791931286, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 420, "loss": 6.672943592071533, "lr": 0.00016800000000000002, "elapsed_sec": 3459.9926238059998, "step_time_sec": 8.237416783929802, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 421, "loss": 6.724233627319336, "lr": 0.00016840000000000003, "elapsed_sec": 3468.2286937236786, "step_time_sec": 8.23591761989519, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 422, "loss": 6.542011737823486, "lr": 0.0001688, "elapsed_sec": 3476.46604847908, "step_time_sec": 8.237255592946894, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 423, "loss": 6.779899597167969, "lr": 0.00016920000000000002, "elapsed_sec": 3484.7029128074646, "step_time_sec": 8.236666855053045, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 424, "loss": 6.658920764923096, "lr": 0.0001696, "elapsed_sec": 3492.940266609192, "step_time_sec": 8.237174296053126, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 425, "loss": 6.650941848754883, "lr": 0.00017, "elapsed_sec": 3501.179856777191, "step_time_sec": 8.239446652005427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 426, "loss": 6.770207405090332, "lr": 0.0001704, "elapsed_sec": 3509.4165856838226, "step_time_sec": 8.236555262003094, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 427, "loss": 6.580413341522217, "lr": 0.0001708, "elapsed_sec": 3517.6542994976044, "step_time_sec": 8.237547955010086, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 428, "loss": 6.585578918457031, "lr": 0.00017120000000000001, "elapsed_sec": 3525.8910303115845, "step_time_sec": 8.236618557944894, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 429, "loss": 6.615407943725586, "lr": 0.0001716, "elapsed_sec": 3534.129508972168, "step_time_sec": 8.238256328972057, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 430, "loss": 6.658230304718018, "lr": 0.000172, "elapsed_sec": 3542.367203950882, "step_time_sec": 8.23752655100543, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 431, "loss": 6.605128288269043, "lr": 0.0001724, "elapsed_sec": 3550.6048979759216, "step_time_sec": 8.23752958001569, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 432, "loss": 6.66192102432251, "lr": 0.0001728, "elapsed_sec": 3558.844058275223, "step_time_sec": 8.23905323795043, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 433, "loss": 6.48598051071167, "lr": 0.0001732, "elapsed_sec": 3567.0804064273834, "step_time_sec": 8.23613163002301, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 434, "loss": 6.706630706787109, "lr": 0.0001736, "elapsed_sec": 3575.3183109760284, "step_time_sec": 8.237757197930478, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 435, "loss": 6.610428333282471, "lr": 0.00017400000000000003, "elapsed_sec": 3583.5542402267456, "step_time_sec": 8.235746005899273, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 436, "loss": 6.731845378875732, "lr": 0.0001744, "elapsed_sec": 3591.7916882038116, "step_time_sec": 8.23727327794768, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 437, "loss": 6.638106346130371, "lr": 0.00017480000000000002, "elapsed_sec": 3600.029107570648, "step_time_sec": 8.237234455998987, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 438, "loss": 6.6113739013671875, "lr": 0.0001752, "elapsed_sec": 3608.2667286396027, "step_time_sec": 8.237464561942033, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 439, "loss": 6.680849075317383, "lr": 0.0001756, "elapsed_sec": 3616.5030887126923, "step_time_sec": 8.236267609056085, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 440, "loss": 6.749250411987305, "lr": 0.00017600000000000002, "elapsed_sec": 3624.740333557129, "step_time_sec": 8.237081312923692, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 441, "loss": 6.551058292388916, "lr": 0.0001764, "elapsed_sec": 3632.9789431095123, "step_time_sec": 8.238442153087817, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 442, "loss": 6.659440040588379, "lr": 0.00017680000000000001, "elapsed_sec": 3641.2159807682037, "step_time_sec": 8.236875939997844, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 443, "loss": 6.601305961608887, "lr": 0.0001772, "elapsed_sec": 3649.453693151474, "step_time_sec": 8.237547565018758, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 444, "loss": 6.693547248840332, "lr": 0.0001776, "elapsed_sec": 3657.689493894577, "step_time_sec": 8.235683089937083, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 445, "loss": 6.58941125869751, "lr": 0.00017800000000000002, "elapsed_sec": 3665.926843881607, "step_time_sec": 8.237173410947435, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 446, "loss": 6.663581371307373, "lr": 0.0001784, "elapsed_sec": 3674.1657061576843, "step_time_sec": 8.238627863000147, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 447, "loss": 6.676753520965576, "lr": 0.0001788, "elapsed_sec": 3682.402194738388, "step_time_sec": 8.23633661097847, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 448, "loss": 6.594208717346191, "lr": 0.0001792, "elapsed_sec": 3690.640021085739, "step_time_sec": 8.23769040289335, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 449, "loss": 6.60386323928833, "lr": 0.0001796, "elapsed_sec": 3698.876903295517, "step_time_sec": 8.236698170891032, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 450, "loss": 6.4923996925354, "lr": 0.00018, "elapsed_sec": 3707.114138364792, "step_time_sec": 8.2370421490632, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 451, "loss": 6.507741451263428, "lr": 0.0001804, "elapsed_sec": 3715.3511457443237, "step_time_sec": 8.236845872947015, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 452, "loss": 6.554675102233887, "lr": 0.0001808, "elapsed_sec": 3723.5889773368835, "step_time_sec": 8.23772471502889, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 453, "loss": 6.7188825607299805, "lr": 0.0001812, "elapsed_sec": 3731.8271868228912, "step_time_sec": 8.237989973975345, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 454, "loss": 6.58380651473999, "lr": 0.00018160000000000002, "elapsed_sec": 3740.064383506775, "step_time_sec": 8.237041000043973, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 455, "loss": 6.752152919769287, "lr": 0.000182, "elapsed_sec": 3748.3028070926666, "step_time_sec": 8.238263722974807, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 456, "loss": 6.655521392822266, "lr": 0.00018240000000000002, "elapsed_sec": 3756.539277791977, "step_time_sec": 8.236286166938953, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 457, "loss": 6.630475044250488, "lr": 0.00018280000000000003, "elapsed_sec": 3764.777759075165, "step_time_sec": 8.2383828680031, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 458, "loss": 6.611438274383545, "lr": 0.0001832, "elapsed_sec": 3773.0141355991364, "step_time_sec": 8.236204036977142, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 459, "loss": 6.615652561187744, "lr": 0.00018360000000000002, "elapsed_sec": 3781.2517580986023, "step_time_sec": 8.237430483917706, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 460, "loss": 6.65726900100708, "lr": 0.000184, "elapsed_sec": 3789.4893407821655, "step_time_sec": 8.23743151803501, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 461, "loss": 6.609244346618652, "lr": 0.0001844, "elapsed_sec": 3797.7284286022186, "step_time_sec": 8.238920636940747, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 462, "loss": 6.593471527099609, "lr": 0.00018480000000000002, "elapsed_sec": 3805.9669704437256, "step_time_sec": 8.238410096964799, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 463, "loss": 6.614142894744873, "lr": 0.0001852, "elapsed_sec": 3814.2045543193817, "step_time_sec": 8.237348973983899, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 464, "loss": 6.800837993621826, "lr": 0.0001856, "elapsed_sec": 3822.4405064582825, "step_time_sec": 8.235798405949026, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 465, "loss": 6.755174160003662, "lr": 0.000186, "elapsed_sec": 3830.6778769493103, "step_time_sec": 8.23717127495911, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 466, "loss": 6.710708141326904, "lr": 0.0001864, "elapsed_sec": 3838.915302991867, "step_time_sec": 8.237259018002078, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 467, "loss": 6.553229808807373, "lr": 0.00018680000000000001, "elapsed_sec": 3847.153188943863, "step_time_sec": 8.237715456052683, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 468, "loss": 6.46530818939209, "lr": 0.0001872, "elapsed_sec": 3855.3903200626373, "step_time_sec": 8.236978903994896, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 469, "loss": 6.731119632720947, "lr": 0.0001876, "elapsed_sec": 3863.6290316581726, "step_time_sec": 8.238535222946666, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 470, "loss": 6.609559535980225, "lr": 0.000188, "elapsed_sec": 3871.866031885147, "step_time_sec": 8.236837951932102, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 471, "loss": 6.58168363571167, "lr": 0.0001884, "elapsed_sec": 3880.1037678718567, "step_time_sec": 8.237616178928874, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 472, "loss": 6.5081658363342285, "lr": 0.0001888, "elapsed_sec": 3888.3422911167145, "step_time_sec": 8.238325899932534, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 473, "loss": 6.67346715927124, "lr": 0.00018920000000000002, "elapsed_sec": 3896.580059528351, "step_time_sec": 8.237585981027223, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 474, "loss": 6.53734827041626, "lr": 0.00018960000000000003, "elapsed_sec": 3904.817964076996, "step_time_sec": 8.237774859997444, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 475, "loss": 6.566558837890625, "lr": 0.00019, "elapsed_sec": 3913.055436372757, "step_time_sec": 8.237356013036333, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 476, "loss": 6.522854804992676, "lr": 0.00019040000000000002, "elapsed_sec": 3921.2947840690613, "step_time_sec": 8.239185780053958, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 477, "loss": 6.529120922088623, "lr": 0.0001908, "elapsed_sec": 3929.5323538780212, "step_time_sec": 8.237343915039673, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 478, "loss": 6.5800018310546875, "lr": 0.0001912, "elapsed_sec": 3937.770190000534, "step_time_sec": 8.237659493926913, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 479, "loss": 7.084271430969238, "lr": 0.00019160000000000002, "elapsed_sec": 3946.008640050888, "step_time_sec": 8.238366162986495, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 480, "loss": 6.49458122253418, "lr": 0.000192, "elapsed_sec": 3954.24450135231, "step_time_sec": 8.235651663970202, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 481, "loss": 6.696572303771973, "lr": 0.00019240000000000001, "elapsed_sec": 3962.481709957123, "step_time_sec": 8.237048660987057, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 482, "loss": 6.495048999786377, "lr": 0.0001928, "elapsed_sec": 3970.7183713912964, "step_time_sec": 8.236456456012093, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 483, "loss": 6.417520046234131, "lr": 0.0001932, "elapsed_sec": 3978.956844806671, "step_time_sec": 8.23832174600102, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 484, "loss": 6.540748596191406, "lr": 0.00019360000000000002, "elapsed_sec": 3987.193925142288, "step_time_sec": 8.236889500985853, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 485, "loss": 6.528688907623291, "lr": 0.000194, "elapsed_sec": 3995.4307107925415, "step_time_sec": 8.236623472068459, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 486, "loss": 6.535941123962402, "lr": 0.0001944, "elapsed_sec": 4003.6696293354034, "step_time_sec": 8.238781758933328, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 487, "loss": 6.380816459655762, "lr": 0.0001948, "elapsed_sec": 4011.9059493541718, "step_time_sec": 8.236129657016136, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 488, "loss": 6.4871439933776855, "lr": 0.0001952, "elapsed_sec": 4020.143747329712, "step_time_sec": 8.23764172999654, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 489, "loss": 6.5016913414001465, "lr": 0.00019559999999999998, "elapsed_sec": 4028.3807854652405, "step_time_sec": 8.236879248986952, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 490, "loss": 6.484105110168457, "lr": 0.00019600000000000002, "elapsed_sec": 4036.6189460754395, "step_time_sec": 8.23803834093269, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 491, "loss": 6.511674404144287, "lr": 0.00019640000000000003, "elapsed_sec": 4044.85586977005, "step_time_sec": 8.23673828295432, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 492, "loss": 6.644986629486084, "lr": 0.0001968, "elapsed_sec": 4053.093138217926, "step_time_sec": 8.2371202489594, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 493, "loss": 6.4693121910095215, "lr": 0.00019720000000000002, "elapsed_sec": 4061.33034992218, "step_time_sec": 8.23706543596927, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 494, "loss": 6.499820709228516, "lr": 0.0001976, "elapsed_sec": 4069.5675206184387, "step_time_sec": 8.236967414966784, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 495, "loss": 6.506015777587891, "lr": 0.00019800000000000002, "elapsed_sec": 4077.8057243824005, "step_time_sec": 8.238046462996863, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 496, "loss": 6.438326358795166, "lr": 0.00019840000000000002, "elapsed_sec": 4086.0445499420166, "step_time_sec": 8.238719242042862, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 497, "loss": 6.541029930114746, "lr": 0.0001988, "elapsed_sec": 4094.2828662395477, "step_time_sec": 8.23811116698198, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 498, "loss": 6.450479030609131, "lr": 0.00019920000000000002, "elapsed_sec": 4102.519869804382, "step_time_sec": 8.236820499994792, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 499, "loss": 6.402902126312256, "lr": 0.0001996, "elapsed_sec": 4110.756897211075, "step_time_sec": 8.236890517058782, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 500, "loss": 6.523006439208984, "lr": 0.0002, "elapsed_sec": 4118.994724750519, "step_time_sec": 25.920445994008332, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": true, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 501, "loss": 6.468405246734619, "lr": 0.0002, "elapsed_sec": 4144.925773859024, "step_time_sec": 8.248024147003889, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 502, "loss": 6.4422287940979, "lr": 0.0002, "elapsed_sec": 4153.159796953201, "step_time_sec": 8.233858450083062, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 503, "loss": 6.551697254180908, "lr": 0.0002, "elapsed_sec": 4161.393047571182, "step_time_sec": 8.233099151984788, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 504, "loss": 6.5763468742370605, "lr": 0.0002, "elapsed_sec": 4169.623754739761, "step_time_sec": 8.230554734007455, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 505, "loss": 6.5189080238342285, "lr": 0.0002, "elapsed_sec": 4177.8561379909515, "step_time_sec": 8.232176118995994, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 506, "loss": 6.3837151527404785, "lr": 0.0002, "elapsed_sec": 4186.0945999622345, "step_time_sec": 8.238359406008385, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 507, "loss": 6.4521074295043945, "lr": 0.0002, "elapsed_sec": 4194.332618236542, "step_time_sec": 8.237832340062596, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 508, "loss": 6.587106227874756, "lr": 0.0002, "elapsed_sec": 4202.569493293762, "step_time_sec": 8.236679601948708, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 509, "loss": 6.551645755767822, "lr": 0.0002, "elapsed_sec": 4210.807247400284, "step_time_sec": 8.237560629029758, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 510, "loss": 6.571903705596924, "lr": 0.0002, "elapsed_sec": 4219.043550491333, "step_time_sec": 8.236250943038613, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 511, "loss": 6.392305850982666, "lr": 0.0002, "elapsed_sec": 4227.28097486496, "step_time_sec": 8.23712573305238, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 512, "loss": 6.442780017852783, "lr": 0.0002, "elapsed_sec": 4235.51842546463, "step_time_sec": 8.237314986996353, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 513, "loss": 6.452914714813232, "lr": 0.0002, "elapsed_sec": 4243.756639957428, "step_time_sec": 8.238051419961266, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 514, "loss": 6.449318885803223, "lr": 0.0002, "elapsed_sec": 4251.995003461838, "step_time_sec": 8.238170808064751, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 515, "loss": 6.735537052154541, "lr": 0.0002, "elapsed_sec": 4260.232397079468, "step_time_sec": 8.237232472980395, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 516, "loss": 6.479465007781982, "lr": 0.0002, "elapsed_sec": 4268.4686460494995, "step_time_sec": 8.23611019202508, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 517, "loss": 6.547534465789795, "lr": 0.0002, "elapsed_sec": 4276.705729007721, "step_time_sec": 8.236932961968705, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 518, "loss": 6.57729959487915, "lr": 0.0002, "elapsed_sec": 4284.942853450775, "step_time_sec": 8.236918270005845, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 519, "loss": 6.512720108032227, "lr": 0.0002, "elapsed_sec": 4293.180322170258, "step_time_sec": 8.237349263043143, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 520, "loss": 6.504668712615967, "lr": 0.0002, "elapsed_sec": 4301.417792797089, "step_time_sec": 8.237364919041283, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 521, "loss": 6.391428470611572, "lr": 0.0002, "elapsed_sec": 4309.654449701309, "step_time_sec": 8.236415695981123, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 522, "loss": 6.531021595001221, "lr": 0.0002, "elapsed_sec": 4317.891617536545, "step_time_sec": 8.23699198209215, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 523, "loss": 6.585330963134766, "lr": 0.0002, "elapsed_sec": 4326.129477024078, "step_time_sec": 8.237749829073437, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 524, "loss": 6.518202781677246, "lr": 0.0002, "elapsed_sec": 4334.366513967514, "step_time_sec": 8.236826309002936, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 525, "loss": 6.395922660827637, "lr": 0.0002, "elapsed_sec": 4342.60448551178, "step_time_sec": 8.237767452956177, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 526, "loss": 6.567442417144775, "lr": 0.0002, "elapsed_sec": 4350.841058731079, "step_time_sec": 8.236460065934807, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 527, "loss": 6.494566440582275, "lr": 0.0002, "elapsed_sec": 4359.078958749771, "step_time_sec": 8.237685762927867, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 528, "loss": 6.391151428222656, "lr": 0.0002, "elapsed_sec": 4367.316475868225, "step_time_sec": 8.237397643970326, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 529, "loss": 6.556124687194824, "lr": 0.0002, "elapsed_sec": 4375.5539655685425, "step_time_sec": 8.237320268061012, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 530, "loss": 6.524158477783203, "lr": 0.0002, "elapsed_sec": 4383.792784452438, "step_time_sec": 8.23861979902722, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 531, "loss": 6.547112464904785, "lr": 0.0002, "elapsed_sec": 4392.028524875641, "step_time_sec": 8.235581472981721, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 532, "loss": 6.533005714416504, "lr": 0.0002, "elapsed_sec": 4400.266336917877, "step_time_sec": 8.237669375957921, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 533, "loss": 6.38336181640625, "lr": 0.0002, "elapsed_sec": 4408.503808498383, "step_time_sec": 8.237339223036543, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 534, "loss": 6.397772789001465, "lr": 0.0002, "elapsed_sec": 4416.742375612259, "step_time_sec": 8.238420296111144, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 535, "loss": 6.442944049835205, "lr": 0.0002, "elapsed_sec": 4424.980876684189, "step_time_sec": 8.238312300061807, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 536, "loss": 6.391616344451904, "lr": 0.0002, "elapsed_sec": 4433.218366384506, "step_time_sec": 8.237364414031617, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 537, "loss": 6.355607032775879, "lr": 0.0002, "elapsed_sec": 4441.456681966782, "step_time_sec": 8.238161963992752, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 538, "loss": 6.646839141845703, "lr": 0.0002, "elapsed_sec": 4449.693997383118, "step_time_sec": 8.237144888029434, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 539, "loss": 6.515507698059082, "lr": 0.0002, "elapsed_sec": 4457.932445049286, "step_time_sec": 8.238274373929016, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 540, "loss": 6.520565509796143, "lr": 0.0002, "elapsed_sec": 4466.169109582901, "step_time_sec": 8.236541691003367, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 541, "loss": 6.7852630615234375, "lr": 0.0002, "elapsed_sec": 4474.406775712967, "step_time_sec": 8.2374548970256, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 542, "loss": 6.421764850616455, "lr": 0.0002, "elapsed_sec": 4482.645686626434, "step_time_sec": 8.238757898914628, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 543, "loss": 6.557107448577881, "lr": 0.0002, "elapsed_sec": 4490.882736444473, "step_time_sec": 8.236887253005989, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 544, "loss": 6.350106239318848, "lr": 0.0002, "elapsed_sec": 4499.119964599609, "step_time_sec": 8.237068700022064, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 545, "loss": 6.372979164123535, "lr": 0.0002, "elapsed_sec": 4507.358246564865, "step_time_sec": 8.238143887021579, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 546, "loss": 6.508896350860596, "lr": 0.0002, "elapsed_sec": 4515.598155260086, "step_time_sec": 8.239768197061494, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 547, "loss": 6.404356956481934, "lr": 0.0002, "elapsed_sec": 4523.836637496948, "step_time_sec": 8.238324491074309, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 548, "loss": 6.340446472167969, "lr": 0.0002, "elapsed_sec": 4532.074182033539, "step_time_sec": 8.237330924952403, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 549, "loss": 6.480678558349609, "lr": 0.0002, "elapsed_sec": 4540.312288761139, "step_time_sec": 8.238004571991041, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 550, "loss": 6.401907444000244, "lr": 0.0002, "elapsed_sec": 4548.548884868622, "step_time_sec": 8.236430165008642, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 551, "loss": 6.4684271812438965, "lr": 0.0002, "elapsed_sec": 4556.787290811539, "step_time_sec": 8.238250957918353, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 552, "loss": 6.4298810958862305, "lr": 0.0002, "elapsed_sec": 4565.025985002518, "step_time_sec": 8.238484281930141, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 553, "loss": 6.389103889465332, "lr": 0.0002, "elapsed_sec": 4573.263678073883, "step_time_sec": 8.237563726026565, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 554, "loss": 6.445688724517822, "lr": 0.0002, "elapsed_sec": 4581.5017812252045, "step_time_sec": 8.237932845950127, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 555, "loss": 6.395557880401611, "lr": 0.0002, "elapsed_sec": 4589.739332914352, "step_time_sec": 8.237386511056684, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 556, "loss": 6.431025505065918, "lr": 0.0002, "elapsed_sec": 4597.976989030838, "step_time_sec": 8.237537046894431, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 557, "loss": 6.412583827972412, "lr": 0.0002, "elapsed_sec": 4606.214241027832, "step_time_sec": 8.237113198032603, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 558, "loss": 6.414778709411621, "lr": 0.0002, "elapsed_sec": 4614.4515171051025, "step_time_sec": 8.237075172015466, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 559, "loss": 6.443361282348633, "lr": 0.0002, "elapsed_sec": 4622.689720869064, "step_time_sec": 8.238053758977912, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 560, "loss": 6.377180099487305, "lr": 0.0002, "elapsed_sec": 4630.926853179932, "step_time_sec": 8.236971194972284, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 561, "loss": 6.40825080871582, "lr": 0.0002, "elapsed_sec": 4639.164032697678, "step_time_sec": 8.237023107009009, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 562, "loss": 6.447500705718994, "lr": 0.0002, "elapsed_sec": 4647.4020001888275, "step_time_sec": 8.237811730010435, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 563, "loss": 6.33694314956665, "lr": 0.0002, "elapsed_sec": 4655.640691995621, "step_time_sec": 8.238598332041875, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 564, "loss": 6.338626384735107, "lr": 0.0002, "elapsed_sec": 4663.877191781998, "step_time_sec": 8.236292868037708, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 565, "loss": 6.327141284942627, "lr": 0.0002, "elapsed_sec": 4672.114409685135, "step_time_sec": 8.237036744947545, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 566, "loss": 6.450070858001709, "lr": 0.0002, "elapsed_sec": 4680.353561162949, "step_time_sec": 8.239008539007045, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 567, "loss": 6.346721172332764, "lr": 0.0002, "elapsed_sec": 4688.5907390117645, "step_time_sec": 8.237008456955664, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 568, "loss": 6.341851711273193, "lr": 0.0002, "elapsed_sec": 4696.828194856644, "step_time_sec": 8.237333523924462, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 569, "loss": 6.325719833374023, "lr": 0.0002, "elapsed_sec": 4705.06532907486, "step_time_sec": 8.236924326978624, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 570, "loss": 6.423584938049316, "lr": 0.0002, "elapsed_sec": 4713.303280353546, "step_time_sec": 8.237795207998715, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 571, "loss": 6.363653182983398, "lr": 0.0002, "elapsed_sec": 4721.540140628815, "step_time_sec": 8.236719758016989, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 572, "loss": 6.477969646453857, "lr": 0.0002, "elapsed_sec": 4729.778518676758, "step_time_sec": 8.238222587038763, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 573, "loss": 6.3352437019348145, "lr": 0.0002, "elapsed_sec": 4738.01685500145, "step_time_sec": 8.238188114017248, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 574, "loss": 6.36262845993042, "lr": 0.0002, "elapsed_sec": 4746.253304481506, "step_time_sec": 8.236306278966367, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 575, "loss": 6.29413366317749, "lr": 0.0002, "elapsed_sec": 4754.492511034012, "step_time_sec": 8.23899284505751, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 576, "loss": 6.438640117645264, "lr": 0.0002, "elapsed_sec": 4762.729480266571, "step_time_sec": 8.236803426058032, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 577, "loss": 6.31358003616333, "lr": 0.0002, "elapsed_sec": 4770.9668345451355, "step_time_sec": 8.237214324995875, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 578, "loss": 6.286072731018066, "lr": 0.0002, "elapsed_sec": 4779.204317808151, "step_time_sec": 8.237316249054857, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 579, "loss": 6.396847248077393, "lr": 0.0002, "elapsed_sec": 4787.441140651703, "step_time_sec": 8.236679858062416, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 580, "loss": 6.3122334480285645, "lr": 0.0002, "elapsed_sec": 4795.679235458374, "step_time_sec": 8.237922348082066, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 581, "loss": 6.315621852874756, "lr": 0.0002, "elapsed_sec": 4803.916691064835, "step_time_sec": 8.237348444992676, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 582, "loss": 6.478233814239502, "lr": 0.0002, "elapsed_sec": 4812.155619621277, "step_time_sec": 8.238760039908811, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 583, "loss": 6.373844146728516, "lr": 0.0002, "elapsed_sec": 4820.395653009415, "step_time_sec": 8.239849728997797, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 584, "loss": 6.466450214385986, "lr": 0.0002, "elapsed_sec": 4828.633867740631, "step_time_sec": 8.238044081022963, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 585, "loss": 6.377491474151611, "lr": 0.0002, "elapsed_sec": 4836.872085809708, "step_time_sec": 8.238086774013937, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 586, "loss": 6.358365058898926, "lr": 0.0002, "elapsed_sec": 4845.109172821045, "step_time_sec": 8.236857244977728, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 587, "loss": 6.2998046875, "lr": 0.0002, "elapsed_sec": 4853.346892595291, "step_time_sec": 8.237587076961063, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 588, "loss": 6.406045913696289, "lr": 0.0002, "elapsed_sec": 4861.58384346962, "step_time_sec": 8.23677675309591, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 589, "loss": 6.400595664978027, "lr": 0.0002, "elapsed_sec": 4869.821844100952, "step_time_sec": 8.237827511969954, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 590, "loss": 6.47178316116333, "lr": 0.0002, "elapsed_sec": 4878.059656620026, "step_time_sec": 8.237669328111224, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 591, "loss": 6.427749156951904, "lr": 0.0002, "elapsed_sec": 4886.296728610992, "step_time_sec": 8.2369560750667, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 592, "loss": 6.346658229827881, "lr": 0.0002, "elapsed_sec": 4894.535289525986, "step_time_sec": 8.23834814003203, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 593, "loss": 6.3932342529296875, "lr": 0.0002, "elapsed_sec": 4902.772211790085, "step_time_sec": 8.236771989031695, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 594, "loss": 6.308912754058838, "lr": 0.0002, "elapsed_sec": 4911.010903596878, "step_time_sec": 8.238526568980888, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 595, "loss": 6.398285388946533, "lr": 0.0002, "elapsed_sec": 4919.247922182083, "step_time_sec": 8.236868557985872, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 596, "loss": 6.357035160064697, "lr": 0.0002, "elapsed_sec": 4927.485997438431, "step_time_sec": 8.23790711199399, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 597, "loss": 6.287331581115723, "lr": 0.0002, "elapsed_sec": 4935.723507881165, "step_time_sec": 8.237394183990546, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 598, "loss": 6.27802038192749, "lr": 0.0002, "elapsed_sec": 4943.960999965668, "step_time_sec": 8.237334338016808, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 599, "loss": 6.341982841491699, "lr": 0.0002, "elapsed_sec": 4952.199055433273, "step_time_sec": 8.237899850006215, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 600, "loss": 6.263479709625244, "lr": 0.0002, "elapsed_sec": 4960.436144113541, "step_time_sec": 8.236935660010204, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 601, "loss": 6.268757343292236, "lr": 0.0002, "elapsed_sec": 4968.674427032471, "step_time_sec": 8.23814545199275, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 602, "loss": 6.11967134475708, "lr": 0.0002, "elapsed_sec": 4976.912340402603, "step_time_sec": 8.237690311972983, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 603, "loss": 6.3733015060424805, "lr": 0.0002, "elapsed_sec": 4985.150814533234, "step_time_sec": 8.238333321991377, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 604, "loss": 6.226396083831787, "lr": 0.0002, "elapsed_sec": 4993.388849496841, "step_time_sec": 8.237882188986987, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 605, "loss": 6.273202896118164, "lr": 0.0002, "elapsed_sec": 5001.626268148422, "step_time_sec": 8.237258602050133, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 606, "loss": 6.391324043273926, "lr": 0.0002, "elapsed_sec": 5009.86336183548, "step_time_sec": 8.236927754012868, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 607, "loss": 6.316005706787109, "lr": 0.0002, "elapsed_sec": 5018.102142333984, "step_time_sec": 8.238611335051246, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 608, "loss": 6.345031261444092, "lr": 0.0002, "elapsed_sec": 5026.339348316193, "step_time_sec": 8.237088915077038, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 609, "loss": 6.316992282867432, "lr": 0.0002, "elapsed_sec": 5034.577061653137, "step_time_sec": 8.237508774036542, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 610, "loss": 6.318918704986572, "lr": 0.0002, "elapsed_sec": 5042.814735174179, "step_time_sec": 8.237561022047885, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 611, "loss": 6.201412677764893, "lr": 0.0002, "elapsed_sec": 5051.053006887436, "step_time_sec": 8.238065431942232, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 612, "loss": 6.311325550079346, "lr": 0.0002, "elapsed_sec": 5059.290847063065, "step_time_sec": 8.237684405990876, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 613, "loss": 6.23984956741333, "lr": 0.0002, "elapsed_sec": 5067.527289152145, "step_time_sec": 8.236323130084202, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 614, "loss": 6.281065940856934, "lr": 0.0002, "elapsed_sec": 5075.765269041061, "step_time_sec": 8.237758835079148, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 615, "loss": 6.296566486358643, "lr": 0.0002, "elapsed_sec": 5084.002127408981, "step_time_sec": 8.236746991984546, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 616, "loss": 6.29313850402832, "lr": 0.0002, "elapsed_sec": 5092.240045785904, "step_time_sec": 8.237710879999213, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 617, "loss": 6.311260223388672, "lr": 0.0002, "elapsed_sec": 5100.476576089859, "step_time_sec": 8.236383936950006, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 618, "loss": 6.23062801361084, "lr": 0.0002, "elapsed_sec": 5108.713953971863, "step_time_sec": 8.237186626996845, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 619, "loss": 6.154551982879639, "lr": 0.0002, "elapsed_sec": 5116.9517550468445, "step_time_sec": 8.237683068029583, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 620, "loss": 6.2958784103393555, "lr": 0.0002, "elapsed_sec": 5125.188981294632, "step_time_sec": 8.237043704022653, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 621, "loss": 6.395510196685791, "lr": 0.0002, "elapsed_sec": 5133.42712521553, "step_time_sec": 8.237976472009905, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 622, "loss": 6.2201924324035645, "lr": 0.0002, "elapsed_sec": 5141.664551258087, "step_time_sec": 8.237264296971262, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 623, "loss": 6.299444198608398, "lr": 0.0002, "elapsed_sec": 5149.901915550232, "step_time_sec": 8.237204783945344, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 624, "loss": 6.3000054359436035, "lr": 0.0002, "elapsed_sec": 5158.137913227081, "step_time_sec": 8.235830316902138, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 625, "loss": 6.202073097229004, "lr": 0.0002, "elapsed_sec": 5166.375958442688, "step_time_sec": 8.237928117043339, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 626, "loss": 6.2626633644104, "lr": 0.0002, "elapsed_sec": 5174.647463083267, "step_time_sec": 8.246308743953705, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 627, "loss": 6.309337139129639, "lr": 0.0002, "elapsed_sec": 5182.885310173035, "step_time_sec": 8.237676245975308, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 628, "loss": 6.486106872558594, "lr": 0.0002, "elapsed_sec": 5191.122522592545, "step_time_sec": 8.237106016022153, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 629, "loss": 6.422849178314209, "lr": 0.0002, "elapsed_sec": 5199.359511613846, "step_time_sec": 8.236823085928336, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 630, "loss": 6.304366588592529, "lr": 0.0002, "elapsed_sec": 5207.596152067184, "step_time_sec": 8.236412213998847, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 631, "loss": 6.330345630645752, "lr": 0.0002, "elapsed_sec": 5215.834602355957, "step_time_sec": 8.238307829014957, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 632, "loss": 6.178158760070801, "lr": 0.0002, "elapsed_sec": 5224.072457313538, "step_time_sec": 8.237741091987118, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 633, "loss": 6.338474750518799, "lr": 0.0002, "elapsed_sec": 5232.309603691101, "step_time_sec": 8.236945376964286, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 634, "loss": 6.268700122833252, "lr": 0.0002, "elapsed_sec": 5240.547782897949, "step_time_sec": 8.238022312056273, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 635, "loss": 6.3579912185668945, "lr": 0.0002, "elapsed_sec": 5248.785370111465, "step_time_sec": 8.237402230035514, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 636, "loss": 6.171693801879883, "lr": 0.0002, "elapsed_sec": 5257.024683475494, "step_time_sec": 8.239175871014595, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 637, "loss": 6.3403496742248535, "lr": 0.0002, "elapsed_sec": 5265.26250743866, "step_time_sec": 8.237658774014562, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 638, "loss": 6.4564714431762695, "lr": 0.0002, "elapsed_sec": 5273.500470399857, "step_time_sec": 8.237777968985029, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 639, "loss": 6.24722957611084, "lr": 0.0002, "elapsed_sec": 5281.737441301346, "step_time_sec": 8.236807301989757, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 640, "loss": 6.2653117179870605, "lr": 0.0002, "elapsed_sec": 5289.974567651749, "step_time_sec": 8.236977113061585, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 641, "loss": 6.282746315002441, "lr": 0.0002, "elapsed_sec": 5298.212821722031, "step_time_sec": 8.238092613988556, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 642, "loss": 6.2402143478393555, "lr": 0.0002, "elapsed_sec": 5306.45025229454, "step_time_sec": 8.237317410996184, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 643, "loss": 6.265903472900391, "lr": 0.0002, "elapsed_sec": 5314.688047409058, "step_time_sec": 8.237626302056015, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 644, "loss": 6.185680389404297, "lr": 0.0002, "elapsed_sec": 5322.925229310989, "step_time_sec": 8.23698915191926, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 645, "loss": 6.218804836273193, "lr": 0.0002, "elapsed_sec": 5331.162197113037, "step_time_sec": 8.236830821959302, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 646, "loss": 6.231677055358887, "lr": 0.0002, "elapsed_sec": 5339.399767160416, "step_time_sec": 8.237462690914981, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 647, "loss": 6.31437873840332, "lr": 0.0002, "elapsed_sec": 5347.637273311615, "step_time_sec": 8.237305648974143, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 648, "loss": 6.335740566253662, "lr": 0.0002, "elapsed_sec": 5355.875513076782, "step_time_sec": 8.238063343102112, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 649, "loss": 6.2346906661987305, "lr": 0.0002, "elapsed_sec": 5364.112564325333, "step_time_sec": 8.23694239591714, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 650, "loss": 6.296419143676758, "lr": 0.0002, "elapsed_sec": 5372.350614786148, "step_time_sec": 8.237904515000992, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 651, "loss": 6.274501323699951, "lr": 0.0002, "elapsed_sec": 5380.5889682769775, "step_time_sec": 8.238165447022766, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 652, "loss": 6.35142183303833, "lr": 0.0002, "elapsed_sec": 5388.826939105988, "step_time_sec": 8.23781036096625, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 653, "loss": 6.164392948150635, "lr": 0.0002, "elapsed_sec": 5397.064736366272, "step_time_sec": 8.237612016964704, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 654, "loss": 6.331472396850586, "lr": 0.0002, "elapsed_sec": 5405.301520347595, "step_time_sec": 8.236612442065962, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 655, "loss": 6.310410022735596, "lr": 0.0002, "elapsed_sec": 5413.53968334198, "step_time_sec": 8.238028635038063, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 656, "loss": 6.213617324829102, "lr": 0.0002, "elapsed_sec": 5421.77814412117, "step_time_sec": 8.238231554976664, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 657, "loss": 6.262372016906738, "lr": 0.0002, "elapsed_sec": 5430.015172243118, "step_time_sec": 8.236868196981959, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 658, "loss": 6.158961296081543, "lr": 0.0002, "elapsed_sec": 5438.2521595954895, "step_time_sec": 8.236829777946696, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 659, "loss": 6.1660051345825195, "lr": 0.0002, "elapsed_sec": 5446.490282058716, "step_time_sec": 8.237991538946517, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 660, "loss": 6.254706382751465, "lr": 0.0002, "elapsed_sec": 5454.728493690491, "step_time_sec": 8.238012771005742, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 661, "loss": 6.263604164123535, "lr": 0.0002, "elapsed_sec": 5462.9651663303375, "step_time_sec": 8.23648919397965, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 662, "loss": 6.193299293518066, "lr": 0.0002, "elapsed_sec": 5471.203182458878, "step_time_sec": 8.237870710087009, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 663, "loss": 6.2022809982299805, "lr": 0.0002, "elapsed_sec": 5479.441395044327, "step_time_sec": 8.238065646961331, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 664, "loss": 6.142171382904053, "lr": 0.0002, "elapsed_sec": 5487.6791915893555, "step_time_sec": 8.237640904961154, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 665, "loss": 6.194819450378418, "lr": 0.0002, "elapsed_sec": 5495.915777921677, "step_time_sec": 8.236426534946077, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 666, "loss": 6.297486305236816, "lr": 0.0002, "elapsed_sec": 5504.151978731155, "step_time_sec": 8.236001270939596, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 667, "loss": 6.316463947296143, "lr": 0.0002, "elapsed_sec": 5512.390634775162, "step_time_sec": 8.238514686003327, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 668, "loss": 6.161529064178467, "lr": 0.0002, "elapsed_sec": 5520.626982450485, "step_time_sec": 8.236150774057023, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 669, "loss": 6.298561096191406, "lr": 0.0002, "elapsed_sec": 5528.864930391312, "step_time_sec": 8.23780316207558, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 670, "loss": 6.174597263336182, "lr": 0.0002, "elapsed_sec": 5537.103204727173, "step_time_sec": 8.238063895027153, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 671, "loss": 6.214995384216309, "lr": 0.0002, "elapsed_sec": 5545.3409948349, "step_time_sec": 8.237636792007834, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 672, "loss": 6.3517866134643555, "lr": 0.0002, "elapsed_sec": 5553.578587055206, "step_time_sec": 8.23743061197456, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 673, "loss": 6.257817268371582, "lr": 0.0002, "elapsed_sec": 5561.8149156570435, "step_time_sec": 8.236166500020772, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 674, "loss": 6.238895893096924, "lr": 0.0002, "elapsed_sec": 5570.054030656815, "step_time_sec": 8.238997310050763, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 675, "loss": 6.114950180053711, "lr": 0.0002, "elapsed_sec": 5578.2910187244415, "step_time_sec": 8.236774140037596, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 676, "loss": 6.196351528167725, "lr": 0.0002, "elapsed_sec": 5586.5299689769745, "step_time_sec": 8.238851319998503, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 677, "loss": 6.175091743469238, "lr": 0.0002, "elapsed_sec": 5594.768530130386, "step_time_sec": 8.23833597090561, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 678, "loss": 6.054113864898682, "lr": 0.0002, "elapsed_sec": 5603.006112337112, "step_time_sec": 8.237433437025174, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 679, "loss": 6.269126892089844, "lr": 0.0002, "elapsed_sec": 5611.243664264679, "step_time_sec": 8.237438995041884, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 680, "loss": 6.167649269104004, "lr": 0.0002, "elapsed_sec": 5619.480889558792, "step_time_sec": 8.237061504041776, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 681, "loss": 6.242992401123047, "lr": 0.0002, "elapsed_sec": 5627.718690156937, "step_time_sec": 8.237631463096477, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 682, "loss": 6.136038780212402, "lr": 0.0002, "elapsed_sec": 5635.955255031586, "step_time_sec": 8.236433811020106, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 683, "loss": 6.240017414093018, "lr": 0.0002, "elapsed_sec": 5644.192089319229, "step_time_sec": 8.2366575099295, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 684, "loss": 6.158687114715576, "lr": 0.0002, "elapsed_sec": 5652.429567337036, "step_time_sec": 8.237271182006225, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 685, "loss": 6.197188377380371, "lr": 0.0002, "elapsed_sec": 5660.667020559311, "step_time_sec": 8.237301103072241, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 686, "loss": 6.2327680587768555, "lr": 0.0002, "elapsed_sec": 5668.904272317886, "step_time_sec": 8.237066886969842, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 687, "loss": 6.2205305099487305, "lr": 0.0002, "elapsed_sec": 5677.141108989716, "step_time_sec": 8.23668915394228, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 688, "loss": 6.18537712097168, "lr": 0.0002, "elapsed_sec": 5685.378534078598, "step_time_sec": 8.23724626202602, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 689, "loss": 6.194740295410156, "lr": 0.0002, "elapsed_sec": 5693.61613368988, "step_time_sec": 8.237436890020035, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 690, "loss": 6.292270183563232, "lr": 0.0002, "elapsed_sec": 5701.853718996048, "step_time_sec": 8.237414457951672, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 691, "loss": 6.071863174438477, "lr": 0.0002, "elapsed_sec": 5710.090859651566, "step_time_sec": 8.236966000054963, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 692, "loss": 6.274838924407959, "lr": 0.0002, "elapsed_sec": 5718.3289270401, "step_time_sec": 8.237917482969351, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 693, "loss": 6.175107479095459, "lr": 0.0002, "elapsed_sec": 5726.566437005997, "step_time_sec": 8.237343777902424, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 694, "loss": 6.158961296081543, "lr": 0.0002, "elapsed_sec": 5734.8031713962555, "step_time_sec": 8.236588453059085, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 695, "loss": 6.194558620452881, "lr": 0.0002, "elapsed_sec": 5743.041093111038, "step_time_sec": 8.237756717950106, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 696, "loss": 6.190471649169922, "lr": 0.0002, "elapsed_sec": 5751.278228759766, "step_time_sec": 8.236971230944619, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 697, "loss": 6.33750057220459, "lr": 0.0002, "elapsed_sec": 5759.516445398331, "step_time_sec": 8.238072529085912, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 698, "loss": 6.313809871673584, "lr": 0.0002, "elapsed_sec": 5767.753284454346, "step_time_sec": 8.236681882059202, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 699, "loss": 6.201420783996582, "lr": 0.0002, "elapsed_sec": 5775.989432096481, "step_time_sec": 8.236017302959226, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 700, "loss": 6.084692001342773, "lr": 0.0002, "elapsed_sec": 5784.226478815079, "step_time_sec": 8.236813950934447, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 701, "loss": 6.124661445617676, "lr": 0.0002, "elapsed_sec": 5792.464127779007, "step_time_sec": 8.237465832033195, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 702, "loss": 6.011321067810059, "lr": 0.0002, "elapsed_sec": 5800.701748371124, "step_time_sec": 8.237507296027616, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 703, "loss": 6.230221271514893, "lr": 0.0002, "elapsed_sec": 5808.939455270767, "step_time_sec": 8.23751541308593, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 704, "loss": 6.04632568359375, "lr": 0.0002, "elapsed_sec": 5817.178137779236, "step_time_sec": 8.238504129927605, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 705, "loss": 6.141980171203613, "lr": 0.0002, "elapsed_sec": 5825.416681051254, "step_time_sec": 8.23844421596732, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 706, "loss": 6.272759914398193, "lr": 0.0002, "elapsed_sec": 5833.653913259506, "step_time_sec": 8.237075464101508, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 707, "loss": 6.169012546539307, "lr": 0.0002, "elapsed_sec": 5841.893109321594, "step_time_sec": 8.23898767598439, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 708, "loss": 6.201030254364014, "lr": 0.0002, "elapsed_sec": 5850.130608558655, "step_time_sec": 8.23733442206867, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 709, "loss": 6.148060321807861, "lr": 0.0002, "elapsed_sec": 5858.368618011475, "step_time_sec": 8.237839228939265, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 710, "loss": 6.214583873748779, "lr": 0.0002, "elapsed_sec": 5866.607157230377, "step_time_sec": 8.23841779702343, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 711, "loss": 6.237536907196045, "lr": 0.0002, "elapsed_sec": 5874.844784021378, "step_time_sec": 8.237478791968897, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 712, "loss": 6.0767822265625, "lr": 0.0002, "elapsed_sec": 5883.081855535507, "step_time_sec": 8.236885474994779, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 713, "loss": 6.195584774017334, "lr": 0.0002, "elapsed_sec": 5891.319139957428, "step_time_sec": 8.237073066062294, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 714, "loss": 6.096136569976807, "lr": 0.0002, "elapsed_sec": 5899.559824466705, "step_time_sec": 8.24055478000082, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 715, "loss": 6.216453552246094, "lr": 0.0002, "elapsed_sec": 5907.79820394516, "step_time_sec": 8.238238242920488, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 716, "loss": 6.23708438873291, "lr": 0.0002, "elapsed_sec": 5916.035541296005, "step_time_sec": 8.237200879026204, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 717, "loss": 6.217616081237793, "lr": 0.0002, "elapsed_sec": 5924.273731946945, "step_time_sec": 8.237945109023713, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 718, "loss": 6.152338981628418, "lr": 0.0002, "elapsed_sec": 5932.510385513306, "step_time_sec": 8.236513415002264, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 719, "loss": 6.131681442260742, "lr": 0.0002, "elapsed_sec": 5940.747960090637, "step_time_sec": 8.237416626070626, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 720, "loss": 6.146206378936768, "lr": 0.0002, "elapsed_sec": 5948.986280679703, "step_time_sec": 8.23808842303697, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 721, "loss": 6.198376178741455, "lr": 0.0002, "elapsed_sec": 5957.223737716675, "step_time_sec": 8.237306905095465, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 722, "loss": 6.221726417541504, "lr": 0.0002, "elapsed_sec": 5965.46076965332, "step_time_sec": 8.236878836993128, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 723, "loss": 6.0691046714782715, "lr": 0.0002, "elapsed_sec": 5973.698550462723, "step_time_sec": 8.237619610968977, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 724, "loss": 6.118704795837402, "lr": 0.0002, "elapsed_sec": 5981.935991048813, "step_time_sec": 8.237261921982281, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 725, "loss": 6.079608917236328, "lr": 0.0002, "elapsed_sec": 5990.1734001636505, "step_time_sec": 8.237223332980648, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 726, "loss": 6.160645484924316, "lr": 0.0002, "elapsed_sec": 5998.411450624466, "step_time_sec": 8.237892762059346, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 727, "loss": 6.190502643585205, "lr": 0.0002, "elapsed_sec": 6006.647592306137, "step_time_sec": 8.23602539789863, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 728, "loss": 6.134383678436279, "lr": 0.0002, "elapsed_sec": 6014.886656999588, "step_time_sec": 8.238996929954737, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 729, "loss": 6.131450176239014, "lr": 0.0002, "elapsed_sec": 6023.124797821045, "step_time_sec": 8.23785430798307, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 730, "loss": 6.152989387512207, "lr": 0.0002, "elapsed_sec": 6031.361748456955, "step_time_sec": 8.236772066913545, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 731, "loss": 6.13836145401001, "lr": 0.0002, "elapsed_sec": 6039.599908351898, "step_time_sec": 8.237981854006648, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 732, "loss": 6.057058334350586, "lr": 0.0002, "elapsed_sec": 6047.83530330658, "step_time_sec": 8.235275918035768, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 733, "loss": 5.994527816772461, "lr": 0.0002, "elapsed_sec": 6056.072263002396, "step_time_sec": 8.236839478951879, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 734, "loss": 6.0243659019470215, "lr": 0.0002, "elapsed_sec": 6064.309215784073, "step_time_sec": 8.236721056047827, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 735, "loss": 6.093164920806885, "lr": 0.0002, "elapsed_sec": 6072.547761440277, "step_time_sec": 8.238395847030915, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 736, "loss": 6.114623069763184, "lr": 0.0002, "elapsed_sec": 6080.785515069962, "step_time_sec": 8.237577709020115, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 737, "loss": 6.07257604598999, "lr": 0.0002, "elapsed_sec": 6089.023990392685, "step_time_sec": 8.238313777022995, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 738, "loss": 6.1835432052612305, "lr": 0.0002, "elapsed_sec": 6097.262173175812, "step_time_sec": 8.237992287031375, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 739, "loss": 6.112321853637695, "lr": 0.0002, "elapsed_sec": 6105.498450994492, "step_time_sec": 8.236128409975208, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 740, "loss": 6.17924690246582, "lr": 0.0002, "elapsed_sec": 6113.735591411591, "step_time_sec": 8.236983663984574, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 741, "loss": 6.002848148345947, "lr": 0.0002, "elapsed_sec": 6121.972446203232, "step_time_sec": 8.236686588963494, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 742, "loss": 6.108661651611328, "lr": 0.0002, "elapsed_sec": 6130.210436820984, "step_time_sec": 8.237818905035965, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 743, "loss": 6.099065780639648, "lr": 0.0002, "elapsed_sec": 6138.448141336441, "step_time_sec": 8.237575898994692, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 744, "loss": 6.076186180114746, "lr": 0.0002, "elapsed_sec": 6146.685481309891, "step_time_sec": 8.237130688037723, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 745, "loss": 6.088324069976807, "lr": 0.0002, "elapsed_sec": 6154.9238114356995, "step_time_sec": 8.238162957015447, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 746, "loss": 6.092095375061035, "lr": 0.0002, "elapsed_sec": 6163.160054206848, "step_time_sec": 8.23608476901427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 747, "loss": 6.080025672912598, "lr": 0.0002, "elapsed_sec": 6171.39590716362, "step_time_sec": 8.235727681894787, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 748, "loss": 6.088971138000488, "lr": 0.0002, "elapsed_sec": 6179.634287118912, "step_time_sec": 8.238203968037851, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 749, "loss": 6.304935455322266, "lr": 0.0002, "elapsed_sec": 6187.871594905853, "step_time_sec": 8.237139815930277, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 750, "loss": 6.116238594055176, "lr": 0.0002, "elapsed_sec": 6196.111206531525, "step_time_sec": 8.239625799003989, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 751, "loss": 6.1464433670043945, "lr": 0.0002, "elapsed_sec": 6204.348260164261, "step_time_sec": 8.236706010065973, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 752, "loss": 6.107797622680664, "lr": 0.0002, "elapsed_sec": 6212.586247205734, "step_time_sec": 8.237863759044558, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 753, "loss": 6.101063251495361, "lr": 0.0002, "elapsed_sec": 6220.822571754456, "step_time_sec": 8.236114838975482, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 754, "loss": 6.041410446166992, "lr": 0.0002, "elapsed_sec": 6229.060378789902, "step_time_sec": 8.237680512946099, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 755, "loss": 6.085630416870117, "lr": 0.0002, "elapsed_sec": 6237.297501325607, "step_time_sec": 8.236934517975897, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 756, "loss": 6.112347602844238, "lr": 0.0002, "elapsed_sec": 6245.5347690582275, "step_time_sec": 8.237061790074222, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 757, "loss": 5.973820209503174, "lr": 0.0002, "elapsed_sec": 6253.772670984268, "step_time_sec": 8.23780438292306, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 758, "loss": 6.203385353088379, "lr": 0.0002, "elapsed_sec": 6262.011000156403, "step_time_sec": 8.238137089996599, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 759, "loss": 6.033566474914551, "lr": 0.0002, "elapsed_sec": 6270.249605894089, "step_time_sec": 8.238462807028554, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 760, "loss": 6.117372035980225, "lr": 0.0002, "elapsed_sec": 6278.48611330986, "step_time_sec": 8.236320425057784, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 761, "loss": 5.927336692810059, "lr": 0.0002, "elapsed_sec": 6286.724291563034, "step_time_sec": 8.237998480093665, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 762, "loss": 6.059098720550537, "lr": 0.0002, "elapsed_sec": 6294.962168931961, "step_time_sec": 8.237748989020474, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 763, "loss": 6.071199417114258, "lr": 0.0002, "elapsed_sec": 6303.20001077652, "step_time_sec": 8.237695405026898, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 764, "loss": 6.0362749099731445, "lr": 0.0002, "elapsed_sec": 6311.437297344208, "step_time_sec": 8.237081355997361, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 765, "loss": 6.116610527038574, "lr": 0.0002, "elapsed_sec": 6319.675038576126, "step_time_sec": 8.237572718062438, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 766, "loss": 6.125187873840332, "lr": 0.0002, "elapsed_sec": 6327.912637472153, "step_time_sec": 8.237464476027526, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 767, "loss": 6.060197830200195, "lr": 0.0002, "elapsed_sec": 6336.150464773178, "step_time_sec": 8.237691101967357, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 768, "loss": 6.053282737731934, "lr": 0.0002, "elapsed_sec": 6344.389939069748, "step_time_sec": 8.239251979044639, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 769, "loss": 6.155305862426758, "lr": 0.0002, "elapsed_sec": 6352.62863445282, "step_time_sec": 8.238540312973782, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 770, "loss": 6.043744087219238, "lr": 0.0002, "elapsed_sec": 6360.86580657959, "step_time_sec": 8.236995305051096, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 771, "loss": 5.918773651123047, "lr": 0.0002, "elapsed_sec": 6369.1041922569275, "step_time_sec": 8.238252497976646, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 772, "loss": 6.15257453918457, "lr": 0.0002, "elapsed_sec": 6377.341028928757, "step_time_sec": 8.236708387965336, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 773, "loss": 6.069919586181641, "lr": 0.0002, "elapsed_sec": 6385.5795521736145, "step_time_sec": 8.238323734025471, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 774, "loss": 6.028438091278076, "lr": 0.0002, "elapsed_sec": 6393.816381216049, "step_time_sec": 8.236641444964334, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 775, "loss": 6.285318374633789, "lr": 0.0002, "elapsed_sec": 6402.054224014282, "step_time_sec": 8.237702119979076, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 776, "loss": 6.038539886474609, "lr": 0.0002, "elapsed_sec": 6410.292217493057, "step_time_sec": 8.237849460914731, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 777, "loss": 6.039074897766113, "lr": 0.0002, "elapsed_sec": 6418.529735088348, "step_time_sec": 8.237290080985986, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 778, "loss": 6.0344038009643555, "lr": 0.0002, "elapsed_sec": 6426.767446041107, "step_time_sec": 8.237541869981214, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 779, "loss": 6.054577350616455, "lr": 0.0002, "elapsed_sec": 6435.004857301712, "step_time_sec": 8.237315451027825, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 780, "loss": 6.070359706878662, "lr": 0.0002, "elapsed_sec": 6443.243051290512, "step_time_sec": 8.238021575962193, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 781, "loss": 6.027656078338623, "lr": 0.0002, "elapsed_sec": 6451.482239961624, "step_time_sec": 8.23900095897261, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 782, "loss": 5.941071033477783, "lr": 0.0002, "elapsed_sec": 6459.719628810883, "step_time_sec": 8.237221806077287, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 783, "loss": 6.083758354187012, "lr": 0.0002, "elapsed_sec": 6467.9567766189575, "step_time_sec": 8.237002293928526, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 784, "loss": 6.054502010345459, "lr": 0.0002, "elapsed_sec": 6476.194204330444, "step_time_sec": 8.237336285063066, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 785, "loss": 6.060385227203369, "lr": 0.0002, "elapsed_sec": 6484.433017253876, "step_time_sec": 8.238553108065389, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 786, "loss": 5.976353168487549, "lr": 0.0002, "elapsed_sec": 6492.671231746674, "step_time_sec": 8.238061219919473, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 787, "loss": 6.108620643615723, "lr": 0.0002, "elapsed_sec": 6500.909479379654, "step_time_sec": 8.238072140957229, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 788, "loss": 6.073378086090088, "lr": 0.0002, "elapsed_sec": 6509.148575544357, "step_time_sec": 8.238940619979985, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 789, "loss": 5.985324859619141, "lr": 0.0002, "elapsed_sec": 6517.385843038559, "step_time_sec": 8.23709536192473, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 790, "loss": 5.985738277435303, "lr": 0.0002, "elapsed_sec": 6525.623551607132, "step_time_sec": 8.237536887987517, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 791, "loss": 6.025042533874512, "lr": 0.0002, "elapsed_sec": 6533.86164355278, "step_time_sec": 8.23791144893039, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 792, "loss": 5.9865875244140625, "lr": 0.0002, "elapsed_sec": 6542.10022187233, "step_time_sec": 8.23841489397455, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 793, "loss": 6.146791458129883, "lr": 0.0002, "elapsed_sec": 6550.338629484177, "step_time_sec": 8.238281453028321, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 794, "loss": 5.976851463317871, "lr": 0.0002, "elapsed_sec": 6558.577611446381, "step_time_sec": 8.238843692932278, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 795, "loss": 5.996965408325195, "lr": 0.0002, "elapsed_sec": 6566.81583571434, "step_time_sec": 8.238001457066275, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 796, "loss": 6.024150848388672, "lr": 0.0002, "elapsed_sec": 6575.053506374359, "step_time_sec": 8.237522153998725, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 797, "loss": 6.007694721221924, "lr": 0.0002, "elapsed_sec": 6583.291656970978, "step_time_sec": 8.237981122918427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 798, "loss": 6.045016765594482, "lr": 0.0002, "elapsed_sec": 6591.52921295166, "step_time_sec": 8.23742180597037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 799, "loss": 6.1163129806518555, "lr": 0.0002, "elapsed_sec": 6599.766577720642, "step_time_sec": 8.237199677038006, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 800, "loss": 6.0257248878479, "lr": 0.0002, "elapsed_sec": 6608.0050983428955, "step_time_sec": 8.238349872059189, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 801, "loss": 6.0749101638793945, "lr": 0.0002, "elapsed_sec": 6616.242610454559, "step_time_sec": 8.237345371977426, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 802, "loss": 6.03798770904541, "lr": 0.0002, "elapsed_sec": 6624.48122549057, "step_time_sec": 8.238439855980687, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 803, "loss": 6.0156354904174805, "lr": 0.0002, "elapsed_sec": 6632.719591856003, "step_time_sec": 8.238206075970083, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 804, "loss": 5.967479228973389, "lr": 0.0002, "elapsed_sec": 6640.9578268527985, "step_time_sec": 8.238045250996947, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 805, "loss": 6.0288166999816895, "lr": 0.0002, "elapsed_sec": 6649.1954870224, "step_time_sec": 8.23750931606628, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 806, "loss": 6.12540864944458, "lr": 0.0002, "elapsed_sec": 6657.4312608242035, "step_time_sec": 8.235637957928702, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 807, "loss": 6.044407367706299, "lr": 0.0002, "elapsed_sec": 6665.669077634811, "step_time_sec": 8.237605895032175, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 808, "loss": 6.063558578491211, "lr": 0.0002, "elapsed_sec": 6673.9068171978, "step_time_sec": 8.237635364988819, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 809, "loss": 5.98024320602417, "lr": 0.0002, "elapsed_sec": 6682.144653081894, "step_time_sec": 8.237631470081396, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 810, "loss": 5.997333526611328, "lr": 0.0002, "elapsed_sec": 6690.381639719009, "step_time_sec": 8.236857722047716, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 811, "loss": 6.043732166290283, "lr": 0.0002, "elapsed_sec": 6698.619679927826, "step_time_sec": 8.237890584976412, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 812, "loss": 6.1148176193237305, "lr": 0.0002, "elapsed_sec": 6720.113300561905, "step_time_sec": 21.493420100887306, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 813, "loss": 6.010879993438721, "lr": 0.0002, "elapsed_sec": 6728.344411373138, "step_time_sec": 8.230997606995516, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 814, "loss": 5.969446182250977, "lr": 0.0002, "elapsed_sec": 6736.582879543304, "step_time_sec": 8.238257012097165, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 815, "loss": 5.9458417892456055, "lr": 0.0002, "elapsed_sec": 6744.819652795792, "step_time_sec": 8.236613217974082, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 816, "loss": 5.869807243347168, "lr": 0.0002, "elapsed_sec": 6753.05756688118, "step_time_sec": 8.237772937049158, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 817, "loss": 6.005959510803223, "lr": 0.0002, "elapsed_sec": 6761.29637336731, "step_time_sec": 8.238650663988665, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 818, "loss": 6.04245662689209, "lr": 0.0002, "elapsed_sec": 6769.534161806107, "step_time_sec": 8.23759669798892, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 819, "loss": 6.0981621742248535, "lr": 0.0002, "elapsed_sec": 6777.772333621979, "step_time_sec": 8.238003221922554, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 820, "loss": 5.9660186767578125, "lr": 0.0002, "elapsed_sec": 6786.009234666824, "step_time_sec": 8.236793912015855, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 821, "loss": 6.030314922332764, "lr": 0.0002, "elapsed_sec": 6794.249428272247, "step_time_sec": 8.239962648949586, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 822, "loss": 5.980444431304932, "lr": 0.0002, "elapsed_sec": 6802.486706495285, "step_time_sec": 8.237135351053439, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 823, "loss": 6.112166881561279, "lr": 0.0002, "elapsed_sec": 6810.725183725357, "step_time_sec": 8.238319030031562, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 824, "loss": 6.02931022644043, "lr": 0.0002, "elapsed_sec": 6818.963086843491, "step_time_sec": 8.237696430063806, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 825, "loss": 5.987240791320801, "lr": 0.0002, "elapsed_sec": 6827.2002389431, "step_time_sec": 8.236998977023177, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 826, "loss": 5.942033290863037, "lr": 0.0002, "elapsed_sec": 6835.43961763382, "step_time_sec": 8.23926411406137, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 827, "loss": 6.016929626464844, "lr": 0.0002, "elapsed_sec": 6843.678462982178, "step_time_sec": 8.238684979034588, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 828, "loss": 6.015319347381592, "lr": 0.0002, "elapsed_sec": 6851.916852712631, "step_time_sec": 8.23817056894768, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 829, "loss": 6.15001916885376, "lr": 0.0002, "elapsed_sec": 6860.155663728714, "step_time_sec": 8.238666688092053, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 830, "loss": 5.975167274475098, "lr": 0.0002, "elapsed_sec": 6868.392299175262, "step_time_sec": 8.236474797013216, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 831, "loss": 6.083776950836182, "lr": 0.0002, "elapsed_sec": 6876.630429506302, "step_time_sec": 8.237977485987358, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 832, "loss": 6.047504425048828, "lr": 0.0002, "elapsed_sec": 6884.868465185165, "step_time_sec": 8.237826211028732, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 833, "loss": 6.010073661804199, "lr": 0.0002, "elapsed_sec": 6893.107593536377, "step_time_sec": 8.239008652977645, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 834, "loss": 6.065292835235596, "lr": 0.0002, "elapsed_sec": 6901.345417737961, "step_time_sec": 8.237625106004998, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 835, "loss": 6.007452964782715, "lr": 0.0002, "elapsed_sec": 6909.583571195602, "step_time_sec": 8.237978979945183, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 836, "loss": 6.005126953125, "lr": 0.0002, "elapsed_sec": 6917.8226001262665, "step_time_sec": 8.238858749973588, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 837, "loss": 5.9723920822143555, "lr": 0.0002, "elapsed_sec": 6926.059741973877, "step_time_sec": 8.23703686799854, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 838, "loss": 6.008437156677246, "lr": 0.0002, "elapsed_sec": 6934.299334764481, "step_time_sec": 8.239365022978745, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 839, "loss": 6.081417083740234, "lr": 0.0002, "elapsed_sec": 6942.53835272789, "step_time_sec": 8.238868845975958, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 840, "loss": 6.00496768951416, "lr": 0.0002, "elapsed_sec": 6950.776679992676, "step_time_sec": 8.238177727092989, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 841, "loss": 5.994686603546143, "lr": 0.0002, "elapsed_sec": 6959.015497207642, "step_time_sec": 8.238580506062135, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 842, "loss": 5.949928283691406, "lr": 0.0002, "elapsed_sec": 6967.25270652771, "step_time_sec": 8.237052156007849, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 843, "loss": 5.950872898101807, "lr": 0.0002, "elapsed_sec": 6975.492799758911, "step_time_sec": 8.239981523016468, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 844, "loss": 5.99911642074585, "lr": 0.0002, "elapsed_sec": 6983.731450557709, "step_time_sec": 8.238501952029765, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 845, "loss": 5.922022819519043, "lr": 0.0002, "elapsed_sec": 6991.96933221817, "step_time_sec": 8.23761068994645, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 846, "loss": 5.8550262451171875, "lr": 0.0002, "elapsed_sec": 7000.207219839096, "step_time_sec": 8.237722847028635, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 847, "loss": 5.944210052490234, "lr": 0.0002, "elapsed_sec": 7008.443944454193, "step_time_sec": 8.23657850897871, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 848, "loss": 6.002410411834717, "lr": 0.0002, "elapsed_sec": 7016.6812880039215, "step_time_sec": 8.237175168003887, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 849, "loss": 5.982828617095947, "lr": 0.0002, "elapsed_sec": 7024.919766664505, "step_time_sec": 8.238324486999772, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 850, "loss": 5.909904479980469, "lr": 0.0002, "elapsed_sec": 7033.158730745316, "step_time_sec": 8.238768156035803, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 851, "loss": 5.882016181945801, "lr": 0.0002, "elapsed_sec": 7041.397875785828, "step_time_sec": 8.238998105982319, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 852, "loss": 5.951806545257568, "lr": 0.0002, "elapsed_sec": 7049.634431600571, "step_time_sec": 8.236407135031186, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 853, "loss": 6.009247303009033, "lr": 0.0002, "elapsed_sec": 7057.871587753296, "step_time_sec": 8.236968327080831, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 854, "loss": 5.919415473937988, "lr": 0.0002, "elapsed_sec": 7066.10937166214, "step_time_sec": 8.2376725099748, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 855, "loss": 5.901643753051758, "lr": 0.0002, "elapsed_sec": 7074.346884965897, "step_time_sec": 8.237284469068982, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 856, "loss": 6.0031867027282715, "lr": 0.0002, "elapsed_sec": 7082.584753990173, "step_time_sec": 8.237726422026753, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 857, "loss": 5.953615665435791, "lr": 0.0002, "elapsed_sec": 7090.821866750717, "step_time_sec": 8.236963259987533, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 858, "loss": 6.068852424621582, "lr": 0.0002, "elapsed_sec": 7099.060608625412, "step_time_sec": 8.2385663370369, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 859, "loss": 6.048716068267822, "lr": 0.0002, "elapsed_sec": 7107.297917366028, "step_time_sec": 8.237126966938376, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 860, "loss": 5.971651077270508, "lr": 0.0002, "elapsed_sec": 7115.53707742691, "step_time_sec": 8.238975825952366, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 861, "loss": 5.971935749053955, "lr": 0.0002, "elapsed_sec": 7123.773876667023, "step_time_sec": 8.236713940976188, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 862, "loss": 5.907235622406006, "lr": 0.0002, "elapsed_sec": 7132.012315273285, "step_time_sec": 8.238219022983685, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 863, "loss": 5.934559345245361, "lr": 0.0002, "elapsed_sec": 7140.251106023788, "step_time_sec": 8.238629854982719, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 864, "loss": 5.741912364959717, "lr": 0.0002, "elapsed_sec": 7148.488182067871, "step_time_sec": 8.236915534012951, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 865, "loss": 5.868271350860596, "lr": 0.0002, "elapsed_sec": 7156.728110074997, "step_time_sec": 8.239775239024311, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 866, "loss": 6.02858829498291, "lr": 0.0002, "elapsed_sec": 7164.965713262558, "step_time_sec": 8.237468245089985, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 867, "loss": 6.01663064956665, "lr": 0.0002, "elapsed_sec": 7173.204967021942, "step_time_sec": 8.239055362064391, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 868, "loss": 5.966097354888916, "lr": 0.0002, "elapsed_sec": 7181.44379901886, "step_time_sec": 8.238693243009038, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 869, "loss": 5.983015537261963, "lr": 0.0002, "elapsed_sec": 7189.681015014648, "step_time_sec": 8.237069205963053, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 870, "loss": 5.981625556945801, "lr": 0.0002, "elapsed_sec": 7197.919695615768, "step_time_sec": 8.238532745046541, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 871, "loss": 6.0026936531066895, "lr": 0.0002, "elapsed_sec": 7206.156133174896, "step_time_sec": 8.236289454973303, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 872, "loss": 5.898515701293945, "lr": 0.0002, "elapsed_sec": 7214.3947060108185, "step_time_sec": 8.238362727919593, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 873, "loss": 5.909167766571045, "lr": 0.0002, "elapsed_sec": 7222.633576154709, "step_time_sec": 8.238680143957026, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 874, "loss": 5.882967472076416, "lr": 0.0002, "elapsed_sec": 7230.870971918106, "step_time_sec": 8.237254770006984, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 875, "loss": 5.9020161628723145, "lr": 0.0002, "elapsed_sec": 7239.108985424042, "step_time_sec": 8.237861553090625, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 876, "loss": 5.913640022277832, "lr": 0.0002, "elapsed_sec": 7247.345508575439, "step_time_sec": 8.236308405990712, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 877, "loss": 5.894476890563965, "lr": 0.0002, "elapsed_sec": 7255.58348441124, "step_time_sec": 8.237792951054871, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 878, "loss": 5.912365436553955, "lr": 0.0002, "elapsed_sec": 7263.821268320084, "step_time_sec": 8.237665889086202, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 879, "loss": 5.991419792175293, "lr": 0.0002, "elapsed_sec": 7272.060249090195, "step_time_sec": 8.23878016800154, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 880, "loss": 5.777850151062012, "lr": 0.0002, "elapsed_sec": 7280.299470663071, "step_time_sec": 8.239045531954616, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 881, "loss": 5.826026916503906, "lr": 0.0002, "elapsed_sec": 7288.537138462067, "step_time_sec": 8.237502455944195, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 882, "loss": 5.908718585968018, "lr": 0.0002, "elapsed_sec": 7296.776069402695, "step_time_sec": 8.238761550048366, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 883, "loss": 5.966797351837158, "lr": 0.0002, "elapsed_sec": 7305.0143094062805, "step_time_sec": 8.238087603007443, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 884, "loss": 5.967588424682617, "lr": 0.0002, "elapsed_sec": 7313.253370285034, "step_time_sec": 8.23888097901363, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 885, "loss": 5.969395637512207, "lr": 0.0002, "elapsed_sec": 7321.4909908771515, "step_time_sec": 8.237466586986557, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 886, "loss": 5.939696311950684, "lr": 0.0002, "elapsed_sec": 7329.727941274643, "step_time_sec": 8.236806837958284, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 887, "loss": 5.930794715881348, "lr": 0.0002, "elapsed_sec": 7337.965780496597, "step_time_sec": 8.237671704031527, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 888, "loss": 5.963259220123291, "lr": 0.0002, "elapsed_sec": 7346.202470779419, "step_time_sec": 8.236570860957727, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 889, "loss": 5.8551483154296875, "lr": 0.0002, "elapsed_sec": 7354.440848350525, "step_time_sec": 8.238208167022094, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 890, "loss": 5.887155532836914, "lr": 0.0002, "elapsed_sec": 7362.677775859833, "step_time_sec": 8.23673573799897, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 891, "loss": 5.937141418457031, "lr": 0.0002, "elapsed_sec": 7370.916134119034, "step_time_sec": 8.238187459064648, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 892, "loss": 5.942077159881592, "lr": 0.0002, "elapsed_sec": 7379.153606176376, "step_time_sec": 8.23733157699462, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 893, "loss": 5.905531406402588, "lr": 0.0002, "elapsed_sec": 7387.390289068222, "step_time_sec": 8.236477624974214, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 894, "loss": 5.875247001647949, "lr": 0.0002, "elapsed_sec": 7395.629148483276, "step_time_sec": 8.238750207005069, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 895, "loss": 5.790702819824219, "lr": 0.0002, "elapsed_sec": 7403.866429567337, "step_time_sec": 8.237110878922977, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 896, "loss": 5.84922981262207, "lr": 0.0002, "elapsed_sec": 7412.105697154999, "step_time_sec": 8.238999538938515, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 897, "loss": 5.861799716949463, "lr": 0.0002, "elapsed_sec": 7420.34393119812, "step_time_sec": 8.238078709924594, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 898, "loss": 5.900511264801025, "lr": 0.0002, "elapsed_sec": 7428.5806403160095, "step_time_sec": 8.236578662064858, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 899, "loss": 5.812155246734619, "lr": 0.0002, "elapsed_sec": 7436.8179812431335, "step_time_sec": 8.237116212956607, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 900, "loss": 6.071177959442139, "lr": 0.0002, "elapsed_sec": 7445.055208206177, "step_time_sec": 8.237085410975851, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 901, "loss": 5.848961353302002, "lr": 0.0002, "elapsed_sec": 7453.2932505607605, "step_time_sec": 8.237856753985398, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 902, "loss": 5.925421714782715, "lr": 0.0002, "elapsed_sec": 7461.530319929123, "step_time_sec": 8.236928189056925, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 903, "loss": 5.91256046295166, "lr": 0.0002, "elapsed_sec": 7469.768927812576, "step_time_sec": 8.238465016009286, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 904, "loss": 5.849853515625, "lr": 0.0002, "elapsed_sec": 7478.006819486618, "step_time_sec": 8.237719601020217, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 905, "loss": 6.165440559387207, "lr": 0.0002, "elapsed_sec": 7486.243924617767, "step_time_sec": 8.236986223026179, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 906, "loss": 5.963357448577881, "lr": 0.0002, "elapsed_sec": 7494.483829259872, "step_time_sec": 8.239718763972633, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 907, "loss": 5.929203033447266, "lr": 0.0002, "elapsed_sec": 7502.721491098404, "step_time_sec": 8.237459815922193, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 908, "loss": 5.899240016937256, "lr": 0.0002, "elapsed_sec": 7510.95915722847, "step_time_sec": 8.2374961579917, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 909, "loss": 6.0820207595825195, "lr": 0.0002, "elapsed_sec": 7519.197004318237, "step_time_sec": 8.237714845919982, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 910, "loss": 5.816178321838379, "lr": 0.0002, "elapsed_sec": 7527.434819936752, "step_time_sec": 8.23761900397949, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 911, "loss": 5.888670921325684, "lr": 0.0002, "elapsed_sec": 7535.6722576618195, "step_time_sec": 8.237252179998904, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 912, "loss": 5.893936634063721, "lr": 0.0002, "elapsed_sec": 7543.90972161293, "step_time_sec": 8.237360239960253, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 913, "loss": 5.917688846588135, "lr": 0.0002, "elapsed_sec": 7552.148231744766, "step_time_sec": 8.238295501098037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 914, "loss": 5.929945945739746, "lr": 0.0002, "elapsed_sec": 7560.385306596756, "step_time_sec": 8.236927788937464, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 915, "loss": 5.897341251373291, "lr": 0.0002, "elapsed_sec": 7568.623270750046, "step_time_sec": 8.23781965195667, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 916, "loss": 5.931339740753174, "lr": 0.0002, "elapsed_sec": 7576.864795207977, "step_time_sec": 8.241352273966186, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 917, "loss": 5.758861064910889, "lr": 0.0002, "elapsed_sec": 7585.102675676346, "step_time_sec": 8.237752557964996, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 918, "loss": 5.839747905731201, "lr": 0.0002, "elapsed_sec": 7593.341241836548, "step_time_sec": 8.238359320908785, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 919, "loss": 6.014077663421631, "lr": 0.0002, "elapsed_sec": 7601.577215909958, "step_time_sec": 8.235790325910784, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 920, "loss": 5.844781875610352, "lr": 0.0002, "elapsed_sec": 7609.816956043243, "step_time_sec": 8.239582792040892, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 921, "loss": 5.792951583862305, "lr": 0.0002, "elapsed_sec": 7618.055676698685, "step_time_sec": 8.238600666983984, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 922, "loss": 5.917839050292969, "lr": 0.0002, "elapsed_sec": 7626.293685913086, "step_time_sec": 8.237865353003144, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 923, "loss": 5.90166711807251, "lr": 0.0002, "elapsed_sec": 7634.532178401947, "step_time_sec": 8.23828620603308, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 924, "loss": 5.8478569984436035, "lr": 0.0002, "elapsed_sec": 7642.76908993721, "step_time_sec": 8.23675738601014, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 925, "loss": 5.911428928375244, "lr": 0.0002, "elapsed_sec": 7651.0063123703, "step_time_sec": 8.237085895962082, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 926, "loss": 5.788697719573975, "lr": 0.0002, "elapsed_sec": 7659.242713928223, "step_time_sec": 8.236238649929874, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 927, "loss": 5.836609840393066, "lr": 0.0002, "elapsed_sec": 7667.480749607086, "step_time_sec": 8.23789292993024, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 928, "loss": 5.785946846008301, "lr": 0.0002, "elapsed_sec": 7675.716620922089, "step_time_sec": 8.235714779002592, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 929, "loss": 5.839075088500977, "lr": 0.0002, "elapsed_sec": 7683.954112768173, "step_time_sec": 8.237295557977632, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 930, "loss": 5.828258037567139, "lr": 0.0002, "elapsed_sec": 7692.193266153336, "step_time_sec": 8.23905191407539, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 931, "loss": 5.87502384185791, "lr": 0.0002, "elapsed_sec": 7700.430664777756, "step_time_sec": 8.237226970959455, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 932, "loss": 5.862271308898926, "lr": 0.0002, "elapsed_sec": 7708.669075489044, "step_time_sec": 8.238199600949883, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 933, "loss": 5.886616230010986, "lr": 0.0002, "elapsed_sec": 7716.906831502914, "step_time_sec": 8.237574212020263, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 934, "loss": 5.728899955749512, "lr": 0.0002, "elapsed_sec": 7725.144451379776, "step_time_sec": 8.237509187078103, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 935, "loss": 5.7943549156188965, "lr": 0.0002, "elapsed_sec": 7733.382192134857, "step_time_sec": 8.237545280950144, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 936, "loss": 5.972867488861084, "lr": 0.0002, "elapsed_sec": 7741.6211178302765, "step_time_sec": 8.238780208979733, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 937, "loss": 5.822423934936523, "lr": 0.0002, "elapsed_sec": 7749.860026836395, "step_time_sec": 8.238750254968181, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 938, "loss": 5.828032970428467, "lr": 0.0002, "elapsed_sec": 7758.097690582275, "step_time_sec": 8.237519994960167, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 939, "loss": 5.822328090667725, "lr": 0.0002, "elapsed_sec": 7766.335638999939, "step_time_sec": 8.237849190016277, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 940, "loss": 5.8014349937438965, "lr": 0.0002, "elapsed_sec": 7774.573006391525, "step_time_sec": 8.237148691085167, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 941, "loss": 5.853812217712402, "lr": 0.0002, "elapsed_sec": 7782.810358047485, "step_time_sec": 8.237205263925716, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 942, "loss": 5.925595283508301, "lr": 0.0002, "elapsed_sec": 7791.04886174202, "step_time_sec": 8.238277178024873, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 943, "loss": 5.82641077041626, "lr": 0.0002, "elapsed_sec": 7799.285872220993, "step_time_sec": 8.236884252051823, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 944, "loss": 5.750348091125488, "lr": 0.0002, "elapsed_sec": 7807.524301290512, "step_time_sec": 8.238234840915538, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 945, "loss": 5.819097995758057, "lr": 0.0002, "elapsed_sec": 7815.761797904968, "step_time_sec": 8.237296652048826, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 946, "loss": 5.9237823486328125, "lr": 0.0002, "elapsed_sec": 7823.999230861664, "step_time_sec": 8.237351409043185, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 947, "loss": 5.793138027191162, "lr": 0.0002, "elapsed_sec": 7832.2372262477875, "step_time_sec": 8.237801098031923, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 948, "loss": 5.97797155380249, "lr": 0.0002, "elapsed_sec": 7840.474617242813, "step_time_sec": 8.237234718981199, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 949, "loss": 5.864076137542725, "lr": 0.0002, "elapsed_sec": 7848.713709831238, "step_time_sec": 8.238948169979267, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 950, "loss": 5.782242774963379, "lr": 0.0002, "elapsed_sec": 7856.950192689896, "step_time_sec": 8.236246239976026, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 951, "loss": 5.882319927215576, "lr": 0.0002, "elapsed_sec": 7865.187894582748, "step_time_sec": 8.237538121989928, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 952, "loss": 5.803074359893799, "lr": 0.0002, "elapsed_sec": 7873.427133321762, "step_time_sec": 8.239092924981378, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 953, "loss": 5.73696756362915, "lr": 0.0002, "elapsed_sec": 7881.665318012238, "step_time_sec": 8.237995123956352, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 954, "loss": 5.840216159820557, "lr": 0.0002, "elapsed_sec": 7889.904288768768, "step_time_sec": 8.23889987799339, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 955, "loss": 5.77152681350708, "lr": 0.0002, "elapsed_sec": 7898.140475511551, "step_time_sec": 8.235976313939318, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 956, "loss": 5.798323631286621, "lr": 0.0002, "elapsed_sec": 7906.379149198532, "step_time_sec": 8.238534418051131, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 957, "loss": 5.811199188232422, "lr": 0.0002, "elapsed_sec": 7914.617398738861, "step_time_sec": 8.238053894019686, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 958, "loss": 5.828604698181152, "lr": 0.0002, "elapsed_sec": 7922.8553285598755, "step_time_sec": 8.237727073021233, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 959, "loss": 5.88892936706543, "lr": 0.0002, "elapsed_sec": 7931.093770503998, "step_time_sec": 8.238275639945641, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 960, "loss": 5.8064985275268555, "lr": 0.0002, "elapsed_sec": 7939.332002401352, "step_time_sec": 8.238064490957186, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 961, "loss": 5.861074924468994, "lr": 0.0002, "elapsed_sec": 7947.570410490036, "step_time_sec": 8.238239152939059, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 962, "loss": 5.846914768218994, "lr": 0.0002, "elapsed_sec": 7955.808208703995, "step_time_sec": 8.2376632100204, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 963, "loss": 5.86182165145874, "lr": 0.0002, "elapsed_sec": 7964.04595375061, "step_time_sec": 8.23753929103259, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 964, "loss": 5.880737781524658, "lr": 0.0002, "elapsed_sec": 7972.285177230835, "step_time_sec": 8.23908465297427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 965, "loss": 5.775676727294922, "lr": 0.0002, "elapsed_sec": 7980.523393154144, "step_time_sec": 8.238076532958075, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 966, "loss": 5.824635982513428, "lr": 0.0002, "elapsed_sec": 7988.7632002830505, "step_time_sec": 8.239599735010415, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 967, "loss": 5.831626892089844, "lr": 0.0002, "elapsed_sec": 7997.002539157867, "step_time_sec": 8.239150766981766, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 968, "loss": 5.798460483551025, "lr": 0.0002, "elapsed_sec": 8005.240513801575, "step_time_sec": 8.237897632992826, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 969, "loss": 5.917320728302002, "lr": 0.0002, "elapsed_sec": 8013.479212522507, "step_time_sec": 8.238471540040337, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 970, "loss": 5.794641971588135, "lr": 0.0002, "elapsed_sec": 8021.716880083084, "step_time_sec": 8.237517073052004, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 971, "loss": 5.695338726043701, "lr": 0.0002, "elapsed_sec": 8029.955845117569, "step_time_sec": 8.2387773609953, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 972, "loss": 5.701529502868652, "lr": 0.0002, "elapsed_sec": 8038.19301533699, "step_time_sec": 8.236999418935739, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 973, "loss": 5.763298034667969, "lr": 0.0002, "elapsed_sec": 8046.430865526199, "step_time_sec": 8.237736523966305, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 974, "loss": 5.825243949890137, "lr": 0.0002, "elapsed_sec": 8054.669268846512, "step_time_sec": 8.23823509097565, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 975, "loss": 5.893764495849609, "lr": 0.0002, "elapsed_sec": 8062.906628847122, "step_time_sec": 8.237179182004184, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 976, "loss": 5.805239677429199, "lr": 0.0002, "elapsed_sec": 8071.144792556763, "step_time_sec": 8.237988553009927, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 977, "loss": 5.834702968597412, "lr": 0.0002, "elapsed_sec": 8079.382258176804, "step_time_sec": 8.237336704973131, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 978, "loss": 5.880005359649658, "lr": 0.0002, "elapsed_sec": 8087.620985507965, "step_time_sec": 8.238532667979598, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 979, "loss": 5.748429775238037, "lr": 0.0002, "elapsed_sec": 8095.859195947647, "step_time_sec": 8.23805037199054, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 980, "loss": 5.954296112060547, "lr": 0.0002, "elapsed_sec": 8104.097659111023, "step_time_sec": 8.238352170912549, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 981, "loss": 5.855659484863281, "lr": 0.0002, "elapsed_sec": 8112.336683511734, "step_time_sec": 8.238814022042789, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 982, "loss": 5.744431018829346, "lr": 0.0002, "elapsed_sec": 8120.573960065842, "step_time_sec": 8.23712444503326, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 983, "loss": 5.8260908126831055, "lr": 0.0002, "elapsed_sec": 8128.81224489212, "step_time_sec": 8.23812288697809, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 984, "loss": 5.912789821624756, "lr": 0.0002, "elapsed_sec": 8137.049486398697, "step_time_sec": 8.237079038051888, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 985, "loss": 5.6796064376831055, "lr": 0.0002, "elapsed_sec": 8145.2870221138, "step_time_sec": 8.237365382956341, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 986, "loss": 5.91783332824707, "lr": 0.0002, "elapsed_sec": 8153.524560689926, "step_time_sec": 8.237401360063814, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 987, "loss": 5.774699687957764, "lr": 0.0002, "elapsed_sec": 8161.76121711731, "step_time_sec": 8.236492458963767, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 988, "loss": 5.790182113647461, "lr": 0.0002, "elapsed_sec": 8170.000390768051, "step_time_sec": 8.238990849000402, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 989, "loss": 5.740967273712158, "lr": 0.0002, "elapsed_sec": 8178.237340450287, "step_time_sec": 8.236832401016727, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 990, "loss": 5.888001441955566, "lr": 0.0002, "elapsed_sec": 8186.476206302643, "step_time_sec": 8.238714108942077, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 991, "loss": 5.726206302642822, "lr": 0.0002, "elapsed_sec": 8194.7135887146, "step_time_sec": 8.237203545984812, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 992, "loss": 5.738180637359619, "lr": 0.0002, "elapsed_sec": 8202.951888084412, "step_time_sec": 8.238100940012373, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 993, "loss": 5.793858051300049, "lr": 0.0002, "elapsed_sec": 8211.190938949585, "step_time_sec": 8.238871339010075, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 994, "loss": 5.7728071212768555, "lr": 0.0002, "elapsed_sec": 8219.428007125854, "step_time_sec": 8.236924955039285, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 995, "loss": 5.844843864440918, "lr": 0.0002, "elapsed_sec": 8227.666462421417, "step_time_sec": 8.238283218001015, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 996, "loss": 5.78871488571167, "lr": 0.0002, "elapsed_sec": 8235.906438112259, "step_time_sec": 8.239823770010844, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 997, "loss": 5.798758029937744, "lr": 0.0002, "elapsed_sec": 8244.144566774368, "step_time_sec": 8.237941292929463, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 998, "loss": 5.792169094085693, "lr": 0.0002, "elapsed_sec": 8252.383379220963, "step_time_sec": 8.238674217951484, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 999, "loss": 5.775876045227051, "lr": 0.0002, "elapsed_sec": 8260.620340585709, "step_time_sec": 8.236801724066027, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1000, "loss": 5.831096172332764, "lr": 0.0002, "elapsed_sec": 8268.860056400299, "step_time_sec": 56.532095120986924, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": true, "validation_ran": true, "probe_ran": true, "probe_elapsed_sec": 1.0408399919979274, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1001, "loss": 5.7492876052856445, "lr": 0.0002, "elapsed_sec": 8325.396635770798, "step_time_sec": 8.243939982028678, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1002, "loss": 5.762201309204102, "lr": 0.0002, "elapsed_sec": 8333.627455234528, "step_time_sec": 8.230634235078469, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1003, "loss": 5.847352981567383, "lr": 0.0002, "elapsed_sec": 8341.86117863655, "step_time_sec": 8.233536534942687, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1004, "loss": 5.825220584869385, "lr": 0.0002, "elapsed_sec": 8350.092835903168, "step_time_sec": 8.231499720015563, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1005, "loss": 5.700362205505371, "lr": 0.0002, "elapsed_sec": 8358.325621843338, "step_time_sec": 8.23260908992961, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1006, "loss": 5.785674095153809, "lr": 0.0002, "elapsed_sec": 8366.563427686691, "step_time_sec": 8.237679223995656, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1007, "loss": 5.860411643981934, "lr": 0.0002, "elapsed_sec": 8374.80281829834, "step_time_sec": 8.239209591061808, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1008, "loss": 5.789721965789795, "lr": 0.0002, "elapsed_sec": 8383.040262699127, "step_time_sec": 8.237263408023864, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1009, "loss": 5.823358058929443, "lr": 0.0002, "elapsed_sec": 8391.27789068222, "step_time_sec": 8.237439482007176, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1010, "loss": 5.747453212738037, "lr": 0.0002, "elapsed_sec": 8399.516308546066, "step_time_sec": 8.238270801026374, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1011, "loss": 5.709474086761475, "lr": 0.0002, "elapsed_sec": 8407.75447511673, "step_time_sec": 8.237985846004449, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1012, "loss": 5.716081142425537, "lr": 0.0002, "elapsed_sec": 8415.993784666061, "step_time_sec": 8.239196565002203, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1013, "loss": 5.7133283615112305, "lr": 0.0002, "elapsed_sec": 8424.232558727264, "step_time_sec": 8.238542675040662, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1014, "loss": 5.7637128829956055, "lr": 0.0002, "elapsed_sec": 8432.471717119217, "step_time_sec": 8.238985714968294, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1015, "loss": 5.724640846252441, "lr": 0.0002, "elapsed_sec": 8440.710587739944, "step_time_sec": 8.238717960892245, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1016, "loss": 5.7167439460754395, "lr": 0.0002, "elapsed_sec": 8448.948226690292, "step_time_sec": 8.237462676945142, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1017, "loss": 5.740604877471924, "lr": 0.0002, "elapsed_sec": 8457.187085151672, "step_time_sec": 8.238678688998334, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1018, "loss": 5.977583885192871, "lr": 0.0002, "elapsed_sec": 8465.423656702042, "step_time_sec": 8.23639739700593, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1019, "loss": 5.609865665435791, "lr": 0.0002, "elapsed_sec": 8473.66333937645, "step_time_sec": 8.239567875978537, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1020, "loss": 5.772155284881592, "lr": 0.0002, "elapsed_sec": 8481.902292728424, "step_time_sec": 8.238755750004202, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1021, "loss": 5.841010570526123, "lr": 0.0002, "elapsed_sec": 8490.139613628387, "step_time_sec": 8.23720831295941, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1022, "loss": 5.7856245040893555, "lr": 0.0002, "elapsed_sec": 8498.377004146576, "step_time_sec": 8.237157876021229, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1023, "loss": 5.6620283126831055, "lr": 0.0002, "elapsed_sec": 8506.614711523056, "step_time_sec": 8.237592600053176, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1024, "loss": 5.78640604019165, "lr": 0.0002, "elapsed_sec": 8514.854366064072, "step_time_sec": 8.239467539009638, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1025, "loss": 5.801955223083496, "lr": 0.0002, "elapsed_sec": 8523.092609882355, "step_time_sec": 8.238043248071335, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1026, "loss": 5.780502796173096, "lr": 0.0002, "elapsed_sec": 8531.331217050552, "step_time_sec": 8.238405293086544, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1027, "loss": 5.626460075378418, "lr": 0.0002, "elapsed_sec": 8539.568985462189, "step_time_sec": 8.237609314033762, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1028, "loss": 5.606596946716309, "lr": 0.0002, "elapsed_sec": 8547.80682349205, "step_time_sec": 8.237663346924819, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1029, "loss": 5.653594017028809, "lr": 0.0002, "elapsed_sec": 8556.044843912125, "step_time_sec": 8.237911181990057, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1030, "loss": 5.708229064941406, "lr": 0.0002, "elapsed_sec": 8564.281670570374, "step_time_sec": 8.23665702494327, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1031, "loss": 5.807880401611328, "lr": 0.0002, "elapsed_sec": 8572.51974105835, "step_time_sec": 8.237835369887762, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1032, "loss": 5.743180751800537, "lr": 0.0002, "elapsed_sec": 8580.75858926773, "step_time_sec": 8.238736505038105, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1033, "loss": 5.777063369750977, "lr": 0.0002, "elapsed_sec": 8588.997186660767, "step_time_sec": 8.238408991950564, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1034, "loss": 5.706149101257324, "lr": 0.0002, "elapsed_sec": 8597.236988782883, "step_time_sec": 8.23966639500577, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1035, "loss": 5.738175392150879, "lr": 0.0002, "elapsed_sec": 8605.475350618362, "step_time_sec": 8.238135698018596, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1036, "loss": 5.8937482833862305, "lr": 0.0002, "elapsed_sec": 8613.714386701584, "step_time_sec": 8.238873871974647, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1037, "loss": 5.699069023132324, "lr": 0.0002, "elapsed_sec": 8621.951421499252, "step_time_sec": 8.236892033019103, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1038, "loss": 5.683004379272461, "lr": 0.0002, "elapsed_sec": 8630.189002752304, "step_time_sec": 8.237425941042602, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1039, "loss": 5.836729526519775, "lr": 0.0002, "elapsed_sec": 8638.427104949951, "step_time_sec": 8.237902098917402, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1040, "loss": 5.712884426116943, "lr": 0.0002, "elapsed_sec": 8646.663284301758, "step_time_sec": 8.236008169944398, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1041, "loss": 5.680383682250977, "lr": 0.0002, "elapsed_sec": 8654.900275707245, "step_time_sec": 8.236905156983994, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1042, "loss": 5.718149185180664, "lr": 0.0002, "elapsed_sec": 8663.13762664795, "step_time_sec": 8.237131973030046, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1043, "loss": 5.702749252319336, "lr": 0.0002, "elapsed_sec": 8671.376321554184, "step_time_sec": 8.238522688974626, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1044, "loss": 5.721015453338623, "lr": 0.0002, "elapsed_sec": 8679.615804672241, "step_time_sec": 8.23931822006125, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1045, "loss": 5.679660320281982, "lr": 0.0002, "elapsed_sec": 8687.853465080261, "step_time_sec": 8.237492181011476, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1046, "loss": 5.799445629119873, "lr": 0.0002, "elapsed_sec": 8696.092912912369, "step_time_sec": 8.239307236042805, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1047, "loss": 5.768051624298096, "lr": 0.0002, "elapsed_sec": 8704.329663038254, "step_time_sec": 8.236530096037313, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1048, "loss": 5.776209354400635, "lr": 0.0002, "elapsed_sec": 8712.568897008896, "step_time_sec": 8.239057422964834, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1049, "loss": 5.802773475646973, "lr": 0.0002, "elapsed_sec": 8720.807306528091, "step_time_sec": 8.238255426986143, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1050, "loss": 5.7917938232421875, "lr": 0.0002, "elapsed_sec": 8729.044823646545, "step_time_sec": 8.237364692962728, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1051, "loss": 5.638964653015137, "lr": 0.0002, "elapsed_sec": 8737.283270835876, "step_time_sec": 8.238293537055142, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1052, "loss": 5.5744829177856445, "lr": 0.0002, "elapsed_sec": 8745.521617174149, "step_time_sec": 8.238170370925218, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1053, "loss": 5.855215072631836, "lr": 0.0002, "elapsed_sec": 8753.760397434235, "step_time_sec": 8.238604249083437, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1054, "loss": 5.779230117797852, "lr": 0.0002, "elapsed_sec": 8761.998337984085, "step_time_sec": 8.237787392921746, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1055, "loss": 5.697941780090332, "lr": 0.0002, "elapsed_sec": 8770.23636841774, "step_time_sec": 8.237899835105054, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1056, "loss": 5.736359119415283, "lr": 0.0002, "elapsed_sec": 8778.475170373917, "step_time_sec": 8.23865349998232, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1057, "loss": 5.4006476402282715, "lr": 0.0002, "elapsed_sec": 8786.712545871735, "step_time_sec": 8.237147633102722, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1058, "loss": 5.775607109069824, "lr": 0.0002, "elapsed_sec": 8794.952246427536, "step_time_sec": 8.239545829012059, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1059, "loss": 5.666363716125488, "lr": 0.0002, "elapsed_sec": 8803.190341949463, "step_time_sec": 8.237953471951187, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1060, "loss": 5.6657915115356445, "lr": 0.0002, "elapsed_sec": 8811.42844581604, "step_time_sec": 8.237922480911948, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1061, "loss": 5.874473571777344, "lr": 0.0002, "elapsed_sec": 8819.66692352295, "step_time_sec": 8.238339951029047, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1062, "loss": 5.729091644287109, "lr": 0.0002, "elapsed_sec": 8827.904589653015, "step_time_sec": 8.237516100052744, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1063, "loss": 5.762699127197266, "lr": 0.0002, "elapsed_sec": 8836.144398450851, "step_time_sec": 8.239633618039079, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1064, "loss": 5.717343807220459, "lr": 0.0002, "elapsed_sec": 8844.383245944977, "step_time_sec": 8.238627644022927, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1065, "loss": 5.669381618499756, "lr": 0.0002, "elapsed_sec": 8852.621701955795, "step_time_sec": 8.238275045063347, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1066, "loss": 5.762794494628906, "lr": 0.0002, "elapsed_sec": 8860.860172986984, "step_time_sec": 8.238320057978854, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1067, "loss": 5.901045799255371, "lr": 0.0002, "elapsed_sec": 8869.098737001419, "step_time_sec": 8.238421263056807, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1068, "loss": 5.643359661102295, "lr": 0.0002, "elapsed_sec": 8877.33758020401, "step_time_sec": 8.238713318947703, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1069, "loss": 5.572392463684082, "lr": 0.0002, "elapsed_sec": 8885.576766252518, "step_time_sec": 8.238941189018078, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1070, "loss": 5.699666500091553, "lr": 0.0002, "elapsed_sec": 8893.81755900383, "step_time_sec": 8.240681488998234, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1071, "loss": 5.559523105621338, "lr": 0.0002, "elapsed_sec": 8902.055938959122, "step_time_sec": 8.23813195200637, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1072, "loss": 5.7430315017700195, "lr": 0.0002, "elapsed_sec": 8910.294046163559, "step_time_sec": 8.237957716919482, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1073, "loss": 5.642056465148926, "lr": 0.0002, "elapsed_sec": 8918.533954620361, "step_time_sec": 8.239710196037777, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1074, "loss": 5.69500732421875, "lr": 0.0002, "elapsed_sec": 8926.772737026215, "step_time_sec": 8.238634299952537, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1075, "loss": 5.710224628448486, "lr": 0.0002, "elapsed_sec": 8935.01040482521, "step_time_sec": 8.237506591016427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1076, "loss": 5.719419956207275, "lr": 0.0002, "elapsed_sec": 8943.248626708984, "step_time_sec": 8.238036549999379, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1077, "loss": 5.709854602813721, "lr": 0.0002, "elapsed_sec": 8951.485495567322, "step_time_sec": 8.236692523933016, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1078, "loss": 5.641151428222656, "lr": 0.0002, "elapsed_sec": 8959.724113941193, "step_time_sec": 8.238443860085681, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1079, "loss": 5.635919094085693, "lr": 0.0002, "elapsed_sec": 8967.961610078812, "step_time_sec": 8.237365762004629, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1080, "loss": 5.683271408081055, "lr": 0.0002, "elapsed_sec": 8976.20024895668, "step_time_sec": 8.238477978040464, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1081, "loss": 5.580080986022949, "lr": 0.0002, "elapsed_sec": 8984.438214302063, "step_time_sec": 8.23777508805506, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1082, "loss": 5.645860195159912, "lr": 0.0002, "elapsed_sec": 8992.675903558731, "step_time_sec": 8.237504675984383, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1083, "loss": 5.710867881774902, "lr": 0.0002, "elapsed_sec": 9000.914854764938, "step_time_sec": 8.238787406939082, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1084, "loss": 5.682284355163574, "lr": 0.0002, "elapsed_sec": 9009.153403520584, "step_time_sec": 8.238392828032374, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1085, "loss": 5.697488307952881, "lr": 0.0002, "elapsed_sec": 9017.393147706985, "step_time_sec": 8.23959685896989, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1086, "loss": 5.644958972930908, "lr": 0.0002, "elapsed_sec": 9025.63291311264, "step_time_sec": 8.239606651943177, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1087, "loss": 5.800923824310303, "lr": 0.0002, "elapsed_sec": 9033.870075941086, "step_time_sec": 8.236980307963677, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1088, "loss": 5.705289363861084, "lr": 0.0002, "elapsed_sec": 9042.108505010605, "step_time_sec": 8.23825920803938, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1089, "loss": 5.665680885314941, "lr": 0.0002, "elapsed_sec": 9050.346052408218, "step_time_sec": 8.237425280967727, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1090, "loss": 5.882720947265625, "lr": 0.0002, "elapsed_sec": 9058.584619283676, "step_time_sec": 8.238393884035759, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1091, "loss": 5.7360663414001465, "lr": 0.0002, "elapsed_sec": 9066.822761058807, "step_time_sec": 8.237959627062082, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1092, "loss": 5.598388671875, "lr": 0.0002, "elapsed_sec": 9075.06047487259, "step_time_sec": 8.23755472106859, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1093, "loss": 5.602817058563232, "lr": 0.0002, "elapsed_sec": 9083.299062013626, "step_time_sec": 8.23842795798555, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1094, "loss": 5.7282185554504395, "lr": 0.0002, "elapsed_sec": 9091.536708593369, "step_time_sec": 8.237477682996541, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1095, "loss": 5.618457317352295, "lr": 0.0002, "elapsed_sec": 9099.775452375412, "step_time_sec": 8.23857018596027, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1096, "loss": 5.66197395324707, "lr": 0.0002, "elapsed_sec": 9108.014026641846, "step_time_sec": 8.238382088951766, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1097, "loss": 5.582663059234619, "lr": 0.0002, "elapsed_sec": 9116.25189256668, "step_time_sec": 8.237776483991183, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1098, "loss": 5.634335994720459, "lr": 0.0002, "elapsed_sec": 9124.49050283432, "step_time_sec": 8.238451301003806, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1099, "loss": 5.6697468757629395, "lr": 0.0002, "elapsed_sec": 9132.728292942047, "step_time_sec": 8.237562494934537, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1100, "loss": 5.608015060424805, "lr": 0.0002, "elapsed_sec": 9140.968186378479, "step_time_sec": 8.239706578082405, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1101, "loss": 5.785818576812744, "lr": 0.0002, "elapsed_sec": 9149.205464601517, "step_time_sec": 8.237135605071671, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1102, "loss": 5.702581405639648, "lr": 0.0002, "elapsed_sec": 9157.444728136063, "step_time_sec": 8.239114130032249, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1103, "loss": 5.70742130279541, "lr": 0.0002, "elapsed_sec": 9165.683538675308, "step_time_sec": 8.23865803598892, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1104, "loss": 5.728655815124512, "lr": 0.0002, "elapsed_sec": 9173.921593904495, "step_time_sec": 8.23783816897776, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1105, "loss": 5.675436496734619, "lr": 0.0002, "elapsed_sec": 9182.160805940628, "step_time_sec": 8.23907891206909, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1106, "loss": 5.565113544464111, "lr": 0.0002, "elapsed_sec": 9190.39851307869, "step_time_sec": 8.237567189964466, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1107, "loss": 5.643481254577637, "lr": 0.0002, "elapsed_sec": 9198.63789010048, "step_time_sec": 8.239223685930483, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1108, "loss": 5.726022720336914, "lr": 0.0002, "elapsed_sec": 9206.876207351685, "step_time_sec": 8.238097036024556, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1109, "loss": 5.684201240539551, "lr": 0.0002, "elapsed_sec": 9215.113835811615, "step_time_sec": 8.237471658037975, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1110, "loss": 5.633396148681641, "lr": 0.0002, "elapsed_sec": 9223.353573560715, "step_time_sec": 8.239589978940785, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1111, "loss": 5.57214879989624, "lr": 0.0002, "elapsed_sec": 9231.592665195465, "step_time_sec": 8.238887262996286, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1112, "loss": 5.672004699707031, "lr": 0.0002, "elapsed_sec": 9239.831012487411, "step_time_sec": 8.238213359960355, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1113, "loss": 5.60535192489624, "lr": 0.0002, "elapsed_sec": 9248.070299863815, "step_time_sec": 8.239100999897346, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1114, "loss": 5.502193450927734, "lr": 0.0002, "elapsed_sec": 9256.30870771408, "step_time_sec": 8.238282499019988, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1115, "loss": 5.751976490020752, "lr": 0.0002, "elapsed_sec": 9264.549221038818, "step_time_sec": 8.240303923957981, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1116, "loss": 5.67785120010376, "lr": 0.0002, "elapsed_sec": 9272.787736177444, "step_time_sec": 8.23835591902025, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1117, "loss": 5.682644844055176, "lr": 0.0002, "elapsed_sec": 9281.02619600296, "step_time_sec": 8.238304015947506, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1118, "loss": 5.532739639282227, "lr": 0.0002, "elapsed_sec": 9289.264891147614, "step_time_sec": 8.238512012991123, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1119, "loss": 5.588464736938477, "lr": 0.0002, "elapsed_sec": 9297.503093004227, "step_time_sec": 8.238068888080306, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1120, "loss": 5.844012260437012, "lr": 0.0002, "elapsed_sec": 9305.74342417717, "step_time_sec": 8.240175173967145, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1121, "loss": 5.613182067871094, "lr": 0.0002, "elapsed_sec": 9313.981181621552, "step_time_sec": 8.23755181604065, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1122, "loss": 5.607848167419434, "lr": 0.0002, "elapsed_sec": 9322.219217061996, "step_time_sec": 8.237850954057649, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1123, "loss": 5.592252731323242, "lr": 0.0002, "elapsed_sec": 9330.45757484436, "step_time_sec": 8.238238068064675, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1124, "loss": 5.689435958862305, "lr": 0.0002, "elapsed_sec": 9338.69433426857, "step_time_sec": 8.236541792051867, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1125, "loss": 5.593691349029541, "lr": 0.0002, "elapsed_sec": 9346.933329820633, "step_time_sec": 8.238894512993284, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1126, "loss": 5.533607482910156, "lr": 0.0002, "elapsed_sec": 9355.170352220535, "step_time_sec": 8.236782395048067, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1127, "loss": 5.642741680145264, "lr": 0.0002, "elapsed_sec": 9363.409605503082, "step_time_sec": 8.2391136350343, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1128, "loss": 5.6197004318237305, "lr": 0.0002, "elapsed_sec": 9371.64877462387, "step_time_sec": 8.238991383928806, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1129, "loss": 5.632290363311768, "lr": 0.0002, "elapsed_sec": 9379.887059688568, "step_time_sec": 8.23812814091798, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1130, "loss": 5.568246364593506, "lr": 0.0002, "elapsed_sec": 9388.125826835632, "step_time_sec": 8.238650672952645, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1131, "loss": 5.705937385559082, "lr": 0.0002, "elapsed_sec": 9396.362433433533, "step_time_sec": 8.236440561013296, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1132, "loss": 5.516078472137451, "lr": 0.0002, "elapsed_sec": 9404.602372646332, "step_time_sec": 8.23971483798232, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1133, "loss": 5.552435398101807, "lr": 0.0002, "elapsed_sec": 9412.841320037842, "step_time_sec": 8.238785302033648, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1134, "loss": 5.708035469055176, "lr": 0.0002, "elapsed_sec": 9421.078213214874, "step_time_sec": 8.2367199070286, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1135, "loss": 5.599870204925537, "lr": 0.0002, "elapsed_sec": 9429.316802740097, "step_time_sec": 8.238452716032043, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1136, "loss": 5.665782928466797, "lr": 0.0002, "elapsed_sec": 9437.553086519241, "step_time_sec": 8.236084164003842, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1137, "loss": 5.6676201820373535, "lr": 0.0002, "elapsed_sec": 9445.790401697159, "step_time_sec": 8.237181973992847, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1138, "loss": 5.583587169647217, "lr": 0.0002, "elapsed_sec": 9454.029131412506, "step_time_sec": 8.238538771984167, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1139, "loss": 5.598339557647705, "lr": 0.0002, "elapsed_sec": 9462.267575025558, "step_time_sec": 8.238281204015948, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1140, "loss": 5.5983405113220215, "lr": 0.0002, "elapsed_sec": 9470.506167650223, "step_time_sec": 8.23845226794947, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1141, "loss": 5.7054667472839355, "lr": 0.0002, "elapsed_sec": 9478.743470907211, "step_time_sec": 8.237141466932371, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1142, "loss": 5.593905448913574, "lr": 0.0002, "elapsed_sec": 9486.982508659363, "step_time_sec": 8.23885398695711, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1143, "loss": 5.586242198944092, "lr": 0.0002, "elapsed_sec": 9495.221694231033, "step_time_sec": 8.239003574941307, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1144, "loss": 5.544682502746582, "lr": 0.0002, "elapsed_sec": 9503.461050271988, "step_time_sec": 8.23918070399668, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1145, "loss": 5.692378044128418, "lr": 0.0002, "elapsed_sec": 9511.698272228241, "step_time_sec": 8.23705393599812, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1146, "loss": 5.648906230926514, "lr": 0.0002, "elapsed_sec": 9519.936476945877, "step_time_sec": 8.238038150942884, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1147, "loss": 5.761913299560547, "lr": 0.0002, "elapsed_sec": 9528.174594163895, "step_time_sec": 8.237960970960557, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1148, "loss": 5.627458095550537, "lr": 0.0002, "elapsed_sec": 9536.411885976791, "step_time_sec": 8.237165806000121, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1149, "loss": 5.447813034057617, "lr": 0.0002, "elapsed_sec": 9544.6505920887, "step_time_sec": 8.238521269056946, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1150, "loss": 5.548807144165039, "lr": 0.0002, "elapsed_sec": 9552.889462709427, "step_time_sec": 8.238704540999606, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1151, "loss": 5.653830051422119, "lr": 0.0002, "elapsed_sec": 9561.128755807877, "step_time_sec": 8.239110232912935, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1152, "loss": 5.60993766784668, "lr": 0.0002, "elapsed_sec": 9569.366733312607, "step_time_sec": 8.237819790025242, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1153, "loss": 5.56334924697876, "lr": 0.0002, "elapsed_sec": 9577.605751514435, "step_time_sec": 8.23883016698528, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1154, "loss": 5.591949462890625, "lr": 0.0002, "elapsed_sec": 9585.844182252884, "step_time_sec": 8.23829540796578, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1155, "loss": 5.592220306396484, "lr": 0.0002, "elapsed_sec": 9594.08143210411, "step_time_sec": 8.237066603964195, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1156, "loss": 5.587693214416504, "lr": 0.0002, "elapsed_sec": 9602.319128274918, "step_time_sec": 8.237509847967885, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1157, "loss": 5.5301055908203125, "lr": 0.0002, "elapsed_sec": 9610.558579683304, "step_time_sec": 8.239297552965581, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1158, "loss": 5.595036029815674, "lr": 0.0002, "elapsed_sec": 9618.795891046524, "step_time_sec": 8.237193756038323, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1159, "loss": 5.563210487365723, "lr": 0.0002, "elapsed_sec": 9627.035104036331, "step_time_sec": 8.23903525294736, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1160, "loss": 5.5221638679504395, "lr": 0.0002, "elapsed_sec": 9635.273544311523, "step_time_sec": 8.238246769062243, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1161, "loss": 5.58657169342041, "lr": 0.0002, "elapsed_sec": 9643.510751008987, "step_time_sec": 8.2370462550316, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1162, "loss": 5.6212944984436035, "lr": 0.0002, "elapsed_sec": 9651.74847316742, "step_time_sec": 8.23755748406984, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1163, "loss": 5.634706020355225, "lr": 0.0002, "elapsed_sec": 9659.986400604248, "step_time_sec": 8.237790261977352, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1164, "loss": 5.548557281494141, "lr": 0.0002, "elapsed_sec": 9668.224517345428, "step_time_sec": 8.2379283180926, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1165, "loss": 5.453871250152588, "lr": 0.0002, "elapsed_sec": 9676.463726520538, "step_time_sec": 8.239039235981181, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1166, "loss": 5.639713287353516, "lr": 0.0002, "elapsed_sec": 9684.703727006912, "step_time_sec": 8.239816189976409, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1167, "loss": 5.674442768096924, "lr": 0.0002, "elapsed_sec": 9692.942476272583, "step_time_sec": 8.238570312969387, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1168, "loss": 5.515896320343018, "lr": 0.0002, "elapsed_sec": 9701.180406093597, "step_time_sec": 8.23774479306303, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1169, "loss": 5.686110973358154, "lr": 0.0002, "elapsed_sec": 9709.419205188751, "step_time_sec": 8.238649741979316, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1170, "loss": 5.663030624389648, "lr": 0.0002, "elapsed_sec": 9717.657663583755, "step_time_sec": 8.238304475904442, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1171, "loss": 5.475349426269531, "lr": 0.0002, "elapsed_sec": 9725.895910024643, "step_time_sec": 8.238111888989806, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1172, "loss": 5.720089912414551, "lr": 0.0002, "elapsed_sec": 9734.135465860367, "step_time_sec": 8.239361237967387, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1173, "loss": 5.471502304077148, "lr": 0.0002, "elapsed_sec": 9742.373144626617, "step_time_sec": 8.237499632989056, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1174, "loss": 5.527563571929932, "lr": 0.0002, "elapsed_sec": 9750.610379219055, "step_time_sec": 8.23708050197456, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1175, "loss": 5.559852123260498, "lr": 0.0002, "elapsed_sec": 9758.847537994385, "step_time_sec": 8.236974266008474, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1176, "loss": 5.689246654510498, "lr": 0.0002, "elapsed_sec": 9767.087243318558, "step_time_sec": 8.239556927001104, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1177, "loss": 5.51239013671875, "lr": 0.0002, "elapsed_sec": 9775.325684309006, "step_time_sec": 8.238286803010851, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1178, "loss": 5.424934387207031, "lr": 0.0002, "elapsed_sec": 9783.562710285187, "step_time_sec": 8.236904102028348, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1179, "loss": 5.552272796630859, "lr": 0.0002, "elapsed_sec": 9791.80062508583, "step_time_sec": 8.237701000995003, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1180, "loss": 5.640812397003174, "lr": 0.0002, "elapsed_sec": 9800.03731250763, "step_time_sec": 8.236581915058196, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1181, "loss": 5.642898082733154, "lr": 0.0002, "elapsed_sec": 9808.275411128998, "step_time_sec": 8.237917341059074, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1182, "loss": 5.552578449249268, "lr": 0.0002, "elapsed_sec": 9816.51351428032, "step_time_sec": 8.23794854793232, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1183, "loss": 5.611141204833984, "lr": 0.0002, "elapsed_sec": 9824.752137184143, "step_time_sec": 8.238404919044115, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1184, "loss": 5.6233649253845215, "lr": 0.0002, "elapsed_sec": 9832.990623235703, "step_time_sec": 8.238349477062002, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1185, "loss": 5.748378753662109, "lr": 0.0002, "elapsed_sec": 9841.22885298729, "step_time_sec": 8.238010981935076, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1186, "loss": 5.513117790222168, "lr": 0.0002, "elapsed_sec": 9849.467611312866, "step_time_sec": 8.23859481792897, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1187, "loss": 5.580532550811768, "lr": 0.0002, "elapsed_sec": 9857.704820156097, "step_time_sec": 8.237049733055755, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1188, "loss": 5.544003009796143, "lr": 0.0002, "elapsed_sec": 9865.944783449173, "step_time_sec": 8.239849805017002, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1189, "loss": 5.672271728515625, "lr": 0.0002, "elapsed_sec": 9874.183552026749, "step_time_sec": 8.238543475978076, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1190, "loss": 5.404423713684082, "lr": 0.0002, "elapsed_sec": 9882.421608686447, "step_time_sec": 8.23788326093927, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1191, "loss": 5.551297664642334, "lr": 0.0002, "elapsed_sec": 9890.660259485245, "step_time_sec": 8.238483963999897, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1192, "loss": 5.67394495010376, "lr": 0.0002, "elapsed_sec": 9898.89783000946, "step_time_sec": 8.237413854105398, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1193, "loss": 5.712590217590332, "lr": 0.0002, "elapsed_sec": 9907.137356042862, "step_time_sec": 8.23936866002623, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1194, "loss": 5.521029472351074, "lr": 0.0002, "elapsed_sec": 9915.376763105392, "step_time_sec": 8.239302437053993, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1195, "loss": 5.5034918785095215, "lr": 0.0002, "elapsed_sec": 9923.61539196968, "step_time_sec": 8.238420377019793, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1196, "loss": 5.526491641998291, "lr": 0.0002, "elapsed_sec": 9931.852265357971, "step_time_sec": 8.236704070004635, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1197, "loss": 5.624948024749756, "lr": 0.0002, "elapsed_sec": 9940.091222286224, "step_time_sec": 8.238788828952238, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1198, "loss": 5.558993339538574, "lr": 0.0002, "elapsed_sec": 9948.330075263977, "step_time_sec": 8.238741582958028, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1199, "loss": 5.522086143493652, "lr": 0.0002, "elapsed_sec": 9956.568368911743, "step_time_sec": 8.238130310899578, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1200, "loss": 5.720624923706055, "lr": 0.0002, "elapsed_sec": 9964.80612564087, "step_time_sec": 8.237533709034324, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1201, "loss": 5.478696346282959, "lr": 0.0002, "elapsed_sec": 9973.043827533722, "step_time_sec": 8.237544902018271, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1202, "loss": 5.580352783203125, "lr": 0.0002, "elapsed_sec": 9981.280463695526, "step_time_sec": 8.236511281924322, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1203, "loss": 5.600078582763672, "lr": 0.0002, "elapsed_sec": 9989.519984960556, "step_time_sec": 8.239319392945617, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1204, "loss": 5.5187201499938965, "lr": 0.0002, "elapsed_sec": 9997.758190631866, "step_time_sec": 8.23803464602679, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1205, "loss": 5.587357044219971, "lr": 0.0002, "elapsed_sec": 10005.997463226318, "step_time_sec": 8.239201387972571, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1206, "loss": 5.5623674392700195, "lr": 0.0002, "elapsed_sec": 10014.2378282547, "step_time_sec": 8.240124189993367, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1207, "loss": 5.510306358337402, "lr": 0.0002, "elapsed_sec": 10022.476666212082, "step_time_sec": 8.238681769929826, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1208, "loss": 5.624374866485596, "lr": 0.0002, "elapsed_sec": 10030.716410398483, "step_time_sec": 8.239612839999609, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1209, "loss": 5.525883197784424, "lr": 0.0002, "elapsed_sec": 10038.955513477325, "step_time_sec": 8.238928514067084, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1210, "loss": 5.449577808380127, "lr": 0.0002, "elapsed_sec": 10047.193769693375, "step_time_sec": 8.238084470969625, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1211, "loss": 5.504542827606201, "lr": 0.0002, "elapsed_sec": 10055.432257413864, "step_time_sec": 8.238324932055548, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1212, "loss": 5.643374919891357, "lr": 0.0002, "elapsed_sec": 10063.670455932617, "step_time_sec": 8.238087489968166, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1213, "loss": 5.458497047424316, "lr": 0.0002, "elapsed_sec": 10071.91118812561, "step_time_sec": 8.240505194058642, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1214, "loss": 5.5609283447265625, "lr": 0.0002, "elapsed_sec": 10080.149883031845, "step_time_sec": 8.238577604060993, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1215, "loss": 5.733659267425537, "lr": 0.0002, "elapsed_sec": 10088.387378454208, "step_time_sec": 8.237335646990687, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1216, "loss": 5.583243370056152, "lr": 0.0002, "elapsed_sec": 10096.627247571945, "step_time_sec": 8.239716197014786, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1217, "loss": 5.548736095428467, "lr": 0.0002, "elapsed_sec": 10104.864953517914, "step_time_sec": 8.237538674031384, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1218, "loss": 5.5911736488342285, "lr": 0.0002, "elapsed_sec": 10113.104197025299, "step_time_sec": 8.239059604005888, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1219, "loss": 5.541533470153809, "lr": 0.0002, "elapsed_sec": 10121.342638254166, "step_time_sec": 8.238316418952309, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1220, "loss": 5.61403751373291, "lr": 0.0002, "elapsed_sec": 10129.580894708633, "step_time_sec": 8.238073713029735, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1221, "loss": 5.51821756362915, "lr": 0.0002, "elapsed_sec": 10137.82086467743, "step_time_sec": 8.239772420958616, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1222, "loss": 5.568461894989014, "lr": 0.0002, "elapsed_sec": 10146.059748411179, "step_time_sec": 8.23880543501582, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1223, "loss": 5.513922214508057, "lr": 0.0002, "elapsed_sec": 10154.298800945282, "step_time_sec": 8.238844734034501, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1224, "loss": 5.558408737182617, "lr": 0.0002, "elapsed_sec": 10162.538010597229, "step_time_sec": 8.239032641984522, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1225, "loss": 5.447139739990234, "lr": 0.0002, "elapsed_sec": 10170.77533364296, "step_time_sec": 8.237184178084135, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1226, "loss": 5.617728233337402, "lr": 0.0002, "elapsed_sec": 10179.014496088028, "step_time_sec": 8.238977859029546, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1227, "loss": 5.585525989532471, "lr": 0.0002, "elapsed_sec": 10187.252007722855, "step_time_sec": 8.237359703984112, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1228, "loss": 5.527303218841553, "lr": 0.0002, "elapsed_sec": 10195.491225481033, "step_time_sec": 8.239077845937572, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1229, "loss": 5.524670124053955, "lr": 0.0002, "elapsed_sec": 10203.729283094406, "step_time_sec": 8.237879010965116, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1230, "loss": 5.527406692504883, "lr": 0.0002, "elapsed_sec": 10211.966289520264, "step_time_sec": 8.236828927998431, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1231, "loss": 5.54246187210083, "lr": 0.0002, "elapsed_sec": 10220.20645070076, "step_time_sec": 8.239996972028166, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1232, "loss": 5.55739688873291, "lr": 0.0002, "elapsed_sec": 10228.445111989975, "step_time_sec": 8.238537825993262, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1233, "loss": 5.550010681152344, "lr": 0.0002, "elapsed_sec": 10236.68547487259, "step_time_sec": 8.240165087976493, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1234, "loss": 5.533030986785889, "lr": 0.0002, "elapsed_sec": 10244.923355579376, "step_time_sec": 8.237715087016113, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1235, "loss": 5.592411041259766, "lr": 0.0002, "elapsed_sec": 10253.160654306412, "step_time_sec": 8.237183014047332, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1236, "loss": 5.568636417388916, "lr": 0.0002, "elapsed_sec": 10261.399726629257, "step_time_sec": 8.23892600799445, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1237, "loss": 5.587735176086426, "lr": 0.0002, "elapsed_sec": 10269.63728094101, "step_time_sec": 8.237340334919281, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1238, "loss": 5.465068817138672, "lr": 0.0002, "elapsed_sec": 10277.877655267715, "step_time_sec": 8.240202057990246, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1239, "loss": 5.493142127990723, "lr": 0.0002, "elapsed_sec": 10286.11576128006, "step_time_sec": 8.238001347053796, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1240, "loss": 5.560571193695068, "lr": 0.0002, "elapsed_sec": 10294.354341745377, "step_time_sec": 8.238391335005872, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1241, "loss": 5.494041919708252, "lr": 0.0002, "elapsed_sec": 10302.592682600021, "step_time_sec": 8.238131974008866, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1242, "loss": 5.562025547027588, "lr": 0.0002, "elapsed_sec": 10310.831002950668, "step_time_sec": 8.238192689954303, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1243, "loss": 5.424881458282471, "lr": 0.0002, "elapsed_sec": 10319.070250749588, "step_time_sec": 8.239085473003797, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1244, "loss": 5.559995651245117, "lr": 0.0002, "elapsed_sec": 10327.307626247406, "step_time_sec": 8.237190876970999, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1245, "loss": 5.508750915527344, "lr": 0.0002, "elapsed_sec": 10335.545326948166, "step_time_sec": 8.237575740902685, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1246, "loss": 5.585342884063721, "lr": 0.0002, "elapsed_sec": 10343.844603538513, "step_time_sec": 8.254509846912697, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1247, "loss": 5.4791154861450195, "lr": 0.0002, "elapsed_sec": 10352.083258628845, "step_time_sec": 8.238427555072121, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1248, "loss": 5.527259826660156, "lr": 0.0002, "elapsed_sec": 10360.321732759476, "step_time_sec": 8.238321734010242, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1249, "loss": 5.49730110168457, "lr": 0.0002, "elapsed_sec": 10368.560568332672, "step_time_sec": 8.238724141963758, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1250, "loss": 5.619856357574463, "lr": 0.0002, "elapsed_sec": 10376.797721624374, "step_time_sec": 8.236990347038954, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1251, "loss": 5.589800834655762, "lr": 0.0002, "elapsed_sec": 10385.036635637283, "step_time_sec": 8.238692663959228, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1252, "loss": 5.537324905395508, "lr": 0.0002, "elapsed_sec": 10393.275536775589, "step_time_sec": 8.238761086016893, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1253, "loss": 5.5793046951293945, "lr": 0.0002, "elapsed_sec": 10401.513476371765, "step_time_sec": 8.237797793932259, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1254, "loss": 5.569583892822266, "lr": 0.0002, "elapsed_sec": 10409.753329277039, "step_time_sec": 8.239642306929454, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1255, "loss": 5.569737434387207, "lr": 0.0002, "elapsed_sec": 10417.99124288559, "step_time_sec": 8.237737271003425, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1256, "loss": 5.490070819854736, "lr": 0.0002, "elapsed_sec": 10426.229978084564, "step_time_sec": 8.23862941202242, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1257, "loss": 5.695608615875244, "lr": 0.0002, "elapsed_sec": 10434.47021317482, "step_time_sec": 8.24006629793439, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1258, "loss": 5.510426998138428, "lr": 0.0002, "elapsed_sec": 10442.709002017975, "step_time_sec": 8.23860278900247, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1259, "loss": 5.555447578430176, "lr": 0.0002, "elapsed_sec": 10450.94906282425, "step_time_sec": 8.239898275001906, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1260, "loss": 5.5805768966674805, "lr": 0.0002, "elapsed_sec": 10459.188297510147, "step_time_sec": 8.23901721299626, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1261, "loss": 5.506722450256348, "lr": 0.0002, "elapsed_sec": 10467.427495718002, "step_time_sec": 8.239063935005106, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1262, "loss": 5.4783830642700195, "lr": 0.0002, "elapsed_sec": 10475.66820716858, "step_time_sec": 8.240505084046163, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1263, "loss": 5.544122219085693, "lr": 0.0002, "elapsed_sec": 10483.906551837921, "step_time_sec": 8.238213259959593, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1264, "loss": 5.469147205352783, "lr": 0.0002, "elapsed_sec": 10492.14646077156, "step_time_sec": 8.23969213408418, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1265, "loss": 5.5416436195373535, "lr": 0.0002, "elapsed_sec": 10500.383992910385, "step_time_sec": 8.237381384009495, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1266, "loss": 5.4174041748046875, "lr": 0.0002, "elapsed_sec": 10508.622056484222, "step_time_sec": 8.237928456044756, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1267, "loss": 5.484126091003418, "lr": 0.0002, "elapsed_sec": 10516.859956264496, "step_time_sec": 8.237748745945282, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1268, "loss": 5.524650573730469, "lr": 0.0002, "elapsed_sec": 10525.097375154495, "step_time_sec": 8.237236544955522, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1269, "loss": 5.676763534545898, "lr": 0.0002, "elapsed_sec": 10533.337229728699, "step_time_sec": 8.239705728017725, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1270, "loss": 5.4195876121521, "lr": 0.0002, "elapsed_sec": 10541.576475381851, "step_time_sec": 8.239082424901426, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1271, "loss": 5.528754234313965, "lr": 0.0002, "elapsed_sec": 10549.815928936005, "step_time_sec": 8.239234621985815, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1272, "loss": 5.49541711807251, "lr": 0.0002, "elapsed_sec": 10558.055676460266, "step_time_sec": 8.239600055036135, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1273, "loss": 5.437655925750732, "lr": 0.0002, "elapsed_sec": 10566.292736530304, "step_time_sec": 8.23695263301488, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1274, "loss": 5.484482288360596, "lr": 0.0002, "elapsed_sec": 10574.531753063202, "step_time_sec": 8.238832194008864, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1275, "loss": 5.633157730102539, "lr": 0.0002, "elapsed_sec": 10582.769815444946, "step_time_sec": 8.237891362980008, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1276, "loss": 5.428874492645264, "lr": 0.0002, "elapsed_sec": 10591.008511781693, "step_time_sec": 8.23855345195625, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1277, "loss": 5.5708417892456055, "lr": 0.0002, "elapsed_sec": 10599.248799800873, "step_time_sec": 8.240082797012292, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1278, "loss": 5.428497314453125, "lr": 0.0002, "elapsed_sec": 10607.487931013107, "step_time_sec": 8.238986689946614, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1279, "loss": 5.468911170959473, "lr": 0.0002, "elapsed_sec": 10615.727841377258, "step_time_sec": 8.239762317971326, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1280, "loss": 5.484607219696045, "lr": 0.0002, "elapsed_sec": 10623.966395378113, "step_time_sec": 8.238401362090372, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1281, "loss": 5.475583076477051, "lr": 0.0002, "elapsed_sec": 10632.205168247223, "step_time_sec": 8.238627155078575, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1282, "loss": 5.435963153839111, "lr": 0.0002, "elapsed_sec": 10640.446283340454, "step_time_sec": 8.240959099959582, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1283, "loss": 5.533639907836914, "lr": 0.0002, "elapsed_sec": 10648.683224916458, "step_time_sec": 8.236752265947871, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1284, "loss": 5.457824230194092, "lr": 0.0002, "elapsed_sec": 10656.922546863556, "step_time_sec": 8.239167476072907, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1285, "loss": 5.610834121704102, "lr": 0.0002, "elapsed_sec": 10665.16039943695, "step_time_sec": 8.237648654030636, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1286, "loss": 5.440496921539307, "lr": 0.0002, "elapsed_sec": 10673.397726297379, "step_time_sec": 8.237212387961335, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1287, "loss": 5.6252336502075195, "lr": 0.0002, "elapsed_sec": 10681.637550354004, "step_time_sec": 8.239684086991474, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1288, "loss": 5.517919540405273, "lr": 0.0002, "elapsed_sec": 10689.874809741974, "step_time_sec": 8.237074351985939, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1289, "loss": 5.565744400024414, "lr": 0.0002, "elapsed_sec": 10698.114179611206, "step_time_sec": 8.239205630961806, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1290, "loss": 5.464184761047363, "lr": 0.0002, "elapsed_sec": 10706.354062318802, "step_time_sec": 8.239772193017416, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1291, "loss": 5.438892841339111, "lr": 0.0002, "elapsed_sec": 10714.59319806099, "step_time_sec": 8.238916054018773, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1292, "loss": 5.4836602210998535, "lr": 0.0002, "elapsed_sec": 10722.833622694016, "step_time_sec": 8.24026088300161, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1293, "loss": 5.536186218261719, "lr": 0.0002, "elapsed_sec": 10731.071986675262, "step_time_sec": 8.238200508058071, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1294, "loss": 5.405031681060791, "lr": 0.0002, "elapsed_sec": 10739.311183691025, "step_time_sec": 8.239018248976208, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1295, "loss": 5.431406021118164, "lr": 0.0002, "elapsed_sec": 10747.550919055939, "step_time_sec": 8.239570814999752, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1296, "loss": 5.425921440124512, "lr": 0.0002, "elapsed_sec": 10755.788084030151, "step_time_sec": 8.23704134603031, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1297, "loss": 5.4990153312683105, "lr": 0.0002, "elapsed_sec": 10764.027345657349, "step_time_sec": 8.239128691027872, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1298, "loss": 5.574397563934326, "lr": 0.0002, "elapsed_sec": 10772.26547408104, "step_time_sec": 8.237960514961742, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1299, "loss": 5.480361461639404, "lr": 0.0002, "elapsed_sec": 10780.504342079163, "step_time_sec": 8.238671183004044, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1300, "loss": 5.466460704803467, "lr": 0.0002, "elapsed_sec": 10788.742087602615, "step_time_sec": 8.237594557111152, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1301, "loss": 5.496545314788818, "lr": 0.0002, "elapsed_sec": 10796.979551076889, "step_time_sec": 8.237315524951555, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1302, "loss": 5.480012893676758, "lr": 0.0002, "elapsed_sec": 10805.21814584732, "step_time_sec": 8.23836750490591, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1303, "loss": 5.596334934234619, "lr": 0.0002, "elapsed_sec": 10813.455733776093, "step_time_sec": 8.237453506095335, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1304, "loss": 5.4864935874938965, "lr": 0.0002, "elapsed_sec": 10821.694825172424, "step_time_sec": 8.238945791032165, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1305, "loss": 5.406485080718994, "lr": 0.0002, "elapsed_sec": 10829.933277368546, "step_time_sec": 8.238253085059114, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1306, "loss": 5.499701023101807, "lr": 0.0002, "elapsed_sec": 10838.171364307404, "step_time_sec": 8.237916106940247, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1307, "loss": 5.508763790130615, "lr": 0.0002, "elapsed_sec": 10846.410970687866, "step_time_sec": 8.239480520016514, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1308, "loss": 5.496114253997803, "lr": 0.0002, "elapsed_sec": 10854.64966082573, "step_time_sec": 8.238487923983485, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1309, "loss": 5.403907299041748, "lr": 0.0002, "elapsed_sec": 10862.88904619217, "step_time_sec": 8.239242824958637, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1310, "loss": 5.466287612915039, "lr": 0.0002, "elapsed_sec": 10871.12758731842, "step_time_sec": 8.238361662020907, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1311, "loss": 5.410333633422852, "lr": 0.0002, "elapsed_sec": 10879.366548538208, "step_time_sec": 8.238795452052727, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1312, "loss": 5.428287982940674, "lr": 0.0002, "elapsed_sec": 10887.607232809067, "step_time_sec": 8.240534350043163, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1313, "loss": 5.38070011138916, "lr": 0.0002, "elapsed_sec": 10895.84444284439, "step_time_sec": 8.237047377042472, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1314, "loss": 5.520674705505371, "lr": 0.0002, "elapsed_sec": 10904.083591222763, "step_time_sec": 8.238987094024196, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1315, "loss": 5.597845077514648, "lr": 0.0002, "elapsed_sec": 10912.321400880814, "step_time_sec": 8.237673557945527, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1316, "loss": 5.464401721954346, "lr": 0.0002, "elapsed_sec": 10920.55897474289, "step_time_sec": 8.237361432984471, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1317, "loss": 5.612661838531494, "lr": 0.0002, "elapsed_sec": 10928.799290180206, "step_time_sec": 8.240133367944509, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1318, "loss": 5.470823764801025, "lr": 0.0002, "elapsed_sec": 10937.038222551346, "step_time_sec": 8.238803418935277, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1319, "loss": 5.385998249053955, "lr": 0.0002, "elapsed_sec": 10945.277640342712, "step_time_sec": 8.239248021040112, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1320, "loss": 5.391565799713135, "lr": 0.0002, "elapsed_sec": 10953.515797376633, "step_time_sec": 8.23797827004455, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1321, "loss": 5.49131965637207, "lr": 0.0002, "elapsed_sec": 10961.753129720688, "step_time_sec": 8.237202035961673, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1322, "loss": 5.485743045806885, "lr": 0.0002, "elapsed_sec": 10969.991447687149, "step_time_sec": 8.238150192075409, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1323, "loss": 5.449018955230713, "lr": 0.0002, "elapsed_sec": 10978.229239463806, "step_time_sec": 8.237600844935514, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1324, "loss": 5.371077537536621, "lr": 0.0002, "elapsed_sec": 10986.465688943863, "step_time_sec": 8.236338670016266, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1325, "loss": 5.504054546356201, "lr": 0.0002, "elapsed_sec": 10994.70385503769, "step_time_sec": 8.237935171928257, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1326, "loss": 5.5066752433776855, "lr": 0.0002, "elapsed_sec": 11002.942980766296, "step_time_sec": 8.239021513960324, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1327, "loss": 5.504491329193115, "lr": 0.0002, "elapsed_sec": 11011.182639837265, "step_time_sec": 8.239491857006215, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1328, "loss": 5.417233943939209, "lr": 0.0002, "elapsed_sec": 11019.420685052872, "step_time_sec": 8.237836476997472, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1329, "loss": 5.468966960906982, "lr": 0.0002, "elapsed_sec": 11027.661313295364, "step_time_sec": 8.240455633960664, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1330, "loss": 5.4856743812561035, "lr": 0.0002, "elapsed_sec": 11035.899408102036, "step_time_sec": 8.237938320962712, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1331, "loss": 5.436383247375488, "lr": 0.0002, "elapsed_sec": 11044.13820552826, "step_time_sec": 8.238640949013643, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1332, "loss": 5.478428840637207, "lr": 0.0002, "elapsed_sec": 11052.378201007843, "step_time_sec": 8.239822670002468, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1333, "loss": 5.529623985290527, "lr": 0.0002, "elapsed_sec": 11060.616036653519, "step_time_sec": 8.237709080101922, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1334, "loss": 5.497187614440918, "lr": 0.0002, "elapsed_sec": 11068.854995250702, "step_time_sec": 8.238744243979454, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1335, "loss": 5.4808735847473145, "lr": 0.0002, "elapsed_sec": 11077.093857049942, "step_time_sec": 8.238745189039037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1336, "loss": 5.511308670043945, "lr": 0.0002, "elapsed_sec": 11085.331374406815, "step_time_sec": 8.237297221901827, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1337, "loss": 5.496774196624756, "lr": 0.0002, "elapsed_sec": 11093.57016658783, "step_time_sec": 8.238635685993358, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1338, "loss": 5.450881481170654, "lr": 0.0002, "elapsed_sec": 11101.808250904083, "step_time_sec": 8.23794632696081, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1339, "loss": 5.31377649307251, "lr": 0.0002, "elapsed_sec": 11110.047871112823, "step_time_sec": 8.239430358982645, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1340, "loss": 5.363471508026123, "lr": 0.0002, "elapsed_sec": 11118.286857843399, "step_time_sec": 8.238839005003683, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1341, "loss": 5.507796764373779, "lr": 0.0002, "elapsed_sec": 11126.526154518127, "step_time_sec": 8.239158286014572, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1342, "loss": 5.358801364898682, "lr": 0.0002, "elapsed_sec": 11134.765453100204, "step_time_sec": 8.239073653006926, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1343, "loss": 5.463801860809326, "lr": 0.0002, "elapsed_sec": 11143.003489017487, "step_time_sec": 8.237880158005282, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1344, "loss": 5.450141906738281, "lr": 0.0002, "elapsed_sec": 11151.242991685867, "step_time_sec": 8.239346283022314, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1345, "loss": 5.464013576507568, "lr": 0.0002, "elapsed_sec": 11159.480628490448, "step_time_sec": 8.23751288303174, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1346, "loss": 5.428010940551758, "lr": 0.0002, "elapsed_sec": 11167.718876838684, "step_time_sec": 8.238050288055092, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1347, "loss": 5.43237829208374, "lr": 0.0002, "elapsed_sec": 11175.958169698715, "step_time_sec": 8.23917156807147, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1348, "loss": 5.429555892944336, "lr": 0.0002, "elapsed_sec": 11184.194720029831, "step_time_sec": 8.236360584036447, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1349, "loss": 5.420926570892334, "lr": 0.0002, "elapsed_sec": 11192.432886600494, "step_time_sec": 8.237978306016885, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1350, "loss": 5.435458660125732, "lr": 0.0002, "elapsed_sec": 11200.670253276825, "step_time_sec": 8.237209431012161, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1351, "loss": 5.4863152503967285, "lr": 0.0002, "elapsed_sec": 11208.908893823624, "step_time_sec": 8.23850824998226, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1352, "loss": 5.5270256996154785, "lr": 0.0002, "elapsed_sec": 11217.148950576782, "step_time_sec": 8.23989565006923, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1353, "loss": 5.455540657043457, "lr": 0.0002, "elapsed_sec": 11225.387613296509, "step_time_sec": 8.238493005977944, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1354, "loss": 5.407631874084473, "lr": 0.0002, "elapsed_sec": 11233.62811589241, "step_time_sec": 8.240372427972034, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1355, "loss": 5.451292037963867, "lr": 0.0002, "elapsed_sec": 11241.86633849144, "step_time_sec": 8.238008672953583, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1356, "loss": 5.299489974975586, "lr": 0.0002, "elapsed_sec": 11250.105055809021, "step_time_sec": 8.238559914985672, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1357, "loss": 5.454590320587158, "lr": 0.0002, "elapsed_sec": 11258.343161582947, "step_time_sec": 8.237939311075024, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1358, "loss": 5.4432148933410645, "lr": 0.0002, "elapsed_sec": 11266.57924580574, "step_time_sec": 8.2359570650151, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1359, "loss": 5.457027435302734, "lr": 0.0002, "elapsed_sec": 11274.818428516388, "step_time_sec": 8.238990588928573, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1360, "loss": 5.348787307739258, "lr": 0.0002, "elapsed_sec": 11283.055754899979, "step_time_sec": 8.237151985988021, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1361, "loss": 5.30654764175415, "lr": 0.0002, "elapsed_sec": 11291.29480266571, "step_time_sec": 8.23890252201818, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1362, "loss": 5.379645824432373, "lr": 0.0002, "elapsed_sec": 11299.535093545914, "step_time_sec": 8.240154560073279, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1363, "loss": 5.250107288360596, "lr": 0.0002, "elapsed_sec": 11307.771894693375, "step_time_sec": 8.236598232993856, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1364, "loss": 5.344667911529541, "lr": 0.0002, "elapsed_sec": 11316.009538888931, "step_time_sec": 8.237569054006599, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1365, "loss": 5.519890308380127, "lr": 0.0002, "elapsed_sec": 11324.24829363823, "step_time_sec": 8.238549161935225, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1366, "loss": 5.4620585441589355, "lr": 0.0002, "elapsed_sec": 11332.487004995346, "step_time_sec": 8.238527481094934, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1367, "loss": 5.24788761138916, "lr": 0.0002, "elapsed_sec": 11340.725041627884, "step_time_sec": 8.237879727967083, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1368, "loss": 5.410823822021484, "lr": 0.0002, "elapsed_sec": 11348.96162724495, "step_time_sec": 8.236391952028498, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1369, "loss": 5.423832893371582, "lr": 0.0002, "elapsed_sec": 11357.200317621231, "step_time_sec": 8.238553371978924, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1370, "loss": 5.487708568572998, "lr": 0.0002, "elapsed_sec": 11365.437041521072, "step_time_sec": 8.236541492049582, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1371, "loss": 5.366307258605957, "lr": 0.0002, "elapsed_sec": 11373.676221370697, "step_time_sec": 8.239015595056117, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1372, "loss": 5.3396315574646, "lr": 0.0002, "elapsed_sec": 11381.913630485535, "step_time_sec": 8.237236837972887, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1373, "loss": 5.43282413482666, "lr": 0.0002, "elapsed_sec": 11390.153567314148, "step_time_sec": 8.239782514981925, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1374, "loss": 5.492530822753906, "lr": 0.0002, "elapsed_sec": 11398.393765687943, "step_time_sec": 8.240028442000039, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1375, "loss": 5.399801731109619, "lr": 0.0002, "elapsed_sec": 11406.632104873657, "step_time_sec": 8.23819063603878, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1376, "loss": 5.440124988555908, "lr": 0.0002, "elapsed_sec": 11414.871645212173, "step_time_sec": 8.239442804944701, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1377, "loss": 5.409465312957764, "lr": 0.0002, "elapsed_sec": 11423.111773490906, "step_time_sec": 8.239889158052392, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1378, "loss": 5.511183738708496, "lr": 0.0002, "elapsed_sec": 11431.350628376007, "step_time_sec": 8.23867908702232, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1379, "loss": 5.409555435180664, "lr": 0.0002, "elapsed_sec": 11439.590248823166, "step_time_sec": 8.23949253198225, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1380, "loss": 5.460183143615723, "lr": 0.0002, "elapsed_sec": 11447.828958272934, "step_time_sec": 8.238524951972067, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1381, "loss": 5.508200645446777, "lr": 0.0002, "elapsed_sec": 11456.068503856659, "step_time_sec": 8.239437430980615, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1382, "loss": 5.429851055145264, "lr": 0.0002, "elapsed_sec": 11464.306794404984, "step_time_sec": 8.238099838024937, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1383, "loss": 5.336330413818359, "lr": 0.0002, "elapsed_sec": 11472.543818712234, "step_time_sec": 8.23683469707612, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1384, "loss": 5.437856197357178, "lr": 0.0002, "elapsed_sec": 11480.782893896103, "step_time_sec": 8.23890661704354, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1385, "loss": 5.409663677215576, "lr": 0.0002, "elapsed_sec": 11489.020700931549, "step_time_sec": 8.237614452955313, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1386, "loss": 5.270754337310791, "lr": 0.0002, "elapsed_sec": 11497.260169267654, "step_time_sec": 8.23933323903475, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1387, "loss": 5.361631393432617, "lr": 0.0002, "elapsed_sec": 11505.49949145317, "step_time_sec": 8.239139578072354, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1388, "loss": 5.249669075012207, "lr": 0.0002, "elapsed_sec": 11513.737778186798, "step_time_sec": 8.238104815012775, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1389, "loss": 5.4424519538879395, "lr": 0.0002, "elapsed_sec": 11521.977991819382, "step_time_sec": 8.240026398096234, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1390, "loss": 5.406835556030273, "lr": 0.0002, "elapsed_sec": 11530.215967655182, "step_time_sec": 8.237788370926864, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1391, "loss": 5.592264175415039, "lr": 0.0002, "elapsed_sec": 11538.454863071442, "step_time_sec": 8.238798484904692, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1392, "loss": 5.387890338897705, "lr": 0.0002, "elapsed_sec": 11546.6939432621, "step_time_sec": 8.238865712075494, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1393, "loss": 5.377303123474121, "lr": 0.0002, "elapsed_sec": 11554.931800842285, "step_time_sec": 8.237667224020697, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1394, "loss": 5.402002334594727, "lr": 0.0002, "elapsed_sec": 11563.169424533844, "step_time_sec": 8.23743831005413, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1395, "loss": 5.362390041351318, "lr": 0.0002, "elapsed_sec": 11571.406019210815, "step_time_sec": 8.236457226099446, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1396, "loss": 5.402854919433594, "lr": 0.0002, "elapsed_sec": 11579.644859552383, "step_time_sec": 8.238703011069447, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1397, "loss": 5.425107479095459, "lr": 0.0002, "elapsed_sec": 11587.882837057114, "step_time_sec": 8.23778897500597, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1398, "loss": 5.302182674407959, "lr": 0.0002, "elapsed_sec": 11596.121878385544, "step_time_sec": 8.238915088935755, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1399, "loss": 5.198119640350342, "lr": 0.0002, "elapsed_sec": 11604.36044049263, "step_time_sec": 8.238350645988248, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1400, "loss": 5.385730266571045, "lr": 0.0002, "elapsed_sec": 11612.597628831863, "step_time_sec": 8.236983945942484, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1401, "loss": 5.329002380371094, "lr": 0.0002, "elapsed_sec": 11620.836401939392, "step_time_sec": 8.238604463054799, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1402, "loss": 5.325942039489746, "lr": 0.0002, "elapsed_sec": 11629.073767662048, "step_time_sec": 8.23722948692739, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1403, "loss": 5.373832702636719, "lr": 0.0002, "elapsed_sec": 11637.312569856644, "step_time_sec": 8.238632395048626, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1404, "loss": 5.478309154510498, "lr": 0.0002, "elapsed_sec": 11645.551816225052, "step_time_sec": 8.239063474000432, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1405, "loss": 5.451113700866699, "lr": 0.0002, "elapsed_sec": 11653.789666891098, "step_time_sec": 8.237711791996844, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1406, "loss": 5.3075270652771, "lr": 0.0002, "elapsed_sec": 11662.02992272377, "step_time_sec": 8.240036744973622, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1407, "loss": 5.2425384521484375, "lr": 0.0002, "elapsed_sec": 11670.267935037613, "step_time_sec": 8.237859335029498, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1408, "loss": 5.462650299072266, "lr": 0.0002, "elapsed_sec": 11678.505934715271, "step_time_sec": 8.237841664929874, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1409, "loss": 5.341048717498779, "lr": 0.0002, "elapsed_sec": 11686.744127750397, "step_time_sec": 8.23803539806977, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1410, "loss": 5.368292331695557, "lr": 0.0002, "elapsed_sec": 11694.982164382935, "step_time_sec": 8.23793804005254, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1411, "loss": 5.322094440460205, "lr": 0.0002, "elapsed_sec": 11703.22164940834, "step_time_sec": 8.239231473999098, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1412, "loss": 5.350639343261719, "lr": 0.0002, "elapsed_sec": 11711.461319684982, "step_time_sec": 8.239494631998241, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1413, "loss": 5.3906707763671875, "lr": 0.0002, "elapsed_sec": 11719.700526237488, "step_time_sec": 8.239105306915008, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1414, "loss": 5.360734939575195, "lr": 0.0002, "elapsed_sec": 11727.938517570496, "step_time_sec": 8.23779331392143, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1415, "loss": 5.474775314331055, "lr": 0.0002, "elapsed_sec": 11736.174851179123, "step_time_sec": 8.236190570984036, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1416, "loss": 5.388302326202393, "lr": 0.0002, "elapsed_sec": 11744.413175821304, "step_time_sec": 8.238098394009285, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1417, "loss": 5.39961576461792, "lr": 0.0002, "elapsed_sec": 11752.650885343552, "step_time_sec": 8.23758081998676, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1418, "loss": 5.364084243774414, "lr": 0.0002, "elapsed_sec": 11760.890275239944, "step_time_sec": 8.239165027975105, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1419, "loss": 5.425032615661621, "lr": 0.0002, "elapsed_sec": 11769.127558708191, "step_time_sec": 8.23711819795426, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1420, "loss": 5.461302757263184, "lr": 0.0002, "elapsed_sec": 11777.365478277206, "step_time_sec": 8.237788482103497, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1421, "loss": 5.42449426651001, "lr": 0.0002, "elapsed_sec": 11785.604853630066, "step_time_sec": 8.239180478034541, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1422, "loss": 5.36832332611084, "lr": 0.0002, "elapsed_sec": 11793.842698574066, "step_time_sec": 8.237690087989904, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1423, "loss": 5.2810821533203125, "lr": 0.0002, "elapsed_sec": 11802.082614183426, "step_time_sec": 8.239747648010962, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1424, "loss": 5.255654335021973, "lr": 0.0002, "elapsed_sec": 11810.322564601898, "step_time_sec": 8.239794773049653, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1425, "loss": 5.307953834533691, "lr": 0.0002, "elapsed_sec": 11818.561421871185, "step_time_sec": 8.23866732697934, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1426, "loss": 5.404363632202148, "lr": 0.0002, "elapsed_sec": 11826.800635099411, "step_time_sec": 8.239056297927164, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1427, "loss": 5.433319091796875, "lr": 0.0002, "elapsed_sec": 11835.039598464966, "step_time_sec": 8.238829080946743, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1428, "loss": 5.42782735824585, "lr": 0.0002, "elapsed_sec": 11843.27910065651, "step_time_sec": 8.239293382968754, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1429, "loss": 5.424367427825928, "lr": 0.0002, "elapsed_sec": 11851.518246889114, "step_time_sec": 8.238961850991473, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1430, "loss": 5.343879699707031, "lr": 0.0002, "elapsed_sec": 11859.755171775818, "step_time_sec": 8.236792165087536, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1431, "loss": 5.419560432434082, "lr": 0.0002, "elapsed_sec": 11867.993515253067, "step_time_sec": 8.238171368953772, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1432, "loss": 5.240066051483154, "lr": 0.0002, "elapsed_sec": 11876.231961727142, "step_time_sec": 8.238288140040822, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1433, "loss": 5.411776542663574, "lr": 0.0002, "elapsed_sec": 11884.470877408981, "step_time_sec": 8.238762069959193, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1434, "loss": 5.295833110809326, "lr": 0.0002, "elapsed_sec": 11892.709846496582, "step_time_sec": 8.23879036703147, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1435, "loss": 5.447446346282959, "lr": 0.0002, "elapsed_sec": 11900.948993682861, "step_time_sec": 8.238935394096188, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1436, "loss": 5.3618316650390625, "lr": 0.0002, "elapsed_sec": 11909.188205242157, "step_time_sec": 8.239054074976593, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1437, "loss": 5.426701545715332, "lr": 0.0002, "elapsed_sec": 11917.42646741867, "step_time_sec": 8.238141501089558, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1438, "loss": 5.336164474487305, "lr": 0.0002, "elapsed_sec": 11925.664937734604, "step_time_sec": 8.238315680995584, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1439, "loss": 5.305149078369141, "lr": 0.0002, "elapsed_sec": 11933.903960943222, "step_time_sec": 8.238808984984644, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1440, "loss": 5.397982597351074, "lr": 0.0002, "elapsed_sec": 11942.142566204071, "step_time_sec": 8.238459820044227, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1441, "loss": 5.326592445373535, "lr": 0.0002, "elapsed_sec": 11950.38253903389, "step_time_sec": 8.239775081980042, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1442, "loss": 5.481414794921875, "lr": 0.0002, "elapsed_sec": 11958.62130856514, "step_time_sec": 8.238653077976778, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1443, "loss": 5.318925857543945, "lr": 0.0002, "elapsed_sec": 11966.860127210617, "step_time_sec": 8.238618847099133, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1444, "loss": 5.282801628112793, "lr": 0.0002, "elapsed_sec": 11975.098995685577, "step_time_sec": 8.238718577078544, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1445, "loss": 5.449173450469971, "lr": 0.0002, "elapsed_sec": 11983.334770679474, "step_time_sec": 8.235589376999997, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1446, "loss": 5.459173202514648, "lr": 0.0002, "elapsed_sec": 11991.573935508728, "step_time_sec": 8.238974966108799, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1447, "loss": 5.2680253982543945, "lr": 0.0002, "elapsed_sec": 11999.812514781952, "step_time_sec": 8.238458595005795, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1448, "loss": 5.227980136871338, "lr": 0.0002, "elapsed_sec": 12008.051653862, "step_time_sec": 8.238985422067344, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1449, "loss": 5.365966320037842, "lr": 0.0002, "elapsed_sec": 12016.291489839554, "step_time_sec": 8.239678762969561, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1450, "loss": 5.282115936279297, "lr": 0.0002, "elapsed_sec": 12024.528626441956, "step_time_sec": 8.236945600016043, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1451, "loss": 5.283285140991211, "lr": 0.0002, "elapsed_sec": 12032.767950296402, "step_time_sec": 8.239145130035467, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1452, "loss": 5.559798717498779, "lr": 0.0002, "elapsed_sec": 12041.005231380463, "step_time_sec": 8.23708259104751, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1453, "loss": 5.312982082366943, "lr": 0.0002, "elapsed_sec": 12049.244740962982, "step_time_sec": 8.239354853983968, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1454, "loss": 5.352067947387695, "lr": 0.0002, "elapsed_sec": 12057.483411312103, "step_time_sec": 8.238483141059987, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1455, "loss": 5.351036071777344, "lr": 0.0002, "elapsed_sec": 12065.72066116333, "step_time_sec": 8.237149814958684, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1456, "loss": 5.326874732971191, "lr": 0.0002, "elapsed_sec": 12073.95890879631, "step_time_sec": 8.238055631052703, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1457, "loss": 5.300807952880859, "lr": 0.0002, "elapsed_sec": 12082.196080207825, "step_time_sec": 8.236989269033074, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1458, "loss": 5.49784517288208, "lr": 0.0002, "elapsed_sec": 12090.433705568314, "step_time_sec": 8.237483906093985, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1459, "loss": 5.119076728820801, "lr": 0.0002, "elapsed_sec": 12098.672594308853, "step_time_sec": 8.23869242100045, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1460, "loss": 5.342711925506592, "lr": 0.0002, "elapsed_sec": 12106.911284208298, "step_time_sec": 8.238557870965451, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1461, "loss": 5.331569671630859, "lr": 0.0002, "elapsed_sec": 12115.15016913414, "step_time_sec": 8.238683533039875, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1462, "loss": 5.488272190093994, "lr": 0.0002, "elapsed_sec": 12123.387996912003, "step_time_sec": 8.237640575971454, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1463, "loss": 5.381036758422852, "lr": 0.0002, "elapsed_sec": 12131.62844991684, "step_time_sec": 8.240282683051191, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1464, "loss": 5.468867778778076, "lr": 0.0002, "elapsed_sec": 12139.867409944534, "step_time_sec": 8.238818209036253, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1465, "loss": 5.398169994354248, "lr": 0.0002, "elapsed_sec": 12148.105781793594, "step_time_sec": 8.23817223298829, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1466, "loss": 5.3201446533203125, "lr": 0.0002, "elapsed_sec": 12156.344267368317, "step_time_sec": 8.23835065192543, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1467, "loss": 5.234644889831543, "lr": 0.0002, "elapsed_sec": 12164.582132339478, "step_time_sec": 8.237666003988124, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1468, "loss": 5.193217754364014, "lr": 0.0002, "elapsed_sec": 12172.821118593216, "step_time_sec": 8.238848753971979, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1469, "loss": 5.495113849639893, "lr": 0.0002, "elapsed_sec": 12181.059602022171, "step_time_sec": 8.238295204006135, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1470, "loss": 5.3159260749816895, "lr": 0.0002, "elapsed_sec": 12189.29818892479, "step_time_sec": 8.238428973942064, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1471, "loss": 5.436243057250977, "lr": 0.0002, "elapsed_sec": 12197.537415027618, "step_time_sec": 8.239054230973125, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1472, "loss": 5.3635735511779785, "lr": 0.0002, "elapsed_sec": 12205.77415561676, "step_time_sec": 8.236560808029026, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1473, "loss": 5.45658016204834, "lr": 0.0002, "elapsed_sec": 12214.012140274048, "step_time_sec": 8.237831110018305, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1474, "loss": 5.352171421051025, "lr": 0.0002, "elapsed_sec": 12222.250983476639, "step_time_sec": 8.23867406800855, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1475, "loss": 5.485755920410156, "lr": 0.0002, "elapsed_sec": 12230.490661859512, "step_time_sec": 8.239513195003383, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1476, "loss": 5.3073859214782715, "lr": 0.0002, "elapsed_sec": 12238.7298848629, "step_time_sec": 8.239041569991969, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1477, "loss": 5.299824237823486, "lr": 0.0002, "elapsed_sec": 12246.968228578568, "step_time_sec": 8.238176190992817, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1478, "loss": 5.255133628845215, "lr": 0.0002, "elapsed_sec": 12255.209262609482, "step_time_sec": 8.240871349000372, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1479, "loss": 5.268468379974365, "lr": 0.0002, "elapsed_sec": 12263.447562456131, "step_time_sec": 8.238135735038668, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1480, "loss": 5.425084590911865, "lr": 0.0002, "elapsed_sec": 12271.686266899109, "step_time_sec": 8.238524607033469, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1481, "loss": 5.258395195007324, "lr": 0.0002, "elapsed_sec": 12279.92613530159, "step_time_sec": 8.239706249907613, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1482, "loss": 5.344427585601807, "lr": 0.0002, "elapsed_sec": 12288.164629459381, "step_time_sec": 8.238349640974775, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1483, "loss": 5.238300323486328, "lr": 0.0002, "elapsed_sec": 12296.402829408646, "step_time_sec": 8.238060730043799, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1484, "loss": 5.366832256317139, "lr": 0.0002, "elapsed_sec": 12304.642248868942, "step_time_sec": 8.239212263026275, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1485, "loss": 5.322914123535156, "lr": 0.0002, "elapsed_sec": 12312.880418539047, "step_time_sec": 8.238047079066746, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1486, "loss": 5.2253546714782715, "lr": 0.0002, "elapsed_sec": 12321.120151042938, "step_time_sec": 8.239493530942127, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1487, "loss": 5.272468566894531, "lr": 0.0002, "elapsed_sec": 12329.359208583832, "step_time_sec": 8.238914199988358, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1488, "loss": 5.297149181365967, "lr": 0.0002, "elapsed_sec": 12337.59860420227, "step_time_sec": 8.239210939966142, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1489, "loss": 5.211629390716553, "lr": 0.0002, "elapsed_sec": 12345.837908029556, "step_time_sec": 8.239145476021804, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1490, "loss": 5.430429935455322, "lr": 0.0002, "elapsed_sec": 12354.075442790985, "step_time_sec": 8.237362180952914, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1491, "loss": 5.267597675323486, "lr": 0.0002, "elapsed_sec": 12362.313776493073, "step_time_sec": 8.238196435035206, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1492, "loss": 5.257351398468018, "lr": 0.0002, "elapsed_sec": 12370.55128955841, "step_time_sec": 8.237344815046526, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1493, "loss": 5.365518093109131, "lr": 0.0002, "elapsed_sec": 12378.79048705101, "step_time_sec": 8.239066933980212, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1494, "loss": 5.3378448486328125, "lr": 0.0002, "elapsed_sec": 12387.029817581177, "step_time_sec": 8.239171935012564, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1495, "loss": 5.312827110290527, "lr": 0.0002, "elapsed_sec": 12395.268729448318, "step_time_sec": 8.238708366989158, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1496, "loss": 5.110752582550049, "lr": 0.0002, "elapsed_sec": 12403.506913900375, "step_time_sec": 8.237994231050834, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1497, "loss": 5.272260665893555, "lr": 0.0002, "elapsed_sec": 12411.744955301285, "step_time_sec": 8.237859932007268, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1498, "loss": 5.1982645988464355, "lr": 0.0002, "elapsed_sec": 12419.982629299164, "step_time_sec": 8.237513855099678, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1499, "loss": 5.405348300933838, "lr": 0.0002, "elapsed_sec": 12428.222156763077, "step_time_sec": 8.23943792108912, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1500, "loss": 5.305271148681641, "lr": 0.0002, "elapsed_sec": 12436.460653305054, "step_time_sec": 24.428988436004147, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": true, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1501, "loss": 5.372705936431885, "lr": 0.0002, "elapsed_sec": 12460.8963701725, "step_time_sec": 8.244473133934662, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1502, "loss": 5.259506702423096, "lr": 0.0002, "elapsed_sec": 12469.128470420837, "step_time_sec": 8.231974474037997, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1503, "loss": 5.439561367034912, "lr": 0.0002, "elapsed_sec": 12477.36069059372, "step_time_sec": 8.232084317947738, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1504, "loss": 5.361197471618652, "lr": 0.0002, "elapsed_sec": 12485.596000432968, "step_time_sec": 8.235121871926822, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1505, "loss": 5.374300003051758, "lr": 0.0002, "elapsed_sec": 12493.828011274338, "step_time_sec": 8.231854586978443, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1506, "loss": 5.34525728225708, "lr": 0.0002, "elapsed_sec": 12502.05984210968, "step_time_sec": 8.231625152984634, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1507, "loss": 5.291436195373535, "lr": 0.0002, "elapsed_sec": 12510.29170012474, "step_time_sec": 8.231709243962541, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1508, "loss": 5.24602746963501, "lr": 0.0002, "elapsed_sec": 12518.529034376144, "step_time_sec": 8.23716374405194, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1509, "loss": 5.4388861656188965, "lr": 0.0002, "elapsed_sec": 12526.768498420715, "step_time_sec": 8.239295359933749, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1510, "loss": 5.2852349281311035, "lr": 0.0002, "elapsed_sec": 12535.007248401642, "step_time_sec": 8.23856462794356, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1511, "loss": 5.3160505294799805, "lr": 0.0002, "elapsed_sec": 12543.245392560959, "step_time_sec": 8.238020236021839, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1512, "loss": 5.415809154510498, "lr": 0.0002, "elapsed_sec": 12551.484572172165, "step_time_sec": 8.238985792035237, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1513, "loss": 5.176618576049805, "lr": 0.0002, "elapsed_sec": 12559.723803520203, "step_time_sec": 8.239115181029774, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1514, "loss": 5.396275520324707, "lr": 0.0002, "elapsed_sec": 12567.963255405426, "step_time_sec": 8.239290768047795, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1515, "loss": 5.186276435852051, "lr": 0.0002, "elapsed_sec": 12576.202083826065, "step_time_sec": 8.238680984941311, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1516, "loss": 5.337910175323486, "lr": 0.0002, "elapsed_sec": 12584.440206766129, "step_time_sec": 8.237976930919103, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1517, "loss": 5.128233432769775, "lr": 0.0002, "elapsed_sec": 12592.679629802704, "step_time_sec": 8.239197980961762, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1518, "loss": 5.391410827636719, "lr": 0.0002, "elapsed_sec": 12600.916306734085, "step_time_sec": 8.236496057012118, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1519, "loss": 5.200239181518555, "lr": 0.0002, "elapsed_sec": 12609.156369447708, "step_time_sec": 8.239892972982489, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1520, "loss": 5.2814621925354, "lr": 0.0002, "elapsed_sec": 12617.395281076431, "step_time_sec": 8.238809269038029, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1521, "loss": 5.302255153656006, "lr": 0.0002, "elapsed_sec": 12625.633658885956, "step_time_sec": 8.238196468097158, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1522, "loss": 5.255687236785889, "lr": 0.0002, "elapsed_sec": 12633.874204397202, "step_time_sec": 8.24034505605232, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1523, "loss": 5.137172698974609, "lr": 0.0002, "elapsed_sec": 12642.11166357994, "step_time_sec": 8.237303382018581, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1524, "loss": 5.3676629066467285, "lr": 0.0002, "elapsed_sec": 12650.350485086441, "step_time_sec": 8.238689344958402, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1525, "loss": 5.254824161529541, "lr": 0.0002, "elapsed_sec": 12658.589973211288, "step_time_sec": 8.239306339994073, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1526, "loss": 5.116167068481445, "lr": 0.0002, "elapsed_sec": 12666.827317476273, "step_time_sec": 8.237168028019369, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1527, "loss": 5.309978485107422, "lr": 0.0002, "elapsed_sec": 12675.067072153091, "step_time_sec": 8.23957947792951, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1528, "loss": 5.279431343078613, "lr": 0.0002, "elapsed_sec": 12683.305738925934, "step_time_sec": 8.238562579965219, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1529, "loss": 5.281795024871826, "lr": 0.0002, "elapsed_sec": 12691.544843673706, "step_time_sec": 8.238874537986703, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1530, "loss": 5.125613689422607, "lr": 0.0002, "elapsed_sec": 12699.78387761116, "step_time_sec": 8.238875207025558, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1531, "loss": 5.310143947601318, "lr": 0.0002, "elapsed_sec": 12708.020564317703, "step_time_sec": 8.236522538005374, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1532, "loss": 5.233020305633545, "lr": 0.0002, "elapsed_sec": 12716.259459972382, "step_time_sec": 8.238800232997164, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1533, "loss": 5.349806785583496, "lr": 0.0002, "elapsed_sec": 12724.49711894989, "step_time_sec": 8.237453323090449, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1534, "loss": 5.333541393280029, "lr": 0.0002, "elapsed_sec": 12732.73652625084, "step_time_sec": 8.239222055999562, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1535, "loss": 5.141005516052246, "lr": 0.0002, "elapsed_sec": 12740.976519823074, "step_time_sec": 8.239818934933282, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1536, "loss": 5.4111175537109375, "lr": 0.0002, "elapsed_sec": 12749.21483874321, "step_time_sec": 8.238152384059504, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1537, "loss": 5.15864372253418, "lr": 0.0002, "elapsed_sec": 12757.453821182251, "step_time_sec": 8.238866120926104, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1538, "loss": 5.356489181518555, "lr": 0.0002, "elapsed_sec": 12765.693291187286, "step_time_sec": 8.239295345032588, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1539, "loss": 5.155674934387207, "lr": 0.0002, "elapsed_sec": 12773.93299460411, "step_time_sec": 8.239517949987203, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1540, "loss": 5.261252403259277, "lr": 0.0002, "elapsed_sec": 12782.172533273697, "step_time_sec": 8.239364687004127, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1541, "loss": 5.1207356452941895, "lr": 0.0002, "elapsed_sec": 12790.409580469131, "step_time_sec": 8.236913420958444, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1542, "loss": 5.258553504943848, "lr": 0.0002, "elapsed_sec": 12798.648681402206, "step_time_sec": 8.238997729960829, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1543, "loss": 5.182117462158203, "lr": 0.0002, "elapsed_sec": 12806.886984348297, "step_time_sec": 8.23807381594088, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1544, "loss": 5.361036777496338, "lr": 0.0002, "elapsed_sec": 12815.125638484955, "step_time_sec": 8.238492358010262, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1545, "loss": 5.332447528839111, "lr": 0.0002, "elapsed_sec": 12823.363917827606, "step_time_sec": 8.238146140007302, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1546, "loss": 5.2749176025390625, "lr": 0.0002, "elapsed_sec": 12831.602015018463, "step_time_sec": 8.237890254007652, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1547, "loss": 5.46523380279541, "lr": 0.0002, "elapsed_sec": 12839.83951640129, "step_time_sec": 8.237341998959891, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1548, "loss": 5.205567836761475, "lr": 0.0002, "elapsed_sec": 12848.076669931412, "step_time_sec": 8.23701905098278, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1549, "loss": 5.301726818084717, "lr": 0.0002, "elapsed_sec": 12856.318008184433, "step_time_sec": 8.241195018985309, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1550, "loss": 5.313621997833252, "lr": 0.0002, "elapsed_sec": 12864.555981636047, "step_time_sec": 8.237771560903639, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1551, "loss": 5.222544193267822, "lr": 0.0002, "elapsed_sec": 12872.792603969574, "step_time_sec": 8.236435628961772, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1552, "loss": 5.3081159591674805, "lr": 0.0002, "elapsed_sec": 12881.030855894089, "step_time_sec": 8.238090793951415, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1553, "loss": 5.204155921936035, "lr": 0.0002, "elapsed_sec": 12889.269043207169, "step_time_sec": 8.238034239970148, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1554, "loss": 5.234918117523193, "lr": 0.0002, "elapsed_sec": 12897.508698225021, "step_time_sec": 8.239497354952618, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1555, "loss": 5.190131187438965, "lr": 0.0002, "elapsed_sec": 12905.748087882996, "step_time_sec": 8.239251508959569, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1556, "loss": 5.19423246383667, "lr": 0.0002, "elapsed_sec": 12913.986278772354, "step_time_sec": 8.238023465964943, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1557, "loss": 5.249810695648193, "lr": 0.0002, "elapsed_sec": 12922.225839853287, "step_time_sec": 8.239365299930796, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1558, "loss": 5.113672733306885, "lr": 0.0002, "elapsed_sec": 12930.463045597076, "step_time_sec": 8.237056307960302, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1559, "loss": 5.212614059448242, "lr": 0.0002, "elapsed_sec": 12938.703253030777, "step_time_sec": 8.240062500932254, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1560, "loss": 5.234808444976807, "lr": 0.0002, "elapsed_sec": 12946.941006660461, "step_time_sec": 8.237585181952454, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1561, "loss": 5.266904354095459, "lr": 0.0002, "elapsed_sec": 12955.17913198471, "step_time_sec": 8.237968637025915, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1562, "loss": 5.240671157836914, "lr": 0.0002, "elapsed_sec": 12963.417326927185, "step_time_sec": 8.238025591010228, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1563, "loss": 5.314881801605225, "lr": 0.0002, "elapsed_sec": 12971.655279397964, "step_time_sec": 8.237772721913643, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1564, "loss": 5.226607799530029, "lr": 0.0002, "elapsed_sec": 12979.894918680191, "step_time_sec": 8.239519777009264, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1565, "loss": 5.247251987457275, "lr": 0.0002, "elapsed_sec": 12988.132802963257, "step_time_sec": 8.237717523938045, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1566, "loss": 5.340978145599365, "lr": 0.0002, "elapsed_sec": 12996.37119936943, "step_time_sec": 8.238235555938445, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1567, "loss": 5.193894863128662, "lr": 0.0002, "elapsed_sec": 13004.608659267426, "step_time_sec": 8.237258165027015, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1568, "loss": 5.149604797363281, "lr": 0.0002, "elapsed_sec": 13012.845751047134, "step_time_sec": 8.236914129927754, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1569, "loss": 5.24086332321167, "lr": 0.0002, "elapsed_sec": 13021.084048986435, "step_time_sec": 8.23814879800193, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1570, "loss": 5.174211502075195, "lr": 0.0002, "elapsed_sec": 13029.322068691254, "step_time_sec": 8.237828579964116, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1571, "loss": 5.323789119720459, "lr": 0.0002, "elapsed_sec": 13037.560950040817, "step_time_sec": 8.238720039953478, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1572, "loss": 5.206663608551025, "lr": 0.0002, "elapsed_sec": 13045.80059170723, "step_time_sec": 8.239489260944538, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1573, "loss": 5.44080924987793, "lr": 0.0002, "elapsed_sec": 13054.039415359497, "step_time_sec": 8.238646178040653, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1574, "loss": 5.25555944442749, "lr": 0.0002, "elapsed_sec": 13062.278268814087, "step_time_sec": 8.23867240396794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1575, "loss": 5.201363563537598, "lr": 0.0002, "elapsed_sec": 13070.516572475433, "step_time_sec": 8.238190711010247, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1576, "loss": 5.3128275871276855, "lr": 0.0002, "elapsed_sec": 13078.755360364914, "step_time_sec": 8.238613073015586, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1577, "loss": 5.179062366485596, "lr": 0.0002, "elapsed_sec": 13086.994441986084, "step_time_sec": 8.238905781996436, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1578, "loss": 5.183904647827148, "lr": 0.0002, "elapsed_sec": 13095.232573986053, "step_time_sec": 8.237928731017746, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1579, "loss": 5.286752223968506, "lr": 0.0002, "elapsed_sec": 13103.471087217331, "step_time_sec": 8.238386042998172, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1580, "loss": 5.373210906982422, "lr": 0.0002, "elapsed_sec": 13111.711275815964, "step_time_sec": 8.240009299013764, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1581, "loss": 5.138172626495361, "lr": 0.0002, "elapsed_sec": 13119.949575185776, "step_time_sec": 8.238131811027415, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1582, "loss": 5.225396633148193, "lr": 0.0002, "elapsed_sec": 13128.18841958046, "step_time_sec": 8.238684771000408, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1583, "loss": 5.236358165740967, "lr": 0.0002, "elapsed_sec": 13136.425505161285, "step_time_sec": 8.236953628947958, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1584, "loss": 5.229609966278076, "lr": 0.0002, "elapsed_sec": 13144.664954662323, "step_time_sec": 8.239261224982329, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1585, "loss": 5.246957302093506, "lr": 0.0002, "elapsed_sec": 13152.903836965561, "step_time_sec": 8.23869144602213, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1586, "loss": 5.25031042098999, "lr": 0.0002, "elapsed_sec": 13161.143830776215, "step_time_sec": 8.239860398927703, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1587, "loss": 5.138912677764893, "lr": 0.0002, "elapsed_sec": 13169.383910894394, "step_time_sec": 8.239927401067689, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1588, "loss": 5.241077899932861, "lr": 0.0002, "elapsed_sec": 13177.622481584549, "step_time_sec": 8.238366101984866, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1589, "loss": 5.204085826873779, "lr": 0.0002, "elapsed_sec": 13185.861407756805, "step_time_sec": 8.238785760942847, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1590, "loss": 5.241603851318359, "lr": 0.0002, "elapsed_sec": 13194.09931921959, "step_time_sec": 8.237712492933497, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1591, "loss": 5.181911945343018, "lr": 0.0002, "elapsed_sec": 13202.33691072464, "step_time_sec": 8.237444186001085, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1592, "loss": 5.290770530700684, "lr": 0.0002, "elapsed_sec": 13210.575334072113, "step_time_sec": 8.238288032938726, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1593, "loss": 5.310023307800293, "lr": 0.0002, "elapsed_sec": 13218.81322479248, "step_time_sec": 8.237761448021047, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1594, "loss": 5.154257774353027, "lr": 0.0002, "elapsed_sec": 13227.052139043808, "step_time_sec": 8.238741229986772, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1595, "loss": 5.405939102172852, "lr": 0.0002, "elapsed_sec": 13235.290630578995, "step_time_sec": 8.238320551929064, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1596, "loss": 5.26571798324585, "lr": 0.0002, "elapsed_sec": 13243.529073238373, "step_time_sec": 8.238278068020009, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1597, "loss": 5.250385761260986, "lr": 0.0002, "elapsed_sec": 13251.768464565277, "step_time_sec": 8.239175911061466, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1598, "loss": 5.219749450683594, "lr": 0.0002, "elapsed_sec": 13260.0060172081, "step_time_sec": 8.237396943033673, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1599, "loss": 5.20824670791626, "lr": 0.0002, "elapsed_sec": 13268.244535207748, "step_time_sec": 8.238380777067505, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1600, "loss": 5.143388748168945, "lr": 0.0002, "elapsed_sec": 13276.482745170593, "step_time_sec": 8.238061053911224, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1601, "loss": 5.2923970222473145, "lr": 0.0002, "elapsed_sec": 13284.72170996666, "step_time_sec": 8.238777315011248, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1602, "loss": 5.314229965209961, "lr": 0.0002, "elapsed_sec": 13292.960411071777, "step_time_sec": 8.238528880989179, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1603, "loss": 5.211843967437744, "lr": 0.0002, "elapsed_sec": 13301.198921203613, "step_time_sec": 8.238319914089516, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1604, "loss": 5.104038715362549, "lr": 0.0002, "elapsed_sec": 13309.437254667282, "step_time_sec": 8.23813853401225, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1605, "loss": 5.137861728668213, "lr": 0.0002, "elapsed_sec": 13317.67583656311, "step_time_sec": 8.238440586952493, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1606, "loss": 5.117508411407471, "lr": 0.0002, "elapsed_sec": 13325.915044784546, "step_time_sec": 8.239013829967007, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1607, "loss": 5.261559963226318, "lr": 0.0002, "elapsed_sec": 13334.152871847153, "step_time_sec": 8.237688891007565, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1608, "loss": 5.117210388183594, "lr": 0.0002, "elapsed_sec": 13342.389693498611, "step_time_sec": 8.236657998058945, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1609, "loss": 5.281485557556152, "lr": 0.0002, "elapsed_sec": 13350.628436088562, "step_time_sec": 8.238614786998369, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1610, "loss": 5.142040729522705, "lr": 0.0002, "elapsed_sec": 13358.865427732468, "step_time_sec": 8.236817482043989, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1611, "loss": 7.443183422088623, "lr": 0.0002, "elapsed_sec": 13367.1032371521, "step_time_sec": 8.237612952012569, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1612, "loss": 5.190044403076172, "lr": 0.0002, "elapsed_sec": 13375.341661453247, "step_time_sec": 8.238268585992046, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1613, "loss": 5.411050796508789, "lr": 0.0002, "elapsed_sec": 13383.579967975616, "step_time_sec": 8.238166223978624, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1614, "loss": 5.247221946716309, "lr": 0.0002, "elapsed_sec": 13391.81892323494, "step_time_sec": 8.238756446982734, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1615, "loss": 5.180058479309082, "lr": 0.0002, "elapsed_sec": 13400.056891918182, "step_time_sec": 8.237848410964943, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1616, "loss": 5.287159442901611, "lr": 0.0002, "elapsed_sec": 13408.297450065613, "step_time_sec": 8.240373702952638, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1617, "loss": 5.157254695892334, "lr": 0.0002, "elapsed_sec": 13416.536500692368, "step_time_sec": 8.238917732029222, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1618, "loss": 5.236556529998779, "lr": 0.0002, "elapsed_sec": 13424.775538682938, "step_time_sec": 8.238840167061426, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1619, "loss": 5.225929260253906, "lr": 0.0002, "elapsed_sec": 13433.01478099823, "step_time_sec": 8.239079401013441, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1620, "loss": 5.160877704620361, "lr": 0.0002, "elapsed_sec": 13441.252117872238, "step_time_sec": 8.237191656953655, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1621, "loss": 5.130638599395752, "lr": 0.0002, "elapsed_sec": 13449.491515636444, "step_time_sec": 8.239249787991866, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1622, "loss": 5.384891986846924, "lr": 0.0002, "elapsed_sec": 13457.729566812515, "step_time_sec": 8.237861218047328, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1623, "loss": 5.184060096740723, "lr": 0.0002, "elapsed_sec": 13479.110015630722, "step_time_sec": 21.380292601999827, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1624, "loss": 5.2469635009765625, "lr": 0.0002, "elapsed_sec": 13487.341346740723, "step_time_sec": 8.231171566061676, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1625, "loss": 5.297922134399414, "lr": 0.0002, "elapsed_sec": 13495.57371044159, "step_time_sec": 8.232217264943756, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1626, "loss": 5.240572452545166, "lr": 0.0002, "elapsed_sec": 13503.810771942139, "step_time_sec": 8.236918703070842, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1627, "loss": 5.208600997924805, "lr": 0.0002, "elapsed_sec": 13512.049510002136, "step_time_sec": 8.238547309068963, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1628, "loss": 5.19317102432251, "lr": 0.0002, "elapsed_sec": 13520.28868842125, "step_time_sec": 8.239010923076421, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1629, "loss": 5.156378269195557, "lr": 0.0002, "elapsed_sec": 13528.525930643082, "step_time_sec": 8.237082238076255, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1630, "loss": 5.192430019378662, "lr": 0.0002, "elapsed_sec": 13536.765046834946, "step_time_sec": 8.238963073003106, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1631, "loss": 5.319798946380615, "lr": 0.0002, "elapsed_sec": 13545.002768039703, "step_time_sec": 8.237555659958161, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1632, "loss": 5.23776912689209, "lr": 0.0002, "elapsed_sec": 13553.242647647858, "step_time_sec": 8.239744582097046, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1633, "loss": 5.236049652099609, "lr": 0.0002, "elapsed_sec": 13561.482072353363, "step_time_sec": 8.239286642987281, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1634, "loss": 5.241647720336914, "lr": 0.0002, "elapsed_sec": 13569.721679210663, "step_time_sec": 8.23946089099627, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1635, "loss": 5.2576470375061035, "lr": 0.0002, "elapsed_sec": 13577.961075305939, "step_time_sec": 8.239189176936634, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1636, "loss": 5.2864532470703125, "lr": 0.0002, "elapsed_sec": 13586.19949054718, "step_time_sec": 8.23830111103598, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1637, "loss": 5.2520623207092285, "lr": 0.0002, "elapsed_sec": 13594.439038276672, "step_time_sec": 8.239310760982335, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1638, "loss": 5.069469928741455, "lr": 0.0002, "elapsed_sec": 13602.677492141724, "step_time_sec": 8.238294878974557, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1639, "loss": 5.204467296600342, "lr": 0.0002, "elapsed_sec": 13610.915428876877, "step_time_sec": 8.23775755299721, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1640, "loss": 5.201766014099121, "lr": 0.0002, "elapsed_sec": 13619.156012773514, "step_time_sec": 8.24044559604954, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1641, "loss": 5.180256366729736, "lr": 0.0002, "elapsed_sec": 13627.395279884338, "step_time_sec": 8.239077599951997, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1642, "loss": 5.354915618896484, "lr": 0.0002, "elapsed_sec": 13635.635779857635, "step_time_sec": 8.240406361059286, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1643, "loss": 5.274636745452881, "lr": 0.0002, "elapsed_sec": 13643.876260995865, "step_time_sec": 8.240287260035984, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1644, "loss": 5.126242637634277, "lr": 0.0002, "elapsed_sec": 13652.113758563995, "step_time_sec": 8.23730932699982, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1645, "loss": 5.156620979309082, "lr": 0.0002, "elapsed_sec": 13660.353659629822, "step_time_sec": 8.239756957977079, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1646, "loss": 4.954676628112793, "lr": 0.0002, "elapsed_sec": 13668.592853307724, "step_time_sec": 8.239017880987376, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1647, "loss": 5.196861267089844, "lr": 0.0002, "elapsed_sec": 13676.83179283142, "step_time_sec": 8.23879030905664, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1648, "loss": 5.115072250366211, "lr": 0.0002, "elapsed_sec": 13685.070053815842, "step_time_sec": 8.23807218496222, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1649, "loss": 5.202167510986328, "lr": 0.0002, "elapsed_sec": 13693.307575702667, "step_time_sec": 8.237406663014553, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1650, "loss": 5.2276835441589355, "lr": 0.0002, "elapsed_sec": 13701.547785282135, "step_time_sec": 8.240023653954268, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1651, "loss": 5.181291103363037, "lr": 0.0002, "elapsed_sec": 13709.78794503212, "step_time_sec": 8.239969573100097, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1652, "loss": 5.324850559234619, "lr": 0.0002, "elapsed_sec": 13718.026696681976, "step_time_sec": 8.238618912990205, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1653, "loss": 5.267448425292969, "lr": 0.0002, "elapsed_sec": 13726.266257047653, "step_time_sec": 8.239436956937425, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1654, "loss": 5.336171627044678, "lr": 0.0002, "elapsed_sec": 13734.504393577576, "step_time_sec": 8.237946758046746, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1655, "loss": 5.172586441040039, "lr": 0.0002, "elapsed_sec": 13742.742968559265, "step_time_sec": 8.23837807693053, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1656, "loss": 5.2614545822143555, "lr": 0.0002, "elapsed_sec": 13750.982261419296, "step_time_sec": 8.23914794693701, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1657, "loss": 5.181759834289551, "lr": 0.0002, "elapsed_sec": 13759.22108244896, "step_time_sec": 8.23867791495286, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1658, "loss": 5.223881244659424, "lr": 0.0002, "elapsed_sec": 13767.459530353546, "step_time_sec": 8.23827419592999, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1659, "loss": 5.03251838684082, "lr": 0.0002, "elapsed_sec": 13775.698229551315, "step_time_sec": 8.238522451953031, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1660, "loss": 5.2009711265563965, "lr": 0.0002, "elapsed_sec": 13783.936789989471, "step_time_sec": 8.238432071986608, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1661, "loss": 5.200300216674805, "lr": 0.0002, "elapsed_sec": 13792.175663471222, "step_time_sec": 8.238671393017285, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1662, "loss": 5.147902011871338, "lr": 0.0002, "elapsed_sec": 13800.412875890732, "step_time_sec": 8.237064294982702, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1663, "loss": 5.250774383544922, "lr": 0.0002, "elapsed_sec": 13808.652191877365, "step_time_sec": 8.239181863027625, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1664, "loss": 5.244476795196533, "lr": 0.0002, "elapsed_sec": 13816.889863491058, "step_time_sec": 8.23754662706051, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1665, "loss": 5.290403842926025, "lr": 0.0002, "elapsed_sec": 13825.129737854004, "step_time_sec": 8.239662115927786, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1666, "loss": 5.140355110168457, "lr": 0.0002, "elapsed_sec": 13833.36866903305, "step_time_sec": 8.238792021060362, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1667, "loss": 5.137167453765869, "lr": 0.0002, "elapsed_sec": 13841.606965065002, "step_time_sec": 8.238138007931411, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1668, "loss": 5.120224952697754, "lr": 0.0002, "elapsed_sec": 13849.846913337708, "step_time_sec": 8.23975733795669, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1669, "loss": 5.2820024490356445, "lr": 0.0002, "elapsed_sec": 13858.084914922714, "step_time_sec": 8.23785366001539, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1670, "loss": 5.119441509246826, "lr": 0.0002, "elapsed_sec": 13866.3253698349, "step_time_sec": 8.240300558973104, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1671, "loss": 5.162012577056885, "lr": 0.0002, "elapsed_sec": 13874.565086603165, "step_time_sec": 8.239523006952368, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1672, "loss": 5.2748942375183105, "lr": 0.0002, "elapsed_sec": 13882.803031682968, "step_time_sec": 8.237771557061933, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1673, "loss": 5.230474472045898, "lr": 0.0002, "elapsed_sec": 13891.04089641571, "step_time_sec": 8.237750186002813, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1674, "loss": 5.182007312774658, "lr": 0.0002, "elapsed_sec": 13899.28003692627, "step_time_sec": 8.238962372997776, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1675, "loss": 5.218392372131348, "lr": 0.0002, "elapsed_sec": 13907.51995229721, "step_time_sec": 8.239785758894868, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1676, "loss": 5.391070365905762, "lr": 0.0002, "elapsed_sec": 13915.757574558258, "step_time_sec": 8.237424642895348, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1677, "loss": 5.144912242889404, "lr": 0.0002, "elapsed_sec": 13923.9955804348, "step_time_sec": 8.237857045023702, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1678, "loss": 5.2217817306518555, "lr": 0.0002, "elapsed_sec": 13932.23480629921, "step_time_sec": 8.239026519004256, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1679, "loss": 5.194705963134766, "lr": 0.0002, "elapsed_sec": 13940.473438739777, "step_time_sec": 8.2385068211006, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1680, "loss": 5.179593563079834, "lr": 0.0002, "elapsed_sec": 13948.712995290756, "step_time_sec": 8.239414309966378, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1681, "loss": 5.1988325119018555, "lr": 0.0002, "elapsed_sec": 13956.952007293701, "step_time_sec": 8.238797083962709, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1682, "loss": 5.144072532653809, "lr": 0.0002, "elapsed_sec": 13965.18974852562, "step_time_sec": 8.237563848961145, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1683, "loss": 5.249324798583984, "lr": 0.0002, "elapsed_sec": 13973.428510189056, "step_time_sec": 8.23861176893115, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1684, "loss": 5.240026950836182, "lr": 0.0002, "elapsed_sec": 13981.666532278061, "step_time_sec": 8.237889521988109, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1685, "loss": 5.294333457946777, "lr": 0.0002, "elapsed_sec": 13989.902562379837, "step_time_sec": 8.235847842996009, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1686, "loss": 5.359785079956055, "lr": 0.0002, "elapsed_sec": 13998.140773296356, "step_time_sec": 8.23803566605784, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1687, "loss": 5.040730953216553, "lr": 0.0002, "elapsed_sec": 14006.37910747528, "step_time_sec": 8.238196511054412, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1688, "loss": 5.266003608703613, "lr": 0.0002, "elapsed_sec": 14014.618025064468, "step_time_sec": 8.238709343946539, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1689, "loss": 5.209016799926758, "lr": 0.0002, "elapsed_sec": 14022.855475187302, "step_time_sec": 8.237297793035395, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1690, "loss": 5.224618434906006, "lr": 0.0002, "elapsed_sec": 14031.094721078873, "step_time_sec": 8.239096226054244, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1691, "loss": 5.214476585388184, "lr": 0.0002, "elapsed_sec": 14039.334117412567, "step_time_sec": 8.239238498033956, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1692, "loss": 5.4452738761901855, "lr": 0.0002, "elapsed_sec": 14047.572373867035, "step_time_sec": 8.238073369022459, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1693, "loss": 5.2439117431640625, "lr": 0.0002, "elapsed_sec": 14055.811720848083, "step_time_sec": 8.239216923946515, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1694, "loss": 5.1589837074279785, "lr": 0.0002, "elapsed_sec": 14064.049496650696, "step_time_sec": 8.23761806904804, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1695, "loss": 5.20598840713501, "lr": 0.0002, "elapsed_sec": 14072.28776216507, "step_time_sec": 8.238064146018587, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1696, "loss": 5.1515212059021, "lr": 0.0002, "elapsed_sec": 14080.526167154312, "step_time_sec": 8.238289038999937, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1697, "loss": 5.108178615570068, "lr": 0.0002, "elapsed_sec": 14088.764654159546, "step_time_sec": 8.238329011015594, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1698, "loss": 5.226228713989258, "lr": 0.0002, "elapsed_sec": 14097.00425863266, "step_time_sec": 8.239427768043242, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1699, "loss": 5.116121292114258, "lr": 0.0002, "elapsed_sec": 14105.242379903793, "step_time_sec": 8.237928322982043, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1700, "loss": 5.214570045471191, "lr": 0.0002, "elapsed_sec": 14113.482519626617, "step_time_sec": 8.240026253974065, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1701, "loss": 5.26440954208374, "lr": 0.0002, "elapsed_sec": 14121.720636844635, "step_time_sec": 8.237952108960599, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1702, "loss": 5.061459541320801, "lr": 0.0002, "elapsed_sec": 14129.958993911743, "step_time_sec": 8.238155611907132, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1703, "loss": 5.30269718170166, "lr": 0.0002, "elapsed_sec": 14138.196858882904, "step_time_sec": 8.237705737003125, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1704, "loss": 5.259503364562988, "lr": 0.0002, "elapsed_sec": 14146.433007955551, "step_time_sec": 8.2360555739142, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1705, "loss": 5.186519622802734, "lr": 0.0002, "elapsed_sec": 14154.671795368195, "step_time_sec": 8.238591526052915, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1706, "loss": 5.124289512634277, "lr": 0.0002, "elapsed_sec": 14162.909226894379, "step_time_sec": 8.237258333945647, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1707, "loss": 5.186017990112305, "lr": 0.0002, "elapsed_sec": 14171.14827466011, "step_time_sec": 8.238922088989057, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1708, "loss": 5.269554615020752, "lr": 0.0002, "elapsed_sec": 14179.38816189766, "step_time_sec": 8.239708745037206, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1709, "loss": 5.095022201538086, "lr": 0.0002, "elapsed_sec": 14187.627004384995, "step_time_sec": 8.23864226799924, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1710, "loss": 5.135054111480713, "lr": 0.0002, "elapsed_sec": 14195.8643181324, "step_time_sec": 8.237153765978292, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1711, "loss": 5.164511680603027, "lr": 0.0002, "elapsed_sec": 14204.101544618607, "step_time_sec": 8.23706828604918, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1712, "loss": 5.130083084106445, "lr": 0.0002, "elapsed_sec": 14212.341444969177, "step_time_sec": 8.239734978997149, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1713, "loss": 4.9768524169921875, "lr": 0.0002, "elapsed_sec": 14220.580385923386, "step_time_sec": 8.238806195906363, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1714, "loss": 5.234082221984863, "lr": 0.0002, "elapsed_sec": 14228.818585157394, "step_time_sec": 8.238046490005217, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1715, "loss": 5.223586082458496, "lr": 0.0002, "elapsed_sec": 14237.057059049606, "step_time_sec": 8.238328885054216, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1716, "loss": 5.169064521789551, "lr": 0.0002, "elapsed_sec": 14245.295459747314, "step_time_sec": 8.238200954045169, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1717, "loss": 5.159966945648193, "lr": 0.0002, "elapsed_sec": 14253.536056280136, "step_time_sec": 8.240438060020097, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1718, "loss": 5.209287166595459, "lr": 0.0002, "elapsed_sec": 14261.776301145554, "step_time_sec": 8.240130524965934, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1719, "loss": 5.006153106689453, "lr": 0.0002, "elapsed_sec": 14270.013482809067, "step_time_sec": 8.236987896030769, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1720, "loss": 5.075064182281494, "lr": 0.0002, "elapsed_sec": 14278.252339839935, "step_time_sec": 8.23869468702469, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1721, "loss": 5.143362522125244, "lr": 0.0002, "elapsed_sec": 14286.490798473358, "step_time_sec": 8.238322988967411, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1722, "loss": 5.161161422729492, "lr": 0.0002, "elapsed_sec": 14294.730026483536, "step_time_sec": 8.239044782007113, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1723, "loss": 5.211104393005371, "lr": 0.0002, "elapsed_sec": 14302.96922659874, "step_time_sec": 8.23902288696263, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1724, "loss": 5.100063800811768, "lr": 0.0002, "elapsed_sec": 14311.206600666046, "step_time_sec": 8.237248279037885, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1725, "loss": 5.145021438598633, "lr": 0.0002, "elapsed_sec": 14319.444967985153, "step_time_sec": 8.238199350074865, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1726, "loss": 5.20469856262207, "lr": 0.0002, "elapsed_sec": 14327.683723926544, "step_time_sec": 8.238558259909041, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1727, "loss": 5.0932111740112305, "lr": 0.0002, "elapsed_sec": 14335.922879934311, "step_time_sec": 8.239010610966943, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1728, "loss": 5.027061939239502, "lr": 0.0002, "elapsed_sec": 14344.162450551987, "step_time_sec": 8.239426904940046, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1729, "loss": 5.157999515533447, "lr": 0.0002, "elapsed_sec": 14352.400527238846, "step_time_sec": 8.237892025965266, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1730, "loss": 5.215456485748291, "lr": 0.0002, "elapsed_sec": 14360.639957666397, "step_time_sec": 8.239287041011266, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1731, "loss": 5.027440071105957, "lr": 0.0002, "elapsed_sec": 14368.879880428314, "step_time_sec": 8.239760446012951, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1732, "loss": 5.094335556030273, "lr": 0.0002, "elapsed_sec": 14377.119723796844, "step_time_sec": 8.23968954605516, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1733, "loss": 5.224240303039551, "lr": 0.0002, "elapsed_sec": 14385.358511686325, "step_time_sec": 8.238602536963299, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1734, "loss": 5.114319801330566, "lr": 0.0002, "elapsed_sec": 14393.595419168472, "step_time_sec": 8.236766858957708, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1735, "loss": 5.091533660888672, "lr": 0.0002, "elapsed_sec": 14401.835005044937, "step_time_sec": 8.239409848931246, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1736, "loss": 5.214577674865723, "lr": 0.0002, "elapsed_sec": 14410.073887586594, "step_time_sec": 8.238712374004535, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1737, "loss": 5.091328144073486, "lr": 0.0002, "elapsed_sec": 14418.313146352768, "step_time_sec": 8.239126138971187, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1738, "loss": 5.129321098327637, "lr": 0.0002, "elapsed_sec": 14426.551238775253, "step_time_sec": 8.237935925950296, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1739, "loss": 5.13237190246582, "lr": 0.0002, "elapsed_sec": 14434.788961172104, "step_time_sec": 8.237534979940392, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1740, "loss": 5.101476192474365, "lr": 0.0002, "elapsed_sec": 14443.028664112091, "step_time_sec": 8.23951896908693, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1741, "loss": 5.180190086364746, "lr": 0.0002, "elapsed_sec": 14451.26725077629, "step_time_sec": 8.238445198046975, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1742, "loss": 4.982423305511475, "lr": 0.0002, "elapsed_sec": 14459.5062789917, "step_time_sec": 8.238873664056882, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1743, "loss": 5.1395392417907715, "lr": 0.0002, "elapsed_sec": 14467.745015144348, "step_time_sec": 8.238521674997173, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1744, "loss": 5.18058443069458, "lr": 0.0002, "elapsed_sec": 14475.98332810402, "step_time_sec": 8.238177597988397, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1745, "loss": 5.184622764587402, "lr": 0.0002, "elapsed_sec": 14484.22344326973, "step_time_sec": 8.239954696968198, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1746, "loss": 5.22385311126709, "lr": 0.0002, "elapsed_sec": 14492.46180677414, "step_time_sec": 8.238201398984529, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1747, "loss": 5.146822929382324, "lr": 0.0002, "elapsed_sec": 14500.702024936676, "step_time_sec": 8.240074843983166, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1748, "loss": 5.133185386657715, "lr": 0.0002, "elapsed_sec": 14508.941256046295, "step_time_sec": 8.239039308973588, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1749, "loss": 5.210466384887695, "lr": 0.0002, "elapsed_sec": 14517.178845882416, "step_time_sec": 8.237415274954401, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1750, "loss": 5.138887882232666, "lr": 0.0002, "elapsed_sec": 14525.41837143898, "step_time_sec": 8.239354116958566, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1751, "loss": 5.13860559463501, "lr": 0.0002, "elapsed_sec": 14533.658264398575, "step_time_sec": 8.239788296050392, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1752, "loss": 5.182357311248779, "lr": 0.0002, "elapsed_sec": 14541.896805047989, "step_time_sec": 8.238331433967687, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1753, "loss": 5.127917766571045, "lr": 0.0002, "elapsed_sec": 14550.135834217072, "step_time_sec": 8.238859941018745, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1754, "loss": 5.2135396003723145, "lr": 0.0002, "elapsed_sec": 14558.372629404068, "step_time_sec": 8.236639969982207, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1755, "loss": 5.030778884887695, "lr": 0.0002, "elapsed_sec": 14566.612276792526, "step_time_sec": 8.23955317994114, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1756, "loss": 5.149448394775391, "lr": 0.0002, "elapsed_sec": 14574.851328611374, "step_time_sec": 8.23885902191978, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1757, "loss": 5.2507004737854, "lr": 0.0002, "elapsed_sec": 14583.089243412018, "step_time_sec": 8.237744236947037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1758, "loss": 5.008941173553467, "lr": 0.0002, "elapsed_sec": 14591.328735351562, "step_time_sec": 8.239347561029717, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1759, "loss": 5.1228814125061035, "lr": 0.0002, "elapsed_sec": 14599.567410707474, "step_time_sec": 8.23853838199284, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1760, "loss": 5.10296106338501, "lr": 0.0002, "elapsed_sec": 14607.80552315712, "step_time_sec": 8.237891175085679, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1761, "loss": 5.213764190673828, "lr": 0.0002, "elapsed_sec": 14616.044161081314, "step_time_sec": 8.2385294130072, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1762, "loss": 5.017508506774902, "lr": 0.0002, "elapsed_sec": 14624.282619476318, "step_time_sec": 8.238235995988362, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1763, "loss": 5.000399112701416, "lr": 0.0002, "elapsed_sec": 14632.521570682526, "step_time_sec": 8.238828629022464, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1764, "loss": 5.187417507171631, "lr": 0.0002, "elapsed_sec": 14640.760177373886, "step_time_sec": 8.23840258596465, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1765, "loss": 5.1702446937561035, "lr": 0.0002, "elapsed_sec": 14648.999294996262, "step_time_sec": 8.238939389004372, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1766, "loss": 5.188361167907715, "lr": 0.0002, "elapsed_sec": 14657.238112688065, "step_time_sec": 8.238674985943362, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1767, "loss": 5.183229446411133, "lr": 0.0002, "elapsed_sec": 14665.47587442398, "step_time_sec": 8.237608865019865, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1768, "loss": 5.115788459777832, "lr": 0.0002, "elapsed_sec": 14673.71461892128, "step_time_sec": 8.238562808954157, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1769, "loss": 5.166517734527588, "lr": 0.0002, "elapsed_sec": 14681.952328681946, "step_time_sec": 8.237540832022205, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1770, "loss": 5.10120153427124, "lr": 0.0002, "elapsed_sec": 14690.191388130188, "step_time_sec": 8.238901951001026, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1771, "loss": 5.197303771972656, "lr": 0.0002, "elapsed_sec": 14698.429433822632, "step_time_sec": 8.237909524934366, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1772, "loss": 5.153285503387451, "lr": 0.0002, "elapsed_sec": 14706.667750835419, "step_time_sec": 8.238124015042558, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1773, "loss": 5.064289093017578, "lr": 0.0002, "elapsed_sec": 14714.907726049423, "step_time_sec": 8.239857968059368, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1774, "loss": 5.093472957611084, "lr": 0.0002, "elapsed_sec": 14723.146080493927, "step_time_sec": 8.238183164969087, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1775, "loss": 5.090137481689453, "lr": 0.0002, "elapsed_sec": 14731.38771867752, "step_time_sec": 8.241428659996018, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1776, "loss": 5.021397590637207, "lr": 0.0002, "elapsed_sec": 14739.626819133759, "step_time_sec": 8.238936700974591, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1777, "loss": 5.11257791519165, "lr": 0.0002, "elapsed_sec": 14747.864959478378, "step_time_sec": 8.23796979605686, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1778, "loss": 5.05686092376709, "lr": 0.0002, "elapsed_sec": 14756.103212118149, "step_time_sec": 8.238146284013055, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1779, "loss": 5.060764789581299, "lr": 0.0002, "elapsed_sec": 14764.34150481224, "step_time_sec": 8.238086221041158, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1780, "loss": 5.244223117828369, "lr": 0.0002, "elapsed_sec": 14772.580812692642, "step_time_sec": 8.239133941940963, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1781, "loss": 5.084570407867432, "lr": 0.0002, "elapsed_sec": 14780.820100069046, "step_time_sec": 8.239133881987073, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1782, "loss": 5.21177864074707, "lr": 0.0002, "elapsed_sec": 14789.057914733887, "step_time_sec": 8.237649200018495, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1783, "loss": 5.035623073577881, "lr": 0.0002, "elapsed_sec": 14797.295897483826, "step_time_sec": 8.237872469937429, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1784, "loss": 5.174994945526123, "lr": 0.0002, "elapsed_sec": 14805.532793521881, "step_time_sec": 8.236691910075024, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1785, "loss": 5.108764171600342, "lr": 0.0002, "elapsed_sec": 14813.771504163742, "step_time_sec": 8.23852720996365, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1786, "loss": 5.137070655822754, "lr": 0.0002, "elapsed_sec": 14822.010128736496, "step_time_sec": 8.238465611939318, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1787, "loss": 4.989394664764404, "lr": 0.0002, "elapsed_sec": 14830.248177289963, "step_time_sec": 8.237889798008837, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1788, "loss": 5.112579822540283, "lr": 0.0002, "elapsed_sec": 14838.486598968506, "step_time_sec": 8.238311827066354, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1789, "loss": 4.981183052062988, "lr": 0.0002, "elapsed_sec": 14846.723909139633, "step_time_sec": 8.237069576047361, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1790, "loss": 5.247637748718262, "lr": 0.0002, "elapsed_sec": 14854.96273446083, "step_time_sec": 8.238724467926659, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1791, "loss": 5.114705562591553, "lr": 0.0002, "elapsed_sec": 14863.201133728027, "step_time_sec": 8.238218531012535, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1792, "loss": 5.121738433837891, "lr": 0.0002, "elapsed_sec": 14871.441043376923, "step_time_sec": 8.239763577003032, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1793, "loss": 5.0414042472839355, "lr": 0.0002, "elapsed_sec": 14879.680007696152, "step_time_sec": 8.238750083954073, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1794, "loss": 5.013720989227295, "lr": 0.0002, "elapsed_sec": 14887.917677640915, "step_time_sec": 8.23754609900061, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1795, "loss": 5.183631896972656, "lr": 0.0002, "elapsed_sec": 14896.156620502472, "step_time_sec": 8.238837428041734, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1796, "loss": 5.023614883422852, "lr": 0.0002, "elapsed_sec": 14904.394050121307, "step_time_sec": 8.237196524976753, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1797, "loss": 5.105917453765869, "lr": 0.0002, "elapsed_sec": 14912.632889986038, "step_time_sec": 8.238683218020014, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1798, "loss": 5.116061687469482, "lr": 0.0002, "elapsed_sec": 14920.871889591217, "step_time_sec": 8.238813780946657, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1799, "loss": 5.150115489959717, "lr": 0.0002, "elapsed_sec": 14929.109393835068, "step_time_sec": 8.237358929007314, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1800, "loss": 5.087799549102783, "lr": 0.0002, "elapsed_sec": 14937.348590373993, "step_time_sec": 8.239036819082685, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1801, "loss": 5.128295421600342, "lr": 0.0002, "elapsed_sec": 14945.587807416916, "step_time_sec": 8.239044699934311, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1802, "loss": 5.031057357788086, "lr": 0.0002, "elapsed_sec": 14953.828044891357, "step_time_sec": 8.240092152031139, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1803, "loss": 5.14749002456665, "lr": 0.0002, "elapsed_sec": 14962.067514657974, "step_time_sec": 8.239275783998892, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1804, "loss": 5.016457557678223, "lr": 0.0002, "elapsed_sec": 14970.305588960648, "step_time_sec": 8.237915802979842, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1805, "loss": 5.044208526611328, "lr": 0.0002, "elapsed_sec": 14978.54528427124, "step_time_sec": 8.239541213959455, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1806, "loss": 5.068687438964844, "lr": 0.0002, "elapsed_sec": 14986.783425092697, "step_time_sec": 8.237971646944061, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1807, "loss": 4.951067924499512, "lr": 0.0002, "elapsed_sec": 14995.021988153458, "step_time_sec": 8.23842083197087, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1808, "loss": 4.94087028503418, "lr": 0.0002, "elapsed_sec": 15003.261131048203, "step_time_sec": 8.238942402997054, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1809, "loss": 5.025082111358643, "lr": 0.0002, "elapsed_sec": 15011.499576330185, "step_time_sec": 8.238308933912776, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1810, "loss": 5.032145023345947, "lr": 0.0002, "elapsed_sec": 15019.739887952805, "step_time_sec": 8.240129427984357, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1811, "loss": 5.131347179412842, "lr": 0.0002, "elapsed_sec": 15027.9784116745, "step_time_sec": 8.238381420029327, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1812, "loss": 5.150102615356445, "lr": 0.0002, "elapsed_sec": 15036.216504573822, "step_time_sec": 8.237990704015829, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1813, "loss": 4.979641437530518, "lr": 0.0002, "elapsed_sec": 15044.454260349274, "step_time_sec": 8.237555372994393, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1814, "loss": 5.116952896118164, "lr": 0.0002, "elapsed_sec": 15052.692249298096, "step_time_sec": 8.23783518502023, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1815, "loss": 5.029140949249268, "lr": 0.0002, "elapsed_sec": 15060.931371688843, "step_time_sec": 8.238924828008749, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1816, "loss": 5.057284355163574, "lr": 0.0002, "elapsed_sec": 15069.16851758957, "step_time_sec": 8.2369985220721, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1817, "loss": 5.232088088989258, "lr": 0.0002, "elapsed_sec": 15077.407175779343, "step_time_sec": 8.238505913992412, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1818, "loss": 5.056023597717285, "lr": 0.0002, "elapsed_sec": 15085.646386861801, "step_time_sec": 8.239044976071455, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1819, "loss": 4.994409561157227, "lr": 0.0002, "elapsed_sec": 15093.883559465408, "step_time_sec": 8.237051295000128, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1820, "loss": 5.024669647216797, "lr": 0.0002, "elapsed_sec": 15102.122924566269, "step_time_sec": 8.239165549981408, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1821, "loss": 5.094742298126221, "lr": 0.0002, "elapsed_sec": 15110.36006140709, "step_time_sec": 8.237018055049703, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1822, "loss": 5.172407627105713, "lr": 0.0002, "elapsed_sec": 15118.598708868027, "step_time_sec": 8.238476724014618, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1823, "loss": 5.039097785949707, "lr": 0.0002, "elapsed_sec": 15126.838203191757, "step_time_sec": 8.239327522926033, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1824, "loss": 4.948460578918457, "lr": 0.0002, "elapsed_sec": 15135.076911449432, "step_time_sec": 8.238573178998195, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1825, "loss": 5.160662651062012, "lr": 0.0002, "elapsed_sec": 15143.316391468048, "step_time_sec": 8.239275185042061, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1826, "loss": 5.144077301025391, "lr": 0.0002, "elapsed_sec": 15151.553725242615, "step_time_sec": 8.23719098197762, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1827, "loss": 5.052887439727783, "lr": 0.0002, "elapsed_sec": 15159.79254579544, "step_time_sec": 8.238640349009074, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1828, "loss": 5.220520496368408, "lr": 0.0002, "elapsed_sec": 15168.031307935715, "step_time_sec": 8.238594261929393, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1829, "loss": 5.008574485778809, "lr": 0.0002, "elapsed_sec": 15176.269061803818, "step_time_sec": 8.237638911930844, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1830, "loss": 5.045375347137451, "lr": 0.0002, "elapsed_sec": 15184.507437705994, "step_time_sec": 8.23819075198844, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1831, "loss": 5.118607521057129, "lr": 0.0002, "elapsed_sec": 15192.745970487595, "step_time_sec": 8.238375155022368, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1832, "loss": 5.192874908447266, "lr": 0.0002, "elapsed_sec": 15200.986553192139, "step_time_sec": 8.24039025802631, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1833, "loss": 5.100021839141846, "lr": 0.0002, "elapsed_sec": 15209.225466251373, "step_time_sec": 8.23876424995251, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1834, "loss": 5.1026482582092285, "lr": 0.0002, "elapsed_sec": 15217.46342754364, "step_time_sec": 8.237795007997192, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1835, "loss": 5.023089408874512, "lr": 0.0002, "elapsed_sec": 15225.702739715576, "step_time_sec": 8.239174194051884, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1836, "loss": 4.971211910247803, "lr": 0.0002, "elapsed_sec": 15233.94228553772, "step_time_sec": 8.239385396009311, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1837, "loss": 5.022545337677002, "lr": 0.0002, "elapsed_sec": 15242.182443618774, "step_time_sec": 8.239958174992353, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1838, "loss": 5.2247209548950195, "lr": 0.0002, "elapsed_sec": 15250.422024011612, "step_time_sec": 8.239460073062219, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1839, "loss": 4.9337849617004395, "lr": 0.0002, "elapsed_sec": 15258.661349773407, "step_time_sec": 8.239173301961273, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1840, "loss": 5.06734561920166, "lr": 0.0002, "elapsed_sec": 15266.900898456573, "step_time_sec": 8.23934608802665, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1841, "loss": 5.173981189727783, "lr": 0.0002, "elapsed_sec": 15275.139065265656, "step_time_sec": 8.238024302991107, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1842, "loss": 5.1523051261901855, "lr": 0.0002, "elapsed_sec": 15283.376974344254, "step_time_sec": 8.237728370004334, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1843, "loss": 5.012359142303467, "lr": 0.0002, "elapsed_sec": 15291.614937067032, "step_time_sec": 8.23781901691109, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1844, "loss": 5.1725077629089355, "lr": 0.0002, "elapsed_sec": 15299.852373600006, "step_time_sec": 8.23728433100041, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1845, "loss": 5.084303379058838, "lr": 0.0002, "elapsed_sec": 15308.09232211113, "step_time_sec": 8.239792251959443, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1846, "loss": 5.187169075012207, "lr": 0.0002, "elapsed_sec": 15316.330038547516, "step_time_sec": 8.237587940995581, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1847, "loss": 4.941648483276367, "lr": 0.0002, "elapsed_sec": 15324.569081068039, "step_time_sec": 8.238872030051425, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1848, "loss": 5.114786148071289, "lr": 0.0002, "elapsed_sec": 15332.808457374573, "step_time_sec": 8.23919457802549, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1849, "loss": 5.088532447814941, "lr": 0.0002, "elapsed_sec": 15341.047547101974, "step_time_sec": 8.23888130695559, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1850, "loss": 5.038227081298828, "lr": 0.0002, "elapsed_sec": 15349.287895679474, "step_time_sec": 8.24018691899255, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1851, "loss": 5.053714752197266, "lr": 0.0002, "elapsed_sec": 15357.526723384857, "step_time_sec": 8.23867125005927, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1852, "loss": 5.092273235321045, "lr": 0.0002, "elapsed_sec": 15365.765686511993, "step_time_sec": 8.238848321023397, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1853, "loss": 5.0419697761535645, "lr": 0.0002, "elapsed_sec": 15374.005450725555, "step_time_sec": 8.2395863130223, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1854, "loss": 5.062037467956543, "lr": 0.0002, "elapsed_sec": 15382.243708133698, "step_time_sec": 8.238056006957777, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1855, "loss": 5.094284534454346, "lr": 0.0002, "elapsed_sec": 15390.484549045563, "step_time_sec": 8.240718028973788, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1856, "loss": 5.131939888000488, "lr": 0.0002, "elapsed_sec": 15398.723008871078, "step_time_sec": 8.23827040498145, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1857, "loss": 5.1191582679748535, "lr": 0.0002, "elapsed_sec": 15406.960286140442, "step_time_sec": 8.237115309922956, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1858, "loss": 4.949476718902588, "lr": 0.0002, "elapsed_sec": 15415.198437929153, "step_time_sec": 8.238013508031145, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1859, "loss": 5.00692081451416, "lr": 0.0002, "elapsed_sec": 15423.434937953949, "step_time_sec": 8.236294943955727, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1860, "loss": 5.105703830718994, "lr": 0.0002, "elapsed_sec": 15431.67421746254, "step_time_sec": 8.239108963985927, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1861, "loss": 5.082329750061035, "lr": 0.0002, "elapsed_sec": 15439.91314959526, "step_time_sec": 8.238755002967082, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1862, "loss": 5.045751571655273, "lr": 0.0002, "elapsed_sec": 15448.15221619606, "step_time_sec": 8.238930854946375, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1863, "loss": 5.091421127319336, "lr": 0.0002, "elapsed_sec": 15456.390528917313, "step_time_sec": 8.238195864018053, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1864, "loss": 5.235377311706543, "lr": 0.0002, "elapsed_sec": 15464.626933336258, "step_time_sec": 8.236229341942817, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1865, "loss": 5.0638322830200195, "lr": 0.0002, "elapsed_sec": 15472.866626739502, "step_time_sec": 8.239501795964316, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1866, "loss": 5.093868732452393, "lr": 0.0002, "elapsed_sec": 15481.10591340065, "step_time_sec": 8.239115010015666, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1867, "loss": 5.134389877319336, "lr": 0.0002, "elapsed_sec": 15489.345239162445, "step_time_sec": 8.239182493067347, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1868, "loss": 5.074323654174805, "lr": 0.0002, "elapsed_sec": 15497.58561205864, "step_time_sec": 8.240196929080412, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1869, "loss": 5.121004581451416, "lr": 0.0002, "elapsed_sec": 15505.823954343796, "step_time_sec": 8.238175502978265, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1870, "loss": 5.001566410064697, "lr": 0.0002, "elapsed_sec": 15514.062943935394, "step_time_sec": 8.238841993035749, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1871, "loss": 4.877310276031494, "lr": 0.0002, "elapsed_sec": 15522.372935771942, "step_time_sec": 8.249912315048277, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1872, "loss": 5.106325626373291, "lr": 0.0002, "elapsed_sec": 15530.611428499222, "step_time_sec": 8.238365705008619, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1873, "loss": 5.014179229736328, "lr": 0.0002, "elapsed_sec": 15538.843343257904, "step_time_sec": 8.231723646982573, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1874, "loss": 4.9896135330200195, "lr": 0.0002, "elapsed_sec": 15547.075958251953, "step_time_sec": 8.232448989059776, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1875, "loss": 4.97173547744751, "lr": 0.0002, "elapsed_sec": 15555.308852910995, "step_time_sec": 8.232773339026608, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1876, "loss": 5.133885860443115, "lr": 0.0002, "elapsed_sec": 15563.548372030258, "step_time_sec": 8.239340616040863, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1877, "loss": 5.265945911407471, "lr": 0.0002, "elapsed_sec": 15571.786990880966, "step_time_sec": 8.238409520941786, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1878, "loss": 4.968745231628418, "lr": 0.0002, "elapsed_sec": 15580.025842428207, "step_time_sec": 8.238705590018071, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1879, "loss": 5.13311243057251, "lr": 0.0002, "elapsed_sec": 15588.264506101608, "step_time_sec": 8.238530621980317, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1880, "loss": 5.108828544616699, "lr": 0.0002, "elapsed_sec": 15596.50344324112, "step_time_sec": 8.238795852987096, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1881, "loss": 5.0775322914123535, "lr": 0.0002, "elapsed_sec": 15604.743536233902, "step_time_sec": 8.239876874955371, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1882, "loss": 5.116812705993652, "lr": 0.0002, "elapsed_sec": 15612.982292890549, "step_time_sec": 8.238620538963005, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1883, "loss": 5.032761096954346, "lr": 0.0002, "elapsed_sec": 15621.223185777664, "step_time_sec": 8.240683914045803, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1884, "loss": 5.1670613288879395, "lr": 0.0002, "elapsed_sec": 15629.461803674698, "step_time_sec": 8.238443864043802, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1885, "loss": 4.953621864318848, "lr": 0.0002, "elapsed_sec": 15637.700251579285, "step_time_sec": 8.238294136011973, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1886, "loss": 5.024170398712158, "lr": 0.0002, "elapsed_sec": 15645.939270019531, "step_time_sec": 8.238838738994673, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1887, "loss": 5.069611072540283, "lr": 0.0002, "elapsed_sec": 15654.177443742752, "step_time_sec": 8.238069990067743, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1888, "loss": 5.074798583984375, "lr": 0.0002, "elapsed_sec": 15662.41787981987, "step_time_sec": 8.240237243007869, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1889, "loss": 5.0247111320495605, "lr": 0.0002, "elapsed_sec": 15670.656305074692, "step_time_sec": 8.238298177951947, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1890, "loss": 5.05764627456665, "lr": 0.0002, "elapsed_sec": 15678.894666671753, "step_time_sec": 8.238183464040048, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1891, "loss": 5.090008735656738, "lr": 0.0002, "elapsed_sec": 15687.133350372314, "step_time_sec": 8.2384690400213, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1892, "loss": 5.182150840759277, "lr": 0.0002, "elapsed_sec": 15695.372062683105, "step_time_sec": 8.238564796047285, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1893, "loss": 5.056201457977295, "lr": 0.0002, "elapsed_sec": 15703.611901044846, "step_time_sec": 8.23967005405575, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1894, "loss": 5.082965850830078, "lr": 0.0002, "elapsed_sec": 15711.851435899734, "step_time_sec": 8.239333934965543, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1895, "loss": 5.052402019500732, "lr": 0.0002, "elapsed_sec": 15720.087662696838, "step_time_sec": 8.236072815023363, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1896, "loss": 4.983088970184326, "lr": 0.0002, "elapsed_sec": 15728.327157735825, "step_time_sec": 8.239347445080057, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1897, "loss": 4.8941755294799805, "lr": 0.0002, "elapsed_sec": 15736.5652551651, "step_time_sec": 8.237975077936426, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1898, "loss": 5.110940933227539, "lr": 0.0002, "elapsed_sec": 15744.804550409317, "step_time_sec": 8.239112014998682, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1899, "loss": 4.936049461364746, "lr": 0.0002, "elapsed_sec": 15753.043376922607, "step_time_sec": 8.238645876990631, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1900, "loss": 5.021721363067627, "lr": 0.0002, "elapsed_sec": 15761.28113746643, "step_time_sec": 8.23759577004239, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1901, "loss": 5.059112071990967, "lr": 0.0002, "elapsed_sec": 15769.519824504852, "step_time_sec": 8.238596477080137, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1902, "loss": 5.0559515953063965, "lr": 0.0002, "elapsed_sec": 15777.757711172104, "step_time_sec": 8.237672468996607, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1903, "loss": 5.0089335441589355, "lr": 0.0002, "elapsed_sec": 15785.997066259384, "step_time_sec": 8.23918689507991, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1904, "loss": 5.12724494934082, "lr": 0.0002, "elapsed_sec": 15794.235372066498, "step_time_sec": 8.238139503984712, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1905, "loss": 4.961965084075928, "lr": 0.0002, "elapsed_sec": 15802.473704576492, "step_time_sec": 8.238175894948654, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1906, "loss": 4.9474663734436035, "lr": 0.0002, "elapsed_sec": 15810.712182998657, "step_time_sec": 8.238365805940703, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1907, "loss": 5.055987358093262, "lr": 0.0002, "elapsed_sec": 15818.949771404266, "step_time_sec": 8.237387930043042, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1908, "loss": 5.073975086212158, "lr": 0.0002, "elapsed_sec": 15827.189607143402, "step_time_sec": 8.239679691032507, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1909, "loss": 4.885948181152344, "lr": 0.0002, "elapsed_sec": 15835.428916692734, "step_time_sec": 8.239187921979465, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1910, "loss": 5.031152248382568, "lr": 0.0002, "elapsed_sec": 15843.666934251785, "step_time_sec": 8.237823076895438, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1911, "loss": 5.046281337738037, "lr": 0.0002, "elapsed_sec": 15851.905858039856, "step_time_sec": 8.238766839029267, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1912, "loss": 5.017050743103027, "lr": 0.0002, "elapsed_sec": 15860.144946336746, "step_time_sec": 8.23893789399881, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1913, "loss": 5.037510871887207, "lr": 0.0002, "elapsed_sec": 15868.383611440659, "step_time_sec": 8.238514091004618, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1914, "loss": 4.993141174316406, "lr": 0.0002, "elapsed_sec": 15876.622117519379, "step_time_sec": 8.23836121801287, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1915, "loss": 5.032851696014404, "lr": 0.0002, "elapsed_sec": 15884.860812664032, "step_time_sec": 8.23848776705563, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1916, "loss": 5.067348003387451, "lr": 0.0002, "elapsed_sec": 15893.100486040115, "step_time_sec": 8.239509730017744, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1917, "loss": 5.027154445648193, "lr": 0.0002, "elapsed_sec": 15901.337882995605, "step_time_sec": 8.237236648914404, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1918, "loss": 5.001223564147949, "lr": 0.0002, "elapsed_sec": 15909.57715344429, "step_time_sec": 8.239120031008497, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1919, "loss": 5.0937113761901855, "lr": 0.0002, "elapsed_sec": 15917.817625761032, "step_time_sec": 8.240318731055595, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1920, "loss": 5.045166969299316, "lr": 0.0002, "elapsed_sec": 15926.055975914001, "step_time_sec": 8.238256257027388, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1921, "loss": 4.973743438720703, "lr": 0.0002, "elapsed_sec": 15934.295032024384, "step_time_sec": 8.23882229696028, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1922, "loss": 5.019031524658203, "lr": 0.0002, "elapsed_sec": 15942.53389453888, "step_time_sec": 8.238720554974861, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1923, "loss": 5.01767110824585, "lr": 0.0002, "elapsed_sec": 15950.772841453552, "step_time_sec": 8.238808780093677, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1924, "loss": 4.924198627471924, "lr": 0.0002, "elapsed_sec": 15959.011910200119, "step_time_sec": 8.238829759997316, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1925, "loss": 5.069791316986084, "lr": 0.0002, "elapsed_sec": 15967.2486410141, "step_time_sec": 8.236590789048932, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1926, "loss": 5.076047897338867, "lr": 0.0002, "elapsed_sec": 15975.487238645554, "step_time_sec": 8.238464219029993, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1927, "loss": 5.0579118728637695, "lr": 0.0002, "elapsed_sec": 15983.724676132202, "step_time_sec": 8.237273836042732, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1928, "loss": 5.132526874542236, "lr": 0.0002, "elapsed_sec": 15991.964022397995, "step_time_sec": 8.239144132006913, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1929, "loss": 4.967777729034424, "lr": 0.0002, "elapsed_sec": 16000.20257282257, "step_time_sec": 8.238362944102846, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1930, "loss": 4.926032066345215, "lr": 0.0002, "elapsed_sec": 16008.44022655487, "step_time_sec": 8.237516118097119, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1931, "loss": 5.055834770202637, "lr": 0.0002, "elapsed_sec": 16016.679771661758, "step_time_sec": 8.239396370947361, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1932, "loss": 5.022822856903076, "lr": 0.0002, "elapsed_sec": 16024.917883396149, "step_time_sec": 8.23792521096766, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1933, "loss": 4.9802775382995605, "lr": 0.0002, "elapsed_sec": 16033.157573461533, "step_time_sec": 8.239501437987201, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1934, "loss": 4.895246982574463, "lr": 0.0002, "elapsed_sec": 16041.396157741547, "step_time_sec": 8.23845184291713, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1935, "loss": 4.970465183258057, "lr": 0.0002, "elapsed_sec": 16049.633943796158, "step_time_sec": 8.237612623954192, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1936, "loss": 4.964703559875488, "lr": 0.0002, "elapsed_sec": 16057.87239909172, "step_time_sec": 8.238284570979886, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1937, "loss": 4.960286617279053, "lr": 0.0002, "elapsed_sec": 16066.109602212906, "step_time_sec": 8.237099769990891, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1938, "loss": 4.97169828414917, "lr": 0.0002, "elapsed_sec": 16074.349130153656, "step_time_sec": 8.239302471047267, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1939, "loss": 5.170129299163818, "lr": 0.0002, "elapsed_sec": 16082.588027954102, "step_time_sec": 8.238743784022518, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1940, "loss": 5.106093406677246, "lr": 0.0002, "elapsed_sec": 16090.825077056885, "step_time_sec": 8.236931614927016, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1941, "loss": 4.964963912963867, "lr": 0.0002, "elapsed_sec": 16099.063986063004, "step_time_sec": 8.238704203045927, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1942, "loss": 4.9846720695495605, "lr": 0.0002, "elapsed_sec": 16107.301599740982, "step_time_sec": 8.237488560960628, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1943, "loss": 5.014648914337158, "lr": 0.0002, "elapsed_sec": 16115.540733575821, "step_time_sec": 8.23896338394843, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1944, "loss": 5.02877140045166, "lr": 0.0002, "elapsed_sec": 16123.780441522598, "step_time_sec": 8.239553590887226, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1945, "loss": 5.027828216552734, "lr": 0.0002, "elapsed_sec": 16132.019959688187, "step_time_sec": 8.239324411028065, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1946, "loss": 5.043426513671875, "lr": 0.0002, "elapsed_sec": 16140.259438991547, "step_time_sec": 8.239290633006021, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1947, "loss": 4.887012004852295, "lr": 0.0002, "elapsed_sec": 16148.497231960297, "step_time_sec": 8.237652392010204, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1948, "loss": 4.873836994171143, "lr": 0.0002, "elapsed_sec": 16156.736860513687, "step_time_sec": 8.239457614021376, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1949, "loss": 5.036677360534668, "lr": 0.0002, "elapsed_sec": 16164.975911140442, "step_time_sec": 8.238904618076049, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1950, "loss": 5.050561904907227, "lr": 0.0002, "elapsed_sec": 16173.214126825333, "step_time_sec": 8.238055319990963, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1951, "loss": 4.99212646484375, "lr": 0.0002, "elapsed_sec": 16181.452681303024, "step_time_sec": 8.238403621944599, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1952, "loss": 5.035738468170166, "lr": 0.0002, "elapsed_sec": 16189.689814567566, "step_time_sec": 8.236966993077658, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1953, "loss": 4.835819244384766, "lr": 0.0002, "elapsed_sec": 16197.927699327469, "step_time_sec": 8.23772493202705, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1954, "loss": 4.971149921417236, "lr": 0.0002, "elapsed_sec": 16206.166517496109, "step_time_sec": 8.238726845011115, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1955, "loss": 5.112158298492432, "lr": 0.0002, "elapsed_sec": 16214.406185388565, "step_time_sec": 8.239452652051114, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1956, "loss": 4.951517105102539, "lr": 0.0002, "elapsed_sec": 16222.644450426102, "step_time_sec": 8.238113279920071, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1957, "loss": 4.832627773284912, "lr": 0.0002, "elapsed_sec": 16230.882584095001, "step_time_sec": 8.237975122057833, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1958, "loss": 4.879690647125244, "lr": 0.0002, "elapsed_sec": 16239.120293140411, "step_time_sec": 8.237498385016806, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1959, "loss": 4.958316802978516, "lr": 0.0002, "elapsed_sec": 16247.359473466873, "step_time_sec": 8.23902314901352, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1960, "loss": 4.955169677734375, "lr": 0.0002, "elapsed_sec": 16255.598269462585, "step_time_sec": 8.238681801012717, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1961, "loss": 4.906167507171631, "lr": 0.0002, "elapsed_sec": 16263.837211370468, "step_time_sec": 8.238777986029163, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1962, "loss": 4.911837577819824, "lr": 0.0002, "elapsed_sec": 16272.074187994003, "step_time_sec": 8.236790817929432, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1963, "loss": 5.040809631347656, "lr": 0.0002, "elapsed_sec": 16280.313623428345, "step_time_sec": 8.239256075932644, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1964, "loss": 5.0807085037231445, "lr": 0.0002, "elapsed_sec": 16288.552742004395, "step_time_sec": 8.238976766937412, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1965, "loss": 5.003354549407959, "lr": 0.0002, "elapsed_sec": 16296.790912151337, "step_time_sec": 8.238034127978608, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1966, "loss": 4.977692127227783, "lr": 0.0002, "elapsed_sec": 16305.030101537704, "step_time_sec": 8.23900008504279, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1967, "loss": 4.924502849578857, "lr": 0.0002, "elapsed_sec": 16313.26756978035, "step_time_sec": 8.23728928703349, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1968, "loss": 4.970996856689453, "lr": 0.0002, "elapsed_sec": 16321.506492137909, "step_time_sec": 8.23878420307301, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1969, "loss": 5.0894856452941895, "lr": 0.0002, "elapsed_sec": 16329.74450969696, "step_time_sec": 8.237850868026726, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1970, "loss": 4.95419979095459, "lr": 0.0002, "elapsed_sec": 16337.983915090561, "step_time_sec": 8.239239039947279, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1971, "loss": 4.974137783050537, "lr": 0.0002, "elapsed_sec": 16346.22281718254, "step_time_sec": 8.238793801050633, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1972, "loss": 5.046050071716309, "lr": 0.0002, "elapsed_sec": 16354.46028137207, "step_time_sec": 8.237238651956432, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1973, "loss": 5.013371467590332, "lr": 0.0002, "elapsed_sec": 16362.698413133621, "step_time_sec": 8.238005077000707, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1974, "loss": 5.1485137939453125, "lr": 0.0002, "elapsed_sec": 16370.93689918518, "step_time_sec": 8.238337916089222, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1975, "loss": 5.094007968902588, "lr": 0.0002, "elapsed_sec": 16379.175598144531, "step_time_sec": 8.238509062910452, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1976, "loss": 5.009097576141357, "lr": 0.0002, "elapsed_sec": 16387.413304805756, "step_time_sec": 8.237522663082927, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1977, "loss": 5.035758972167969, "lr": 0.0002, "elapsed_sec": 16395.65118098259, "step_time_sec": 8.237769411993213, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1978, "loss": 4.9398112297058105, "lr": 0.0002, "elapsed_sec": 16403.889803171158, "step_time_sec": 8.238464310998097, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1979, "loss": 4.883821964263916, "lr": 0.0002, "elapsed_sec": 16412.127360105515, "step_time_sec": 8.237357674981467, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1980, "loss": 4.984462738037109, "lr": 0.0002, "elapsed_sec": 16420.365221738815, "step_time_sec": 8.237697473959997, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1981, "loss": 4.98663854598999, "lr": 0.0002, "elapsed_sec": 16428.602840662003, "step_time_sec": 8.237502980977297, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1982, "loss": 5.006802558898926, "lr": 0.0002, "elapsed_sec": 16436.840836763382, "step_time_sec": 8.237813640967943, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1983, "loss": 5.081171035766602, "lr": 0.0002, "elapsed_sec": 16445.078847885132, "step_time_sec": 8.237823446979746, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1984, "loss": 4.943243980407715, "lr": 0.0002, "elapsed_sec": 16453.317247867584, "step_time_sec": 8.238258698955178, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1985, "loss": 4.94022274017334, "lr": 0.0002, "elapsed_sec": 16461.556487083435, "step_time_sec": 8.239060855004936, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1986, "loss": 5.017919540405273, "lr": 0.0002, "elapsed_sec": 16469.79483938217, "step_time_sec": 8.238197885919362, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1987, "loss": 4.994764804840088, "lr": 0.0002, "elapsed_sec": 16478.03369450569, "step_time_sec": 8.238693795981817, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1988, "loss": 4.870893955230713, "lr": 0.0002, "elapsed_sec": 16486.272770643234, "step_time_sec": 8.238947859033942, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1989, "loss": 5.4555792808532715, "lr": 0.0002, "elapsed_sec": 16494.509765625, "step_time_sec": 8.236801175051369, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1990, "loss": 4.970239639282227, "lr": 0.0002, "elapsed_sec": 16502.748862981796, "step_time_sec": 8.238913081004284, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1991, "loss": 5.048379898071289, "lr": 0.0002, "elapsed_sec": 16510.98723101616, "step_time_sec": 8.238205465022475, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1992, "loss": 4.919727325439453, "lr": 0.0002, "elapsed_sec": 16519.226324796677, "step_time_sec": 8.238927382975817, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1993, "loss": 5.133878707885742, "lr": 0.0002, "elapsed_sec": 16527.4655752182, "step_time_sec": 8.23907526198309, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1994, "loss": 5.031689643859863, "lr": 0.0002, "elapsed_sec": 16535.70362472534, "step_time_sec": 8.237911241012625, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1995, "loss": 4.80579137802124, "lr": 0.0002, "elapsed_sec": 16543.944277524948, "step_time_sec": 8.240510168951005, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1996, "loss": 4.954228401184082, "lr": 0.0002, "elapsed_sec": 16552.183574438095, "step_time_sec": 8.239128320012242, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1997, "loss": 5.022448539733887, "lr": 0.0002, "elapsed_sec": 16560.42345905304, "step_time_sec": 8.239685097010806, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1998, "loss": 5.013780117034912, "lr": 0.0002, "elapsed_sec": 16568.663101911545, "step_time_sec": 8.239478674950078, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1999, "loss": 4.928530693054199, "lr": 0.0002, "elapsed_sec": 16576.903294801712, "step_time_sec": 8.2400087909773, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2000, "loss": 4.950565338134766, "lr": 0.0002, "elapsed_sec": 16585.14266037941, "step_time_sec": 56.58917029690929, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": true, "validation_ran": true, "probe_ran": true, "probe_elapsed_sec": 0.9771755130495876, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2001, "loss": 4.904843330383301, "lr": 0.0002, "elapsed_sec": 16641.732122659683, "step_time_sec": 8.239410973968916, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2002, "loss": 5.010636329650879, "lr": 0.0002, "elapsed_sec": 16649.96468615532, "step_time_sec": 8.23238494002726, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2003, "loss": 4.859950065612793, "lr": 0.0002, "elapsed_sec": 16658.197249174118, "step_time_sec": 8.232384253060445, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2004, "loss": 4.9286017417907715, "lr": 0.0002, "elapsed_sec": 16666.429627895355, "step_time_sec": 8.232282018987462, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2005, "loss": 4.925117015838623, "lr": 0.0002, "elapsed_sec": 16674.662600517273, "step_time_sec": 8.232779593090527, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2006, "loss": 5.024764060974121, "lr": 0.0002, "elapsed_sec": 16682.90027666092, "step_time_sec": 8.237484687939286, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2007, "loss": 4.957297325134277, "lr": 0.0002, "elapsed_sec": 16691.139452695847, "step_time_sec": 8.23903109098319, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2008, "loss": 5.140630722045898, "lr": 0.0002, "elapsed_sec": 16699.378475427628, "step_time_sec": 8.238852582057007, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2009, "loss": 4.9238433837890625, "lr": 0.0002, "elapsed_sec": 16707.61099100113, "step_time_sec": 8.23238213302102, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2010, "loss": 4.919071197509766, "lr": 0.0002, "elapsed_sec": 16715.85012435913, "step_time_sec": 8.238928931998089, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2011, "loss": 4.943631172180176, "lr": 0.0002, "elapsed_sec": 16724.08722090721, "step_time_sec": 8.23695133603178, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2012, "loss": 4.9293904304504395, "lr": 0.0002, "elapsed_sec": 16732.326459407806, "step_time_sec": 8.239034184021875, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2013, "loss": 5.021457672119141, "lr": 0.0002, "elapsed_sec": 16740.5651140213, "step_time_sec": 8.23851813399233, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2014, "loss": 4.974385738372803, "lr": 0.0002, "elapsed_sec": 16748.802798748016, "step_time_sec": 8.237504452001303, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2015, "loss": 4.935319900512695, "lr": 0.0002, "elapsed_sec": 16757.040749549866, "step_time_sec": 8.237796258064918, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2016, "loss": 5.020432949066162, "lr": 0.0002, "elapsed_sec": 16765.27912044525, "step_time_sec": 8.23820941708982, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2017, "loss": 4.955202579498291, "lr": 0.0002, "elapsed_sec": 16773.51919388771, "step_time_sec": 8.239931678050198, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2018, "loss": 4.999904632568359, "lr": 0.0002, "elapsed_sec": 16781.758784770966, "step_time_sec": 8.239428595989011, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2019, "loss": 4.916734218597412, "lr": 0.0002, "elapsed_sec": 16789.996696710587, "step_time_sec": 8.237740376032889, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2020, "loss": 4.946156978607178, "lr": 0.0002, "elapsed_sec": 16798.235854625702, "step_time_sec": 8.23902450595051, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2021, "loss": 4.880692958831787, "lr": 0.0002, "elapsed_sec": 16806.473680257797, "step_time_sec": 8.237642153981142, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2022, "loss": 4.911525249481201, "lr": 0.0002, "elapsed_sec": 16814.71482372284, "step_time_sec": 8.240985712967813, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2023, "loss": 4.859903335571289, "lr": 0.0002, "elapsed_sec": 16822.954691886902, "step_time_sec": 8.239667164976709, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2024, "loss": 4.970463275909424, "lr": 0.0002, "elapsed_sec": 16831.19165444374, "step_time_sec": 8.236795935081318, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2025, "loss": 4.923557758331299, "lr": 0.0002, "elapsed_sec": 16839.431970596313, "step_time_sec": 8.240154249011539, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2026, "loss": 4.87526798248291, "lr": 0.0002, "elapsed_sec": 16847.67082953453, "step_time_sec": 8.238702769973315, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2027, "loss": 4.953964710235596, "lr": 0.0002, "elapsed_sec": 16855.91006422043, "step_time_sec": 8.239105446031317, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2028, "loss": 4.925581455230713, "lr": 0.0002, "elapsed_sec": 16864.1495950222, "step_time_sec": 8.239399223937653, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2029, "loss": 4.981436252593994, "lr": 0.0002, "elapsed_sec": 16872.38709783554, "step_time_sec": 8.237263021990657, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2030, "loss": 4.901494979858398, "lr": 0.0002, "elapsed_sec": 16880.627922534943, "step_time_sec": 8.240729164914228, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2031, "loss": 4.8356614112854, "lr": 0.0002, "elapsed_sec": 16888.866607904434, "step_time_sec": 8.238479470019229, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2032, "loss": 4.953188419342041, "lr": 0.0002, "elapsed_sec": 16897.105867385864, "step_time_sec": 8.23911846196279, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2033, "loss": 4.926445960998535, "lr": 0.0002, "elapsed_sec": 16905.34520959854, "step_time_sec": 8.239181490032934, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2034, "loss": 4.987549304962158, "lr": 0.0002, "elapsed_sec": 16913.584057092667, "step_time_sec": 8.238704631919973, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2035, "loss": 4.94258451461792, "lr": 0.0002, "elapsed_sec": 16921.823670625687, "step_time_sec": 8.23941498703789, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2036, "loss": 5.047349452972412, "lr": 0.0002, "elapsed_sec": 16930.062378168106, "step_time_sec": 8.23853207693901, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2037, "loss": 5.118734836578369, "lr": 0.0002, "elapsed_sec": 16938.30221939087, "step_time_sec": 8.239730719011277, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2038, "loss": 5.0296244621276855, "lr": 0.0002, "elapsed_sec": 16946.542948007584, "step_time_sec": 8.240574408089742, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2039, "loss": 4.928855895996094, "lr": 0.0002, "elapsed_sec": 16954.78252840042, "step_time_sec": 8.239370647002943, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2040, "loss": 4.942481994628906, "lr": 0.0002, "elapsed_sec": 16963.021738052368, "step_time_sec": 8.23906641791109, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2041, "loss": 4.939273834228516, "lr": 0.0002, "elapsed_sec": 16971.26154279709, "step_time_sec": 8.239641848951578, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2042, "loss": 5.033320426940918, "lr": 0.0002, "elapsed_sec": 16979.49952363968, "step_time_sec": 8.237889020005241, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2043, "loss": 4.927199363708496, "lr": 0.0002, "elapsed_sec": 16987.739898204803, "step_time_sec": 8.240099819027819, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2044, "loss": 4.9426727294921875, "lr": 0.0002, "elapsed_sec": 16995.979932785034, "step_time_sec": 8.239931219024584, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2045, "loss": 5.095893383026123, "lr": 0.0002, "elapsed_sec": 17004.219092607498, "step_time_sec": 8.238985868985765, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2046, "loss": 5.045770168304443, "lr": 0.0002, "elapsed_sec": 17012.4586956501, "step_time_sec": 8.2393934589345, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2047, "loss": 4.84870719909668, "lr": 0.0002, "elapsed_sec": 17020.698666095734, "step_time_sec": 8.23984100006055, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2048, "loss": 4.903833389282227, "lr": 0.0002, "elapsed_sec": 17028.93841791153, "step_time_sec": 8.239585219998844, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2049, "loss": 4.958150863647461, "lr": 0.0002, "elapsed_sec": 17037.17835044861, "step_time_sec": 8.239742542966269, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2050, "loss": 5.058640003204346, "lr": 0.0002, "elapsed_sec": 17045.41604948044, "step_time_sec": 8.237556206993759, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2051, "loss": 4.9243597984313965, "lr": 0.0002, "elapsed_sec": 17053.656685113907, "step_time_sec": 8.240487854927778, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2052, "loss": 4.980607032775879, "lr": 0.0002, "elapsed_sec": 17061.895134687424, "step_time_sec": 8.238297396106645, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2053, "loss": 4.909254550933838, "lr": 0.0002, "elapsed_sec": 17070.13525247574, "step_time_sec": 8.239924847963266, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2054, "loss": 4.91591739654541, "lr": 0.0002, "elapsed_sec": 17078.374962568283, "step_time_sec": 8.23952791991178, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2055, "loss": 4.892734050750732, "lr": 0.0002, "elapsed_sec": 17086.613917589188, "step_time_sec": 8.238811659044586, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2056, "loss": 4.815364360809326, "lr": 0.0002, "elapsed_sec": 17094.85386657715, "step_time_sec": 8.239775402937084, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2057, "loss": 4.759912490844727, "lr": 0.0002, "elapsed_sec": 17103.09359049797, "step_time_sec": 8.239542062045075, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2058, "loss": 5.000146865844727, "lr": 0.0002, "elapsed_sec": 17111.332124710083, "step_time_sec": 8.238364098942839, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2059, "loss": 5.017196178436279, "lr": 0.0002, "elapsed_sec": 17119.572496652603, "step_time_sec": 8.240234380005859, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2060, "loss": 4.986390590667725, "lr": 0.0002, "elapsed_sec": 17127.81026506424, "step_time_sec": 8.23759470100049, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2061, "loss": 5.076152324676514, "lr": 0.0002, "elapsed_sec": 17136.05091691017, "step_time_sec": 8.240547058987431, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2062, "loss": 4.882349967956543, "lr": 0.0002, "elapsed_sec": 17144.291101694107, "step_time_sec": 8.239987538079731, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2063, "loss": 4.907930850982666, "lr": 0.0002, "elapsed_sec": 17152.528109788895, "step_time_sec": 8.23686640302185, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2064, "loss": 4.900290012359619, "lr": 0.0002, "elapsed_sec": 17160.76823616028, "step_time_sec": 8.23997477593366, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2065, "loss": 5.086536884307861, "lr": 0.0002, "elapsed_sec": 17169.006175994873, "step_time_sec": 8.237713730079122, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2066, "loss": 4.89257287979126, "lr": 0.0002, "elapsed_sec": 17177.2452981472, "step_time_sec": 8.238943955977447, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2067, "loss": 4.9206743240356445, "lr": 0.0002, "elapsed_sec": 17185.485197782516, "step_time_sec": 8.239759592921473, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2068, "loss": 4.7415947914123535, "lr": 0.0002, "elapsed_sec": 17193.72265124321, "step_time_sec": 8.237304887035862, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2069, "loss": 4.9369964599609375, "lr": 0.0002, "elapsed_sec": 17201.961235761642, "step_time_sec": 8.23844433692284, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2070, "loss": 5.049741744995117, "lr": 0.0002, "elapsed_sec": 17210.19920039177, "step_time_sec": 8.237780948984437, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2071, "loss": 4.823147296905518, "lr": 0.0002, "elapsed_sec": 17218.438204050064, "step_time_sec": 8.238839620957151, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2072, "loss": 4.928022384643555, "lr": 0.0002, "elapsed_sec": 17226.678111553192, "step_time_sec": 8.239791175932623, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2073, "loss": 4.9676923751831055, "lr": 0.0002, "elapsed_sec": 17234.917715787888, "step_time_sec": 8.239405578933656, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2074, "loss": 4.978610038757324, "lr": 0.0002, "elapsed_sec": 17243.156096935272, "step_time_sec": 8.238187162904069, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2075, "loss": 4.874540328979492, "lr": 0.0002, "elapsed_sec": 17251.39530801773, "step_time_sec": 8.239055498037487, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2076, "loss": 5.0473785400390625, "lr": 0.0002, "elapsed_sec": 17259.63501238823, "step_time_sec": 8.239547687000595, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2077, "loss": 5.022682189941406, "lr": 0.0002, "elapsed_sec": 17267.874462127686, "step_time_sec": 8.239294178085402, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2078, "loss": 4.941969394683838, "lr": 0.0002, "elapsed_sec": 17276.112682819366, "step_time_sec": 8.238091564038768, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2079, "loss": 4.901291847229004, "lr": 0.0002, "elapsed_sec": 17284.352116584778, "step_time_sec": 8.239202940021642, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2080, "loss": 4.760471820831299, "lr": 0.0002, "elapsed_sec": 17292.59188723564, "step_time_sec": 8.239598348038271, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2081, "loss": 4.929256916046143, "lr": 0.0002, "elapsed_sec": 17300.831326961517, "step_time_sec": 8.239280975889415, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2082, "loss": 5.034232139587402, "lr": 0.0002, "elapsed_sec": 17309.068780899048, "step_time_sec": 8.237369903945364, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2083, "loss": 4.988807678222656, "lr": 0.0002, "elapsed_sec": 17317.306642770767, "step_time_sec": 8.237623929046094, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2084, "loss": 4.934765338897705, "lr": 0.0002, "elapsed_sec": 17325.54617547989, "step_time_sec": 8.239351741969585, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2085, "loss": 4.986397743225098, "lr": 0.0002, "elapsed_sec": 17333.7832300663, "step_time_sec": 8.23686153406743, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2086, "loss": 4.8579301834106445, "lr": 0.0002, "elapsed_sec": 17342.021504878998, "step_time_sec": 8.238164111971855, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2087, "loss": 4.886356830596924, "lr": 0.0002, "elapsed_sec": 17350.260102510452, "step_time_sec": 8.23839359905105, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2088, "loss": 4.7953596115112305, "lr": 0.0002, "elapsed_sec": 17358.4984292984, "step_time_sec": 8.238213118049316, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2089, "loss": 4.884156703948975, "lr": 0.0002, "elapsed_sec": 17366.73931288719, "step_time_sec": 8.240699390997179, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2090, "loss": 4.861311912536621, "lr": 0.0002, "elapsed_sec": 17374.978927373886, "step_time_sec": 8.239406569977291, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2091, "loss": 4.86945104598999, "lr": 0.0002, "elapsed_sec": 17383.218121528625, "step_time_sec": 8.239022335968912, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2092, "loss": 4.906793594360352, "lr": 0.0002, "elapsed_sec": 17391.456352233887, "step_time_sec": 8.238097238005139, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2093, "loss": 4.810602188110352, "lr": 0.0002, "elapsed_sec": 17399.69429588318, "step_time_sec": 8.237817465909757, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2094, "loss": 4.940496444702148, "lr": 0.0002, "elapsed_sec": 17407.934233903885, "step_time_sec": 8.239726627012715, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2095, "loss": 4.841550827026367, "lr": 0.0002, "elapsed_sec": 17416.172947645187, "step_time_sec": 8.238625665078871, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2096, "loss": 4.831644058227539, "lr": 0.0002, "elapsed_sec": 17424.411418437958, "step_time_sec": 8.238194352015853, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2097, "loss": 4.910651683807373, "lr": 0.0002, "elapsed_sec": 17432.64983534813, "step_time_sec": 8.238276990945451, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2098, "loss": 4.917787075042725, "lr": 0.0002, "elapsed_sec": 17440.887090206146, "step_time_sec": 8.237074871896766, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2099, "loss": 4.943055152893066, "lr": 0.0002, "elapsed_sec": 17449.1268055439, "step_time_sec": 8.239578244974837, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2100, "loss": 4.872707843780518, "lr": 0.0002, "elapsed_sec": 17457.36520409584, "step_time_sec": 8.238235874916427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2101, "loss": 4.852238655090332, "lr": 0.0002, "elapsed_sec": 17465.60390663147, "step_time_sec": 8.238562315003946, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2102, "loss": 4.830532550811768, "lr": 0.0002, "elapsed_sec": 17473.844218730927, "step_time_sec": 8.240192663972266, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2103, "loss": 4.847853183746338, "lr": 0.0002, "elapsed_sec": 17482.083259820938, "step_time_sec": 8.238868451095186, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2104, "loss": 5.013761520385742, "lr": 0.0002, "elapsed_sec": 17490.32331109047, "step_time_sec": 8.239882371970452, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2105, "loss": 4.862973213195801, "lr": 0.0002, "elapsed_sec": 17498.56115102768, "step_time_sec": 8.23762594803702, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2106, "loss": 4.885683536529541, "lr": 0.0002, "elapsed_sec": 17506.79952287674, "step_time_sec": 8.238274274975993, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2107, "loss": 4.8680243492126465, "lr": 0.0002, "elapsed_sec": 17515.03758072853, "step_time_sec": 8.237848302000202, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2108, "loss": 4.953861713409424, "lr": 0.0002, "elapsed_sec": 17523.27564930916, "step_time_sec": 8.237897244049236, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2109, "loss": 4.948950290679932, "lr": 0.0002, "elapsed_sec": 17531.514823675156, "step_time_sec": 8.239010727964342, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2110, "loss": 4.895938873291016, "lr": 0.0002, "elapsed_sec": 17539.755089759827, "step_time_sec": 8.240142071037553, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2111, "loss": 5.028153896331787, "lr": 0.0002, "elapsed_sec": 17547.993817329407, "step_time_sec": 8.238548594992608, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2112, "loss": 4.843321800231934, "lr": 0.0002, "elapsed_sec": 17556.232506513596, "step_time_sec": 8.238570883055218, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2113, "loss": 4.934185981750488, "lr": 0.0002, "elapsed_sec": 17564.469804286957, "step_time_sec": 8.237055926932953, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2114, "loss": 4.978735446929932, "lr": 0.0002, "elapsed_sec": 17572.710048675537, "step_time_sec": 8.240120528033003, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2115, "loss": 5.023430824279785, "lr": 0.0002, "elapsed_sec": 17580.94929242134, "step_time_sec": 8.23907206300646, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2116, "loss": 4.726180553436279, "lr": 0.0002, "elapsed_sec": 17589.187883615494, "step_time_sec": 8.238391752005555, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2117, "loss": 4.911501884460449, "lr": 0.0002, "elapsed_sec": 17597.428166627884, "step_time_sec": 8.24011786095798, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2118, "loss": 4.922548294067383, "lr": 0.0002, "elapsed_sec": 17605.667160511017, "step_time_sec": 8.238869271008298, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2119, "loss": 4.949389934539795, "lr": 0.0002, "elapsed_sec": 17613.905701637268, "step_time_sec": 8.238337384071201, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2120, "loss": 4.853734016418457, "lr": 0.0002, "elapsed_sec": 17622.14591407776, "step_time_sec": 8.240120572969317, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2121, "loss": 4.864793300628662, "lr": 0.0002, "elapsed_sec": 17630.383733272552, "step_time_sec": 8.237600709078833, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2122, "loss": 4.73187780380249, "lr": 0.0002, "elapsed_sec": 17638.623086214066, "step_time_sec": 8.23920380405616, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2123, "loss": 4.868164539337158, "lr": 0.0002, "elapsed_sec": 17646.86225795746, "step_time_sec": 8.239018876920454, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2124, "loss": 4.869239330291748, "lr": 0.0002, "elapsed_sec": 17655.101959228516, "step_time_sec": 8.23957415507175, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2125, "loss": 4.955258846282959, "lr": 0.0002, "elapsed_sec": 17663.34250020981, "step_time_sec": 8.24032394203823, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2126, "loss": 4.973252773284912, "lr": 0.0002, "elapsed_sec": 17671.581104278564, "step_time_sec": 8.238419792964123, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2127, "loss": 5.072171688079834, "lr": 0.0002, "elapsed_sec": 17679.821083068848, "step_time_sec": 8.239842574927025, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2128, "loss": 4.943661212921143, "lr": 0.0002, "elapsed_sec": 17688.059275388718, "step_time_sec": 8.237997851916589, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2129, "loss": 4.959967136383057, "lr": 0.0002, "elapsed_sec": 17696.298272371292, "step_time_sec": 8.238868406042457, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2130, "loss": 4.971222400665283, "lr": 0.0002, "elapsed_sec": 17704.536127090454, "step_time_sec": 8.237659464939497, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2131, "loss": 4.7789082527160645, "lr": 0.0002, "elapsed_sec": 17712.774395942688, "step_time_sec": 8.238130533020012, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2132, "loss": 4.841586589813232, "lr": 0.0002, "elapsed_sec": 17721.01475906372, "step_time_sec": 8.240171240991913, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2133, "loss": 4.805880546569824, "lr": 0.0002, "elapsed_sec": 17729.253608703613, "step_time_sec": 8.23869584791828, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2134, "loss": 4.896736145019531, "lr": 0.0002, "elapsed_sec": 17737.49246287346, "step_time_sec": 8.238672326086089, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2135, "loss": 4.80053186416626, "lr": 0.0002, "elapsed_sec": 17745.73211121559, "step_time_sec": 8.239483006997034, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2136, "loss": 4.958138465881348, "lr": 0.0002, "elapsed_sec": 17753.968513727188, "step_time_sec": 8.23625878803432, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2137, "loss": 4.8044939041137695, "lr": 0.0002, "elapsed_sec": 17762.208584547043, "step_time_sec": 8.239941980107687, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2138, "loss": 4.876369476318359, "lr": 0.0002, "elapsed_sec": 17770.448274850845, "step_time_sec": 8.23951327893883, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2139, "loss": 4.959245204925537, "lr": 0.0002, "elapsed_sec": 17778.686453819275, "step_time_sec": 8.238018534029834, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2140, "loss": 4.884952068328857, "lr": 0.0002, "elapsed_sec": 17786.925751447678, "step_time_sec": 8.239081804989837, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2141, "loss": 4.873982906341553, "lr": 0.0002, "elapsed_sec": 17795.163982391357, "step_time_sec": 8.238079325994477, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2142, "loss": 4.832751750946045, "lr": 0.0002, "elapsed_sec": 17803.40531396866, "step_time_sec": 8.241153256036341, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2143, "loss": 4.950920581817627, "lr": 0.0002, "elapsed_sec": 17811.644911289215, "step_time_sec": 8.239441673038527, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2144, "loss": 4.979039192199707, "lr": 0.0002, "elapsed_sec": 17819.8818359375, "step_time_sec": 8.236800718004815, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2145, "loss": 4.875493049621582, "lr": 0.0002, "elapsed_sec": 17828.12169766426, "step_time_sec": 8.239691979950294, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2146, "loss": 4.752832412719727, "lr": 0.0002, "elapsed_sec": 17836.36035823822, "step_time_sec": 8.238532862975262, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2147, "loss": 4.928502559661865, "lr": 0.0002, "elapsed_sec": 17844.60052371025, "step_time_sec": 8.239949795999564, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2148, "loss": 4.935489654541016, "lr": 0.0002, "elapsed_sec": 17852.83998823166, "step_time_sec": 8.23934022092726, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2149, "loss": 4.943672180175781, "lr": 0.0002, "elapsed_sec": 17861.078944444656, "step_time_sec": 8.238751702010632, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2150, "loss": 4.815985679626465, "lr": 0.0002, "elapsed_sec": 17869.319219589233, "step_time_sec": 8.24010828207247, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2151, "loss": 4.867344856262207, "lr": 0.0002, "elapsed_sec": 17877.558815717697, "step_time_sec": 8.23940667300485, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2152, "loss": 4.861866474151611, "lr": 0.0002, "elapsed_sec": 17885.797171115875, "step_time_sec": 8.238200488965958, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2153, "loss": 4.835839748382568, "lr": 0.0002, "elapsed_sec": 17894.03610944748, "step_time_sec": 8.238782830070704, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2154, "loss": 4.775817394256592, "lr": 0.0002, "elapsed_sec": 17902.274740457535, "step_time_sec": 8.238485832000151, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2155, "loss": 4.772789478302002, "lr": 0.0002, "elapsed_sec": 17910.51446247101, "step_time_sec": 8.239555616048165, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2156, "loss": 4.963638782501221, "lr": 0.0002, "elapsed_sec": 17918.75232720375, "step_time_sec": 8.237644960987382, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2157, "loss": 4.6729607582092285, "lr": 0.0002, "elapsed_sec": 17926.990017175674, "step_time_sec": 8.237544753937982, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2158, "loss": 4.997402191162109, "lr": 0.0002, "elapsed_sec": 17935.229630470276, "step_time_sec": 8.239474497037008, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2159, "loss": 4.765285491943359, "lr": 0.0002, "elapsed_sec": 17943.466392993927, "step_time_sec": 8.236542895087041, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2160, "loss": 4.949280738830566, "lr": 0.0002, "elapsed_sec": 17951.705319166183, "step_time_sec": 8.238787668989971, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2161, "loss": 4.901760578155518, "lr": 0.0002, "elapsed_sec": 17959.944000959396, "step_time_sec": 8.238530411967076, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2162, "loss": 4.908939838409424, "lr": 0.0002, "elapsed_sec": 17968.182307720184, "step_time_sec": 8.23812780994922, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2163, "loss": 4.997348785400391, "lr": 0.0002, "elapsed_sec": 17976.42124724388, "step_time_sec": 8.238858896074817, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2164, "loss": 4.845398426055908, "lr": 0.0002, "elapsed_sec": 17984.65936923027, "step_time_sec": 8.237885100999847, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2165, "loss": 4.766848564147949, "lr": 0.0002, "elapsed_sec": 17992.89909219742, "step_time_sec": 8.23956995701883, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2166, "loss": 4.811318874359131, "lr": 0.0002, "elapsed_sec": 18001.137652873993, "step_time_sec": 8.238457041908987, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2167, "loss": 4.826196193695068, "lr": 0.0002, "elapsed_sec": 18009.376836776733, "step_time_sec": 8.238988175056875, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2168, "loss": 4.856212615966797, "lr": 0.0002, "elapsed_sec": 18017.61668395996, "step_time_sec": 8.239701497950591, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2169, "loss": 4.814630508422852, "lr": 0.0002, "elapsed_sec": 18025.855260372162, "step_time_sec": 8.238383891992271, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2170, "loss": 4.808464527130127, "lr": 0.0002, "elapsed_sec": 18034.0935254097, "step_time_sec": 8.238139987923205, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2171, "loss": 4.8082475662231445, "lr": 0.0002, "elapsed_sec": 18042.33131456375, "step_time_sec": 8.23762191098649, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2172, "loss": 4.818228244781494, "lr": 0.0002, "elapsed_sec": 18050.570596456528, "step_time_sec": 8.23911191697698, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2173, "loss": 4.864802360534668, "lr": 0.0002, "elapsed_sec": 18058.8110370636, "step_time_sec": 8.240271868999116, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2174, "loss": 5.001303195953369, "lr": 0.0002, "elapsed_sec": 18067.049024820328, "step_time_sec": 8.23780872207135, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2175, "loss": 4.874752044677734, "lr": 0.0002, "elapsed_sec": 18075.28809118271, "step_time_sec": 8.238942486001179, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2176, "loss": 4.794112205505371, "lr": 0.0002, "elapsed_sec": 18083.528779029846, "step_time_sec": 8.240488125942647, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2177, "loss": 4.936871528625488, "lr": 0.0002, "elapsed_sec": 18091.766909360886, "step_time_sec": 8.237979877041653, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2178, "loss": 4.901277542114258, "lr": 0.0002, "elapsed_sec": 18100.005968809128, "step_time_sec": 8.2389055629028, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2179, "loss": 4.880674362182617, "lr": 0.0002, "elapsed_sec": 18108.244619607925, "step_time_sec": 8.2385242349701, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2180, "loss": 4.806257724761963, "lr": 0.0002, "elapsed_sec": 18116.484577655792, "step_time_sec": 8.239806041005068, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2181, "loss": 4.884339332580566, "lr": 0.0002, "elapsed_sec": 18124.724000692368, "step_time_sec": 8.239210169063881, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2182, "loss": 4.779458999633789, "lr": 0.0002, "elapsed_sec": 18132.962540626526, "step_time_sec": 8.238374997046776, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2183, "loss": 4.792176246643066, "lr": 0.0002, "elapsed_sec": 18141.203880786896, "step_time_sec": 8.241190135013312, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2184, "loss": 4.909801483154297, "lr": 0.0002, "elapsed_sec": 18149.44238114357, "step_time_sec": 8.238355005043559, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2185, "loss": 4.851511478424072, "lr": 0.0002, "elapsed_sec": 18157.68223285675, "step_time_sec": 8.23966942797415, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2186, "loss": 4.912481784820557, "lr": 0.0002, "elapsed_sec": 18165.92191028595, "step_time_sec": 8.239489600993693, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2187, "loss": 4.905263900756836, "lr": 0.0002, "elapsed_sec": 18174.160045146942, "step_time_sec": 8.237973745912313, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2188, "loss": 5.045092582702637, "lr": 0.0002, "elapsed_sec": 18182.401900291443, "step_time_sec": 8.241721839061938, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2189, "loss": 4.879006862640381, "lr": 0.0002, "elapsed_sec": 18190.640544891357, "step_time_sec": 8.23844919691328, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2190, "loss": 4.788090229034424, "lr": 0.0002, "elapsed_sec": 18198.878336429596, "step_time_sec": 8.237673914059997, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2191, "loss": 4.742589950561523, "lr": 0.0002, "elapsed_sec": 18207.11657691002, "step_time_sec": 8.238039913936518, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2192, "loss": 4.878084182739258, "lr": 0.0002, "elapsed_sec": 18215.35327720642, "step_time_sec": 8.236546758096665, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2193, "loss": 4.8599629402160645, "lr": 0.0002, "elapsed_sec": 18223.592567443848, "step_time_sec": 8.239146591047756, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2194, "loss": 4.8016862869262695, "lr": 0.0002, "elapsed_sec": 18231.83138537407, "step_time_sec": 8.238631628919393, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2195, "loss": 4.8210577964782715, "lr": 0.0002, "elapsed_sec": 18240.06959915161, "step_time_sec": 8.238032154040411, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2196, "loss": 4.864475250244141, "lr": 0.0002, "elapsed_sec": 18248.310133695602, "step_time_sec": 8.24042103602551, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2197, "loss": 4.835540771484375, "lr": 0.0002, "elapsed_sec": 18256.547499656677, "step_time_sec": 8.237193809007294, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2198, "loss": 4.691307544708252, "lr": 0.0002, "elapsed_sec": 18264.788313627243, "step_time_sec": 8.240611877990887, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2199, "loss": 4.803158283233643, "lr": 0.0002, "elapsed_sec": 18273.027095794678, "step_time_sec": 8.238610222004354, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2200, "loss": 4.871632099151611, "lr": 0.0002, "elapsed_sec": 18281.265722751617, "step_time_sec": 8.23844663402997, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2201, "loss": 4.8514323234558105, "lr": 0.0002, "elapsed_sec": 18289.506417274475, "step_time_sec": 8.24055462796241, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2202, "loss": 4.90993595123291, "lr": 0.0002, "elapsed_sec": 18297.744971513748, "step_time_sec": 8.238419721019454, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2203, "loss": 4.9127349853515625, "lr": 0.0002, "elapsed_sec": 18305.984692811966, "step_time_sec": 8.239490332081914, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2204, "loss": 4.844099521636963, "lr": 0.0002, "elapsed_sec": 18314.223724126816, "step_time_sec": 8.238864701008424, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2205, "loss": 4.944378852844238, "lr": 0.0002, "elapsed_sec": 18322.462116241455, "step_time_sec": 8.23827329010237, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2206, "loss": 4.7939558029174805, "lr": 0.0002, "elapsed_sec": 18330.70141172409, "step_time_sec": 8.239105563028716, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2207, "loss": 4.810115814208984, "lr": 0.0002, "elapsed_sec": 18338.939497232437, "step_time_sec": 8.237876905011944, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2208, "loss": 4.735218048095703, "lr": 0.0002, "elapsed_sec": 18347.17698264122, "step_time_sec": 8.23730474896729, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2209, "loss": 4.7795023918151855, "lr": 0.0002, "elapsed_sec": 18355.4157435894, "step_time_sec": 8.238653350970708, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2210, "loss": 4.816893100738525, "lr": 0.0002, "elapsed_sec": 18363.654155492783, "step_time_sec": 8.238224697997794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2211, "loss": 4.855317115783691, "lr": 0.0002, "elapsed_sec": 18371.89458322525, "step_time_sec": 8.240266192005947, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2212, "loss": 4.820127964019775, "lr": 0.0002, "elapsed_sec": 18380.132855415344, "step_time_sec": 8.23806063493248, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2213, "loss": 4.771463871002197, "lr": 0.0002, "elapsed_sec": 18388.3731341362, "step_time_sec": 8.240160285960883, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2214, "loss": 4.927245140075684, "lr": 0.0002, "elapsed_sec": 18396.613367795944, "step_time_sec": 8.240056967944838, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2215, "loss": 4.750319480895996, "lr": 0.0002, "elapsed_sec": 18404.85192465782, "step_time_sec": 8.23832290689461, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2216, "loss": 4.849072456359863, "lr": 0.0002, "elapsed_sec": 18413.091048955917, "step_time_sec": 8.238974572974257, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2217, "loss": 4.896397113800049, "lr": 0.0002, "elapsed_sec": 18421.32882809639, "step_time_sec": 8.237627779948525, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2218, "loss": 4.759817600250244, "lr": 0.0002, "elapsed_sec": 18429.568141937256, "step_time_sec": 8.23919070395641, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2219, "loss": 4.731647968292236, "lr": 0.0002, "elapsed_sec": 18437.8080701828, "step_time_sec": 8.239766375976615, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2220, "loss": 4.831627368927002, "lr": 0.0002, "elapsed_sec": 18446.044908761978, "step_time_sec": 8.236687344964594, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2221, "loss": 4.825608730316162, "lr": 0.0002, "elapsed_sec": 18454.285043001175, "step_time_sec": 8.239924009074457, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2222, "loss": 4.9053215980529785, "lr": 0.0002, "elapsed_sec": 18462.523949861526, "step_time_sec": 8.23873465997167, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2223, "loss": 4.757532119750977, "lr": 0.0002, "elapsed_sec": 18470.75674343109, "step_time_sec": 8.232654461055063, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2224, "loss": 4.756869316101074, "lr": 0.0002, "elapsed_sec": 18478.989897489548, "step_time_sec": 8.232987474068068, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2225, "loss": 4.849669933319092, "lr": 0.0002, "elapsed_sec": 18487.22207593918, "step_time_sec": 8.23195801791735, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2226, "loss": 4.823538780212402, "lr": 0.0002, "elapsed_sec": 18495.461033582687, "step_time_sec": 8.238845865009353, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2227, "loss": 4.7919511795043945, "lr": 0.0002, "elapsed_sec": 18503.699538469315, "step_time_sec": 8.238381334929727, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2228, "loss": 4.850564002990723, "lr": 0.0002, "elapsed_sec": 18511.938092947006, "step_time_sec": 8.238341565942392, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2229, "loss": 4.721895217895508, "lr": 0.0002, "elapsed_sec": 18520.175290584564, "step_time_sec": 8.237009553937241, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2230, "loss": 4.8894524574279785, "lr": 0.0002, "elapsed_sec": 18528.414901971817, "step_time_sec": 8.239458369091153, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2231, "loss": 4.763119697570801, "lr": 0.0002, "elapsed_sec": 18536.653653144836, "step_time_sec": 8.238620199030265, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2232, "loss": 4.829745292663574, "lr": 0.0002, "elapsed_sec": 18544.892850399017, "step_time_sec": 8.239001784007996, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2233, "loss": 4.799195766448975, "lr": 0.0002, "elapsed_sec": 18553.1316280365, "step_time_sec": 8.238622139091603, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2234, "loss": 4.794660568237305, "lr": 0.0002, "elapsed_sec": 18561.369313955307, "step_time_sec": 8.237516271066852, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2235, "loss": 4.680501937866211, "lr": 0.0002, "elapsed_sec": 18569.610427379608, "step_time_sec": 8.240938646020368, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2236, "loss": 4.813910007476807, "lr": 0.0002, "elapsed_sec": 18577.84990119934, "step_time_sec": 8.239369941991754, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2237, "loss": 4.788445472717285, "lr": 0.0002, "elapsed_sec": 18586.087510347366, "step_time_sec": 8.237430689972825, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2238, "loss": 4.791430950164795, "lr": 0.0002, "elapsed_sec": 18594.32793354988, "step_time_sec": 8.240212089964189, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2239, "loss": 4.669491767883301, "lr": 0.0002, "elapsed_sec": 18602.56481552124, "step_time_sec": 8.236758637940511, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2240, "loss": 4.82398796081543, "lr": 0.0002, "elapsed_sec": 18610.804371356964, "step_time_sec": 8.23935579508543, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2241, "loss": 4.728701591491699, "lr": 0.0002, "elapsed_sec": 18619.04473233223, "step_time_sec": 8.240189924952574, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2242, "loss": 4.662596702575684, "lr": 0.0002, "elapsed_sec": 18627.283394098282, "step_time_sec": 8.238495249068365, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2243, "loss": 4.845340251922607, "lr": 0.0002, "elapsed_sec": 18635.522880792618, "step_time_sec": 8.23934410198126, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2244, "loss": 4.89625358581543, "lr": 0.0002, "elapsed_sec": 18643.76102924347, "step_time_sec": 8.237996790907346, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2245, "loss": 4.667341709136963, "lr": 0.0002, "elapsed_sec": 18652.00109410286, "step_time_sec": 8.23986411804799, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2246, "loss": 4.805480480194092, "lr": 0.0002, "elapsed_sec": 18660.240384817123, "step_time_sec": 8.239122453029267, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2247, "loss": 4.936251640319824, "lr": 0.0002, "elapsed_sec": 18668.478678941727, "step_time_sec": 8.238149949000217, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2248, "loss": 4.809908390045166, "lr": 0.0002, "elapsed_sec": 18676.718707084656, "step_time_sec": 8.239913059980609, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2249, "loss": 4.871353626251221, "lr": 0.0002, "elapsed_sec": 18684.95815730095, "step_time_sec": 8.239247984951362, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2250, "loss": 4.747663974761963, "lr": 0.0002, "elapsed_sec": 18693.19793176651, "step_time_sec": 8.23959038790781, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2251, "loss": 4.735942363739014, "lr": 0.0002, "elapsed_sec": 18701.43652319908, "step_time_sec": 8.238437806023285, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2252, "loss": 4.809780120849609, "lr": 0.0002, "elapsed_sec": 18709.67388343811, "step_time_sec": 8.237190261948854, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2253, "loss": 4.834644794464111, "lr": 0.0002, "elapsed_sec": 18717.913200855255, "step_time_sec": 8.239189757965505, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2254, "loss": 4.642432689666748, "lr": 0.0002, "elapsed_sec": 18726.152807950974, "step_time_sec": 8.239460577024147, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2255, "loss": 4.838382244110107, "lr": 0.0002, "elapsed_sec": 18734.392466783524, "step_time_sec": 8.239409290021285, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2256, "loss": 4.8295135498046875, "lr": 0.0002, "elapsed_sec": 18742.625881910324, "step_time_sec": 8.233279714011587, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2257, "loss": 4.682231426239014, "lr": 0.0002, "elapsed_sec": 18750.8590528965, "step_time_sec": 8.23300744604785, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2258, "loss": 4.82346248626709, "lr": 0.0002, "elapsed_sec": 18759.093736886978, "step_time_sec": 8.234490024973638, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2259, "loss": 4.792685508728027, "lr": 0.0002, "elapsed_sec": 18767.32729077339, "step_time_sec": 8.233375638024881, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2260, "loss": 4.795947551727295, "lr": 0.0002, "elapsed_sec": 18775.56051182747, "step_time_sec": 8.233073509065434, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2261, "loss": 4.7923994064331055, "lr": 0.0002, "elapsed_sec": 18783.79322910309, "step_time_sec": 8.23255142592825, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2262, "loss": 4.635632514953613, "lr": 0.0002, "elapsed_sec": 18792.032756567, "step_time_sec": 8.239354557008483, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2263, "loss": 4.788885116577148, "lr": 0.0002, "elapsed_sec": 18800.271367549896, "step_time_sec": 8.23845013603568, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2264, "loss": 4.768579483032227, "lr": 0.0002, "elapsed_sec": 18808.51044869423, "step_time_sec": 8.23894804797601, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2265, "loss": 4.778578758239746, "lr": 0.0002, "elapsed_sec": 18816.748874425888, "step_time_sec": 8.238233404932544, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2266, "loss": 4.746028423309326, "lr": 0.0002, "elapsed_sec": 18824.98799419403, "step_time_sec": 8.238959574024193, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2267, "loss": 4.7525200843811035, "lr": 0.0002, "elapsed_sec": 18833.22751379013, "step_time_sec": 8.239381060004234, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2268, "loss": 4.768521785736084, "lr": 0.0002, "elapsed_sec": 18841.466491937637, "step_time_sec": 8.238782164058648, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2269, "loss": 4.945390701293945, "lr": 0.0002, "elapsed_sec": 18849.706412553787, "step_time_sec": 8.23975633806549, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2270, "loss": 4.821176528930664, "lr": 0.0002, "elapsed_sec": 18857.945689439774, "step_time_sec": 8.239131710026413, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2271, "loss": 4.7236104011535645, "lr": 0.0002, "elapsed_sec": 18866.185353040695, "step_time_sec": 8.2395140749868, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2272, "loss": 4.842911720275879, "lr": 0.0002, "elapsed_sec": 18874.42445206642, "step_time_sec": 8.238854703027755, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2273, "loss": 4.815934658050537, "lr": 0.0002, "elapsed_sec": 18882.66237974167, "step_time_sec": 8.237780146999285, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2274, "loss": 4.848452568054199, "lr": 0.0002, "elapsed_sec": 18890.903478622437, "step_time_sec": 8.240955421002582, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2275, "loss": 4.743191242218018, "lr": 0.0002, "elapsed_sec": 18899.142784118652, "step_time_sec": 8.23912809905596, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2276, "loss": 4.704724311828613, "lr": 0.0002, "elapsed_sec": 18907.381741046906, "step_time_sec": 8.238788901944645, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2277, "loss": 4.635433197021484, "lr": 0.0002, "elapsed_sec": 18915.619898557663, "step_time_sec": 8.23805948800873, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2278, "loss": 4.74659538269043, "lr": 0.0002, "elapsed_sec": 18923.858152151108, "step_time_sec": 8.23808863607701, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2279, "loss": 4.770030498504639, "lr": 0.0002, "elapsed_sec": 18932.098796129227, "step_time_sec": 8.240405736025423, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2280, "loss": 4.777792453765869, "lr": 0.0002, "elapsed_sec": 18940.33927464485, "step_time_sec": 8.240321265067905, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2281, "loss": 4.885087490081787, "lr": 0.0002, "elapsed_sec": 18948.57199048996, "step_time_sec": 8.232593614957295, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2282, "loss": 4.727658748626709, "lr": 0.0002, "elapsed_sec": 18956.804475069046, "step_time_sec": 8.232310741092078, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2283, "loss": 4.8185038566589355, "lr": 0.0002, "elapsed_sec": 18965.03684258461, "step_time_sec": 8.23219435790088, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2284, "loss": 4.727950572967529, "lr": 0.0002, "elapsed_sec": 18973.274214982986, "step_time_sec": 8.237234039930627, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2285, "loss": 4.768850326538086, "lr": 0.0002, "elapsed_sec": 18981.51286315918, "step_time_sec": 8.238429257995449, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2286, "loss": 4.771475315093994, "lr": 0.0002, "elapsed_sec": 18989.751195192337, "step_time_sec": 8.238175112986937, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2287, "loss": 4.81589412689209, "lr": 0.0002, "elapsed_sec": 18997.990375757217, "step_time_sec": 8.239060385967605, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2288, "loss": 4.730445384979248, "lr": 0.0002, "elapsed_sec": 19006.229861021042, "step_time_sec": 8.239324633963406, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2289, "loss": 4.681085109710693, "lr": 0.0002, "elapsed_sec": 19014.46980547905, "step_time_sec": 8.239708586013876, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2290, "loss": 4.777790069580078, "lr": 0.0002, "elapsed_sec": 19022.709052801132, "step_time_sec": 8.239107489935122, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2291, "loss": 4.653259754180908, "lr": 0.0002, "elapsed_sec": 19030.948209285736, "step_time_sec": 8.238989304983988, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2292, "loss": 4.737284183502197, "lr": 0.0002, "elapsed_sec": 19039.187435865402, "step_time_sec": 8.239048707997426, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2293, "loss": 4.859063625335693, "lr": 0.0002, "elapsed_sec": 19047.42627763748, "step_time_sec": 8.23866477899719, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2294, "loss": 4.689181804656982, "lr": 0.0002, "elapsed_sec": 19055.664860725403, "step_time_sec": 8.238416836946271, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2295, "loss": 4.853832721710205, "lr": 0.0002, "elapsed_sec": 19063.901887655258, "step_time_sec": 8.2369012509007, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2296, "loss": 4.738153457641602, "lr": 0.0002, "elapsed_sec": 19072.140212535858, "step_time_sec": 8.238143595983274, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2297, "loss": 4.814902305603027, "lr": 0.0002, "elapsed_sec": 19080.37852549553, "step_time_sec": 8.2382153859362, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2298, "loss": 4.799952507019043, "lr": 0.0002, "elapsed_sec": 19088.61731004715, "step_time_sec": 8.238564684055746, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2299, "loss": 4.756276607513428, "lr": 0.0002, "elapsed_sec": 19096.856073617935, "step_time_sec": 8.23862449801527, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2300, "loss": 4.855718612670898, "lr": 0.0002, "elapsed_sec": 19105.09384226799, "step_time_sec": 8.23757118708454, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2301, "loss": 4.7378363609313965, "lr": 0.0002, "elapsed_sec": 19113.33443379402, "step_time_sec": 8.240467465016991, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2302, "loss": 4.719040870666504, "lr": 0.0002, "elapsed_sec": 19121.572825670242, "step_time_sec": 8.238181333988905, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2303, "loss": 4.758952617645264, "lr": 0.0002, "elapsed_sec": 19129.812014341354, "step_time_sec": 8.239015159895644, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2304, "loss": 4.714512825012207, "lr": 0.0002, "elapsed_sec": 19138.0510058403, "step_time_sec": 8.238853763905354, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2305, "loss": 4.887925148010254, "lr": 0.0002, "elapsed_sec": 19146.289359807968, "step_time_sec": 8.23822492302861, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2306, "loss": 4.729275226593018, "lr": 0.0002, "elapsed_sec": 19154.53053212166, "step_time_sec": 8.240981486043893, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2307, "loss": 4.752406120300293, "lr": 0.0002, "elapsed_sec": 19162.769844055176, "step_time_sec": 8.239094863994978, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2308, "loss": 4.72754430770874, "lr": 0.0002, "elapsed_sec": 19171.00923871994, "step_time_sec": 8.239289526944049, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2309, "loss": 4.81292724609375, "lr": 0.0002, "elapsed_sec": 19179.24885392189, "step_time_sec": 8.2394100620877, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2310, "loss": 4.668997764587402, "lr": 0.0002, "elapsed_sec": 19187.488401651382, "step_time_sec": 8.239376503042877, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2311, "loss": 4.898784637451172, "lr": 0.0002, "elapsed_sec": 19195.727470874786, "step_time_sec": 8.238906870014034, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2312, "loss": 4.701839447021484, "lr": 0.0002, "elapsed_sec": 19203.965348243713, "step_time_sec": 8.237762806005776, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2313, "loss": 4.656906604766846, "lr": 0.0002, "elapsed_sec": 19212.20281291008, "step_time_sec": 8.237265349947847, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2314, "loss": 4.796067237854004, "lr": 0.0002, "elapsed_sec": 19220.442669391632, "step_time_sec": 8.2397012100555, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2315, "loss": 4.725703716278076, "lr": 0.0002, "elapsed_sec": 19228.68163871765, "step_time_sec": 8.23879092792049, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2316, "loss": 4.730001449584961, "lr": 0.0002, "elapsed_sec": 19236.92223596573, "step_time_sec": 8.240427310927771, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2317, "loss": 4.638140678405762, "lr": 0.0002, "elapsed_sec": 19245.161232233047, "step_time_sec": 8.238833015901037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2318, "loss": 4.8062310218811035, "lr": 0.0002, "elapsed_sec": 19253.399367809296, "step_time_sec": 8.238017366966233, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2319, "loss": 4.7077507972717285, "lr": 0.0002, "elapsed_sec": 19261.63890314102, "step_time_sec": 8.239378405967727, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2320, "loss": 4.825150012969971, "lr": 0.0002, "elapsed_sec": 19269.877374887466, "step_time_sec": 8.238263655919582, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2321, "loss": 4.832192420959473, "lr": 0.0002, "elapsed_sec": 19278.11687898636, "step_time_sec": 8.239353541051969, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2322, "loss": 4.785722255706787, "lr": 0.0002, "elapsed_sec": 19286.35595202446, "step_time_sec": 8.238946674973704, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2323, "loss": 4.842602729797363, "lr": 0.0002, "elapsed_sec": 19294.594009399414, "step_time_sec": 8.23784695204813, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2324, "loss": 4.736236095428467, "lr": 0.0002, "elapsed_sec": 19302.832533836365, "step_time_sec": 8.238378353998996, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2325, "loss": 4.794769763946533, "lr": 0.0002, "elapsed_sec": 19311.07246041298, "step_time_sec": 8.239761695964262, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2326, "loss": 4.7165937423706055, "lr": 0.0002, "elapsed_sec": 19319.310665607452, "step_time_sec": 8.238046442973427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2327, "loss": 4.670875072479248, "lr": 0.0002, "elapsed_sec": 19327.54948735237, "step_time_sec": 8.238654045038857, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2328, "loss": 4.684144973754883, "lr": 0.0002, "elapsed_sec": 19335.787435531616, "step_time_sec": 8.237781592062674, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2329, "loss": 4.6499199867248535, "lr": 0.0002, "elapsed_sec": 19344.02690052986, "step_time_sec": 8.239355158060789, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2330, "loss": 4.771386623382568, "lr": 0.0002, "elapsed_sec": 19352.264346837997, "step_time_sec": 8.237242325907573, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2331, "loss": 4.763125896453857, "lr": 0.0002, "elapsed_sec": 19360.50315952301, "step_time_sec": 8.238646264071576, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2332, "loss": 4.820764541625977, "lr": 0.0002, "elapsed_sec": 19368.741937160492, "step_time_sec": 8.238652715925127, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2333, "loss": 4.6030683517456055, "lr": 0.0002, "elapsed_sec": 19376.97948861122, "step_time_sec": 8.2373162410222, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2334, "loss": 4.772895812988281, "lr": 0.0002, "elapsed_sec": 19385.219504117966, "step_time_sec": 8.23988113400992, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2335, "loss": 4.69044828414917, "lr": 0.0002, "elapsed_sec": 19393.458232164383, "step_time_sec": 8.238582231104374, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2336, "loss": 4.848067760467529, "lr": 0.0002, "elapsed_sec": 19401.69753217697, "step_time_sec": 8.239125201944262, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2337, "loss": 4.555241584777832, "lr": 0.0002, "elapsed_sec": 19409.930779218674, "step_time_sec": 8.233083092025481, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2338, "loss": 4.759963512420654, "lr": 0.0002, "elapsed_sec": 19418.164813756943, "step_time_sec": 8.233880969928578, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2339, "loss": 4.835060119628906, "lr": 0.0002, "elapsed_sec": 19426.397730112076, "step_time_sec": 8.232811641995795, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2340, "loss": 4.7393646240234375, "lr": 0.0002, "elapsed_sec": 19434.635741710663, "step_time_sec": 8.237824506009929, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2341, "loss": 4.674060821533203, "lr": 0.0002, "elapsed_sec": 19442.875321388245, "step_time_sec": 8.239404955063947, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2342, "loss": 4.745750427246094, "lr": 0.0002, "elapsed_sec": 19451.113515138626, "step_time_sec": 8.238010274944827, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2343, "loss": 4.951932907104492, "lr": 0.0002, "elapsed_sec": 19459.35301756859, "step_time_sec": 8.239334838930517, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2344, "loss": 4.709037780761719, "lr": 0.0002, "elapsed_sec": 19467.590698719025, "step_time_sec": 8.237537989974953, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2345, "loss": 4.707856178283691, "lr": 0.0002, "elapsed_sec": 19475.829358816147, "step_time_sec": 8.238454642007127, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2346, "loss": 4.756815433502197, "lr": 0.0002, "elapsed_sec": 19484.06778550148, "step_time_sec": 8.23827838501893, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2347, "loss": 4.835315227508545, "lr": 0.0002, "elapsed_sec": 19492.30486345291, "step_time_sec": 8.236927833058871, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2348, "loss": 4.948169708251953, "lr": 0.0002, "elapsed_sec": 19500.543663024902, "step_time_sec": 8.238636947004125, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2349, "loss": 4.727671146392822, "lr": 0.0002, "elapsed_sec": 19508.783057451248, "step_time_sec": 8.239256291999482, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2350, "loss": 4.649133682250977, "lr": 0.0002, "elapsed_sec": 19517.022814512253, "step_time_sec": 8.239559727022424, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2351, "loss": 4.646661281585693, "lr": 0.0002, "elapsed_sec": 19525.262182950974, "step_time_sec": 8.239192959968932, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2352, "loss": 4.722386837005615, "lr": 0.0002, "elapsed_sec": 19533.499126195908, "step_time_sec": 8.236931264982559, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2353, "loss": 4.720146179199219, "lr": 0.0002, "elapsed_sec": 19541.737900018692, "step_time_sec": 8.238486371003091, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2354, "loss": 4.725595474243164, "lr": 0.0002, "elapsed_sec": 19549.97659277916, "step_time_sec": 8.238515175995417, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2355, "loss": 4.73041296005249, "lr": 0.0002, "elapsed_sec": 19558.215734243393, "step_time_sec": 8.238992646103725, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2356, "loss": 4.690983772277832, "lr": 0.0002, "elapsed_sec": 19566.45601773262, "step_time_sec": 8.240168757038191, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2357, "loss": 4.724883079528809, "lr": 0.0002, "elapsed_sec": 19574.694398880005, "step_time_sec": 8.238218349055387, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2358, "loss": 4.6710028648376465, "lr": 0.0002, "elapsed_sec": 19582.934068918228, "step_time_sec": 8.239475121023133, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2359, "loss": 4.763197898864746, "lr": 0.0002, "elapsed_sec": 19591.16699886322, "step_time_sec": 8.232788156019524, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2360, "loss": 4.7551093101501465, "lr": 0.0002, "elapsed_sec": 19599.400317907333, "step_time_sec": 8.23314654501155, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2361, "loss": 4.749151229858398, "lr": 0.0002, "elapsed_sec": 19607.631532907486, "step_time_sec": 8.231069568078965, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2362, "loss": 4.700328350067139, "lr": 0.0002, "elapsed_sec": 19615.871533870697, "step_time_sec": 8.239825065014884, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2363, "loss": 4.660091876983643, "lr": 0.0002, "elapsed_sec": 19624.1101231575, "step_time_sec": 8.23843432799913, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2364, "loss": 4.677552223205566, "lr": 0.0002, "elapsed_sec": 19632.34892320633, "step_time_sec": 8.238638023962267, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2365, "loss": 4.806666374206543, "lr": 0.0002, "elapsed_sec": 19640.58810377121, "step_time_sec": 8.239045207039453, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2366, "loss": 4.63339900970459, "lr": 0.0002, "elapsed_sec": 19648.82644510269, "step_time_sec": 8.238188266986981, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2367, "loss": 4.770762920379639, "lr": 0.0002, "elapsed_sec": 19657.06602358818, "step_time_sec": 8.23939295206219, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2368, "loss": 4.647812366485596, "lr": 0.0002, "elapsed_sec": 19665.306300640106, "step_time_sec": 8.240108781028539, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2369, "loss": 4.924630165100098, "lr": 0.0002, "elapsed_sec": 19673.543210029602, "step_time_sec": 8.236767449066974, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2370, "loss": 4.749342918395996, "lr": 0.0002, "elapsed_sec": 19681.78304553032, "step_time_sec": 8.239651871961541, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2371, "loss": 4.587502479553223, "lr": 0.0002, "elapsed_sec": 19690.02129459381, "step_time_sec": 8.238067953032441, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2372, "loss": 4.672367572784424, "lr": 0.0002, "elapsed_sec": 19698.26116991043, "step_time_sec": 8.239718517987058, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2373, "loss": 4.686743259429932, "lr": 0.0002, "elapsed_sec": 19706.50101208687, "step_time_sec": 8.239724593004212, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2374, "loss": 4.805275917053223, "lr": 0.0002, "elapsed_sec": 19714.737956762314, "step_time_sec": 8.236752143013291, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2375, "loss": 4.631917476654053, "lr": 0.0002, "elapsed_sec": 19722.97703719139, "step_time_sec": 8.238889019936323, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2376, "loss": 4.820391654968262, "lr": 0.0002, "elapsed_sec": 19731.21665740013, "step_time_sec": 8.239460007986054, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2377, "loss": 4.696444034576416, "lr": 0.0002, "elapsed_sec": 19739.45695757866, "step_time_sec": 8.24016006407328, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2378, "loss": 4.595555782318115, "lr": 0.0002, "elapsed_sec": 19747.697239875793, "step_time_sec": 8.240118450950831, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2379, "loss": 4.728360176086426, "lr": 0.0002, "elapsed_sec": 19755.93706393242, "step_time_sec": 8.239641541964374, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2380, "loss": 4.784695148468018, "lr": 0.0002, "elapsed_sec": 19764.175411224365, "step_time_sec": 8.238207614980638, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2381, "loss": 4.6687092781066895, "lr": 0.0002, "elapsed_sec": 19772.415380716324, "step_time_sec": 8.239793649991043, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2382, "loss": 4.712374210357666, "lr": 0.0002, "elapsed_sec": 19780.653549671173, "step_time_sec": 8.238038284005597, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2383, "loss": 4.800349235534668, "lr": 0.0002, "elapsed_sec": 19788.89242386818, "step_time_sec": 8.238736387924291, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2384, "loss": 4.900257110595703, "lr": 0.0002, "elapsed_sec": 19797.131855726242, "step_time_sec": 8.239213036024012, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2385, "loss": 4.804378032684326, "lr": 0.0002, "elapsed_sec": 19805.370958805084, "step_time_sec": 8.239026929019019, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2386, "loss": 4.800069332122803, "lr": 0.0002, "elapsed_sec": 19813.6116669178, "step_time_sec": 8.240525790955871, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2387, "loss": 4.605947017669678, "lr": 0.0002, "elapsed_sec": 19821.849953889847, "step_time_sec": 8.238093590945937, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2388, "loss": 4.747931003570557, "lr": 0.0002, "elapsed_sec": 19830.08336520195, "step_time_sec": 8.23321048298385, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2389, "loss": 4.714600086212158, "lr": 0.0002, "elapsed_sec": 19838.32084465027, "step_time_sec": 8.237338523031212, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2390, "loss": 4.726794719696045, "lr": 0.0002, "elapsed_sec": 19846.558984041214, "step_time_sec": 8.238025839906186, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2391, "loss": 4.713479042053223, "lr": 0.0002, "elapsed_sec": 19854.797783613205, "step_time_sec": 8.238596776034683, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2392, "loss": 4.689876556396484, "lr": 0.0002, "elapsed_sec": 19863.036947727203, "step_time_sec": 8.238979817018844, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2393, "loss": 4.772754192352295, "lr": 0.0002, "elapsed_sec": 19871.27665233612, "step_time_sec": 8.239554923959076, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2394, "loss": 4.749240875244141, "lr": 0.0002, "elapsed_sec": 19879.515492916107, "step_time_sec": 8.23865337704774, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2395, "loss": 4.588191509246826, "lr": 0.0002, "elapsed_sec": 19887.748504638672, "step_time_sec": 8.23285382206086, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2396, "loss": 4.768052101135254, "lr": 0.0002, "elapsed_sec": 19895.986409187317, "step_time_sec": 8.23773596493993, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2397, "loss": 4.773197174072266, "lr": 0.0002, "elapsed_sec": 19904.225976228714, "step_time_sec": 8.239383881096728, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2398, "loss": 4.789783477783203, "lr": 0.0002, "elapsed_sec": 19912.464475393295, "step_time_sec": 8.238347447942942, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2399, "loss": 4.764623165130615, "lr": 0.0002, "elapsed_sec": 19920.702699422836, "step_time_sec": 8.23812525998801, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2400, "loss": 4.605006217956543, "lr": 0.0002, "elapsed_sec": 19928.941796779633, "step_time_sec": 8.23892120597884, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2401, "loss": 4.704070091247559, "lr": 0.0002, "elapsed_sec": 19937.179807901382, "step_time_sec": 8.237815105938353, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2402, "loss": 4.722075939178467, "lr": 0.0002, "elapsed_sec": 19945.418601751328, "step_time_sec": 8.238636952009983, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2403, "loss": 4.702727794647217, "lr": 0.0002, "elapsed_sec": 19953.650689840317, "step_time_sec": 8.231934815994464, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2404, "loss": 4.730191707611084, "lr": 0.0002, "elapsed_sec": 19961.88473367691, "step_time_sec": 8.233897564001381, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2405, "loss": 4.78464412689209, "lr": 0.0002, "elapsed_sec": 19970.11837363243, "step_time_sec": 8.233452946995385, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2406, "loss": 4.762113094329834, "lr": 0.0002, "elapsed_sec": 19978.35713815689, "step_time_sec": 8.238598976051435, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2407, "loss": 4.762005805969238, "lr": 0.0002, "elapsed_sec": 19986.5956594944, "step_time_sec": 8.238418142078444, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2408, "loss": 4.639458179473877, "lr": 0.0002, "elapsed_sec": 19994.833862543106, "step_time_sec": 8.238022131030448, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2409, "loss": 4.696110725402832, "lr": 0.0002, "elapsed_sec": 20003.073791742325, "step_time_sec": 8.23975855205208, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2410, "loss": 4.717739582061768, "lr": 0.0002, "elapsed_sec": 20011.311490774155, "step_time_sec": 8.237533587031066, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2411, "loss": 4.641846179962158, "lr": 0.0002, "elapsed_sec": 20019.549765348434, "step_time_sec": 8.238125824020244, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2412, "loss": 4.713456630706787, "lr": 0.0002, "elapsed_sec": 20027.788062095642, "step_time_sec": 8.238091645995155, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2413, "loss": 4.420602798461914, "lr": 0.0002, "elapsed_sec": 20036.025486946106, "step_time_sec": 8.237296769977547, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2414, "loss": 4.660959243774414, "lr": 0.0002, "elapsed_sec": 20044.264590263367, "step_time_sec": 8.238905077101663, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2415, "loss": 4.829292297363281, "lr": 0.0002, "elapsed_sec": 20052.503071069717, "step_time_sec": 8.238311836030334, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2416, "loss": 4.697012424468994, "lr": 0.0002, "elapsed_sec": 20060.742616176605, "step_time_sec": 8.239421288948506, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2417, "loss": 4.658918380737305, "lr": 0.0002, "elapsed_sec": 20068.98112630844, "step_time_sec": 8.238354473025538, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2418, "loss": 4.699184417724609, "lr": 0.0002, "elapsed_sec": 20077.219087600708, "step_time_sec": 8.237763451994397, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2419, "loss": 4.7400593757629395, "lr": 0.0002, "elapsed_sec": 20085.45769572258, "step_time_sec": 8.238461892004125, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2420, "loss": 4.6932268142700195, "lr": 0.0002, "elapsed_sec": 20093.690361976624, "step_time_sec": 8.232486176071689, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2421, "loss": 4.729378700256348, "lr": 0.0002, "elapsed_sec": 20101.928560972214, "step_time_sec": 8.23806209396571, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2422, "loss": 4.693334579467773, "lr": 0.0002, "elapsed_sec": 20110.16618704796, "step_time_sec": 8.237444999977015, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2423, "loss": 4.759965896606445, "lr": 0.0002, "elapsed_sec": 20118.40589118004, "step_time_sec": 8.239612329052761, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2424, "loss": 4.721356391906738, "lr": 0.0002, "elapsed_sec": 20126.644631385803, "step_time_sec": 8.238574524992146, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2425, "loss": 4.736179828643799, "lr": 0.0002, "elapsed_sec": 20134.882668733597, "step_time_sec": 8.237848191987723, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2426, "loss": 4.698638439178467, "lr": 0.0002, "elapsed_sec": 20143.122781038284, "step_time_sec": 8.239938457962126, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2427, "loss": 4.586414813995361, "lr": 0.0002, "elapsed_sec": 20151.361098766327, "step_time_sec": 8.238151972997002, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2428, "loss": 4.738913059234619, "lr": 0.0002, "elapsed_sec": 20159.600083827972, "step_time_sec": 8.238845597952604, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2429, "loss": 4.747646331787109, "lr": 0.0002, "elapsed_sec": 20167.83795952797, "step_time_sec": 8.237744656042196, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2430, "loss": 4.673962593078613, "lr": 0.0002, "elapsed_sec": 20176.077271699905, "step_time_sec": 8.239138028933667, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2431, "loss": 4.69917631149292, "lr": 0.0002, "elapsed_sec": 20184.310368299484, "step_time_sec": 8.232898799935356, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2432, "loss": 4.523250579833984, "lr": 0.0002, "elapsed_sec": 20192.548222064972, "step_time_sec": 8.237602186971344, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2433, "loss": 4.8601298332214355, "lr": 0.0002, "elapsed_sec": 20200.78676009178, "step_time_sec": 8.238388917059638, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2434, "loss": 4.777367115020752, "lr": 0.0002, "elapsed_sec": 20222.467734336853, "step_time_sec": 21.68084370403085, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2435, "loss": 4.723048210144043, "lr": 0.0002, "elapsed_sec": 20230.69995546341, "step_time_sec": 8.232044865028001, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2436, "loss": 4.7414140701293945, "lr": 0.0002, "elapsed_sec": 20238.93312883377, "step_time_sec": 8.233016454963945, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2437, "loss": 4.6533660888671875, "lr": 0.0002, "elapsed_sec": 20247.165353298187, "step_time_sec": 8.232040368020535, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2438, "loss": 4.839407444000244, "lr": 0.0002, "elapsed_sec": 20255.403208971024, "step_time_sec": 8.23769327008631, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2439, "loss": 4.7803955078125, "lr": 0.0002, "elapsed_sec": 20263.63525867462, "step_time_sec": 8.231901741004549, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2440, "loss": 4.843047618865967, "lr": 0.0002, "elapsed_sec": 20271.869432926178, "step_time_sec": 8.234009773936123, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2441, "loss": 4.73293924331665, "lr": 0.0002, "elapsed_sec": 20280.10729122162, "step_time_sec": 8.237683189101517, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2442, "loss": 4.781368255615234, "lr": 0.0002, "elapsed_sec": 20288.3469851017, "step_time_sec": 8.239539965987206, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2443, "loss": 4.691490650177002, "lr": 0.0002, "elapsed_sec": 20296.585837602615, "step_time_sec": 8.238723083981313, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2444, "loss": 4.6640143394470215, "lr": 0.0002, "elapsed_sec": 20304.824663877487, "step_time_sec": 8.238630515988916, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2445, "loss": 4.622775077819824, "lr": 0.0002, "elapsed_sec": 20313.063548088074, "step_time_sec": 8.238733757985756, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2446, "loss": 4.698075294494629, "lr": 0.0002, "elapsed_sec": 20321.302149534225, "step_time_sec": 8.238436246989295, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2447, "loss": 4.708794116973877, "lr": 0.0002, "elapsed_sec": 20329.542573690414, "step_time_sec": 8.240293959970586, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2448, "loss": 4.676326751708984, "lr": 0.0002, "elapsed_sec": 20337.7813475132, "step_time_sec": 8.23858103202656, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2449, "loss": 4.724616050720215, "lr": 0.0002, "elapsed_sec": 20346.021233558655, "step_time_sec": 8.23974384390749, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2450, "loss": 4.803921699523926, "lr": 0.0002, "elapsed_sec": 20354.25983452797, "step_time_sec": 8.238446935079992, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2451, "loss": 4.684576034545898, "lr": 0.0002, "elapsed_sec": 20362.498511314392, "step_time_sec": 8.238476998987608, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2452, "loss": 4.858487129211426, "lr": 0.0002, "elapsed_sec": 20370.73807501793, "step_time_sec": 8.239445780985989, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2453, "loss": 4.685953617095947, "lr": 0.0002, "elapsed_sec": 20378.977635383606, "step_time_sec": 8.239378249971196, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2454, "loss": 4.697542190551758, "lr": 0.0002, "elapsed_sec": 20387.216358423233, "step_time_sec": 8.238568483968265, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2455, "loss": 4.721024990081787, "lr": 0.0002, "elapsed_sec": 20395.4561419487, "step_time_sec": 8.239619772997685, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2456, "loss": 4.695216655731201, "lr": 0.0002, "elapsed_sec": 20403.693578481674, "step_time_sec": 8.237280383007601, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2457, "loss": 4.714582920074463, "lr": 0.0002, "elapsed_sec": 20411.93341445923, "step_time_sec": 8.239697140059434, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2458, "loss": 4.707203388214111, "lr": 0.0002, "elapsed_sec": 20420.172390699387, "step_time_sec": 8.238788452930748, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2459, "loss": 4.797109127044678, "lr": 0.0002, "elapsed_sec": 20428.411487340927, "step_time_sec": 8.238918220973574, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2460, "loss": 4.72135066986084, "lr": 0.0002, "elapsed_sec": 20436.651898622513, "step_time_sec": 8.240280431928113, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2461, "loss": 4.675530433654785, "lr": 0.0002, "elapsed_sec": 20444.889026403427, "step_time_sec": 8.236929647042416, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2462, "loss": 4.592369079589844, "lr": 0.0002, "elapsed_sec": 20453.128561496735, "step_time_sec": 8.239383365958929, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2463, "loss": 4.787352561950684, "lr": 0.0002, "elapsed_sec": 20461.36695575714, "step_time_sec": 8.238262756960467, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2464, "loss": 4.687318801879883, "lr": 0.0002, "elapsed_sec": 20469.605405569077, "step_time_sec": 8.238314206013456, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2465, "loss": 4.703935623168945, "lr": 0.0002, "elapsed_sec": 20477.84637904167, "step_time_sec": 8.240736274048686, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2466, "loss": 4.630680561065674, "lr": 0.0002, "elapsed_sec": 20486.083609580994, "step_time_sec": 8.237131683039479, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2467, "loss": 4.769975185394287, "lr": 0.0002, "elapsed_sec": 20494.32433629036, "step_time_sec": 8.240549416048452, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2468, "loss": 4.816619396209717, "lr": 0.0002, "elapsed_sec": 20502.563760519028, "step_time_sec": 8.239222296979278, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2469, "loss": 4.832223892211914, "lr": 0.0002, "elapsed_sec": 20510.802153110504, "step_time_sec": 8.238228728063405, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2470, "loss": 4.662134647369385, "lr": 0.0002, "elapsed_sec": 20519.04078888893, "step_time_sec": 8.238505710964091, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2471, "loss": 4.807907581329346, "lr": 0.0002, "elapsed_sec": 20527.279129505157, "step_time_sec": 8.23817665502429, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2472, "loss": 4.674191474914551, "lr": 0.0002, "elapsed_sec": 20535.5194542408, "step_time_sec": 8.24016731698066, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2473, "loss": 4.643742084503174, "lr": 0.0002, "elapsed_sec": 20543.757968187332, "step_time_sec": 8.238345884019509, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2474, "loss": 4.78818416595459, "lr": 0.0002, "elapsed_sec": 20551.99555039406, "step_time_sec": 8.237376419012435, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2475, "loss": 4.7462592124938965, "lr": 0.0002, "elapsed_sec": 20560.234237909317, "step_time_sec": 8.238538186065853, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2476, "loss": 4.670506954193115, "lr": 0.0002, "elapsed_sec": 20568.473379850388, "step_time_sec": 8.238977383938618, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2477, "loss": 4.660611629486084, "lr": 0.0002, "elapsed_sec": 20576.713022232056, "step_time_sec": 8.239525642013177, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2478, "loss": 4.664594650268555, "lr": 0.0002, "elapsed_sec": 20584.952723503113, "step_time_sec": 8.239512181957252, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2479, "loss": 4.903988838195801, "lr": 0.0002, "elapsed_sec": 20593.189503908157, "step_time_sec": 8.236630594008602, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2480, "loss": 4.673655033111572, "lr": 0.0002, "elapsed_sec": 20601.429970264435, "step_time_sec": 8.24030533398036, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2481, "loss": 4.676642417907715, "lr": 0.0002, "elapsed_sec": 20609.668177843094, "step_time_sec": 8.238067034049891, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2482, "loss": 4.699193000793457, "lr": 0.0002, "elapsed_sec": 20617.906549930573, "step_time_sec": 8.238158995984122, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2483, "loss": 4.671123504638672, "lr": 0.0002, "elapsed_sec": 20626.146507263184, "step_time_sec": 8.239855546038598, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2484, "loss": 4.70184326171875, "lr": 0.0002, "elapsed_sec": 20634.383879184723, "step_time_sec": 8.237179094925523, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2485, "loss": 4.655069351196289, "lr": 0.0002, "elapsed_sec": 20642.62255549431, "step_time_sec": 8.238498898106627, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2486, "loss": 4.786971092224121, "lr": 0.0002, "elapsed_sec": 20650.861327171326, "step_time_sec": 8.238583068014123, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2487, "loss": 4.696102619171143, "lr": 0.0002, "elapsed_sec": 20659.100511312485, "step_time_sec": 8.239107300993055, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2488, "loss": 4.676632404327393, "lr": 0.0002, "elapsed_sec": 20667.34046149254, "step_time_sec": 8.239731041016057, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2489, "loss": 4.652410984039307, "lr": 0.0002, "elapsed_sec": 20675.57918715477, "step_time_sec": 8.23855273600202, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2490, "loss": 4.683133125305176, "lr": 0.0002, "elapsed_sec": 20683.81902551651, "step_time_sec": 8.239683821913786, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2491, "loss": 4.601347923278809, "lr": 0.0002, "elapsed_sec": 20692.12535095215, "step_time_sec": 8.246365625993349, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2492, "loss": 4.773100852966309, "lr": 0.0002, "elapsed_sec": 20700.35809469223, "step_time_sec": 8.232594450004399, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2493, "loss": 4.666715621948242, "lr": 0.0002, "elapsed_sec": 20708.591249465942, "step_time_sec": 8.233004103996791, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2494, "loss": 4.760243892669678, "lr": 0.0002, "elapsed_sec": 20716.830115795135, "step_time_sec": 8.238731696037576, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2495, "loss": 4.617644786834717, "lr": 0.0002, "elapsed_sec": 20725.069716453552, "step_time_sec": 8.239414103096351, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2496, "loss": 4.703856468200684, "lr": 0.0002, "elapsed_sec": 20733.30877995491, "step_time_sec": 8.238916311995126, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2497, "loss": 4.636803150177002, "lr": 0.0002, "elapsed_sec": 20741.54756641388, "step_time_sec": 8.238649865030311, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2498, "loss": 4.8814544677734375, "lr": 0.0002, "elapsed_sec": 20749.78718495369, "step_time_sec": 8.239453961024992, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2499, "loss": 4.709570407867432, "lr": 0.0002, "elapsed_sec": 20758.024793863297, "step_time_sec": 8.237438925076276, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2500, "loss": 4.673212051391602, "lr": 0.0002, "elapsed_sec": 20766.263300180435, "step_time_sec": 34.216253751073964, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": true, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2501, "loss": 4.77292537689209, "lr": 0.0002, "elapsed_sec": 20800.488747119904, "step_time_sec": 8.24700617196504, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2502, "loss": 4.595791339874268, "lr": 0.0002, "elapsed_sec": 20808.72030520439, "step_time_sec": 8.231413788045757, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2503, "loss": 4.804860591888428, "lr": 0.0002, "elapsed_sec": 20816.95293903351, "step_time_sec": 8.232417708961293, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2504, "loss": 4.6907525062561035, "lr": 0.0002, "elapsed_sec": 20825.18512892723, "step_time_sec": 8.232049865997396, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2505, "loss": 4.7244367599487305, "lr": 0.0002, "elapsed_sec": 20833.417493104935, "step_time_sec": 8.23220767697785, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2506, "loss": 4.705247402191162, "lr": 0.0002, "elapsed_sec": 20841.65013027191, "step_time_sec": 8.232474566088058, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2507, "loss": 4.640117168426514, "lr": 0.0002, "elapsed_sec": 20849.88573884964, "step_time_sec": 8.235442938050255, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2508, "loss": 4.705870151519775, "lr": 0.0002, "elapsed_sec": 20858.120152950287, "step_time_sec": 8.234270478016697, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2509, "loss": 4.68946647644043, "lr": 0.0002, "elapsed_sec": 20866.353219747543, "step_time_sec": 8.232902348972857, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2510, "loss": 4.53928279876709, "lr": 0.0002, "elapsed_sec": 20874.585017204285, "step_time_sec": 8.231625064043328, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2511, "loss": 4.6869072914123535, "lr": 0.0002, "elapsed_sec": 20882.81729054451, "step_time_sec": 8.23209167807363, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2512, "loss": 4.626951694488525, "lr": 0.0002, "elapsed_sec": 20891.0501434803, "step_time_sec": 8.232670343015343, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2513, "loss": 4.72511100769043, "lr": 0.0002, "elapsed_sec": 20899.283197164536, "step_time_sec": 8.232888535014354, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2514, "loss": 4.845163822174072, "lr": 0.0002, "elapsed_sec": 20907.52176475525, "step_time_sec": 8.238411291036755, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2515, "loss": 4.74529504776001, "lr": 0.0002, "elapsed_sec": 20915.759819746017, "step_time_sec": 8.23789625405334, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2516, "loss": 4.695300102233887, "lr": 0.0002, "elapsed_sec": 20923.9971909523, "step_time_sec": 8.237207047990523, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2517, "loss": 4.801213264465332, "lr": 0.0002, "elapsed_sec": 20932.235919237137, "step_time_sec": 8.23854460194707, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2518, "loss": 4.667686939239502, "lr": 0.0002, "elapsed_sec": 20940.473737955093, "step_time_sec": 8.237682456965558, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2519, "loss": 4.591490745544434, "lr": 0.0002, "elapsed_sec": 20948.713559627533, "step_time_sec": 8.23965958005283, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2520, "loss": 4.595898628234863, "lr": 0.0002, "elapsed_sec": 20956.9526386261, "step_time_sec": 8.238890251028351, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2521, "loss": 4.696532249450684, "lr": 0.0002, "elapsed_sec": 20965.18960094452, "step_time_sec": 8.236814035917632, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2522, "loss": 4.647920608520508, "lr": 0.0002, "elapsed_sec": 20973.428868055344, "step_time_sec": 8.239089487004094, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2523, "loss": 4.645381927490234, "lr": 0.0002, "elapsed_sec": 20981.66726899147, "step_time_sec": 8.23823849891778, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2524, "loss": 4.683784008026123, "lr": 0.0002, "elapsed_sec": 20989.906580209732, "step_time_sec": 8.23920808499679, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2525, "loss": 4.747432231903076, "lr": 0.0002, "elapsed_sec": 20998.144730329514, "step_time_sec": 8.237918499973603, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2526, "loss": 4.725982189178467, "lr": 0.0002, "elapsed_sec": 21006.382814645767, "step_time_sec": 8.237978547927924, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2527, "loss": 4.672797203063965, "lr": 0.0002, "elapsed_sec": 21014.62147784233, "step_time_sec": 8.238479972933419, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2528, "loss": 4.813431739807129, "lr": 0.0002, "elapsed_sec": 21022.860268115997, "step_time_sec": 8.238606086932123, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2529, "loss": 4.6263837814331055, "lr": 0.0002, "elapsed_sec": 21031.099850416183, "step_time_sec": 8.239429190987721, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2530, "loss": 4.688011646270752, "lr": 0.0002, "elapsed_sec": 21039.33903503418, "step_time_sec": 8.239011707017198, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2531, "loss": 4.77854585647583, "lr": 0.0002, "elapsed_sec": 21047.577057361603, "step_time_sec": 8.237851774087176, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2532, "loss": 4.64990758895874, "lr": 0.0002, "elapsed_sec": 21055.811295747757, "step_time_sec": 8.234094360959716, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2533, "loss": 4.603376865386963, "lr": 0.0002, "elapsed_sec": 21064.049697875977, "step_time_sec": 8.238222359097563, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2534, "loss": 4.766007900238037, "lr": 0.0002, "elapsed_sec": 21072.288947343826, "step_time_sec": 8.239060049061663, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2535, "loss": 4.67343282699585, "lr": 0.0002, "elapsed_sec": 21080.52742409706, "step_time_sec": 8.238325748010539, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2536, "loss": 4.672544002532959, "lr": 0.0002, "elapsed_sec": 21088.76637005806, "step_time_sec": 8.238799071055837, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2537, "loss": 4.51254415512085, "lr": 0.0002, "elapsed_sec": 21097.005729198456, "step_time_sec": 8.239160829107277, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2538, "loss": 4.653616428375244, "lr": 0.0002, "elapsed_sec": 21105.244034051895, "step_time_sec": 8.238144078059122, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2539, "loss": 4.818573474884033, "lr": 0.0002, "elapsed_sec": 21113.482418060303, "step_time_sec": 8.23822762107011, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2540, "loss": 4.695529937744141, "lr": 0.0002, "elapsed_sec": 21121.72105360031, "step_time_sec": 8.238465176080354, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2541, "loss": 4.606269359588623, "lr": 0.0002, "elapsed_sec": 21129.95906686783, "step_time_sec": 8.23786813300103, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2542, "loss": 4.637726306915283, "lr": 0.0002, "elapsed_sec": 21138.198130846024, "step_time_sec": 8.238846581079997, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2543, "loss": 4.710552215576172, "lr": 0.0002, "elapsed_sec": 21146.43589067459, "step_time_sec": 8.237669560010545, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2544, "loss": 4.683148384094238, "lr": 0.0002, "elapsed_sec": 21154.675144672394, "step_time_sec": 8.239047634997405, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2545, "loss": 4.6009016036987305, "lr": 0.0002, "elapsed_sec": 21162.914056777954, "step_time_sec": 8.23877242708113, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2546, "loss": 4.54088830947876, "lr": 0.0002, "elapsed_sec": 21171.152325868607, "step_time_sec": 8.238046501995996, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2547, "loss": 4.694581031799316, "lr": 0.0002, "elapsed_sec": 21179.39140534401, "step_time_sec": 8.238920700037852, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2548, "loss": 4.7238569259643555, "lr": 0.0002, "elapsed_sec": 21187.628118038177, "step_time_sec": 8.236541493912227, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2549, "loss": 4.706777572631836, "lr": 0.0002, "elapsed_sec": 21195.866876602173, "step_time_sec": 8.238630681997165, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2550, "loss": 4.63810920715332, "lr": 0.0002, "elapsed_sec": 21204.104573726654, "step_time_sec": 8.237579775042832, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2551, "loss": 4.7131853103637695, "lr": 0.0002, "elapsed_sec": 21212.34374165535, "step_time_sec": 8.238924014032818, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2552, "loss": 4.665659427642822, "lr": 0.0002, "elapsed_sec": 21220.582731485367, "step_time_sec": 8.238858610973693, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2553, "loss": 4.661831378936768, "lr": 0.0002, "elapsed_sec": 21228.82119846344, "step_time_sec": 8.238311029970646, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2554, "loss": 4.707553386688232, "lr": 0.0002, "elapsed_sec": 21237.060620069504, "step_time_sec": 8.239207508042455, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2555, "loss": 4.6146368980407715, "lr": 0.0002, "elapsed_sec": 21245.29959177971, "step_time_sec": 8.238836842007004, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2556, "loss": 4.7116827964782715, "lr": 0.0002, "elapsed_sec": 21253.53852367401, "step_time_sec": 8.23876997095067, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2557, "loss": 4.6100873947143555, "lr": 0.0002, "elapsed_sec": 21261.77783894539, "step_time_sec": 8.239112474955618, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2558, "loss": 4.731297492980957, "lr": 0.0002, "elapsed_sec": 21270.014604330063, "step_time_sec": 8.236628975020722, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2559, "loss": 4.70935583114624, "lr": 0.0002, "elapsed_sec": 21278.25424671173, "step_time_sec": 8.239435574039817, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2560, "loss": 4.616344451904297, "lr": 0.0002, "elapsed_sec": 21286.49333500862, "step_time_sec": 8.238971487968229, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2561, "loss": 4.707179546356201, "lr": 0.0002, "elapsed_sec": 21294.73311138153, "step_time_sec": 8.239572040969506, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2562, "loss": 4.542510032653809, "lr": 0.0002, "elapsed_sec": 21302.972098350525, "step_time_sec": 8.238832491915673, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2563, "loss": 4.6498517990112305, "lr": 0.0002, "elapsed_sec": 21311.209649086, "step_time_sec": 8.237375748925842, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2564, "loss": 4.723969459533691, "lr": 0.0002, "elapsed_sec": 21319.44976091385, "step_time_sec": 8.239985340973362, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2565, "loss": 4.578947067260742, "lr": 0.0002, "elapsed_sec": 21327.688519239426, "step_time_sec": 8.238555990974419, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2566, "loss": 4.691047191619873, "lr": 0.0002, "elapsed_sec": 21335.927505016327, "step_time_sec": 8.238807794987224, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2567, "loss": 4.690130233764648, "lr": 0.0002, "elapsed_sec": 21344.16822719574, "step_time_sec": 8.240540080005303, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2568, "loss": 4.611644744873047, "lr": 0.0002, "elapsed_sec": 21352.406668663025, "step_time_sec": 8.238278068951331, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2569, "loss": 4.578601360321045, "lr": 0.0002, "elapsed_sec": 21360.645121335983, "step_time_sec": 8.238281403086148, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2570, "loss": 4.714354038238525, "lr": 0.0002, "elapsed_sec": 21368.882163524628, "step_time_sec": 8.236929389997385, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2571, "loss": 4.60905122756958, "lr": 0.0002, "elapsed_sec": 21377.12070274353, "step_time_sec": 8.238329062936828, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2572, "loss": 4.638650894165039, "lr": 0.0002, "elapsed_sec": 21385.35889005661, "step_time_sec": 8.238037222996354, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2573, "loss": 4.711033821105957, "lr": 0.0002, "elapsed_sec": 21393.59079194069, "step_time_sec": 8.231758946087211, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2574, "loss": 4.630015850067139, "lr": 0.0002, "elapsed_sec": 21401.823962688446, "step_time_sec": 8.232996465987526, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2575, "loss": 4.493292331695557, "lr": 0.0002, "elapsed_sec": 21410.05640935898, "step_time_sec": 8.232324122916907, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2576, "loss": 4.6251678466796875, "lr": 0.0002, "elapsed_sec": 21418.289906024933, "step_time_sec": 8.233304622001015, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2577, "loss": 4.613870143890381, "lr": 0.0002, "elapsed_sec": 21426.523364543915, "step_time_sec": 8.233323872904293, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2578, "loss": 4.551504135131836, "lr": 0.0002, "elapsed_sec": 21434.758452892303, "step_time_sec": 8.234853805974126, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2579, "loss": 4.653430938720703, "lr": 0.0002, "elapsed_sec": 21442.990746974945, "step_time_sec": 8.23214433400426, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2580, "loss": 4.600569248199463, "lr": 0.0002, "elapsed_sec": 21451.223145484924, "step_time_sec": 8.232205283013172, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2581, "loss": 4.669114589691162, "lr": 0.0002, "elapsed_sec": 21459.45548772812, "step_time_sec": 8.232238363940269, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2582, "loss": 4.702721118927002, "lr": 0.0002, "elapsed_sec": 21467.693413972855, "step_time_sec": 8.237707318970934, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2583, "loss": 4.626807689666748, "lr": 0.0002, "elapsed_sec": 21475.931699037552, "step_time_sec": 8.23811511695385, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2584, "loss": 4.760247707366943, "lr": 0.0002, "elapsed_sec": 21484.170772075653, "step_time_sec": 8.238914902904071, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2585, "loss": 4.621838092803955, "lr": 0.0002, "elapsed_sec": 21492.409542560577, "step_time_sec": 8.238592242007144, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2586, "loss": 4.62770414352417, "lr": 0.0002, "elapsed_sec": 21500.64894080162, "step_time_sec": 8.239231006009504, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2587, "loss": 4.624045372009277, "lr": 0.0002, "elapsed_sec": 21508.887672424316, "step_time_sec": 8.238558290991932, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2588, "loss": 4.632867336273193, "lr": 0.0002, "elapsed_sec": 21517.125660181046, "step_time_sec": 8.237823506002314, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2589, "loss": 4.66279935836792, "lr": 0.0002, "elapsed_sec": 21525.364924669266, "step_time_sec": 8.23912275605835, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2590, "loss": 4.707982063293457, "lr": 0.0002, "elapsed_sec": 21533.60234427452, "step_time_sec": 8.23721587099135, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2591, "loss": 4.684623718261719, "lr": 0.0002, "elapsed_sec": 21541.841459274292, "step_time_sec": 8.23896229499951, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2592, "loss": 4.602928638458252, "lr": 0.0002, "elapsed_sec": 21550.08037161827, "step_time_sec": 8.238779773004353, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2593, "loss": 4.590416431427002, "lr": 0.0002, "elapsed_sec": 21558.313578128815, "step_time_sec": 8.232992534991354, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2594, "loss": 4.718428134918213, "lr": 0.0002, "elapsed_sec": 21566.546449661255, "step_time_sec": 8.2327294270508, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2595, "loss": 4.594848155975342, "lr": 0.0002, "elapsed_sec": 21574.77907037735, "step_time_sec": 8.232401018962264, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2596, "loss": 4.688693523406982, "lr": 0.0002, "elapsed_sec": 21583.01713204384, "step_time_sec": 8.237897292943671, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2597, "loss": 4.649708271026611, "lr": 0.0002, "elapsed_sec": 21591.255937576294, "step_time_sec": 8.238654371001758, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2598, "loss": 4.611485004425049, "lr": 0.0002, "elapsed_sec": 21599.495544433594, "step_time_sec": 8.239474836038426, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2599, "loss": 4.624054908752441, "lr": 0.0002, "elapsed_sec": 21607.728129148483, "step_time_sec": 8.23238121997565, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2600, "loss": 4.602052211761475, "lr": 0.0002, "elapsed_sec": 21615.966943264008, "step_time_sec": 8.238685413030908, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2601, "loss": 4.57535457611084, "lr": 0.0002, "elapsed_sec": 21624.205478191376, "step_time_sec": 8.238350340980105, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2602, "loss": 4.641141891479492, "lr": 0.0002, "elapsed_sec": 21632.44382596016, "step_time_sec": 8.238163881003857, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2603, "loss": 4.657835960388184, "lr": 0.0002, "elapsed_sec": 21640.682975053787, "step_time_sec": 8.239031174918637, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2604, "loss": 4.640682220458984, "lr": 0.0002, "elapsed_sec": 21648.91996908188, "step_time_sec": 8.236849979963154, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2605, "loss": 4.631878852844238, "lr": 0.0002, "elapsed_sec": 21657.15679383278, "step_time_sec": 8.236627800972201, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2606, "loss": 4.653157711029053, "lr": 0.0002, "elapsed_sec": 21665.39436364174, "step_time_sec": 8.23741602001246, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2607, "loss": 4.642961025238037, "lr": 0.0002, "elapsed_sec": 21673.63422179222, "step_time_sec": 8.23971408396028, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2608, "loss": 4.666129112243652, "lr": 0.0002, "elapsed_sec": 21681.872516155243, "step_time_sec": 8.238090878003277, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2609, "loss": 4.61911153793335, "lr": 0.0002, "elapsed_sec": 21690.11027097702, "step_time_sec": 8.237632746924646, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2610, "loss": 4.705513954162598, "lr": 0.0002, "elapsed_sec": 21698.349236488342, "step_time_sec": 8.238775828038342, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2611, "loss": 4.577478408813477, "lr": 0.0002, "elapsed_sec": 21706.586468696594, "step_time_sec": 8.237103216932155, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2612, "loss": 4.70490837097168, "lr": 0.0002, "elapsed_sec": 21714.826628684998, "step_time_sec": 8.239942081039771, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2613, "loss": 4.6040940284729, "lr": 0.0002, "elapsed_sec": 21723.0662651062, "step_time_sec": 8.239461428951472, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2614, "loss": 4.60746955871582, "lr": 0.0002, "elapsed_sec": 21731.303688526154, "step_time_sec": 8.237267152988352, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2615, "loss": 4.562106132507324, "lr": 0.0002, "elapsed_sec": 21739.543124437332, "step_time_sec": 8.239263625000603, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2616, "loss": 4.663993835449219, "lr": 0.0002, "elapsed_sec": 21747.780036211014, "step_time_sec": 8.23675798100885, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2617, "loss": 4.5139851570129395, "lr": 0.0002, "elapsed_sec": 21756.01862001419, "step_time_sec": 8.238456405932084, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2618, "loss": 4.523565769195557, "lr": 0.0002, "elapsed_sec": 21764.251217126846, "step_time_sec": 8.23238718509674, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2619, "loss": 4.678747653961182, "lr": 0.0002, "elapsed_sec": 21772.484834432602, "step_time_sec": 8.233440671931021, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2620, "loss": 4.568846225738525, "lr": 0.0002, "elapsed_sec": 21780.716512441635, "step_time_sec": 8.23156507499516, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2621, "loss": 4.749790668487549, "lr": 0.0002, "elapsed_sec": 21788.949990272522, "step_time_sec": 8.233269179938361, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2622, "loss": 4.624752998352051, "lr": 0.0002, "elapsed_sec": 21797.182908058167, "step_time_sec": 8.23274375195615, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2623, "loss": 4.635716915130615, "lr": 0.0002, "elapsed_sec": 21805.422425985336, "step_time_sec": 8.239411239977926, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2624, "loss": 4.593883991241455, "lr": 0.0002, "elapsed_sec": 21813.654585123062, "step_time_sec": 8.232008924009278, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2625, "loss": 4.637870788574219, "lr": 0.0002, "elapsed_sec": 21821.886896133423, "step_time_sec": 8.232097149011679, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2626, "loss": 4.488887786865234, "lr": 0.0002, "elapsed_sec": 21830.119051218033, "step_time_sec": 8.231971082044765, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2627, "loss": 4.656825065612793, "lr": 0.0002, "elapsed_sec": 21838.352127075195, "step_time_sec": 8.23292756092269, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2628, "loss": 4.748608589172363, "lr": 0.0002, "elapsed_sec": 21846.583845853806, "step_time_sec": 8.231593424105085, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2629, "loss": 4.674834728240967, "lr": 0.0002, "elapsed_sec": 21854.82350373268, "step_time_sec": 8.239459417993203, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2630, "loss": 4.682519435882568, "lr": 0.0002, "elapsed_sec": 21863.056035995483, "step_time_sec": 8.232351347920485, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2631, "loss": 4.654355525970459, "lr": 0.0002, "elapsed_sec": 21871.289595365524, "step_time_sec": 8.233364082989283, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2632, "loss": 4.685782432556152, "lr": 0.0002, "elapsed_sec": 21879.521671533585, "step_time_sec": 8.231928384979255, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2633, "loss": 4.575623512268066, "lr": 0.0002, "elapsed_sec": 21887.76086115837, "step_time_sec": 8.239059615996666, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2634, "loss": 4.625538349151611, "lr": 0.0002, "elapsed_sec": 21896.000346183777, "step_time_sec": 8.239313914906234, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2635, "loss": 4.640074729919434, "lr": 0.0002, "elapsed_sec": 21904.238854169846, "step_time_sec": 8.238366712001152, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2636, "loss": 4.683140754699707, "lr": 0.0002, "elapsed_sec": 21912.47232747078, "step_time_sec": 8.233315666904673, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2637, "loss": 4.669519901275635, "lr": 0.0002, "elapsed_sec": 21920.711216449738, "step_time_sec": 8.238747996976599, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2638, "loss": 4.629046440124512, "lr": 0.0002, "elapsed_sec": 21928.950417280197, "step_time_sec": 8.238973643979989, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2639, "loss": 4.544658660888672, "lr": 0.0002, "elapsed_sec": 21937.181676387787, "step_time_sec": 8.231098253047094, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2640, "loss": 4.637231826782227, "lr": 0.0002, "elapsed_sec": 21945.415362358093, "step_time_sec": 8.233548698015511, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2641, "loss": 4.726095676422119, "lr": 0.0002, "elapsed_sec": 21953.64715409279, "step_time_sec": 8.231623721076176, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2642, "loss": 4.620804786682129, "lr": 0.0002, "elapsed_sec": 21961.880603313446, "step_time_sec": 8.2332670269534, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2643, "loss": 4.573581218719482, "lr": 0.0002, "elapsed_sec": 21970.11723423004, "step_time_sec": 8.23645874497015, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2644, "loss": 4.719319820404053, "lr": 0.0002, "elapsed_sec": 21978.355349302292, "step_time_sec": 8.237971721915528, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2645, "loss": 4.499490261077881, "lr": 0.0002, "elapsed_sec": 21986.592862844467, "step_time_sec": 8.237385426997207, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2646, "loss": 4.698122978210449, "lr": 0.0002, "elapsed_sec": 21994.832280635834, "step_time_sec": 8.239211954991333, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2647, "loss": 4.595215320587158, "lr": 0.0002, "elapsed_sec": 22003.070831537247, "step_time_sec": 8.238374745938927, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2648, "loss": 4.60807466506958, "lr": 0.0002, "elapsed_sec": 22011.30890440941, "step_time_sec": 8.237913507036865, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2649, "loss": 4.534450054168701, "lr": 0.0002, "elapsed_sec": 22019.54264473915, "step_time_sec": 8.233605459914543, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2650, "loss": 4.687450408935547, "lr": 0.0002, "elapsed_sec": 22027.77928686142, "step_time_sec": 8.236471524927765, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2651, "loss": 4.655012130737305, "lr": 0.0002, "elapsed_sec": 22036.0185713768, "step_time_sec": 8.239171462948434, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2652, "loss": 4.697533130645752, "lr": 0.0002, "elapsed_sec": 22044.256645202637, "step_time_sec": 8.23787687101867, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2653, "loss": 4.561844825744629, "lr": 0.0002, "elapsed_sec": 22052.48928785324, "step_time_sec": 8.23248222109396, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2654, "loss": 4.55502986907959, "lr": 0.0002, "elapsed_sec": 22060.72140288353, "step_time_sec": 8.231975959031843, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2655, "loss": 4.476539611816406, "lr": 0.0002, "elapsed_sec": 22068.95413684845, "step_time_sec": 8.23253592196852, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2656, "loss": 4.605727672576904, "lr": 0.0002, "elapsed_sec": 22077.186708450317, "step_time_sec": 8.232392478967085, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2657, "loss": 4.61718225479126, "lr": 0.0002, "elapsed_sec": 22085.419981956482, "step_time_sec": 8.23315718700178, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2658, "loss": 4.38611364364624, "lr": 0.0002, "elapsed_sec": 22093.652927160263, "step_time_sec": 8.23279051494319, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2659, "loss": 4.558691024780273, "lr": 0.0002, "elapsed_sec": 22101.892073392868, "step_time_sec": 8.238952999003232, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2660, "loss": 4.723341464996338, "lr": 0.0002, "elapsed_sec": 22110.129730701447, "step_time_sec": 8.237506711971946, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2661, "loss": 4.582016468048096, "lr": 0.0002, "elapsed_sec": 22118.361659526825, "step_time_sec": 8.231791864032857, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2662, "loss": 4.621946811676025, "lr": 0.0002, "elapsed_sec": 22126.598967552185, "step_time_sec": 8.237146517029032, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2663, "loss": 4.613465309143066, "lr": 0.0002, "elapsed_sec": 22134.83707666397, "step_time_sec": 8.237878536921926, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2664, "loss": 4.610342025756836, "lr": 0.0002, "elapsed_sec": 22143.074558258057, "step_time_sec": 8.237307447008789, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2665, "loss": 4.565835475921631, "lr": 0.0002, "elapsed_sec": 22151.30609536171, "step_time_sec": 8.231395279988647, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2666, "loss": 4.661013126373291, "lr": 0.0002, "elapsed_sec": 22159.53941965103, "step_time_sec": 8.233131416956894, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2667, "loss": 4.599739074707031, "lr": 0.0002, "elapsed_sec": 22167.771072149277, "step_time_sec": 8.231511329999194, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2668, "loss": 4.707118034362793, "lr": 0.0002, "elapsed_sec": 22176.003929376602, "step_time_sec": 8.232689457945526, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2669, "loss": 4.6760735511779785, "lr": 0.0002, "elapsed_sec": 22184.235359191895, "step_time_sec": 8.231289039016701, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2670, "loss": 4.67885684967041, "lr": 0.0002, "elapsed_sec": 22192.46807742119, "step_time_sec": 8.23255349707324, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2671, "loss": 4.602485656738281, "lr": 0.0002, "elapsed_sec": 22200.700344085693, "step_time_sec": 8.232132832985371, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2672, "loss": 4.69747257232666, "lr": 0.0002, "elapsed_sec": 22208.933244228363, "step_time_sec": 8.23275869898498, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2673, "loss": 4.518361568450928, "lr": 0.0002, "elapsed_sec": 22217.16708469391, "step_time_sec": 8.233647273969837, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2674, "loss": 4.606334209442139, "lr": 0.0002, "elapsed_sec": 22225.404673576355, "step_time_sec": 8.237449316889979, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2675, "loss": 4.556384086608887, "lr": 0.0002, "elapsed_sec": 22233.64126443863, "step_time_sec": 8.236422864021733, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2676, "loss": 4.581583023071289, "lr": 0.0002, "elapsed_sec": 22241.87681388855, "step_time_sec": 8.235372832044959, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2677, "loss": 4.662025451660156, "lr": 0.0002, "elapsed_sec": 22250.10964012146, "step_time_sec": 8.232661023037508, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2678, "loss": 4.569182872772217, "lr": 0.0002, "elapsed_sec": 22258.34193944931, "step_time_sec": 8.232142466004007, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2679, "loss": 4.637867450714111, "lr": 0.0002, "elapsed_sec": 22266.575244665146, "step_time_sec": 8.23319812794216, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2680, "loss": 4.509982109069824, "lr": 0.0002, "elapsed_sec": 22274.81456542015, "step_time_sec": 8.239135842071846, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2681, "loss": 4.764338493347168, "lr": 0.0002, "elapsed_sec": 22283.053207159042, "step_time_sec": 8.238471986958757, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2682, "loss": 4.52482271194458, "lr": 0.0002, "elapsed_sec": 22291.291676044464, "step_time_sec": 8.238303362973966, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2683, "loss": 4.684296607971191, "lr": 0.0002, "elapsed_sec": 22299.523492097855, "step_time_sec": 8.231659681070596, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2684, "loss": 4.493890285491943, "lr": 0.0002, "elapsed_sec": 22307.756513834, "step_time_sec": 8.232871375977993, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2685, "loss": 4.525544166564941, "lr": 0.0002, "elapsed_sec": 22315.98922920227, "step_time_sec": 8.23257203202229, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2686, "loss": 4.6233720779418945, "lr": 0.0002, "elapsed_sec": 22324.22195315361, "step_time_sec": 8.23254823195748, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2687, "loss": 4.577465534210205, "lr": 0.0002, "elapsed_sec": 22332.45406770706, "step_time_sec": 8.231950325076468, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2688, "loss": 4.399368762969971, "lr": 0.0002, "elapsed_sec": 22340.686930179596, "step_time_sec": 8.232726886984892, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2689, "loss": 4.608211994171143, "lr": 0.0002, "elapsed_sec": 22348.923010110855, "step_time_sec": 8.23593125701882, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2690, "loss": 4.615815162658691, "lr": 0.0002, "elapsed_sec": 22357.159638166428, "step_time_sec": 8.236435619997792, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2691, "loss": 4.492146015167236, "lr": 0.0002, "elapsed_sec": 22365.393613100052, "step_time_sec": 8.233828027034178, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2692, "loss": 4.628127098083496, "lr": 0.0002, "elapsed_sec": 22373.626748085022, "step_time_sec": 8.232968715019524, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2693, "loss": 4.62217903137207, "lr": 0.0002, "elapsed_sec": 22381.864054441452, "step_time_sec": 8.23718545702286, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2694, "loss": 4.638241767883301, "lr": 0.0002, "elapsed_sec": 22390.102729797363, "step_time_sec": 8.238496222998947, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2695, "loss": 4.615408897399902, "lr": 0.0002, "elapsed_sec": 22398.341715812683, "step_time_sec": 8.23882109601982, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2696, "loss": 4.635746479034424, "lr": 0.0002, "elapsed_sec": 22406.574348688126, "step_time_sec": 8.232511538080871, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2697, "loss": 4.5299506187438965, "lr": 0.0002, "elapsed_sec": 22414.80690073967, "step_time_sec": 8.232371910009533, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2698, "loss": 4.597431659698486, "lr": 0.0002, "elapsed_sec": 22423.039699316025, "step_time_sec": 8.23268121399451, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2699, "loss": 4.577731132507324, "lr": 0.0002, "elapsed_sec": 22431.27318739891, "step_time_sec": 8.233265841961838, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2700, "loss": 4.5624823570251465, "lr": 0.0002, "elapsed_sec": 22439.51122736931, "step_time_sec": 8.23789992404636, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2701, "loss": 4.639501094818115, "lr": 0.0002, "elapsed_sec": 22447.745862722397, "step_time_sec": 8.234486469998956, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2702, "loss": 4.745110511779785, "lr": 0.0002, "elapsed_sec": 22455.984150648117, "step_time_sec": 8.238166998955421, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2703, "loss": 4.598275184631348, "lr": 0.0002, "elapsed_sec": 22464.22230911255, "step_time_sec": 8.237968335975893, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2704, "loss": 4.626202583312988, "lr": 0.0002, "elapsed_sec": 22472.459505558014, "step_time_sec": 8.23702023702208, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2705, "loss": 4.538136959075928, "lr": 0.0002, "elapsed_sec": 22480.692172527313, "step_time_sec": 8.232503523933701, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2706, "loss": 4.543801784515381, "lr": 0.0002, "elapsed_sec": 22488.926630735397, "step_time_sec": 8.234336592955515, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2707, "loss": 4.573238849639893, "lr": 0.0002, "elapsed_sec": 22497.16491651535, "step_time_sec": 8.238087709993124, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2708, "loss": 4.563626289367676, "lr": 0.0002, "elapsed_sec": 22505.397117376328, "step_time_sec": 8.232042185030878, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2709, "loss": 4.511753559112549, "lr": 0.0002, "elapsed_sec": 22513.63068175316, "step_time_sec": 8.233453227905557, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2710, "loss": 4.645218849182129, "lr": 0.0002, "elapsed_sec": 22521.86448955536, "step_time_sec": 8.233644936000928, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2711, "loss": 4.806094169616699, "lr": 0.0002, "elapsed_sec": 22530.097516298294, "step_time_sec": 8.232806019019336, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2712, "loss": 4.611745834350586, "lr": 0.0002, "elapsed_sec": 22538.334874391556, "step_time_sec": 8.237191827967763, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2713, "loss": 4.619546890258789, "lr": 0.0002, "elapsed_sec": 22546.57291316986, "step_time_sec": 8.23792916093953, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2714, "loss": 4.554239273071289, "lr": 0.0002, "elapsed_sec": 22554.8089094162, "step_time_sec": 8.235769705963321, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2715, "loss": 4.4658050537109375, "lr": 0.0002, "elapsed_sec": 22563.047573804855, "step_time_sec": 8.238514944096096, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2716, "loss": 4.706512451171875, "lr": 0.0002, "elapsed_sec": 22571.2854244709, "step_time_sec": 8.237685436033644, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2717, "loss": 4.796936988830566, "lr": 0.0002, "elapsed_sec": 22579.523121595383, "step_time_sec": 8.237492601037957, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2718, "loss": 4.580512523651123, "lr": 0.0002, "elapsed_sec": 22587.761064291, "step_time_sec": 8.237805302953348, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2719, "loss": 4.653039455413818, "lr": 0.0002, "elapsed_sec": 22595.996251106262, "step_time_sec": 8.23506046005059, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2720, "loss": 4.58126163482666, "lr": 0.0002, "elapsed_sec": 22604.23429131508, "step_time_sec": 8.237869607983157, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2721, "loss": 4.601686477661133, "lr": 0.0002, "elapsed_sec": 22612.469921827316, "step_time_sec": 8.23543543391861, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2722, "loss": 4.608139514923096, "lr": 0.0002, "elapsed_sec": 22620.70847415924, "step_time_sec": 8.238407408003695, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2723, "loss": 4.608706474304199, "lr": 0.0002, "elapsed_sec": 22628.945614099503, "step_time_sec": 8.2370789720444, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2724, "loss": 4.597056865692139, "lr": 0.0002, "elapsed_sec": 22637.18106341362, "step_time_sec": 8.23516900895629, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2725, "loss": 4.5979156494140625, "lr": 0.0002, "elapsed_sec": 22645.413248300552, "step_time_sec": 8.232034555985592, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2726, "loss": 4.578396797180176, "lr": 0.0002, "elapsed_sec": 22653.64631485939, "step_time_sec": 8.232909742975608, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2727, "loss": 4.705024242401123, "lr": 0.0002, "elapsed_sec": 22661.879989385605, "step_time_sec": 8.233517893007956, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2728, "loss": 4.538601875305176, "lr": 0.0002, "elapsed_sec": 22670.111369371414, "step_time_sec": 8.231211298960261, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2729, "loss": 4.5913166999816895, "lr": 0.0002, "elapsed_sec": 22678.342945575714, "step_time_sec": 8.23139824194368, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2730, "loss": 4.496891498565674, "lr": 0.0002, "elapsed_sec": 22686.576206445694, "step_time_sec": 8.233153166947886, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2731, "loss": 4.531677722930908, "lr": 0.0002, "elapsed_sec": 22694.80885243416, "step_time_sec": 8.232468603062443, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2732, "loss": 4.547509670257568, "lr": 0.0002, "elapsed_sec": 22703.04276752472, "step_time_sec": 8.233759836992249, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2733, "loss": 4.687865734100342, "lr": 0.0002, "elapsed_sec": 22711.275765895844, "step_time_sec": 8.232891016057692, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2734, "loss": 4.614645957946777, "lr": 0.0002, "elapsed_sec": 22719.514325857162, "step_time_sec": 8.238323665922508, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2735, "loss": 4.775249004364014, "lr": 0.0002, "elapsed_sec": 22727.753301143646, "step_time_sec": 8.238849444082007, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2736, "loss": 4.55155611038208, "lr": 0.0002, "elapsed_sec": 22735.991148471832, "step_time_sec": 8.237702735001221, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2737, "loss": 4.5707573890686035, "lr": 0.0002, "elapsed_sec": 22744.23028755188, "step_time_sec": 8.238975252956152, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2738, "loss": 4.53197717666626, "lr": 0.0002, "elapsed_sec": 22752.466414928436, "step_time_sec": 8.235961131053045, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2739, "loss": 4.547797203063965, "lr": 0.0002, "elapsed_sec": 22760.706505060196, "step_time_sec": 8.23990290297661, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2740, "loss": 4.559208869934082, "lr": 0.0002, "elapsed_sec": 22768.94344687462, "step_time_sec": 8.236776374047622, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2741, "loss": 4.486682891845703, "lr": 0.0002, "elapsed_sec": 22777.180188417435, "step_time_sec": 8.236579203046858, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2742, "loss": 4.406516075134277, "lr": 0.0002, "elapsed_sec": 22785.41797041893, "step_time_sec": 8.237625390058383, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2743, "loss": 4.661185264587402, "lr": 0.0002, "elapsed_sec": 22793.65580701828, "step_time_sec": 8.237644296023063, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2744, "loss": 4.582903861999512, "lr": 0.0002, "elapsed_sec": 22801.888934135437, "step_time_sec": 8.232971464982256, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2745, "loss": 4.478359222412109, "lr": 0.0002, "elapsed_sec": 22810.121147871017, "step_time_sec": 8.232065591961145, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2746, "loss": 4.63394832611084, "lr": 0.0002, "elapsed_sec": 22818.35421514511, "step_time_sec": 8.232913126936182, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2747, "loss": 4.52517032623291, "lr": 0.0002, "elapsed_sec": 22826.586124897003, "step_time_sec": 8.231778434943408, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2748, "loss": 4.550296306610107, "lr": 0.0002, "elapsed_sec": 22834.817835330963, "step_time_sec": 8.231505378964357, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2749, "loss": 4.521875858306885, "lr": 0.0002, "elapsed_sec": 22843.049854040146, "step_time_sec": 8.231830231961794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2750, "loss": 4.542964458465576, "lr": 0.0002, "elapsed_sec": 22851.28756213188, "step_time_sec": 8.237587535055354, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2751, "loss": 4.536910533905029, "lr": 0.0002, "elapsed_sec": 22859.525640249252, "step_time_sec": 8.237896200967953, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2752, "loss": 4.3999223709106445, "lr": 0.0002, "elapsed_sec": 22867.764797449112, "step_time_sec": 8.238988562952727, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2753, "loss": 4.51750373840332, "lr": 0.0002, "elapsed_sec": 22875.996232032776, "step_time_sec": 8.231317546102218, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2754, "loss": 4.508795261383057, "lr": 0.0002, "elapsed_sec": 22884.234142541885, "step_time_sec": 8.23771657596808, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2755, "loss": 4.444116115570068, "lr": 0.0002, "elapsed_sec": 22892.471160888672, "step_time_sec": 8.236876736977138, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2756, "loss": 4.529016971588135, "lr": 0.0002, "elapsed_sec": 22900.70292711258, "step_time_sec": 8.231607220950536, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2757, "loss": 4.708949089050293, "lr": 0.0002, "elapsed_sec": 22908.93426465988, "step_time_sec": 8.231192164006643, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2758, "loss": 4.707867622375488, "lr": 0.0002, "elapsed_sec": 22917.167347431183, "step_time_sec": 8.232895051012747, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2759, "loss": 4.605132579803467, "lr": 0.0002, "elapsed_sec": 22925.40342283249, "step_time_sec": 8.235926527995616, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2760, "loss": 4.534983158111572, "lr": 0.0002, "elapsed_sec": 22933.641748905182, "step_time_sec": 8.238177777035162, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2761, "loss": 4.487884521484375, "lr": 0.0002, "elapsed_sec": 22941.879175662994, "step_time_sec": 8.237200481933542, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2762, "loss": 4.625931262969971, "lr": 0.0002, "elapsed_sec": 22950.118369340897, "step_time_sec": 8.239083491964266, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2763, "loss": 4.606260299682617, "lr": 0.0002, "elapsed_sec": 22958.35064792633, "step_time_sec": 8.23207494802773, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2764, "loss": 4.607665061950684, "lr": 0.0002, "elapsed_sec": 22966.583300828934, "step_time_sec": 8.232528195017949, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2765, "loss": 4.561252117156982, "lr": 0.0002, "elapsed_sec": 22974.816076040268, "step_time_sec": 8.232565616024658, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2766, "loss": 4.534127712249756, "lr": 0.0002, "elapsed_sec": 22983.05515742302, "step_time_sec": 8.23896445601713, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2767, "loss": 4.526645183563232, "lr": 0.0002, "elapsed_sec": 22991.293476819992, "step_time_sec": 8.23815916897729, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2768, "loss": 4.7771220207214355, "lr": 0.0002, "elapsed_sec": 22999.529613018036, "step_time_sec": 8.235957314958796, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2769, "loss": 4.579810619354248, "lr": 0.0002, "elapsed_sec": 23007.76284980774, "step_time_sec": 8.233075174037367, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2770, "loss": 4.595240116119385, "lr": 0.0002, "elapsed_sec": 23015.995081186295, "step_time_sec": 8.232057126937434, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2771, "loss": 4.600870609283447, "lr": 0.0002, "elapsed_sec": 23024.228098630905, "step_time_sec": 8.232854804955423, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2772, "loss": 4.701169490814209, "lr": 0.0002, "elapsed_sec": 23032.459353923798, "step_time_sec": 8.231087545049377, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2773, "loss": 4.569125652313232, "lr": 0.0002, "elapsed_sec": 23040.692045927048, "step_time_sec": 8.23253177490551, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2774, "loss": 4.533157825469971, "lr": 0.0002, "elapsed_sec": 23048.92413020134, "step_time_sec": 8.231926770997234, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2775, "loss": 4.61191463470459, "lr": 0.0002, "elapsed_sec": 23057.1570353508, "step_time_sec": 8.232742897002026, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2776, "loss": 4.542243480682373, "lr": 0.0002, "elapsed_sec": 23065.393286943436, "step_time_sec": 8.23610712599475, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2777, "loss": 4.618340969085693, "lr": 0.0002, "elapsed_sec": 23073.632142066956, "step_time_sec": 8.238679439993575, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2778, "loss": 4.737907886505127, "lr": 0.0002, "elapsed_sec": 23081.864679574966, "step_time_sec": 8.232396031962708, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2779, "loss": 4.558767795562744, "lr": 0.0002, "elapsed_sec": 23090.09669327736, "step_time_sec": 8.2318246740615, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2780, "loss": 4.587095737457275, "lr": 0.0002, "elapsed_sec": 23098.329704761505, "step_time_sec": 8.232819361961447, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2781, "loss": 4.553696632385254, "lr": 0.0002, "elapsed_sec": 23106.56591463089, "step_time_sec": 8.236101705930196, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2782, "loss": 4.582793235778809, "lr": 0.0002, "elapsed_sec": 23114.802096128464, "step_time_sec": 8.236008389969356, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2783, "loss": 4.617706298828125, "lr": 0.0002, "elapsed_sec": 23123.040399074554, "step_time_sec": 8.238079537986778, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2784, "loss": 4.575710296630859, "lr": 0.0002, "elapsed_sec": 23131.271816253662, "step_time_sec": 8.231301766005345, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2785, "loss": 4.543135643005371, "lr": 0.0002, "elapsed_sec": 23139.50598692894, "step_time_sec": 8.233983734971844, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2786, "loss": 4.424354076385498, "lr": 0.0002, "elapsed_sec": 23147.743685007095, "step_time_sec": 8.237516594002955, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2787, "loss": 4.730440139770508, "lr": 0.0002, "elapsed_sec": 23155.977268218994, "step_time_sec": 8.23342132603284, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2788, "loss": 4.580309867858887, "lr": 0.0002, "elapsed_sec": 23164.208543539047, "step_time_sec": 8.231168602011167, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2789, "loss": 4.533012866973877, "lr": 0.0002, "elapsed_sec": 23172.44116282463, "step_time_sec": 8.232427928014658, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2790, "loss": 4.717680931091309, "lr": 0.0002, "elapsed_sec": 23180.679154872894, "step_time_sec": 8.237818298046477, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2791, "loss": 4.705749034881592, "lr": 0.0002, "elapsed_sec": 23188.918487548828, "step_time_sec": 8.239153200993314, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2792, "loss": 4.563006401062012, "lr": 0.0002, "elapsed_sec": 23197.154762268066, "step_time_sec": 8.236093080951832, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2793, "loss": 4.560028553009033, "lr": 0.0002, "elapsed_sec": 23205.391627311707, "step_time_sec": 8.236728035029955, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2794, "loss": 4.560571193695068, "lr": 0.0002, "elapsed_sec": 23213.629623889923, "step_time_sec": 8.237884328933433, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2795, "loss": 4.624244689941406, "lr": 0.0002, "elapsed_sec": 23221.867353200912, "step_time_sec": 8.237494240049273, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2796, "loss": 4.479955673217773, "lr": 0.0002, "elapsed_sec": 23230.106213331223, "step_time_sec": 8.238707438926212, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2797, "loss": 4.637572765350342, "lr": 0.0002, "elapsed_sec": 23238.338591337204, "step_time_sec": 8.232216431992128, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2798, "loss": 4.615710735321045, "lr": 0.0002, "elapsed_sec": 23246.572087287903, "step_time_sec": 8.23339600500185, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2799, "loss": 4.497077941894531, "lr": 0.0002, "elapsed_sec": 23254.80454349518, "step_time_sec": 8.232281728065573, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2800, "loss": 4.584300518035889, "lr": 0.0002, "elapsed_sec": 23263.037211179733, "step_time_sec": 8.232468201080337, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2801, "loss": 4.525637149810791, "lr": 0.0002, "elapsed_sec": 23271.269600868225, "step_time_sec": 8.232245099032298, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2802, "loss": 4.52586030960083, "lr": 0.0002, "elapsed_sec": 23279.501472473145, "step_time_sec": 8.231723102973774, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2803, "loss": 4.515237331390381, "lr": 0.0002, "elapsed_sec": 23287.73991203308, "step_time_sec": 8.238285503000952, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2804, "loss": 4.584395408630371, "lr": 0.0002, "elapsed_sec": 23295.978476047516, "step_time_sec": 8.23841874790378, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2805, "loss": 4.612459182739258, "lr": 0.0002, "elapsed_sec": 23304.217046022415, "step_time_sec": 8.238363570999354, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2806, "loss": 4.6666717529296875, "lr": 0.0002, "elapsed_sec": 23312.455098628998, "step_time_sec": 8.23792185401544, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2807, "loss": 4.570528030395508, "lr": 0.0002, "elapsed_sec": 23320.69207549095, "step_time_sec": 8.23680574807804, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2808, "loss": 4.657318115234375, "lr": 0.0002, "elapsed_sec": 23328.93024468422, "step_time_sec": 8.2380022960715, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2809, "loss": 4.63351583480835, "lr": 0.0002, "elapsed_sec": 23337.168317317963, "step_time_sec": 8.237893266021274, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2810, "loss": 4.543303966522217, "lr": 0.0002, "elapsed_sec": 23345.404524564743, "step_time_sec": 8.236061003059149, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2811, "loss": 4.567899227142334, "lr": 0.0002, "elapsed_sec": 23353.64282298088, "step_time_sec": 8.238166003022343, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2812, "loss": 4.496588230133057, "lr": 0.0002, "elapsed_sec": 23361.879769325256, "step_time_sec": 8.236744625959545, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2813, "loss": 4.463970184326172, "lr": 0.0002, "elapsed_sec": 23370.11300444603, "step_time_sec": 8.233045921078883, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2814, "loss": 4.5308518409729, "lr": 0.0002, "elapsed_sec": 23378.34983110428, "step_time_sec": 8.236656239023432, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2815, "loss": 4.681124687194824, "lr": 0.0002, "elapsed_sec": 23386.58850812912, "step_time_sec": 8.238565027015284, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2816, "loss": 4.585830211639404, "lr": 0.0002, "elapsed_sec": 23394.824771881104, "step_time_sec": 8.236088747973554, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2817, "loss": 4.545302391052246, "lr": 0.0002, "elapsed_sec": 23403.062894821167, "step_time_sec": 8.23792407498695, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2818, "loss": 4.751219272613525, "lr": 0.0002, "elapsed_sec": 23411.300993680954, "step_time_sec": 8.237934628967196, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2819, "loss": 4.562956809997559, "lr": 0.0002, "elapsed_sec": 23419.54060816765, "step_time_sec": 8.239450508961454, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2820, "loss": 4.598691463470459, "lr": 0.0002, "elapsed_sec": 23427.77859544754, "step_time_sec": 8.237841136055067, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2821, "loss": 4.588897228240967, "lr": 0.0002, "elapsed_sec": 23436.017243146896, "step_time_sec": 8.23843939602375, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2822, "loss": 4.5486016273498535, "lr": 0.0002, "elapsed_sec": 23444.250252008438, "step_time_sec": 8.232868374092504, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2823, "loss": 4.523272514343262, "lr": 0.0002, "elapsed_sec": 23452.486373901367, "step_time_sec": 8.235914040007629, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2824, "loss": 4.574197292327881, "lr": 0.0002, "elapsed_sec": 23460.71927332878, "step_time_sec": 8.232752972049639, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2825, "loss": 4.463644504547119, "lr": 0.0002, "elapsed_sec": 23468.950484275818, "step_time_sec": 8.231029360089451, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2826, "loss": 4.60476541519165, "lr": 0.0002, "elapsed_sec": 23477.184252738953, "step_time_sec": 8.23360917204991, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2827, "loss": 4.605281829833984, "lr": 0.0002, "elapsed_sec": 23485.42353463173, "step_time_sec": 8.239132993970998, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2828, "loss": 4.461027145385742, "lr": 0.0002, "elapsed_sec": 23493.660312652588, "step_time_sec": 8.23661306698341, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2829, "loss": 4.621217250823975, "lr": 0.0002, "elapsed_sec": 23501.897875785828, "step_time_sec": 8.237400816055015, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2830, "loss": 4.54816198348999, "lr": 0.0002, "elapsed_sec": 23510.13022851944, "step_time_sec": 8.232175368000753, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2831, "loss": 4.546955108642578, "lr": 0.0002, "elapsed_sec": 23518.363129138947, "step_time_sec": 8.23272490000818, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2832, "loss": 4.553386688232422, "lr": 0.0002, "elapsed_sec": 23526.595696926117, "step_time_sec": 8.232463384978473, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2833, "loss": 4.5220537185668945, "lr": 0.0002, "elapsed_sec": 23534.83033680916, "step_time_sec": 8.234464359935373, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2834, "loss": 4.514069080352783, "lr": 0.0002, "elapsed_sec": 23543.066467285156, "step_time_sec": 8.235938757075928, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2835, "loss": 4.58001184463501, "lr": 0.0002, "elapsed_sec": 23551.298452615738, "step_time_sec": 8.231810092926025, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2836, "loss": 4.449817657470703, "lr": 0.0002, "elapsed_sec": 23559.531408786774, "step_time_sec": 8.232798035023734, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2837, "loss": 4.602397918701172, "lr": 0.0002, "elapsed_sec": 23567.76366043091, "step_time_sec": 8.232088705059141, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2838, "loss": 4.475444316864014, "lr": 0.0002, "elapsed_sec": 23575.99697327614, "step_time_sec": 8.233131899964064, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2839, "loss": 4.586663722991943, "lr": 0.0002, "elapsed_sec": 23584.235100984573, "step_time_sec": 8.238000549958088, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2840, "loss": 4.569156646728516, "lr": 0.0002, "elapsed_sec": 23592.471738100052, "step_time_sec": 8.23645387205761, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2841, "loss": 4.536271572113037, "lr": 0.0002, "elapsed_sec": 23600.705508947372, "step_time_sec": 8.233623845968395, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2842, "loss": 4.552751541137695, "lr": 0.0002, "elapsed_sec": 23608.942907094955, "step_time_sec": 8.23722697806079, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2843, "loss": 4.423854827880859, "lr": 0.0002, "elapsed_sec": 23617.18222618103, "step_time_sec": 8.239149093045853, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2844, "loss": 4.633228778839111, "lr": 0.0002, "elapsed_sec": 23625.416283369064, "step_time_sec": 8.233919901074842, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2845, "loss": 4.538827896118164, "lr": 0.0002, "elapsed_sec": 23633.649057626724, "step_time_sec": 8.232605901081115, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2846, "loss": 4.60315465927124, "lr": 0.0002, "elapsed_sec": 23641.88035964966, "step_time_sec": 8.231176896952093, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2847, "loss": 4.431577682495117, "lr": 0.0002, "elapsed_sec": 23650.113597869873, "step_time_sec": 8.233076740987599, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2848, "loss": 4.590188503265381, "lr": 0.0002, "elapsed_sec": 23658.35058927536, "step_time_sec": 8.236795958015136, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2849, "loss": 4.62045955657959, "lr": 0.0002, "elapsed_sec": 23666.583738327026, "step_time_sec": 8.233048345078714, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2850, "loss": 4.519741058349609, "lr": 0.0002, "elapsed_sec": 23674.815529108047, "step_time_sec": 8.23158800008241, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2851, "loss": 4.527105808258057, "lr": 0.0002, "elapsed_sec": 23683.053079366684, "step_time_sec": 8.237362120999023, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2852, "loss": 4.688686847686768, "lr": 0.0002, "elapsed_sec": 23691.290593385696, "step_time_sec": 8.237356933997944, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2853, "loss": 4.480648517608643, "lr": 0.0002, "elapsed_sec": 23699.52941775322, "step_time_sec": 8.23865322000347, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2854, "loss": 4.673759937286377, "lr": 0.0002, "elapsed_sec": 23707.766478061676, "step_time_sec": 8.2368966039503, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2855, "loss": 4.518076419830322, "lr": 0.0002, "elapsed_sec": 23716.0038189888, "step_time_sec": 8.237217928050086, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2856, "loss": 4.580362796783447, "lr": 0.0002, "elapsed_sec": 23724.24349951744, "step_time_sec": 8.23948011896573, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2857, "loss": 4.570213794708252, "lr": 0.0002, "elapsed_sec": 23732.480031728745, "step_time_sec": 8.23635553510394, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2858, "loss": 4.465444087982178, "lr": 0.0002, "elapsed_sec": 23740.713299274445, "step_time_sec": 8.233102572034113, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2859, "loss": 4.641592979431152, "lr": 0.0002, "elapsed_sec": 23748.950000047684, "step_time_sec": 8.23657192196697, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2860, "loss": 4.469133377075195, "lr": 0.0002, "elapsed_sec": 23757.18823480606, "step_time_sec": 8.238030139938928, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2861, "loss": 4.6464524269104, "lr": 0.0002, "elapsed_sec": 23765.42574286461, "step_time_sec": 8.237356005003676, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2862, "loss": 4.470635414123535, "lr": 0.0002, "elapsed_sec": 23773.65881252289, "step_time_sec": 8.232928613899276, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2863, "loss": 4.623138904571533, "lr": 0.0002, "elapsed_sec": 23781.890913009644, "step_time_sec": 8.23196365404874, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2864, "loss": 4.504180908203125, "lr": 0.0002, "elapsed_sec": 23790.123690366745, "step_time_sec": 8.232614026986994, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2865, "loss": 4.604401588439941, "lr": 0.0002, "elapsed_sec": 23798.355380773544, "step_time_sec": 8.231484627933241, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2866, "loss": 4.566237449645996, "lr": 0.0002, "elapsed_sec": 23806.58800816536, "step_time_sec": 8.23251815396361, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2867, "loss": 4.580059051513672, "lr": 0.0002, "elapsed_sec": 23814.820063114166, "step_time_sec": 8.231872985023074, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2868, "loss": 4.540450572967529, "lr": 0.0002, "elapsed_sec": 23823.059003829956, "step_time_sec": 8.238733535981737, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2869, "loss": 4.640474796295166, "lr": 0.0002, "elapsed_sec": 23831.297659158707, "step_time_sec": 8.238497978076339, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2870, "loss": 4.442908763885498, "lr": 0.0002, "elapsed_sec": 23839.5341796875, "step_time_sec": 8.236345266923308, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2871, "loss": 4.510067462921143, "lr": 0.0002, "elapsed_sec": 23847.772965669632, "step_time_sec": 8.238638879032806, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2872, "loss": 4.572822570800781, "lr": 0.0002, "elapsed_sec": 23856.008004665375, "step_time_sec": 8.234929128084332, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2873, "loss": 4.517200469970703, "lr": 0.0002, "elapsed_sec": 23864.24089884758, "step_time_sec": 8.232708944007754, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2874, "loss": 4.504609107971191, "lr": 0.0002, "elapsed_sec": 23872.477499485016, "step_time_sec": 8.236434785998426, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2875, "loss": 4.580841541290283, "lr": 0.0002, "elapsed_sec": 23880.71651816368, "step_time_sec": 8.238863270031288, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2876, "loss": 4.579373359680176, "lr": 0.0002, "elapsed_sec": 23888.95347929001, "step_time_sec": 8.236767039983533, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2877, "loss": 4.634947776794434, "lr": 0.0002, "elapsed_sec": 23897.18617463112, "step_time_sec": 8.23258190404158, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2878, "loss": 5.088428497314453, "lr": 0.0002, "elapsed_sec": 23905.418450832367, "step_time_sec": 8.232091337093152, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2879, "loss": 4.660604476928711, "lr": 0.0002, "elapsed_sec": 23913.651401996613, "step_time_sec": 8.232749828952365, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2880, "loss": 4.694765090942383, "lr": 0.0002, "elapsed_sec": 23921.883348703384, "step_time_sec": 8.231830257922411, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2881, "loss": 4.502465724945068, "lr": 0.0002, "elapsed_sec": 23930.12289261818, "step_time_sec": 8.239376018987969, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2882, "loss": 4.583756923675537, "lr": 0.0002, "elapsed_sec": 23938.35955953598, "step_time_sec": 8.236506044049747, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2883, "loss": 4.586366176605225, "lr": 0.0002, "elapsed_sec": 23946.5977973938, "step_time_sec": 8.238100496935658, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2884, "loss": 4.682885646820068, "lr": 0.0002, "elapsed_sec": 23954.82948231697, "step_time_sec": 8.231475524022244, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2885, "loss": 4.547609806060791, "lr": 0.0002, "elapsed_sec": 23963.06212759018, "step_time_sec": 8.232494049938396, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2886, "loss": 4.540558815002441, "lr": 0.0002, "elapsed_sec": 23971.294414758682, "step_time_sec": 8.232104626018554, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2887, "loss": 4.6022820472717285, "lr": 0.0002, "elapsed_sec": 23979.53340268135, "step_time_sec": 8.238831515074708, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2888, "loss": 4.539041996002197, "lr": 0.0002, "elapsed_sec": 23987.77111005783, "step_time_sec": 8.237521742004901, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2889, "loss": 4.578585624694824, "lr": 0.0002, "elapsed_sec": 23996.008900642395, "step_time_sec": 8.237621954991482, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2890, "loss": 4.511549949645996, "lr": 0.0002, "elapsed_sec": 24004.240912675858, "step_time_sec": 8.231891090981662, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2891, "loss": 4.504777908325195, "lr": 0.0002, "elapsed_sec": 24012.47286438942, "step_time_sec": 8.231772875995375, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2892, "loss": 4.602393627166748, "lr": 0.0002, "elapsed_sec": 24020.706508874893, "step_time_sec": 8.233519371016882, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2893, "loss": 4.4407219886779785, "lr": 0.0002, "elapsed_sec": 24028.941213607788, "step_time_sec": 8.234497967059724, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2894, "loss": 4.580251216888428, "lr": 0.0002, "elapsed_sec": 24037.173687934875, "step_time_sec": 8.23234990495257, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2895, "loss": 4.494787693023682, "lr": 0.0002, "elapsed_sec": 24045.405781507492, "step_time_sec": 8.23193709901534, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2896, "loss": 4.48171854019165, "lr": 0.0002, "elapsed_sec": 24053.63796544075, "step_time_sec": 8.231976348906755, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2897, "loss": 4.449864387512207, "lr": 0.0002, "elapsed_sec": 24061.869841337204, "step_time_sec": 8.231746667996049, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2898, "loss": 4.55666971206665, "lr": 0.0002, "elapsed_sec": 24070.101994514465, "step_time_sec": 8.231980473035946, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2899, "loss": 4.469262599945068, "lr": 0.0002, "elapsed_sec": 24078.341220617294, "step_time_sec": 8.239044431014918, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2900, "loss": 4.436347484588623, "lr": 0.0002, "elapsed_sec": 24086.579191446304, "step_time_sec": 8.237868129974231, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2901, "loss": 4.591381549835205, "lr": 0.0002, "elapsed_sec": 24094.81728720665, "step_time_sec": 8.237904769950546, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2902, "loss": 4.578990936279297, "lr": 0.0002, "elapsed_sec": 24103.04959177971, "step_time_sec": 8.232146073016338, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2903, "loss": 4.468789577484131, "lr": 0.0002, "elapsed_sec": 24111.28142261505, "step_time_sec": 8.231673169066198, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2904, "loss": 4.700981616973877, "lr": 0.0002, "elapsed_sec": 24119.514170885086, "step_time_sec": 8.232578556984663, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2905, "loss": 4.65245246887207, "lr": 0.0002, "elapsed_sec": 24127.74751663208, "step_time_sec": 8.233173883985728, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2906, "loss": 4.474198818206787, "lr": 0.0002, "elapsed_sec": 24135.980273723602, "step_time_sec": 8.232600224087946, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2907, "loss": 4.488051891326904, "lr": 0.0002, "elapsed_sec": 24144.214241027832, "step_time_sec": 8.233810169971548, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2908, "loss": 4.551340103149414, "lr": 0.0002, "elapsed_sec": 24152.4526553154, "step_time_sec": 8.238279976998456, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2909, "loss": 4.5301194190979, "lr": 0.0002, "elapsed_sec": 24160.690913915634, "step_time_sec": 8.238082348019816, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2910, "loss": 4.570316314697266, "lr": 0.0002, "elapsed_sec": 24168.92970252037, "step_time_sec": 8.2386166540673, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2911, "loss": 4.387472152709961, "lr": 0.0002, "elapsed_sec": 24177.167202949524, "step_time_sec": 8.237343317014165, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2912, "loss": 4.5603532791137695, "lr": 0.0002, "elapsed_sec": 24185.406754493713, "step_time_sec": 8.2393923610216, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2913, "loss": 4.569360256195068, "lr": 0.0002, "elapsed_sec": 24193.64593720436, "step_time_sec": 8.239030577009544, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2914, "loss": 4.616202354431152, "lr": 0.0002, "elapsed_sec": 24201.883907079697, "step_time_sec": 8.237768057035282, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2915, "loss": 4.568603515625, "lr": 0.0002, "elapsed_sec": 24210.122901678085, "step_time_sec": 8.23884023190476, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2916, "loss": 4.60083532333374, "lr": 0.0002, "elapsed_sec": 24218.360239982605, "step_time_sec": 8.23718078492675, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2917, "loss": 4.575170040130615, "lr": 0.0002, "elapsed_sec": 24226.599014997482, "step_time_sec": 8.238654762040824, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2918, "loss": 4.695653438568115, "lr": 0.0002, "elapsed_sec": 24234.837104797363, "step_time_sec": 8.237877675099298, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2919, "loss": 4.531646728515625, "lr": 0.0002, "elapsed_sec": 24243.069667816162, "step_time_sec": 8.232409978983924, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2920, "loss": 4.476004600524902, "lr": 0.0002, "elapsed_sec": 24251.301764965057, "step_time_sec": 8.231932873022743, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2921, "loss": 4.538980007171631, "lr": 0.0002, "elapsed_sec": 24259.534422397614, "step_time_sec": 8.23249433701858, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2922, "loss": 4.596266746520996, "lr": 0.0002, "elapsed_sec": 24267.772491693497, "step_time_sec": 8.237889871001244, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2923, "loss": 4.713606834411621, "lr": 0.0002, "elapsed_sec": 24276.01049399376, "step_time_sec": 8.237851312034763, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2924, "loss": 4.268945217132568, "lr": 0.0002, "elapsed_sec": 24284.24758696556, "step_time_sec": 8.236863552010618, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2925, "loss": 4.663768768310547, "lr": 0.0002, "elapsed_sec": 24292.48411011696, "step_time_sec": 8.236379935988225, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2926, "loss": 4.549192428588867, "lr": 0.0002, "elapsed_sec": 24300.72248029709, "step_time_sec": 8.238194991950877, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2927, "loss": 4.543569087982178, "lr": 0.0002, "elapsed_sec": 24308.959488391876, "step_time_sec": 8.236808672081679, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2928, "loss": 4.443147659301758, "lr": 0.0002, "elapsed_sec": 24317.199233531952, "step_time_sec": 8.239606953924522, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2929, "loss": 4.457735061645508, "lr": 0.0002, "elapsed_sec": 24325.431317806244, "step_time_sec": 8.231945825973526, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2930, "loss": 4.604200839996338, "lr": 0.0002, "elapsed_sec": 24333.66473340988, "step_time_sec": 8.233223267016001, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2931, "loss": 4.566765308380127, "lr": 0.0002, "elapsed_sec": 24341.896524190903, "step_time_sec": 8.231594673008658, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2932, "loss": 4.49680757522583, "lr": 0.0002, "elapsed_sec": 24350.135253429413, "step_time_sec": 8.23858415801078, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2933, "loss": 4.404208660125732, "lr": 0.0002, "elapsed_sec": 24358.373763799667, "step_time_sec": 8.238373856060207, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2934, "loss": 4.38621187210083, "lr": 0.0002, "elapsed_sec": 24366.611283779144, "step_time_sec": 8.2373667700449, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2935, "loss": 4.350447654724121, "lr": 0.0002, "elapsed_sec": 24374.850809574127, "step_time_sec": 8.239319627988152, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2936, "loss": 4.6663432121276855, "lr": 0.0002, "elapsed_sec": 24383.082080364227, "step_time_sec": 8.231094124959782, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2937, "loss": 4.575453281402588, "lr": 0.0002, "elapsed_sec": 24391.315224647522, "step_time_sec": 8.23296694306191, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2938, "loss": 4.609690189361572, "lr": 0.0002, "elapsed_sec": 24399.54649233818, "step_time_sec": 8.231108892010525, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2939, "loss": 4.4644575119018555, "lr": 0.0002, "elapsed_sec": 24407.7859416008, "step_time_sec": 8.23928366205655, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2940, "loss": 4.490154266357422, "lr": 0.0002, "elapsed_sec": 24416.023778676987, "step_time_sec": 8.237721796962433, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2941, "loss": 4.439341068267822, "lr": 0.0002, "elapsed_sec": 24424.263069868088, "step_time_sec": 8.239078711019829, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2942, "loss": 4.555778503417969, "lr": 0.0002, "elapsed_sec": 24432.500514030457, "step_time_sec": 8.237288626027294, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2943, "loss": 4.460265159606934, "lr": 0.0002, "elapsed_sec": 24440.739659309387, "step_time_sec": 8.238994652987458, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2944, "loss": 4.37692403793335, "lr": 0.0002, "elapsed_sec": 24448.976630449295, "step_time_sec": 8.236743296030909, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2945, "loss": 4.547465801239014, "lr": 0.0002, "elapsed_sec": 24457.208461999893, "step_time_sec": 8.23173141409643, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2946, "loss": 4.53121280670166, "lr": 0.0002, "elapsed_sec": 24465.441552877426, "step_time_sec": 8.232882095035166, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2947, "loss": 4.404312610626221, "lr": 0.0002, "elapsed_sec": 24473.673273086548, "step_time_sec": 8.231557353050448, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2948, "loss": 4.5166473388671875, "lr": 0.0002, "elapsed_sec": 24481.911256074905, "step_time_sec": 8.237818797002546, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2949, "loss": 4.502830982208252, "lr": 0.0002, "elapsed_sec": 24490.148238897324, "step_time_sec": 8.236824055900797, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2950, "loss": 4.569853782653809, "lr": 0.0002, "elapsed_sec": 24498.3872590065, "step_time_sec": 8.238924699020572, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2951, "loss": 4.546796798706055, "lr": 0.0002, "elapsed_sec": 24506.624009132385, "step_time_sec": 8.236599685042165, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2952, "loss": 4.492567539215088, "lr": 0.0002, "elapsed_sec": 24514.862254858017, "step_time_sec": 8.238024630001746, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2953, "loss": 4.396430492401123, "lr": 0.0002, "elapsed_sec": 24523.10011792183, "step_time_sec": 8.23768406000454, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2954, "loss": 4.596899032592773, "lr": 0.0002, "elapsed_sec": 24531.338190555573, "step_time_sec": 8.237898812978528, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2955, "loss": 4.530564785003662, "lr": 0.0002, "elapsed_sec": 24539.57130765915, "step_time_sec": 8.23294191795867, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2956, "loss": 4.480418682098389, "lr": 0.0002, "elapsed_sec": 24547.808364629745, "step_time_sec": 8.236913923057728, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2957, "loss": 4.531280040740967, "lr": 0.0002, "elapsed_sec": 24556.04104948044, "step_time_sec": 8.232528602005914, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2958, "loss": 4.582812309265137, "lr": 0.0002, "elapsed_sec": 24564.277022600174, "step_time_sec": 8.235795169952326, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2959, "loss": 4.518925189971924, "lr": 0.0002, "elapsed_sec": 24572.516271829605, "step_time_sec": 8.23911608196795, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2960, "loss": 4.421298503875732, "lr": 0.0002, "elapsed_sec": 24580.753163337708, "step_time_sec": 8.236746900016442, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2961, "loss": 4.431709289550781, "lr": 0.0002, "elapsed_sec": 24588.989203453064, "step_time_sec": 8.235870709992014, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2962, "loss": 4.426600933074951, "lr": 0.0002, "elapsed_sec": 24597.22730398178, "step_time_sec": 8.237871055025607, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2963, "loss": 4.616915702819824, "lr": 0.0002, "elapsed_sec": 24605.465441942215, "step_time_sec": 8.237988949986175, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2964, "loss": 4.731898784637451, "lr": 0.0002, "elapsed_sec": 24613.703649282455, "step_time_sec": 8.238089613965712, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2965, "loss": 4.506553649902344, "lr": 0.0002, "elapsed_sec": 24621.934748888016, "step_time_sec": 8.230898123933002, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2966, "loss": 4.628877639770508, "lr": 0.0002, "elapsed_sec": 24630.16803908348, "step_time_sec": 8.233156964997761, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2967, "loss": 4.4979023933410645, "lr": 0.0002, "elapsed_sec": 24638.40025949478, "step_time_sec": 8.23199347499758, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2968, "loss": 4.5015387535095215, "lr": 0.0002, "elapsed_sec": 24646.632348537445, "step_time_sec": 8.231987839913927, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2969, "loss": 4.405902862548828, "lr": 0.0002, "elapsed_sec": 24654.864622592926, "step_time_sec": 8.232082389993593, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2970, "loss": 4.583550930023193, "lr": 0.0002, "elapsed_sec": 24663.097636699677, "step_time_sec": 8.23283088405151, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2971, "loss": 4.562112331390381, "lr": 0.0002, "elapsed_sec": 24671.33009147644, "step_time_sec": 8.232288973056711, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2972, "loss": 4.525388240814209, "lr": 0.0002, "elapsed_sec": 24679.564551591873, "step_time_sec": 8.234354067011736, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2973, "loss": 4.479517459869385, "lr": 0.0002, "elapsed_sec": 24687.797810554504, "step_time_sec": 8.233072406961583, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2974, "loss": 4.655386924743652, "lr": 0.0002, "elapsed_sec": 24696.03497004509, "step_time_sec": 8.236964853014797, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2975, "loss": 4.49717903137207, "lr": 0.0002, "elapsed_sec": 24704.272342205048, "step_time_sec": 8.237196273985319, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2976, "loss": 4.54261589050293, "lr": 0.0002, "elapsed_sec": 24712.510947942734, "step_time_sec": 8.238476207014173, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2977, "loss": 4.452041149139404, "lr": 0.0002, "elapsed_sec": 24720.74175953865, "step_time_sec": 8.230641035013832, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2978, "loss": 4.5120625495910645, "lr": 0.0002, "elapsed_sec": 24728.978042125702, "step_time_sec": 8.236103147966787, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2979, "loss": 4.377809524536133, "lr": 0.0002, "elapsed_sec": 24737.21521449089, "step_time_sec": 8.23702838702593, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2980, "loss": 4.647897720336914, "lr": 0.0002, "elapsed_sec": 24745.44775056839, "step_time_sec": 8.232373396982439, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2981, "loss": 4.759984016418457, "lr": 0.0002, "elapsed_sec": 24753.67954325676, "step_time_sec": 8.231620487989858, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2982, "loss": 4.438902854919434, "lr": 0.0002, "elapsed_sec": 24761.91217660904, "step_time_sec": 8.232455360936001, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2983, "loss": 4.589939117431641, "lr": 0.0002, "elapsed_sec": 24770.14430308342, "step_time_sec": 8.231981859076768, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2984, "loss": 4.559035301208496, "lr": 0.0002, "elapsed_sec": 24778.376333236694, "step_time_sec": 8.23185281699989, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2985, "loss": 4.5039496421813965, "lr": 0.0002, "elapsed_sec": 24786.60948896408, "step_time_sec": 8.233037195983343, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2986, "loss": 4.600565433502197, "lr": 0.0002, "elapsed_sec": 24794.84230542183, "step_time_sec": 8.232598940958269, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2987, "loss": 4.517284870147705, "lr": 0.0002, "elapsed_sec": 24803.078988790512, "step_time_sec": 8.236530867987312, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2988, "loss": 4.468228816986084, "lr": 0.0002, "elapsed_sec": 24811.316752910614, "step_time_sec": 8.237589325988665, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2989, "loss": 4.485027313232422, "lr": 0.0002, "elapsed_sec": 24819.547914505005, "step_time_sec": 8.231011861003935, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2990, "loss": 4.411196708679199, "lr": 0.0002, "elapsed_sec": 24827.78009915352, "step_time_sec": 8.232053871033713, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2991, "loss": 4.451189041137695, "lr": 0.0002, "elapsed_sec": 24836.012746810913, "step_time_sec": 8.232474236981943, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2992, "loss": 4.525766849517822, "lr": 0.0002, "elapsed_sec": 24844.244763612747, "step_time_sec": 8.231838703970425, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2993, "loss": 4.469139575958252, "lr": 0.0002, "elapsed_sec": 24852.48096728325, "step_time_sec": 8.236042615957558, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2994, "loss": 4.424123764038086, "lr": 0.0002, "elapsed_sec": 24860.718559265137, "step_time_sec": 8.237479433068074, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2995, "loss": 4.5482025146484375, "lr": 0.0002, "elapsed_sec": 24868.953104019165, "step_time_sec": 8.234338228940032, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2996, "loss": 4.4410223960876465, "lr": 0.0002, "elapsed_sec": 24877.191420793533, "step_time_sec": 8.23819843295496, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2997, "loss": 4.562047958374023, "lr": 0.0002, "elapsed_sec": 24885.426627635956, "step_time_sec": 8.235014292993583, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2998, "loss": 4.606184005737305, "lr": 0.0002, "elapsed_sec": 24893.663868665695, "step_time_sec": 8.237073490978219, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2999, "loss": 4.425201416015625, "lr": 0.0002, "elapsed_sec": 24901.89974141121, "step_time_sec": 8.235723193967715, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3000, "loss": 4.397818565368652, "lr": 0.0002, "elapsed_sec": 24910.13749551773, "step_time_sec": 46.18747211492155, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": true, "validation_ran": true, "probe_ran": true, "probe_elapsed_sec": 0.9917754379566759, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3001, "loss": 4.451274394989014, "lr": 0.0002, "elapsed_sec": 24956.326476097107, "step_time_sec": 8.23903248901479, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3002, "loss": 4.458016395568848, "lr": 0.0002, "elapsed_sec": 24964.556704998016, "step_time_sec": 8.23000323609449, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3003, "loss": 4.454651355743408, "lr": 0.0002, "elapsed_sec": 24972.789597034454, "step_time_sec": 8.232760251034051, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3004, "loss": 4.550535678863525, "lr": 0.0002, "elapsed_sec": 24981.02158021927, "step_time_sec": 8.231824396993034, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3005, "loss": 4.709826469421387, "lr": 0.0002, "elapsed_sec": 24989.254057884216, "step_time_sec": 8.23230194405187, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3006, "loss": 4.53294038772583, "lr": 0.0002, "elapsed_sec": 24997.485641479492, "step_time_sec": 8.231449996936135, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3007, "loss": 4.523214340209961, "lr": 0.0002, "elapsed_sec": 25005.717119932175, "step_time_sec": 8.231307227048092, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3008, "loss": 4.452311038970947, "lr": 0.0002, "elapsed_sec": 25013.949867486954, "step_time_sec": 8.23252667800989, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3009, "loss": 4.53781795501709, "lr": 0.0002, "elapsed_sec": 25022.184262514114, "step_time_sec": 8.234254813054577, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3010, "loss": 4.596030235290527, "lr": 0.0002, "elapsed_sec": 25030.41894865036, "step_time_sec": 8.234518898068927, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3011, "loss": 4.575403213500977, "lr": 0.0002, "elapsed_sec": 25038.65785384178, "step_time_sec": 8.238780985004269, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3012, "loss": 4.42482328414917, "lr": 0.0002, "elapsed_sec": 25046.891883134842, "step_time_sec": 8.233785895979963, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3013, "loss": 4.480098247528076, "lr": 0.0002, "elapsed_sec": 25055.12374973297, "step_time_sec": 8.231765315053053, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3014, "loss": 4.506313323974609, "lr": 0.0002, "elapsed_sec": 25063.359545469284, "step_time_sec": 8.235629393020645, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3015, "loss": 4.421623706817627, "lr": 0.0002, "elapsed_sec": 25071.59729409218, "step_time_sec": 8.237569137942046, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3016, "loss": 4.537124156951904, "lr": 0.0002, "elapsed_sec": 25079.834140777588, "step_time_sec": 8.236672675004229, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3017, "loss": 4.456048965454102, "lr": 0.0002, "elapsed_sec": 25088.06966805458, "step_time_sec": 8.235393982031383, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3018, "loss": 4.590367317199707, "lr": 0.0002, "elapsed_sec": 25096.30349469185, "step_time_sec": 8.233669638051651, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3019, "loss": 4.5726094245910645, "lr": 0.0002, "elapsed_sec": 25104.53646183014, "step_time_sec": 8.232728107948788, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3020, "loss": 4.317881107330322, "lr": 0.0002, "elapsed_sec": 25112.77279806137, "step_time_sec": 8.236172451055609, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3021, "loss": 4.459748268127441, "lr": 0.0002, "elapsed_sec": 25121.01015996933, "step_time_sec": 8.237199736060575, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3022, "loss": 4.422595500946045, "lr": 0.0002, "elapsed_sec": 25129.248059511185, "step_time_sec": 8.23775862203911, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3023, "loss": 4.3973307609558105, "lr": 0.0002, "elapsed_sec": 25137.485627412796, "step_time_sec": 8.23737568501383, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3024, "loss": 4.4146318435668945, "lr": 0.0002, "elapsed_sec": 25145.721717357635, "step_time_sec": 8.23592460504733, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3025, "loss": 4.753598690032959, "lr": 0.0002, "elapsed_sec": 25153.95774435997, "step_time_sec": 8.23588480707258, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3026, "loss": 4.472655296325684, "lr": 0.0002, "elapsed_sec": 25162.194226264954, "step_time_sec": 8.236311514046974, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3027, "loss": 4.370820045471191, "lr": 0.0002, "elapsed_sec": 25170.427222013474, "step_time_sec": 8.232855864916928, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3028, "loss": 4.407683849334717, "lr": 0.0002, "elapsed_sec": 25178.663858413696, "step_time_sec": 8.236466859932989, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3029, "loss": 4.481119632720947, "lr": 0.0002, "elapsed_sec": 25186.896542072296, "step_time_sec": 8.23249012301676, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3030, "loss": 4.550898551940918, "lr": 0.0002, "elapsed_sec": 25195.12896847725, "step_time_sec": 8.23226048599463, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3031, "loss": 4.542115688323975, "lr": 0.0002, "elapsed_sec": 25203.362772464752, "step_time_sec": 8.233682241057977, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3032, "loss": 4.484879493713379, "lr": 0.0002, "elapsed_sec": 25211.599684476852, "step_time_sec": 8.236759801977314, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3033, "loss": 4.595024585723877, "lr": 0.0002, "elapsed_sec": 25219.833813905716, "step_time_sec": 8.233957915916108, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3034, "loss": 4.408452987670898, "lr": 0.0002, "elapsed_sec": 25228.06635904312, "step_time_sec": 8.232426526024938, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3035, "loss": 4.484935760498047, "lr": 0.0002, "elapsed_sec": 25236.298342227936, "step_time_sec": 8.231848157010972, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3036, "loss": 4.6530256271362305, "lr": 0.0002, "elapsed_sec": 25244.535264015198, "step_time_sec": 8.236666225013323, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3037, "loss": 4.5398712158203125, "lr": 0.0002, "elapsed_sec": 25252.76751589775, "step_time_sec": 8.232123457011767, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3038, "loss": 4.51271915435791, "lr": 0.0002, "elapsed_sec": 25260.9998793602, "step_time_sec": 8.232133690966293, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3039, "loss": 4.506898403167725, "lr": 0.0002, "elapsed_sec": 25269.231236934662, "step_time_sec": 8.231192420003936, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3040, "loss": 4.4736104011535645, "lr": 0.0002, "elapsed_sec": 25277.470452070236, "step_time_sec": 8.239060087944381, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3041, "loss": 4.325530529022217, "lr": 0.0002, "elapsed_sec": 25285.70784330368, "step_time_sec": 8.237205642042682, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3042, "loss": 4.476400375366211, "lr": 0.0002, "elapsed_sec": 25293.94078350067, "step_time_sec": 8.232813259004615, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3043, "loss": 4.356841564178467, "lr": 0.0002, "elapsed_sec": 25302.17752432823, "step_time_sec": 8.236577181029133, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3044, "loss": 4.435427188873291, "lr": 0.0002, "elapsed_sec": 25310.41505908966, "step_time_sec": 8.237390827969648, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3045, "loss": 4.538036346435547, "lr": 0.0002, "elapsed_sec": 25318.6487531662, "step_time_sec": 8.233524801093154, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3046, "loss": 4.432666301727295, "lr": 0.0002, "elapsed_sec": 25326.880777835846, "step_time_sec": 8.231850150041282, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3047, "loss": 4.4088873863220215, "lr": 0.0002, "elapsed_sec": 25335.11334681511, "step_time_sec": 8.232421762892045, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3048, "loss": 4.488629341125488, "lr": 0.0002, "elapsed_sec": 25343.346448659897, "step_time_sec": 8.23298958700616, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3049, "loss": 4.590194225311279, "lr": 0.0002, "elapsed_sec": 25351.578021287918, "step_time_sec": 8.23138518910855, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3050, "loss": 4.473712921142578, "lr": 0.0002, "elapsed_sec": 25359.813611507416, "step_time_sec": 8.235411342931911, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3051, "loss": 4.587047100067139, "lr": 0.0002, "elapsed_sec": 25368.04700732231, "step_time_sec": 8.233261424931698, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3052, "loss": 4.4523210525512695, "lr": 0.0002, "elapsed_sec": 25376.280703783035, "step_time_sec": 8.23355210700538, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3053, "loss": 4.36053991317749, "lr": 0.0002, "elapsed_sec": 25384.517854452133, "step_time_sec": 8.236912260996178, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3054, "loss": 4.5494842529296875, "lr": 0.0002, "elapsed_sec": 25392.752103805542, "step_time_sec": 8.234101317008026, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3055, "loss": 4.410473346710205, "lr": 0.0002, "elapsed_sec": 25400.98551273346, "step_time_sec": 8.233232949045487, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3056, "loss": 4.562653541564941, "lr": 0.0002, "elapsed_sec": 25409.22328543663, "step_time_sec": 8.237604183028452, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3057, "loss": 4.433657646179199, "lr": 0.0002, "elapsed_sec": 25417.457317113876, "step_time_sec": 8.233904550899751, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3058, "loss": 4.561960697174072, "lr": 0.0002, "elapsed_sec": 25425.689870595932, "step_time_sec": 8.23236297594849, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3059, "loss": 4.4659647941589355, "lr": 0.0002, "elapsed_sec": 25433.92135334015, "step_time_sec": 8.23129714699462, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3060, "loss": 4.350471019744873, "lr": 0.0002, "elapsed_sec": 25442.152894973755, "step_time_sec": 8.23134860303253, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3061, "loss": 4.292292594909668, "lr": 0.0002, "elapsed_sec": 25450.385330677032, "step_time_sec": 8.232323023024946, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3062, "loss": 4.462204456329346, "lr": 0.0002, "elapsed_sec": 25458.617219686508, "step_time_sec": 8.231773174949922, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3063, "loss": 4.493217945098877, "lr": 0.0002, "elapsed_sec": 25466.849764347076, "step_time_sec": 8.232333485037088, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3064, "loss": 4.556338787078857, "lr": 0.0002, "elapsed_sec": 25475.081792593002, "step_time_sec": 8.231890498893335, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3065, "loss": 4.467426776885986, "lr": 0.0002, "elapsed_sec": 25483.314103603363, "step_time_sec": 8.232144866953604, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3066, "loss": 4.474644184112549, "lr": 0.0002, "elapsed_sec": 25491.54774570465, "step_time_sec": 8.233471966930665, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3067, "loss": 4.661799907684326, "lr": 0.0002, "elapsed_sec": 25499.781136989594, "step_time_sec": 8.233192954096012, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3068, "loss": 4.504817485809326, "lr": 0.0002, "elapsed_sec": 25508.013963222504, "step_time_sec": 8.232716200989671, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3069, "loss": 4.24566650390625, "lr": 0.0002, "elapsed_sec": 25516.24823474884, "step_time_sec": 8.234094035928138, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3070, "loss": 4.358666896820068, "lr": 0.0002, "elapsed_sec": 25524.48508620262, "step_time_sec": 8.23662488500122, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3071, "loss": 4.546547889709473, "lr": 0.0002, "elapsed_sec": 25532.71661090851, "step_time_sec": 8.231385934050195, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3072, "loss": 4.489814758300781, "lr": 0.0002, "elapsed_sec": 25540.950890541077, "step_time_sec": 8.23409254103899, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3073, "loss": 4.480832099914551, "lr": 0.0002, "elapsed_sec": 25549.185611248016, "step_time_sec": 8.234571452019736, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3074, "loss": 4.473348617553711, "lr": 0.0002, "elapsed_sec": 25557.422290802002, "step_time_sec": 8.23651381698437, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3075, "loss": 4.36333703994751, "lr": 0.0002, "elapsed_sec": 25565.6522834301, "step_time_sec": 8.229831882053986, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3076, "loss": 4.363778591156006, "lr": 0.0002, "elapsed_sec": 25573.885711193085, "step_time_sec": 8.233299638028257, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3077, "loss": 4.433929920196533, "lr": 0.0002, "elapsed_sec": 25582.11682152748, "step_time_sec": 8.230879248003475, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3078, "loss": 4.4030442237854, "lr": 0.0002, "elapsed_sec": 25590.34925889969, "step_time_sec": 8.232340313959867, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3079, "loss": 4.442781448364258, "lr": 0.0002, "elapsed_sec": 25598.581706285477, "step_time_sec": 8.232288222992793, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3080, "loss": 4.379233360290527, "lr": 0.0002, "elapsed_sec": 25606.81351161003, "step_time_sec": 8.231645222054794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3081, "loss": 4.486227512359619, "lr": 0.0002, "elapsed_sec": 25615.045244455338, "step_time_sec": 8.231494528008625, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3082, "loss": 4.533071517944336, "lr": 0.0002, "elapsed_sec": 25623.279000759125, "step_time_sec": 8.233619191101752, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3083, "loss": 4.549759387969971, "lr": 0.0002, "elapsed_sec": 25631.513208150864, "step_time_sec": 8.234039169969037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3084, "loss": 4.371138095855713, "lr": 0.0002, "elapsed_sec": 25639.74419951439, "step_time_sec": 8.230807694024406, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3085, "loss": 4.54526424407959, "lr": 0.0002, "elapsed_sec": 25647.977479934692, "step_time_sec": 8.233109983964823, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3086, "loss": 4.236583232879639, "lr": 0.0002, "elapsed_sec": 25656.209582328796, "step_time_sec": 8.231990589993075, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3087, "loss": 4.54148006439209, "lr": 0.0002, "elapsed_sec": 25664.447065353394, "step_time_sec": 8.237245003925636, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3088, "loss": 4.405120372772217, "lr": 0.0002, "elapsed_sec": 25672.682198286057, "step_time_sec": 8.235031764023006, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3089, "loss": 4.436501979827881, "lr": 0.0002, "elapsed_sec": 25680.919417142868, "step_time_sec": 8.237001521978527, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3090, "loss": 4.314320087432861, "lr": 0.0002, "elapsed_sec": 25689.151580810547, "step_time_sec": 8.232040602015331, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3091, "loss": 4.526482582092285, "lr": 0.0002, "elapsed_sec": 25697.385264396667, "step_time_sec": 8.233483653049916, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3092, "loss": 4.417730808258057, "lr": 0.0002, "elapsed_sec": 25705.616967201233, "step_time_sec": 8.231551822973415, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3093, "loss": 4.48635721206665, "lr": 0.0002, "elapsed_sec": 25713.850244522095, "step_time_sec": 8.233109973021783, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3094, "loss": 4.487398147583008, "lr": 0.0002, "elapsed_sec": 25722.081768512726, "step_time_sec": 8.231339287012815, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3095, "loss": 4.485714912414551, "lr": 0.0002, "elapsed_sec": 25730.314402341843, "step_time_sec": 8.232479966012761, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3096, "loss": 4.439633846282959, "lr": 0.0002, "elapsed_sec": 25738.54726409912, "step_time_sec": 8.232719637919217, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3097, "loss": 4.4691290855407715, "lr": 0.0002, "elapsed_sec": 25746.783751487732, "step_time_sec": 8.236328252940439, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3098, "loss": 4.449773788452148, "lr": 0.0002, "elapsed_sec": 25755.015681505203, "step_time_sec": 8.231751970946789, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3099, "loss": 4.396188259124756, "lr": 0.0002, "elapsed_sec": 25763.250339508057, "step_time_sec": 8.234516922966577, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3100, "loss": 4.521545886993408, "lr": 0.0002, "elapsed_sec": 25771.486775398254, "step_time_sec": 8.23623969394248, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3101, "loss": 4.4768524169921875, "lr": 0.0002, "elapsed_sec": 25779.721574544907, "step_time_sec": 8.234635681961663, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3102, "loss": 4.5029377937316895, "lr": 0.0002, "elapsed_sec": 25787.952710151672, "step_time_sec": 8.230977379949763, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3103, "loss": 4.410736083984375, "lr": 0.0002, "elapsed_sec": 25796.18540072441, "step_time_sec": 8.232569152023643, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3104, "loss": 4.454659938812256, "lr": 0.0002, "elapsed_sec": 25804.417573451996, "step_time_sec": 8.231931081973016, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3105, "loss": 4.337461471557617, "lr": 0.0002, "elapsed_sec": 25812.649927139282, "step_time_sec": 8.23218735307455, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3106, "loss": 4.560267448425293, "lr": 0.0002, "elapsed_sec": 25820.882173538208, "step_time_sec": 8.23206750897225, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3107, "loss": 4.403928279876709, "lr": 0.0002, "elapsed_sec": 25829.114285707474, "step_time_sec": 8.231986048049293, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3108, "loss": 4.415592193603516, "lr": 0.0002, "elapsed_sec": 25837.346721172333, "step_time_sec": 8.232273287023418, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3109, "loss": 4.421982288360596, "lr": 0.0002, "elapsed_sec": 25845.580070972443, "step_time_sec": 8.233166028978303, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3110, "loss": 4.479611873626709, "lr": 0.0002, "elapsed_sec": 25853.81063771248, "step_time_sec": 8.230404733098112, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3111, "loss": 4.516005516052246, "lr": 0.0002, "elapsed_sec": 25862.05550813675, "step_time_sec": 8.237442622077651, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3112, "loss": 4.367657661437988, "lr": 0.0002, "elapsed_sec": 25870.289459466934, "step_time_sec": 8.233743643970229, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3113, "loss": 4.4416046142578125, "lr": 0.0002, "elapsed_sec": 25878.525129318237, "step_time_sec": 8.23556242894847, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3114, "loss": 4.364328384399414, "lr": 0.0002, "elapsed_sec": 25886.757412433624, "step_time_sec": 8.232124370988458, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3115, "loss": 4.332004070281982, "lr": 0.0002, "elapsed_sec": 25894.991570949554, "step_time_sec": 8.233947674976662, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3116, "loss": 4.424645900726318, "lr": 0.0002, "elapsed_sec": 25903.22395682335, "step_time_sec": 8.232217467040755, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3117, "loss": 4.561093330383301, "lr": 0.0002, "elapsed_sec": 25911.458413124084, "step_time_sec": 8.234323340002447, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3118, "loss": 4.482474327087402, "lr": 0.0002, "elapsed_sec": 25919.69245147705, "step_time_sec": 8.233895219978876, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3119, "loss": 4.341700077056885, "lr": 0.0002, "elapsed_sec": 25927.925128936768, "step_time_sec": 8.232489448972046, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3120, "loss": 4.517014026641846, "lr": 0.0002, "elapsed_sec": 25936.161471128464, "step_time_sec": 8.236186310066842, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3121, "loss": 4.439598560333252, "lr": 0.0002, "elapsed_sec": 25944.398110866547, "step_time_sec": 8.236388238961808, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3122, "loss": 4.600067138671875, "lr": 0.0002, "elapsed_sec": 25952.63029193878, "step_time_sec": 8.23203900596127, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3123, "loss": 4.528568744659424, "lr": 0.0002, "elapsed_sec": 25960.86336016655, "step_time_sec": 8.232910598046146, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3124, "loss": 4.560232639312744, "lr": 0.0002, "elapsed_sec": 25969.099371433258, "step_time_sec": 8.235876686987467, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3125, "loss": 4.458303928375244, "lr": 0.0002, "elapsed_sec": 25977.332137584686, "step_time_sec": 8.232529971981421, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3126, "loss": 4.411995887756348, "lr": 0.0002, "elapsed_sec": 25985.564643383026, "step_time_sec": 8.23235837800894, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3127, "loss": 4.391222953796387, "lr": 0.0002, "elapsed_sec": 25993.801414966583, "step_time_sec": 8.236632321961224, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3128, "loss": 4.44504976272583, "lr": 0.0002, "elapsed_sec": 26002.033502578735, "step_time_sec": 8.231910331989639, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3129, "loss": 4.421806335449219, "lr": 0.0002, "elapsed_sec": 26010.26610827446, "step_time_sec": 8.232397974934429, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3130, "loss": 4.435845375061035, "lr": 0.0002, "elapsed_sec": 26018.49957561493, "step_time_sec": 8.23335314693395, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3131, "loss": 4.515201568603516, "lr": 0.0002, "elapsed_sec": 26026.732019901276, "step_time_sec": 8.232264229096472, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3132, "loss": 4.41386604309082, "lr": 0.0002, "elapsed_sec": 26034.967835903168, "step_time_sec": 8.235609570983797, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3133, "loss": 4.380678176879883, "lr": 0.0002, "elapsed_sec": 26043.20518732071, "step_time_sec": 8.237173962057568, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3134, "loss": 4.302052021026611, "lr": 0.0002, "elapsed_sec": 26051.43905520439, "step_time_sec": 8.23370497697033, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3135, "loss": 4.476922512054443, "lr": 0.0002, "elapsed_sec": 26059.67237186432, "step_time_sec": 8.233192331972532, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3136, "loss": 4.455271244049072, "lr": 0.0002, "elapsed_sec": 26067.904804944992, "step_time_sec": 8.232195305987261, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3137, "loss": 4.4306745529174805, "lr": 0.0002, "elapsed_sec": 26076.138793468475, "step_time_sec": 8.233891976065934, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3138, "loss": 4.380963325500488, "lr": 0.0002, "elapsed_sec": 26084.37030482292, "step_time_sec": 8.231279708910733, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3139, "loss": 4.451044082641602, "lr": 0.0002, "elapsed_sec": 26092.604293346405, "step_time_sec": 8.233847219031304, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3140, "loss": 4.46059513092041, "lr": 0.0002, "elapsed_sec": 26100.83732175827, "step_time_sec": 8.232849635998718, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3141, "loss": 4.430739879608154, "lr": 0.0002, "elapsed_sec": 26109.069113254547, "step_time_sec": 8.23164163599722, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3142, "loss": 4.429052829742432, "lr": 0.0002, "elapsed_sec": 26117.30602502823, "step_time_sec": 8.236759471008554, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3143, "loss": 4.443422317504883, "lr": 0.0002, "elapsed_sec": 26125.540689468384, "step_time_sec": 8.234504358028062, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3144, "loss": 4.435821533203125, "lr": 0.0002, "elapsed_sec": 26133.773665905, "step_time_sec": 8.23283861100208, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3145, "loss": 4.456973075866699, "lr": 0.0002, "elapsed_sec": 26142.007462263107, "step_time_sec": 8.23361648095306, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3146, "loss": 4.40169620513916, "lr": 0.0002, "elapsed_sec": 26150.242745876312, "step_time_sec": 8.235093801980838, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3147, "loss": 4.689956188201904, "lr": 0.0002, "elapsed_sec": 26158.475752830505, "step_time_sec": 8.232856886927038, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3148, "loss": 4.364945888519287, "lr": 0.0002, "elapsed_sec": 26166.712104797363, "step_time_sec": 8.236212116084062, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3149, "loss": 4.217365264892578, "lr": 0.0002, "elapsed_sec": 26174.944886446, "step_time_sec": 8.232553850975819, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3150, "loss": 4.488561153411865, "lr": 0.0002, "elapsed_sec": 26183.17719221115, "step_time_sec": 8.232162756030448, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3151, "loss": 4.40017032623291, "lr": 0.0002, "elapsed_sec": 26191.410036802292, "step_time_sec": 8.232690954930149, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3152, "loss": 4.576910972595215, "lr": 0.0002, "elapsed_sec": 26199.644973754883, "step_time_sec": 8.234784240019508, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3153, "loss": 4.443170547485352, "lr": 0.0002, "elapsed_sec": 26207.88000512123, "step_time_sec": 8.234854722977616, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3154, "loss": 4.4814348220825195, "lr": 0.0002, "elapsed_sec": 26216.112761497498, "step_time_sec": 8.232647563097998, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3155, "loss": 4.452919006347656, "lr": 0.0002, "elapsed_sec": 26224.34494829178, "step_time_sec": 8.231995178037323, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3156, "loss": 4.484823226928711, "lr": 0.0002, "elapsed_sec": 26232.57943058014, "step_time_sec": 8.234303898992948, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3157, "loss": 4.3367533683776855, "lr": 0.0002, "elapsed_sec": 26240.8131878376, "step_time_sec": 8.233606197987683, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3158, "loss": 4.525226593017578, "lr": 0.0002, "elapsed_sec": 26249.048862457275, "step_time_sec": 8.235520606045611, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3159, "loss": 4.434479236602783, "lr": 0.0002, "elapsed_sec": 26257.28175187111, "step_time_sec": 8.232743035070598, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3160, "loss": 4.58366060256958, "lr": 0.0002, "elapsed_sec": 26265.516695022583, "step_time_sec": 8.234749841038138, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3161, "loss": 4.306504726409912, "lr": 0.0002, "elapsed_sec": 26273.749205827713, "step_time_sec": 8.232366667012684, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3162, "loss": 4.5171661376953125, "lr": 0.0002, "elapsed_sec": 26281.98193502426, "step_time_sec": 8.232497112010606, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3163, "loss": 4.4906792640686035, "lr": 0.0002, "elapsed_sec": 26290.21454310417, "step_time_sec": 8.232531565940008, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3164, "loss": 4.470058441162109, "lr": 0.0002, "elapsed_sec": 26298.44688987732, "step_time_sec": 8.232189393020235, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3165, "loss": 4.398813724517822, "lr": 0.0002, "elapsed_sec": 26306.6784927845, "step_time_sec": 8.231432161992416, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3166, "loss": 4.504300117492676, "lr": 0.0002, "elapsed_sec": 26314.91587972641, "step_time_sec": 8.237174264038913, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3167, "loss": 4.4987568855285645, "lr": 0.0002, "elapsed_sec": 26323.151074409485, "step_time_sec": 8.235036841011606, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3168, "loss": 4.389163970947266, "lr": 0.0002, "elapsed_sec": 26331.387588262558, "step_time_sec": 8.236366946948692, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3169, "loss": 4.214711666107178, "lr": 0.0002, "elapsed_sec": 26339.619415044785, "step_time_sec": 8.231657469994389, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3170, "loss": 4.319453716278076, "lr": 0.0002, "elapsed_sec": 26347.85153913498, "step_time_sec": 8.231972437002696, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3171, "loss": 4.536755561828613, "lr": 0.0002, "elapsed_sec": 26356.08437013626, "step_time_sec": 8.232623031013645, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3172, "loss": 4.45809268951416, "lr": 0.0002, "elapsed_sec": 26364.321365594864, "step_time_sec": 8.236857778043486, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3173, "loss": 4.38252592086792, "lr": 0.0002, "elapsed_sec": 26372.553335666656, "step_time_sec": 8.231827550916933, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3174, "loss": 4.5069451332092285, "lr": 0.0002, "elapsed_sec": 26380.79171180725, "step_time_sec": 8.238201108993962, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3175, "loss": 4.4914350509643555, "lr": 0.0002, "elapsed_sec": 26389.02732038498, "step_time_sec": 8.235479608993046, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3176, "loss": 4.400797367095947, "lr": 0.0002, "elapsed_sec": 26397.25963449478, "step_time_sec": 8.232136318925768, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3177, "loss": 4.410691261291504, "lr": 0.0002, "elapsed_sec": 26405.491258621216, "step_time_sec": 8.231465549906716, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3178, "loss": 4.562859058380127, "lr": 0.0002, "elapsed_sec": 26413.7235019207, "step_time_sec": 8.232094927923754, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3179, "loss": 4.358163356781006, "lr": 0.0002, "elapsed_sec": 26421.95631146431, "step_time_sec": 8.232604951946996, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3180, "loss": 4.450184345245361, "lr": 0.0002, "elapsed_sec": 26430.188485860825, "step_time_sec": 8.232061802060343, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3181, "loss": 4.533093452453613, "lr": 0.0002, "elapsed_sec": 26438.41967701912, "step_time_sec": 8.230994019075297, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3182, "loss": 4.480890274047852, "lr": 0.0002, "elapsed_sec": 26446.652238368988, "step_time_sec": 8.232417101971805, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3183, "loss": 4.543671607971191, "lr": 0.0002, "elapsed_sec": 26454.8846886158, "step_time_sec": 8.232259685057215, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3184, "loss": 4.414905548095703, "lr": 0.0002, "elapsed_sec": 26463.117325782776, "step_time_sec": 8.232483844039962, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3185, "loss": 4.33827543258667, "lr": 0.0002, "elapsed_sec": 26471.350481510162, "step_time_sec": 8.232999355066568, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3186, "loss": 4.364034652709961, "lr": 0.0002, "elapsed_sec": 26479.58398079872, "step_time_sec": 8.233335135970265, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3187, "loss": 4.494781017303467, "lr": 0.0002, "elapsed_sec": 26487.81583094597, "step_time_sec": 8.231685303035192, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3188, "loss": 4.368886470794678, "lr": 0.0002, "elapsed_sec": 26496.047379016876, "step_time_sec": 8.2313577409368, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3189, "loss": 4.461017608642578, "lr": 0.0002, "elapsed_sec": 26504.278900384903, "step_time_sec": 8.231408247025684, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3190, "loss": 4.390085220336914, "lr": 0.0002, "elapsed_sec": 26512.510360240936, "step_time_sec": 8.231297785998322, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3191, "loss": 4.467686176300049, "lr": 0.0002, "elapsed_sec": 26520.74405002594, "step_time_sec": 8.233535244944505, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3192, "loss": 4.430937767028809, "lr": 0.0002, "elapsed_sec": 26528.976132392883, "step_time_sec": 8.231923060957342, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3193, "loss": 4.390294075012207, "lr": 0.0002, "elapsed_sec": 26537.20897912979, "step_time_sec": 8.232643828960136, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3194, "loss": 4.53483772277832, "lr": 0.0002, "elapsed_sec": 26545.440961122513, "step_time_sec": 8.23186735494528, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3195, "loss": 4.4710917472839355, "lr": 0.0002, "elapsed_sec": 26553.673449277878, "step_time_sec": 8.232290281099267, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3196, "loss": 4.444796562194824, "lr": 0.0002, "elapsed_sec": 26561.90751194954, "step_time_sec": 8.233882311964408, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3197, "loss": 4.432831764221191, "lr": 0.0002, "elapsed_sec": 26570.139176607132, "step_time_sec": 8.231489046942443, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3198, "loss": 4.408693313598633, "lr": 0.0002, "elapsed_sec": 26578.374279022217, "step_time_sec": 8.234941881033592, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3199, "loss": 4.464259147644043, "lr": 0.0002, "elapsed_sec": 26586.606697320938, "step_time_sec": 8.232297297916375, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3200, "loss": 4.395345687866211, "lr": 0.0002, "elapsed_sec": 26594.839288711548, "step_time_sec": 8.232374095008709, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3201, "loss": 4.52642822265625, "lr": 0.0002, "elapsed_sec": 26603.072760105133, "step_time_sec": 8.233338118996471, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3202, "loss": 4.563186168670654, "lr": 0.0002, "elapsed_sec": 26611.304995775223, "step_time_sec": 8.232034001965076, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3203, "loss": 4.44887638092041, "lr": 0.0002, "elapsed_sec": 26619.538558721542, "step_time_sec": 8.233400390017778, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3204, "loss": 4.3831658363342285, "lr": 0.0002, "elapsed_sec": 26627.77007842064, "step_time_sec": 8.231392326997593, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3205, "loss": 4.485721588134766, "lr": 0.0002, "elapsed_sec": 26636.001794338226, "step_time_sec": 8.231561516993679, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3206, "loss": 4.349246501922607, "lr": 0.0002, "elapsed_sec": 26644.234771490097, "step_time_sec": 8.232811619061977, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3207, "loss": 4.329967975616455, "lr": 0.0002, "elapsed_sec": 26652.4664645195, "step_time_sec": 8.231512772035785, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3208, "loss": 4.28619909286499, "lr": 0.0002, "elapsed_sec": 26660.703154325485, "step_time_sec": 8.23651663097553, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3209, "loss": 4.413715362548828, "lr": 0.0002, "elapsed_sec": 26668.938860416412, "step_time_sec": 8.235532694961876, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3210, "loss": 4.488641262054443, "lr": 0.0002, "elapsed_sec": 26677.170966625214, "step_time_sec": 8.231973342015408, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3211, "loss": 4.3926544189453125, "lr": 0.0002, "elapsed_sec": 26685.40297436714, "step_time_sec": 8.231845272006467, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3212, "loss": 4.367194652557373, "lr": 0.0002, "elapsed_sec": 26693.635962724686, "step_time_sec": 8.23278627905529, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3213, "loss": 4.351407527923584, "lr": 0.0002, "elapsed_sec": 26701.86959195137, "step_time_sec": 8.233460741001181, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3214, "loss": 4.418981552124023, "lr": 0.0002, "elapsed_sec": 26710.101130008698, "step_time_sec": 8.231352797942236, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3215, "loss": 4.336595058441162, "lr": 0.0002, "elapsed_sec": 26718.338723421097, "step_time_sec": 8.23742579203099, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3216, "loss": 4.4114990234375, "lr": 0.0002, "elapsed_sec": 26726.571209192276, "step_time_sec": 8.232403047964908, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3217, "loss": 4.435850143432617, "lr": 0.0002, "elapsed_sec": 26734.803275823593, "step_time_sec": 8.231846804963425, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3218, "loss": 4.462010860443115, "lr": 0.0002, "elapsed_sec": 26743.035833358765, "step_time_sec": 8.232412521028891, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3219, "loss": 4.42028284072876, "lr": 0.0002, "elapsed_sec": 26751.27204632759, "step_time_sec": 8.236077297013253, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3220, "loss": 4.418653964996338, "lr": 0.0002, "elapsed_sec": 26759.50608253479, "step_time_sec": 8.233842959976755, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3221, "loss": 4.364592552185059, "lr": 0.0002, "elapsed_sec": 26767.736010551453, "step_time_sec": 8.229769106023014, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3222, "loss": 4.5284833908081055, "lr": 0.0002, "elapsed_sec": 26775.968916654587, "step_time_sec": 8.2327394959284, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3223, "loss": 4.520843982696533, "lr": 0.0002, "elapsed_sec": 26784.203506469727, "step_time_sec": 8.234442010056227, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3224, "loss": 4.491455554962158, "lr": 0.0002, "elapsed_sec": 26792.438179969788, "step_time_sec": 8.234484146931209, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3225, "loss": 4.425684928894043, "lr": 0.0002, "elapsed_sec": 26800.66983985901, "step_time_sec": 8.231516545056365, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3226, "loss": 4.5668182373046875, "lr": 0.0002, "elapsed_sec": 26808.902831315994, "step_time_sec": 8.232844336074777, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3227, "loss": 4.493022441864014, "lr": 0.0002, "elapsed_sec": 26817.134593963623, "step_time_sec": 8.2315435919445, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3228, "loss": 4.40925931930542, "lr": 0.0002, "elapsed_sec": 26825.367389917374, "step_time_sec": 8.232663600938395, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3229, "loss": 4.399245262145996, "lr": 0.0002, "elapsed_sec": 26833.600433588028, "step_time_sec": 8.232896147994325, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3230, "loss": 4.60800838470459, "lr": 0.0002, "elapsed_sec": 26841.837848186493, "step_time_sec": 8.237196996924467, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3231, "loss": 4.40891170501709, "lr": 0.0002, "elapsed_sec": 26850.070637226105, "step_time_sec": 8.232601181021892, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3232, "loss": 4.608367443084717, "lr": 0.0002, "elapsed_sec": 26858.303488254547, "step_time_sec": 8.232675209990703, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3233, "loss": 4.431474208831787, "lr": 0.0002, "elapsed_sec": 26866.533910274506, "step_time_sec": 8.230328485951759, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3234, "loss": 4.412576198577881, "lr": 0.0002, "elapsed_sec": 26874.766159296036, "step_time_sec": 8.232028767000884, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3235, "loss": 4.41252326965332, "lr": 0.0002, "elapsed_sec": 26883.000986099243, "step_time_sec": 8.234703419962898, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3236, "loss": 4.51894998550415, "lr": 0.0002, "elapsed_sec": 26891.236357450485, "step_time_sec": 8.23515970504377, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3237, "loss": 4.452917575836182, "lr": 0.0002, "elapsed_sec": 26899.471333503723, "step_time_sec": 8.234853683039546, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3238, "loss": 4.3513665199279785, "lr": 0.0002, "elapsed_sec": 26907.70772075653, "step_time_sec": 8.236231156042777, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3239, "loss": 4.453488826751709, "lr": 0.0002, "elapsed_sec": 26915.941405773163, "step_time_sec": 8.233486474957317, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3240, "loss": 4.488948822021484, "lr": 0.0002, "elapsed_sec": 26924.173743009567, "step_time_sec": 8.232181509025395, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3241, "loss": 4.46435022354126, "lr": 0.0002, "elapsed_sec": 26932.405792474747, "step_time_sec": 8.231885607005097, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3242, "loss": 4.381466865539551, "lr": 0.0002, "elapsed_sec": 26940.637956380844, "step_time_sec": 8.232014928013086, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3243, "loss": 4.37847900390625, "lr": 0.0002, "elapsed_sec": 26948.86985516548, "step_time_sec": 8.231746413977817, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3244, "loss": 4.399765968322754, "lr": 0.0002, "elapsed_sec": 26957.103276252747, "step_time_sec": 8.23325755505357, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3245, "loss": 4.464969635009766, "lr": 0.0002, "elapsed_sec": 26977.61215877533, "step_time_sec": 20.50872271694243, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3246, "loss": 4.424515724182129, "lr": 0.0002, "elapsed_sec": 26985.84406685829, "step_time_sec": 8.2317630329635, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3247, "loss": 4.391050338745117, "lr": 0.0002, "elapsed_sec": 26994.075751781464, "step_time_sec": 8.231509741046466, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3248, "loss": 4.37669038772583, "lr": 0.0002, "elapsed_sec": 27002.30813217163, "step_time_sec": 8.232180609018542, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3249, "loss": 4.52837610244751, "lr": 0.0002, "elapsed_sec": 27010.539763450623, "step_time_sec": 8.231494610081427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3250, "loss": 4.348695278167725, "lr": 0.0002, "elapsed_sec": 27018.772906780243, "step_time_sec": 8.233009591000155, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3251, "loss": 4.358288764953613, "lr": 0.0002, "elapsed_sec": 27027.00625061989, "step_time_sec": 8.233183754025958, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3252, "loss": 4.478082180023193, "lr": 0.0002, "elapsed_sec": 27035.238293886185, "step_time_sec": 8.23184644000139, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3253, "loss": 4.5624589920043945, "lr": 0.0002, "elapsed_sec": 27043.472073554993, "step_time_sec": 8.233609733055346, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3254, "loss": 4.476986408233643, "lr": 0.0002, "elapsed_sec": 27051.70423102379, "step_time_sec": 8.232025627046824, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3255, "loss": 4.580624580383301, "lr": 0.0002, "elapsed_sec": 27059.93670988083, "step_time_sec": 8.232294880901463, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3256, "loss": 4.380424499511719, "lr": 0.0002, "elapsed_sec": 27068.17280483246, "step_time_sec": 8.235964271007106, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3257, "loss": 4.564871311187744, "lr": 0.0002, "elapsed_sec": 27076.40624499321, "step_time_sec": 8.233276171027683, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3258, "loss": 4.481104850769043, "lr": 0.0002, "elapsed_sec": 27084.640892505646, "step_time_sec": 8.234424623078667, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3259, "loss": 4.633317947387695, "lr": 0.0002, "elapsed_sec": 27092.873155593872, "step_time_sec": 8.232132895034738, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3260, "loss": 4.4361419677734375, "lr": 0.0002, "elapsed_sec": 27101.10488176346, "step_time_sec": 8.231561589054763, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3261, "loss": 4.411470413208008, "lr": 0.0002, "elapsed_sec": 27109.343273878098, "step_time_sec": 8.238224934088066, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3262, "loss": 4.447425842285156, "lr": 0.0002, "elapsed_sec": 27117.57993578911, "step_time_sec": 8.23649529798422, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3263, "loss": 4.452564239501953, "lr": 0.0002, "elapsed_sec": 27125.811829805374, "step_time_sec": 8.231740999035537, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3264, "loss": 4.289203643798828, "lr": 0.0002, "elapsed_sec": 27134.048176050186, "step_time_sec": 8.236195392906666, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3265, "loss": 4.484467029571533, "lr": 0.0002, "elapsed_sec": 27142.28654718399, "step_time_sec": 8.238233141019009, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3266, "loss": 4.344668388366699, "lr": 0.0002, "elapsed_sec": 27150.522998571396, "step_time_sec": 8.236303188023157, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3267, "loss": 4.498162269592285, "lr": 0.0002, "elapsed_sec": 27158.75516963005, "step_time_sec": 8.231973023968749, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3268, "loss": 4.430897235870361, "lr": 0.0002, "elapsed_sec": 27166.987705230713, "step_time_sec": 8.232401836896315, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3269, "loss": 4.461334705352783, "lr": 0.0002, "elapsed_sec": 27175.220409870148, "step_time_sec": 8.232535055023618, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3270, "loss": 4.463438987731934, "lr": 0.0002, "elapsed_sec": 27183.458174943924, "step_time_sec": 8.237581743043847, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3271, "loss": 4.395109176635742, "lr": 0.0002, "elapsed_sec": 27191.695186138153, "step_time_sec": 8.236877184011973, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3272, "loss": 4.434361457824707, "lr": 0.0002, "elapsed_sec": 27199.93398499489, "step_time_sec": 8.238600115990266, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3273, "loss": 4.375602722167969, "lr": 0.0002, "elapsed_sec": 27208.166400432587, "step_time_sec": 8.232282213051803, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3274, "loss": 4.401582717895508, "lr": 0.0002, "elapsed_sec": 27216.39913892746, "step_time_sec": 8.232607460930012, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3275, "loss": 4.39561128616333, "lr": 0.0002, "elapsed_sec": 27224.631006717682, "step_time_sec": 8.231652962043881, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3276, "loss": 4.299432754516602, "lr": 0.0002, "elapsed_sec": 27232.862839460373, "step_time_sec": 8.231672885012813, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3277, "loss": 4.3749213218688965, "lr": 0.0002, "elapsed_sec": 27241.095822572708, "step_time_sec": 8.232823511003517, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3278, "loss": 4.593959808349609, "lr": 0.0002, "elapsed_sec": 27249.32816386223, "step_time_sec": 8.232159264967777, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3279, "loss": 4.620635032653809, "lr": 0.0002, "elapsed_sec": 27257.56056547165, "step_time_sec": 8.232239279896021, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3280, "loss": 4.537662982940674, "lr": 0.0002, "elapsed_sec": 27265.792099952698, "step_time_sec": 8.231374141061679, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3281, "loss": 4.405154228210449, "lr": 0.0002, "elapsed_sec": 27274.02455472946, "step_time_sec": 8.232289602980018, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3282, "loss": 4.4045491218566895, "lr": 0.0002, "elapsed_sec": 27282.25616335869, "step_time_sec": 8.231432721018791, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3283, "loss": 4.452691555023193, "lr": 0.0002, "elapsed_sec": 27290.493045806885, "step_time_sec": 8.236747227958404, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3284, "loss": 4.451304912567139, "lr": 0.0002, "elapsed_sec": 27298.723543167114, "step_time_sec": 8.23035611002706, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3285, "loss": 4.406555652618408, "lr": 0.0002, "elapsed_sec": 27306.96114206314, "step_time_sec": 8.237432788009755, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3286, "loss": 4.610159397125244, "lr": 0.0002, "elapsed_sec": 27315.197979450226, "step_time_sec": 8.236679145949893, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3287, "loss": 4.348730564117432, "lr": 0.0002, "elapsed_sec": 27323.433394908905, "step_time_sec": 8.235204892000183, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3288, "loss": 4.3491621017456055, "lr": 0.0002, "elapsed_sec": 27331.664897203445, "step_time_sec": 8.231347790919244, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3289, "loss": 4.407780647277832, "lr": 0.0002, "elapsed_sec": 27339.897804260254, "step_time_sec": 8.232752509997226, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3290, "loss": 4.472410678863525, "lr": 0.0002, "elapsed_sec": 27348.129323244095, "step_time_sec": 8.231363572995178, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3291, "loss": 4.481769561767578, "lr": 0.0002, "elapsed_sec": 27356.36749458313, "step_time_sec": 8.238041845033877, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3292, "loss": 4.512937068939209, "lr": 0.0002, "elapsed_sec": 27364.599839925766, "step_time_sec": 8.232129993964918, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3293, "loss": 4.419572830200195, "lr": 0.0002, "elapsed_sec": 27372.83263373375, "step_time_sec": 8.232666074065492, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3294, "loss": 4.4989471435546875, "lr": 0.0002, "elapsed_sec": 27381.06545495987, "step_time_sec": 8.232617876958102, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3295, "loss": 4.503006935119629, "lr": 0.0002, "elapsed_sec": 27389.29771089554, "step_time_sec": 8.232118321931921, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3296, "loss": 4.327880859375, "lr": 0.0002, "elapsed_sec": 27397.52965068817, "step_time_sec": 8.231787250027992, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3297, "loss": 4.369223594665527, "lr": 0.0002, "elapsed_sec": 27405.768028497696, "step_time_sec": 8.238186275935732, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3298, "loss": 4.522778511047363, "lr": 0.0002, "elapsed_sec": 27414.00121307373, "step_time_sec": 8.233017744030803, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3299, "loss": 4.579559326171875, "lr": 0.0002, "elapsed_sec": 27422.233988523483, "step_time_sec": 8.23261786904186, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3300, "loss": 4.308288097381592, "lr": 0.0002, "elapsed_sec": 27430.467655181885, "step_time_sec": 8.23353213001974, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3301, "loss": 4.284564018249512, "lr": 0.0002, "elapsed_sec": 27438.70550274849, "step_time_sec": 8.237705019069836, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3302, "loss": 4.485703468322754, "lr": 0.0002, "elapsed_sec": 27446.94258904457, "step_time_sec": 8.236900195013732, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3303, "loss": 4.490586757659912, "lr": 0.0002, "elapsed_sec": 27455.17475795746, "step_time_sec": 8.231975183007307, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3304, "loss": 4.462466239929199, "lr": 0.0002, "elapsed_sec": 27463.4068505764, "step_time_sec": 8.231940001947805, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3305, "loss": 4.397047519683838, "lr": 0.0002, "elapsed_sec": 27471.639946222305, "step_time_sec": 8.232903328025714, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3306, "loss": 4.5571208000183105, "lr": 0.0002, "elapsed_sec": 27479.87357878685, "step_time_sec": 8.233497062930837, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3307, "loss": 4.380002021789551, "lr": 0.0002, "elapsed_sec": 27488.107092618942, "step_time_sec": 8.233328310889192, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3308, "loss": 4.378542423248291, "lr": 0.0002, "elapsed_sec": 27496.33943796158, "step_time_sec": 8.232226046966389, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3309, "loss": 4.379422664642334, "lr": 0.0002, "elapsed_sec": 27504.572858333588, "step_time_sec": 8.233186613069847, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3310, "loss": 4.431548595428467, "lr": 0.0002, "elapsed_sec": 27512.805337429047, "step_time_sec": 8.232314987108111, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3311, "loss": 4.341507911682129, "lr": 0.0002, "elapsed_sec": 27521.03862452507, "step_time_sec": 8.23312759201508, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3312, "loss": 4.424261569976807, "lr": 0.0002, "elapsed_sec": 27529.274753808975, "step_time_sec": 8.235990511951968, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3313, "loss": 4.4585676193237305, "lr": 0.0002, "elapsed_sec": 27537.51207304001, "step_time_sec": 8.23715096001979, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3314, "loss": 4.3152852058410645, "lr": 0.0002, "elapsed_sec": 27545.74368572235, "step_time_sec": 8.231467421050183, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3315, "loss": 4.49558162689209, "lr": 0.0002, "elapsed_sec": 27553.97695159912, "step_time_sec": 8.233114562928677, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3316, "loss": 4.345945835113525, "lr": 0.0002, "elapsed_sec": 27562.208929777145, "step_time_sec": 8.231811312958598, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3317, "loss": 4.462437152862549, "lr": 0.0002, "elapsed_sec": 27570.446964263916, "step_time_sec": 8.237832411075942, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3318, "loss": 4.381821632385254, "lr": 0.0002, "elapsed_sec": 27578.683187007904, "step_time_sec": 8.236069967970252, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3319, "loss": 4.487292766571045, "lr": 0.0002, "elapsed_sec": 27586.921703100204, "step_time_sec": 8.238398712011985, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3320, "loss": 4.453155040740967, "lr": 0.0002, "elapsed_sec": 27595.15821123123, "step_time_sec": 8.236287461942993, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3321, "loss": 4.485411643981934, "lr": 0.0002, "elapsed_sec": 27603.3923535347, "step_time_sec": 8.234009578940459, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3322, "loss": 4.427615642547607, "lr": 0.0002, "elapsed_sec": 27611.623575925827, "step_time_sec": 8.23103641194757, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3323, "loss": 4.467099189758301, "lr": 0.0002, "elapsed_sec": 27619.8566968441, "step_time_sec": 8.232939505018294, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3324, "loss": 4.46379280090332, "lr": 0.0002, "elapsed_sec": 27628.09474015236, "step_time_sec": 8.237892928998917, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3325, "loss": 4.384115695953369, "lr": 0.0002, "elapsed_sec": 27636.331555366516, "step_time_sec": 8.236713298945688, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3326, "loss": 4.361752986907959, "lr": 0.0002, "elapsed_sec": 27644.569205522537, "step_time_sec": 8.237439486081712, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3327, "loss": 4.496685981750488, "lr": 0.0002, "elapsed_sec": 27652.805882692337, "step_time_sec": 8.236521766055375, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3328, "loss": 4.282104015350342, "lr": 0.0002, "elapsed_sec": 27661.03926038742, "step_time_sec": 8.233180747949518, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3329, "loss": 4.477495193481445, "lr": 0.0002, "elapsed_sec": 27669.272174358368, "step_time_sec": 8.23274817597121, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3330, "loss": 4.468421936035156, "lr": 0.0002, "elapsed_sec": 27677.506523370743, "step_time_sec": 8.234212819021195, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3331, "loss": 4.539824485778809, "lr": 0.0002, "elapsed_sec": 27685.738738536835, "step_time_sec": 8.232041003997438, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3332, "loss": 4.456181526184082, "lr": 0.0002, "elapsed_sec": 27693.97171807289, "step_time_sec": 8.23282603698317, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3333, "loss": 4.387608051300049, "lr": 0.0002, "elapsed_sec": 27702.204607009888, "step_time_sec": 8.232745768036693, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3334, "loss": 4.28627347946167, "lr": 0.0002, "elapsed_sec": 27710.438584566116, "step_time_sec": 8.233776438981295, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3335, "loss": 4.349700927734375, "lr": 0.0002, "elapsed_sec": 27718.671726226807, "step_time_sec": 8.233011547010392, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3336, "loss": 4.359137058258057, "lr": 0.0002, "elapsed_sec": 27726.909103870392, "step_time_sec": 8.237164056976326, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3337, "loss": 4.3444600105285645, "lr": 0.0002, "elapsed_sec": 27735.14558982849, "step_time_sec": 8.236345760058612, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3338, "loss": 4.234476566314697, "lr": 0.0002, "elapsed_sec": 27743.383962392807, "step_time_sec": 8.23822001402732, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3339, "loss": 4.305459022521973, "lr": 0.0002, "elapsed_sec": 27751.615514039993, "step_time_sec": 8.231366188032553, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3340, "loss": 4.302170753479004, "lr": 0.0002, "elapsed_sec": 27759.854552030563, "step_time_sec": 8.238844166975468, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3341, "loss": 4.432943820953369, "lr": 0.0002, "elapsed_sec": 27768.088466644287, "step_time_sec": 8.233785207034089, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3342, "loss": 4.464234828948975, "lr": 0.0002, "elapsed_sec": 27776.327832460403, "step_time_sec": 8.239249157020822, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3343, "loss": 4.33333683013916, "lr": 0.0002, "elapsed_sec": 27784.561047315598, "step_time_sec": 8.232979499967769, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3344, "loss": 4.327271938323975, "lr": 0.0002, "elapsed_sec": 27792.799202680588, "step_time_sec": 8.237994807073846, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3345, "loss": 4.410742282867432, "lr": 0.0002, "elapsed_sec": 27801.036722183228, "step_time_sec": 8.23735936009325, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3346, "loss": 4.449737548828125, "lr": 0.0002, "elapsed_sec": 27809.27295780182, "step_time_sec": 8.23613820807077, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3347, "loss": 4.355191230773926, "lr": 0.0002, "elapsed_sec": 27817.509509563446, "step_time_sec": 8.236362342024222, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3348, "loss": 4.364027976989746, "lr": 0.0002, "elapsed_sec": 27825.747292757034, "step_time_sec": 8.237602259032428, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3349, "loss": 4.493017196655273, "lr": 0.0002, "elapsed_sec": 27833.98548936844, "step_time_sec": 8.238023910904303, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3350, "loss": 4.366335391998291, "lr": 0.0002, "elapsed_sec": 27842.217317342758, "step_time_sec": 8.231674694921821, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3351, "loss": 4.473071575164795, "lr": 0.0002, "elapsed_sec": 27850.449251413345, "step_time_sec": 8.231745969038457, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3352, "loss": 4.417202472686768, "lr": 0.0002, "elapsed_sec": 27858.684827566147, "step_time_sec": 8.235459255054593, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3353, "loss": 4.368654251098633, "lr": 0.0002, "elapsed_sec": 27866.917241334915, "step_time_sec": 8.232217230950482, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3354, "loss": 4.442241191864014, "lr": 0.0002, "elapsed_sec": 27875.149443864822, "step_time_sec": 8.232040085014887, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3355, "loss": 4.415924549102783, "lr": 0.0002, "elapsed_sec": 27883.382952928543, "step_time_sec": 8.2333256280981, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3356, "loss": 4.366641521453857, "lr": 0.0002, "elapsed_sec": 27891.615269184113, "step_time_sec": 8.232178511912934, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3357, "loss": 4.365773677825928, "lr": 0.0002, "elapsed_sec": 27899.85388112068, "step_time_sec": 8.238479827996343, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3358, "loss": 4.5409626960754395, "lr": 0.0002, "elapsed_sec": 27908.084986448288, "step_time_sec": 8.230907476041466, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3359, "loss": 4.290529727935791, "lr": 0.0002, "elapsed_sec": 27916.322940587997, "step_time_sec": 8.237844832008705, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3360, "loss": 4.405423641204834, "lr": 0.0002, "elapsed_sec": 27924.55961370468, "step_time_sec": 8.236432212986983, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3361, "loss": 4.453901290893555, "lr": 0.0002, "elapsed_sec": 27932.7936065197, "step_time_sec": 8.233851447002962, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3362, "loss": 4.419208526611328, "lr": 0.0002, "elapsed_sec": 27941.025940179825, "step_time_sec": 8.232143235974945, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3363, "loss": 4.410192489624023, "lr": 0.0002, "elapsed_sec": 27949.26326608658, "step_time_sec": 8.237185562029481, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3364, "loss": 4.536483287811279, "lr": 0.0002, "elapsed_sec": 27957.495000123978, "step_time_sec": 8.231604558066465, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3365, "loss": 4.404520034790039, "lr": 0.0002, "elapsed_sec": 27965.73156452179, "step_time_sec": 8.236383975017816, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3366, "loss": 4.432997226715088, "lr": 0.0002, "elapsed_sec": 27973.96325826645, "step_time_sec": 8.231574550969526, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3367, "loss": 4.544892311096191, "lr": 0.0002, "elapsed_sec": 27982.20071387291, "step_time_sec": 8.2372637699591, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3368, "loss": 4.442546844482422, "lr": 0.0002, "elapsed_sec": 27990.43897175789, "step_time_sec": 8.238127502030693, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3369, "loss": 4.2857184410095215, "lr": 0.0002, "elapsed_sec": 27998.67211508751, "step_time_sec": 8.23294130701106, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3370, "loss": 4.417070388793945, "lr": 0.0002, "elapsed_sec": 28006.904891490936, "step_time_sec": 8.232637827983126, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3371, "loss": 4.33125638961792, "lr": 0.0002, "elapsed_sec": 28015.137311458588, "step_time_sec": 8.232231786940247, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3372, "loss": 4.395755290985107, "lr": 0.0002, "elapsed_sec": 28023.36930847168, "step_time_sec": 8.23181912000291, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3373, "loss": 4.311310768127441, "lr": 0.0002, "elapsed_sec": 28031.60284280777, "step_time_sec": 8.23342631594278, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3374, "loss": 4.502244472503662, "lr": 0.0002, "elapsed_sec": 28039.84064769745, "step_time_sec": 8.237625416019, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3375, "loss": 4.358160972595215, "lr": 0.0002, "elapsed_sec": 28048.076642751694, "step_time_sec": 8.235806673998013, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3376, "loss": 4.477359771728516, "lr": 0.0002, "elapsed_sec": 28056.309092998505, "step_time_sec": 8.23237106599845, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3377, "loss": 4.467742919921875, "lr": 0.0002, "elapsed_sec": 28064.542326688766, "step_time_sec": 8.232994455029257, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3378, "loss": 4.395197868347168, "lr": 0.0002, "elapsed_sec": 28072.777623653412, "step_time_sec": 8.23517159698531, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3379, "loss": 4.311135292053223, "lr": 0.0002, "elapsed_sec": 28081.010565042496, "step_time_sec": 8.232750635012053, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3380, "loss": 4.472747802734375, "lr": 0.0002, "elapsed_sec": 28089.24219083786, "step_time_sec": 8.231485747965053, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3381, "loss": 4.449331283569336, "lr": 0.0002, "elapsed_sec": 28097.47430896759, "step_time_sec": 8.231985909980722, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3382, "loss": 4.4172682762146, "lr": 0.0002, "elapsed_sec": 28105.709300518036, "step_time_sec": 8.234833418042399, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3383, "loss": 4.401401996612549, "lr": 0.0002, "elapsed_sec": 28113.942362070084, "step_time_sec": 8.232913209940307, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3384, "loss": 4.247726917266846, "lr": 0.0002, "elapsed_sec": 28122.173406362534, "step_time_sec": 8.230860952986404, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3385, "loss": 4.541574478149414, "lr": 0.0002, "elapsed_sec": 28130.405567884445, "step_time_sec": 8.231947178021073, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3386, "loss": 4.332236289978027, "lr": 0.0002, "elapsed_sec": 28138.636688947678, "step_time_sec": 8.230998217011802, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3387, "loss": 4.335377216339111, "lr": 0.0002, "elapsed_sec": 28146.86933708191, "step_time_sec": 8.232442263979465, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3388, "loss": 4.291614055633545, "lr": 0.0002, "elapsed_sec": 28155.0997030735, "step_time_sec": 8.230207783984952, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3389, "loss": 4.463893890380859, "lr": 0.0002, "elapsed_sec": 28163.33171224594, "step_time_sec": 8.231833308003843, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3390, "loss": 4.463738441467285, "lr": 0.0002, "elapsed_sec": 28171.566462516785, "step_time_sec": 8.234583624987863, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3391, "loss": 4.352858543395996, "lr": 0.0002, "elapsed_sec": 28179.798694610596, "step_time_sec": 8.23207414499484, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3392, "loss": 4.331420421600342, "lr": 0.0002, "elapsed_sec": 28188.03136754036, "step_time_sec": 8.232495140982792, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3393, "loss": 4.517204761505127, "lr": 0.0002, "elapsed_sec": 28196.26290154457, "step_time_sec": 8.231423752964474, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3394, "loss": 4.410607814788818, "lr": 0.0002, "elapsed_sec": 28204.496736049652, "step_time_sec": 8.23363654105924, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3395, "loss": 4.303769588470459, "lr": 0.0002, "elapsed_sec": 28212.728605508804, "step_time_sec": 8.231722443015315, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3396, "loss": 4.386658191680908, "lr": 0.0002, "elapsed_sec": 28220.96231508255, "step_time_sec": 8.233517621061765, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3397, "loss": 4.407001972198486, "lr": 0.0002, "elapsed_sec": 28229.192911863327, "step_time_sec": 8.230422348948196, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3398, "loss": 4.3668107986450195, "lr": 0.0002, "elapsed_sec": 28237.426092863083, "step_time_sec": 8.233060228056274, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3399, "loss": 4.550961971282959, "lr": 0.0002, "elapsed_sec": 28245.659306764603, "step_time_sec": 8.233030940056778, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3400, "loss": 4.468221664428711, "lr": 0.0002, "elapsed_sec": 28253.891928195953, "step_time_sec": 8.232455682009459, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3401, "loss": 4.331059455871582, "lr": 0.0002, "elapsed_sec": 28262.12432575226, "step_time_sec": 8.232221347978339, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3402, "loss": 4.407310962677002, "lr": 0.0002, "elapsed_sec": 28270.357179641724, "step_time_sec": 8.232700352091342, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3403, "loss": 4.375729560852051, "lr": 0.0002, "elapsed_sec": 28278.589795589447, "step_time_sec": 8.232486029039137, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3404, "loss": 4.475652694702148, "lr": 0.0002, "elapsed_sec": 28286.82275390625, "step_time_sec": 8.232750943978317, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3405, "loss": 4.377934455871582, "lr": 0.0002, "elapsed_sec": 28295.054618120193, "step_time_sec": 8.231729340041056, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3406, "loss": 4.300724983215332, "lr": 0.0002, "elapsed_sec": 28303.286345005035, "step_time_sec": 8.2315410109004, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3407, "loss": 4.373461723327637, "lr": 0.0002, "elapsed_sec": 28311.51913881302, "step_time_sec": 8.232627159915864, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3408, "loss": 4.460932731628418, "lr": 0.0002, "elapsed_sec": 28319.750631332397, "step_time_sec": 8.231350832968019, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3409, "loss": 4.42286491394043, "lr": 0.0002, "elapsed_sec": 28327.983254909515, "step_time_sec": 8.232518453034572, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3410, "loss": 4.373054504394531, "lr": 0.0002, "elapsed_sec": 28336.215434074402, "step_time_sec": 8.232036691973917, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3411, "loss": 4.535395622253418, "lr": 0.0002, "elapsed_sec": 28344.44837284088, "step_time_sec": 8.232655402040109, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3412, "loss": 4.3000874519348145, "lr": 0.0002, "elapsed_sec": 28352.679329633713, "step_time_sec": 8.230834402958862, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3413, "loss": 4.413848400115967, "lr": 0.0002, "elapsed_sec": 28360.91286301613, "step_time_sec": 8.233325763954781, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3414, "loss": 4.410678386688232, "lr": 0.0002, "elapsed_sec": 28369.14471721649, "step_time_sec": 8.231723041972145, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3415, "loss": 4.355880260467529, "lr": 0.0002, "elapsed_sec": 28377.378053188324, "step_time_sec": 8.233187705976889, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3416, "loss": 4.283920764923096, "lr": 0.0002, "elapsed_sec": 28385.610117912292, "step_time_sec": 8.231873392011039, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3417, "loss": 4.392091751098633, "lr": 0.0002, "elapsed_sec": 28393.843354940414, "step_time_sec": 8.233092661947012, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3418, "loss": 4.251224517822266, "lr": 0.0002, "elapsed_sec": 28402.07518529892, "step_time_sec": 8.231611319002695, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3419, "loss": 4.450288772583008, "lr": 0.0002, "elapsed_sec": 28410.307480096817, "step_time_sec": 8.232133558019996, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3420, "loss": 4.50157356262207, "lr": 0.0002, "elapsed_sec": 28418.540254354477, "step_time_sec": 8.232610233011656, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3421, "loss": 4.381917953491211, "lr": 0.0002, "elapsed_sec": 28426.773152828217, "step_time_sec": 8.232764672022313, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3422, "loss": 4.348518371582031, "lr": 0.0002, "elapsed_sec": 28435.005905151367, "step_time_sec": 8.232573344022967, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3423, "loss": 4.304539203643799, "lr": 0.0002, "elapsed_sec": 28443.237370729446, "step_time_sec": 8.23127322900109, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3424, "loss": 4.4165263175964355, "lr": 0.0002, "elapsed_sec": 28451.47077512741, "step_time_sec": 8.233225899050012, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3425, "loss": 4.5133538246154785, "lr": 0.0002, "elapsed_sec": 28459.702829122543, "step_time_sec": 8.231892079929821, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3426, "loss": 4.409897327423096, "lr": 0.0002, "elapsed_sec": 28467.9345433712, "step_time_sec": 8.23159320501145, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3427, "loss": 4.412725448608398, "lr": 0.0002, "elapsed_sec": 28476.167838573456, "step_time_sec": 8.233124608988874, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3428, "loss": 4.330204963684082, "lr": 0.0002, "elapsed_sec": 28484.401430368423, "step_time_sec": 8.233376336982474, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3429, "loss": 4.490208625793457, "lr": 0.0002, "elapsed_sec": 28492.632303237915, "step_time_sec": 8.230742034036666, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3430, "loss": 4.39249324798584, "lr": 0.0002, "elapsed_sec": 28500.86416387558, "step_time_sec": 8.231696675997227, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3431, "loss": 4.307723045349121, "lr": 0.0002, "elapsed_sec": 28509.095633506775, "step_time_sec": 8.231271708966233, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3432, "loss": 4.218557357788086, "lr": 0.0002, "elapsed_sec": 28517.327256917953, "step_time_sec": 8.231428137980402, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3433, "loss": 4.48938512802124, "lr": 0.0002, "elapsed_sec": 28525.558989048004, "step_time_sec": 8.231594975921325, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3434, "loss": 4.226943016052246, "lr": 0.0002, "elapsed_sec": 28533.789816617966, "step_time_sec": 8.230653716018423, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3435, "loss": 4.3722662925720215, "lr": 0.0002, "elapsed_sec": 28542.02100801468, "step_time_sec": 8.231046367087401, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3436, "loss": 4.267021179199219, "lr": 0.0002, "elapsed_sec": 28550.253733873367, "step_time_sec": 8.232560764066875, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3437, "loss": 4.560519695281982, "lr": 0.0002, "elapsed_sec": 28558.48671388626, "step_time_sec": 8.232834329013713, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3438, "loss": 4.343455791473389, "lr": 0.0002, "elapsed_sec": 28566.717188358307, "step_time_sec": 8.23033585306257, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3439, "loss": 4.431280612945557, "lr": 0.0002, "elapsed_sec": 28574.95009660721, "step_time_sec": 8.232720326050185, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3440, "loss": 4.526254177093506, "lr": 0.0002, "elapsed_sec": 28583.182034254074, "step_time_sec": 8.231756206019782, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3441, "loss": 4.263418197631836, "lr": 0.0002, "elapsed_sec": 28591.413732528687, "step_time_sec": 8.231590747949667, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3442, "loss": 4.438264846801758, "lr": 0.0002, "elapsed_sec": 28599.645871162415, "step_time_sec": 8.23192618205212, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3443, "loss": 4.301305770874023, "lr": 0.0002, "elapsed_sec": 28607.878236293793, "step_time_sec": 8.232240532990545, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3444, "loss": 4.344808578491211, "lr": 0.0002, "elapsed_sec": 28616.109738349915, "step_time_sec": 8.231301976018585, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3445, "loss": 4.368781566619873, "lr": 0.0002, "elapsed_sec": 28624.34095978737, "step_time_sec": 8.23107502493076, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3446, "loss": 4.330414295196533, "lr": 0.0002, "elapsed_sec": 28632.57090997696, "step_time_sec": 8.229772715014406, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3447, "loss": 4.447818756103516, "lr": 0.0002, "elapsed_sec": 28640.803249835968, "step_time_sec": 8.232229428947903, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3448, "loss": 4.414568901062012, "lr": 0.0002, "elapsed_sec": 28649.03507375717, "step_time_sec": 8.231637226999737, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3449, "loss": 4.53003454208374, "lr": 0.0002, "elapsed_sec": 28657.266923189163, "step_time_sec": 8.231687967083417, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3450, "loss": 4.408664703369141, "lr": 0.0002, "elapsed_sec": 28665.49999332428, "step_time_sec": 8.232950152945705, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3451, "loss": 4.237273693084717, "lr": 0.0002, "elapsed_sec": 28673.7320394516, "step_time_sec": 8.231834477977827, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3452, "loss": 4.353357315063477, "lr": 0.0002, "elapsed_sec": 28681.96317410469, "step_time_sec": 8.231020238017663, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3453, "loss": 4.250943660736084, "lr": 0.0002, "elapsed_sec": 28690.19392323494, "step_time_sec": 8.230529706925154, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3454, "loss": 4.3621392250061035, "lr": 0.0002, "elapsed_sec": 28698.426156282425, "step_time_sec": 8.232149021001533, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3455, "loss": 4.462343692779541, "lr": 0.0002, "elapsed_sec": 28706.657984018326, "step_time_sec": 8.23164084693417, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3456, "loss": 4.5152363777160645, "lr": 0.0002, "elapsed_sec": 28714.89035797119, "step_time_sec": 8.23216114193201, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3457, "loss": 4.324028968811035, "lr": 0.0002, "elapsed_sec": 28723.123154878616, "step_time_sec": 8.232663589995354, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3458, "loss": 4.2908124923706055, "lr": 0.0002, "elapsed_sec": 28731.354078292847, "step_time_sec": 8.230756638920866, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3459, "loss": 4.308454513549805, "lr": 0.0002, "elapsed_sec": 28739.585745334625, "step_time_sec": 8.231512428959832, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3460, "loss": 4.408417701721191, "lr": 0.0002, "elapsed_sec": 28747.81809401512, "step_time_sec": 8.232180143939331, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3461, "loss": 4.385887622833252, "lr": 0.0002, "elapsed_sec": 28756.050741434097, "step_time_sec": 8.232491142000072, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3462, "loss": 4.320639610290527, "lr": 0.0002, "elapsed_sec": 28764.282219409943, "step_time_sec": 8.231293214019388, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3463, "loss": 4.514185905456543, "lr": 0.0002, "elapsed_sec": 28772.514350891113, "step_time_sec": 8.231974043999799, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3464, "loss": 4.362950325012207, "lr": 0.0002, "elapsed_sec": 28780.745577812195, "step_time_sec": 8.231041593011469, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3465, "loss": 4.328076362609863, "lr": 0.0002, "elapsed_sec": 28788.97648048401, "step_time_sec": 8.23073985404335, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3466, "loss": 4.3872599601745605, "lr": 0.0002, "elapsed_sec": 28797.208899736404, "step_time_sec": 8.232264161924832, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3467, "loss": 4.264732837677002, "lr": 0.0002, "elapsed_sec": 28805.440008163452, "step_time_sec": 8.230934321996756, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3468, "loss": 4.227305889129639, "lr": 0.0002, "elapsed_sec": 28813.673148155212, "step_time_sec": 8.233028189977631, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3469, "loss": 4.381351470947266, "lr": 0.0002, "elapsed_sec": 28821.90481519699, "step_time_sec": 8.231443300959654, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3470, "loss": 4.388479232788086, "lr": 0.0002, "elapsed_sec": 28830.137249708176, "step_time_sec": 8.232287082937546, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3471, "loss": 4.303308010101318, "lr": 0.0002, "elapsed_sec": 28838.36913394928, "step_time_sec": 8.231750316102989, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3472, "loss": 4.263008117675781, "lr": 0.0002, "elapsed_sec": 28846.602609157562, "step_time_sec": 8.233356304001063, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3473, "loss": 4.447658538818359, "lr": 0.0002, "elapsed_sec": 28854.834982156754, "step_time_sec": 8.23215281299781, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3474, "loss": 4.421505928039551, "lr": 0.0002, "elapsed_sec": 28863.06591796875, "step_time_sec": 8.230822212994099, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3475, "loss": 4.239105701446533, "lr": 0.0002, "elapsed_sec": 28871.29687142372, "step_time_sec": 8.230766550055705, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3476, "loss": 4.4345173835754395, "lr": 0.0002, "elapsed_sec": 28879.528331041336, "step_time_sec": 8.231329180998728, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3477, "loss": 4.279122352600098, "lr": 0.0002, "elapsed_sec": 28887.760389089584, "step_time_sec": 8.231873448006809, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3478, "loss": 4.338593482971191, "lr": 0.0002, "elapsed_sec": 28895.99280166626, "step_time_sec": 8.232295888010412, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3479, "loss": 4.398959636688232, "lr": 0.0002, "elapsed_sec": 28904.22579407692, "step_time_sec": 8.232779953978024, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3480, "loss": 4.226840019226074, "lr": 0.0002, "elapsed_sec": 28912.457389831543, "step_time_sec": 8.231485625961795, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3481, "loss": 4.356712818145752, "lr": 0.0002, "elapsed_sec": 28920.69029378891, "step_time_sec": 8.232724442961626, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3482, "loss": 4.377618789672852, "lr": 0.0002, "elapsed_sec": 28928.92085790634, "step_time_sec": 8.23040105192922, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3483, "loss": 4.27191686630249, "lr": 0.0002, "elapsed_sec": 28937.154677152634, "step_time_sec": 8.23368114803452, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3484, "loss": 4.499473571777344, "lr": 0.0002, "elapsed_sec": 28945.38737630844, "step_time_sec": 8.2325575100258, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3485, "loss": 4.404983997344971, "lr": 0.0002, "elapsed_sec": 28953.647145986557, "step_time_sec": 8.2595988989342, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3486, "loss": 4.395076274871826, "lr": 0.0002, "elapsed_sec": 28961.882645130157, "step_time_sec": 8.23524493793957, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3487, "loss": 4.4513139724731445, "lr": 0.0002, "elapsed_sec": 28970.12033700943, "step_time_sec": 8.237548209959641, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3488, "loss": 4.465000629425049, "lr": 0.0002, "elapsed_sec": 28978.358162164688, "step_time_sec": 8.237606299109757, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3489, "loss": 4.295258522033691, "lr": 0.0002, "elapsed_sec": 28986.594457149506, "step_time_sec": 8.236153524019755, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3490, "loss": 4.452977180480957, "lr": 0.0002, "elapsed_sec": 28994.82826733589, "step_time_sec": 8.233619207981974, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3491, "loss": 4.476616382598877, "lr": 0.0002, "elapsed_sec": 29003.064248800278, "step_time_sec": 8.235823643044569, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3492, "loss": 4.363863945007324, "lr": 0.0002, "elapsed_sec": 29011.301478624344, "step_time_sec": 8.237084796070121, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3493, "loss": 4.395393371582031, "lr": 0.0002, "elapsed_sec": 29019.536206960678, "step_time_sec": 8.234500011079945, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3494, "loss": 4.523796558380127, "lr": 0.0002, "elapsed_sec": 29027.768281936646, "step_time_sec": 8.23194249800872, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3495, "loss": 4.257209777832031, "lr": 0.0002, "elapsed_sec": 29036.002307653427, "step_time_sec": 8.233874741010368, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3496, "loss": 4.365158557891846, "lr": 0.0002, "elapsed_sec": 29044.233511447906, "step_time_sec": 8.23104011407122, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3497, "loss": 4.250648021697998, "lr": 0.0002, "elapsed_sec": 29052.46523475647, "step_time_sec": 8.231551553006284, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3498, "loss": 4.303411483764648, "lr": 0.0002, "elapsed_sec": 29060.69564795494, "step_time_sec": 8.23027134896256, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3499, "loss": 4.336148262023926, "lr": 0.0002, "elapsed_sec": 29068.926392555237, "step_time_sec": 8.23049056599848, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3500, "loss": 4.34026575088501, "lr": 0.0002, "elapsed_sec": 29077.15875530243, "step_time_sec": 34.77543511497788, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": true, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3501, "loss": 4.470857620239258, "lr": 0.0002, "elapsed_sec": 29111.94184064865, "step_time_sec": 8.239463389967568, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3502, "loss": 4.4290900230407715, "lr": 0.0002, "elapsed_sec": 29120.16851902008, "step_time_sec": 8.226459480938502, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3503, "loss": 4.284092426300049, "lr": 0.0002, "elapsed_sec": 29128.40060377121, "step_time_sec": 8.231953373993747, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3504, "loss": 4.347887992858887, "lr": 0.0002, "elapsed_sec": 29136.632858753204, "step_time_sec": 8.232113649952225, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3505, "loss": 4.452051639556885, "lr": 0.0002, "elapsed_sec": 29144.864567518234, "step_time_sec": 8.231558692990802, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3506, "loss": 4.405646800994873, "lr": 0.0002, "elapsed_sec": 29153.096557855606, "step_time_sec": 8.231828197021969, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3507, "loss": 4.443693161010742, "lr": 0.0002, "elapsed_sec": 29161.32808613777, "step_time_sec": 8.231335167074576, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3508, "loss": 4.313877582550049, "lr": 0.0002, "elapsed_sec": 29169.560095071793, "step_time_sec": 8.231847985996865, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3509, "loss": 4.302901268005371, "lr": 0.0002, "elapsed_sec": 29177.792837142944, "step_time_sec": 8.232512021088041, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3510, "loss": 4.435062408447266, "lr": 0.0002, "elapsed_sec": 29186.02347803116, "step_time_sec": 8.230494968011044, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3511, "loss": 4.34998083114624, "lr": 0.0002, "elapsed_sec": 29194.254913806915, "step_time_sec": 8.231243446003646, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3512, "loss": 4.307962417602539, "lr": 0.0002, "elapsed_sec": 29202.486511945724, "step_time_sec": 8.231441417941824, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3513, "loss": 4.321359634399414, "lr": 0.0002, "elapsed_sec": 29210.717471837997, "step_time_sec": 8.230817201081663, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3514, "loss": 4.397961616516113, "lr": 0.0002, "elapsed_sec": 29218.94709920883, "step_time_sec": 8.229460194939747, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3515, "loss": 4.275479793548584, "lr": 0.0002, "elapsed_sec": 29227.17658472061, "step_time_sec": 8.229334934032522, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3516, "loss": 4.3802032470703125, "lr": 0.0002, "elapsed_sec": 29235.408757209778, "step_time_sec": 8.232063675066456, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3517, "loss": 4.397853851318359, "lr": 0.0002, "elapsed_sec": 29243.640871286392, "step_time_sec": 8.231918858014978, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3518, "loss": 4.161469459533691, "lr": 0.0002, "elapsed_sec": 29251.873797893524, "step_time_sec": 8.232753489050083, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3519, "loss": 4.366097450256348, "lr": 0.0002, "elapsed_sec": 29260.104199171066, "step_time_sec": 8.230258249910548, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3520, "loss": 4.428991317749023, "lr": 0.0002, "elapsed_sec": 29268.333892822266, "step_time_sec": 8.229547598981299, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3521, "loss": 4.242702484130859, "lr": 0.0002, "elapsed_sec": 29276.5637717247, "step_time_sec": 8.229738578083925, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3522, "loss": 4.311009883880615, "lr": 0.0002, "elapsed_sec": 29284.795466899872, "step_time_sec": 8.231493504019454, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3523, "loss": 4.270780563354492, "lr": 0.0002, "elapsed_sec": 29293.026309490204, "step_time_sec": 8.230655280989595, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3524, "loss": 4.135424613952637, "lr": 0.0002, "elapsed_sec": 29301.25697851181, "step_time_sec": 8.230482639046386, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3525, "loss": 4.375688076019287, "lr": 0.0002, "elapsed_sec": 29309.488577604294, "step_time_sec": 8.23145185294561, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3526, "loss": 4.268374443054199, "lr": 0.0002, "elapsed_sec": 29317.71863603592, "step_time_sec": 8.229909099987708, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3527, "loss": 4.409497261047363, "lr": 0.0002, "elapsed_sec": 29325.95118546486, "step_time_sec": 8.232400709996, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3528, "loss": 4.251886367797852, "lr": 0.0002, "elapsed_sec": 29334.181099414825, "step_time_sec": 8.22972121508792, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3529, "loss": 4.313004016876221, "lr": 0.0002, "elapsed_sec": 29342.411019086838, "step_time_sec": 8.229767516022548, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3530, "loss": 4.5245890617370605, "lr": 0.0002, "elapsed_sec": 29350.641651391983, "step_time_sec": 8.230512456968427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3531, "loss": 4.411394119262695, "lr": 0.0002, "elapsed_sec": 29358.874676704407, "step_time_sec": 8.232874130946584, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3532, "loss": 4.2949676513671875, "lr": 0.0002, "elapsed_sec": 29367.10681128502, "step_time_sec": 8.231967263971455, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3533, "loss": 4.224719047546387, "lr": 0.0002, "elapsed_sec": 29375.33778977394, "step_time_sec": 8.230784077895805, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3534, "loss": 4.2019124031066895, "lr": 0.0002, "elapsed_sec": 29383.569774866104, "step_time_sec": 8.231860086089, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3535, "loss": 4.348132610321045, "lr": 0.0002, "elapsed_sec": 29391.799907922745, "step_time_sec": 8.229955478920601, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3536, "loss": 4.29004430770874, "lr": 0.0002, "elapsed_sec": 29400.02961373329, "step_time_sec": 8.229553253971972, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3537, "loss": 4.387026786804199, "lr": 0.0002, "elapsed_sec": 29408.260050296783, "step_time_sec": 8.230279117939062, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3538, "loss": 4.278601169586182, "lr": 0.0002, "elapsed_sec": 29416.589282751083, "step_time_sec": 8.329112497041933, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3539, "loss": 4.221573829650879, "lr": 0.0002, "elapsed_sec": 29424.819872379303, "step_time_sec": 8.23041081102565, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3540, "loss": 4.356555938720703, "lr": 0.0002, "elapsed_sec": 29433.37828564644, "step_time_sec": 8.558277742937207, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3541, "loss": 4.455869674682617, "lr": 0.0002, "elapsed_sec": 29441.908144950867, "step_time_sec": 8.529667243012227, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3542, "loss": 4.306445598602295, "lr": 0.0002, "elapsed_sec": 29450.13969564438, "step_time_sec": 8.231387680047192, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3543, "loss": 4.388194561004639, "lr": 0.0002, "elapsed_sec": 29458.699935674667, "step_time_sec": 8.56011615798343, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3544, "loss": 4.330454349517822, "lr": 0.0002, "elapsed_sec": 29467.261078596115, "step_time_sec": 8.560970708960667, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3545, "loss": 4.321819305419922, "lr": 0.0002, "elapsed_sec": 29475.492460250854, "step_time_sec": 8.231216558022425, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3546, "loss": 4.327335357666016, "lr": 0.0002, "elapsed_sec": 29483.963677167892, "step_time_sec": 8.471051341970451, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3547, "loss": 4.245993137359619, "lr": 0.0002, "elapsed_sec": 29492.576805114746, "step_time_sec": 8.612924539949745, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3548, "loss": 4.396326065063477, "lr": 0.0002, "elapsed_sec": 29500.808875322342, "step_time_sec": 8.231922172009945, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3549, "loss": 4.49164342880249, "lr": 0.0002, "elapsed_sec": 29509.220351934433, "step_time_sec": 8.411345947999507, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3550, "loss": 4.378093719482422, "lr": 0.0002, "elapsed_sec": 29517.599679470062, "step_time_sec": 8.379145903978497, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3551, "loss": 4.429680347442627, "lr": 0.0002, "elapsed_sec": 29526.159236192703, "step_time_sec": 8.559400469996035, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3552, "loss": 4.426207065582275, "lr": 0.0002, "elapsed_sec": 29534.720138788223, "step_time_sec": 8.560725659946911, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3553, "loss": 4.356632232666016, "lr": 0.0002, "elapsed_sec": 29542.952039003372, "step_time_sec": 8.231710088904947, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3554, "loss": 4.346240520477295, "lr": 0.0002, "elapsed_sec": 29551.189600229263, "step_time_sec": 8.237510116072372, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3555, "loss": 4.349908828735352, "lr": 0.0002, "elapsed_sec": 29559.741991996765, "step_time_sec": 8.552114352933131, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3556, "loss": 4.329570293426514, "lr": 0.0002, "elapsed_sec": 29567.973624944687, "step_time_sec": 8.231448572012596, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3557, "loss": 4.236452102661133, "lr": 0.0002, "elapsed_sec": 29576.205973625183, "step_time_sec": 8.232240030891262, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3558, "loss": 4.193739414215088, "lr": 0.0002, "elapsed_sec": 29584.437462329865, "step_time_sec": 8.231282489025034, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3559, "loss": 4.239534854888916, "lr": 0.0002, "elapsed_sec": 29592.669511556625, "step_time_sec": 8.231863835942931, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3560, "loss": 4.336446762084961, "lr": 0.0002, "elapsed_sec": 29600.90275120735, "step_time_sec": 8.233096780022606, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3561, "loss": 4.414663791656494, "lr": 0.0002, "elapsed_sec": 29609.135124206543, "step_time_sec": 8.23218743596226, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3562, "loss": 4.395718574523926, "lr": 0.0002, "elapsed_sec": 29617.366950273514, "step_time_sec": 8.231676477007568, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3563, "loss": 4.386781692504883, "lr": 0.0002, "elapsed_sec": 29625.599629878998, "step_time_sec": 8.232504357001744, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3564, "loss": 4.227298259735107, "lr": 0.0002, "elapsed_sec": 29633.830278396606, "step_time_sec": 8.23050800897181, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3565, "loss": 4.253028392791748, "lr": 0.0002, "elapsed_sec": 29642.06270289421, "step_time_sec": 8.23223115701694, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3566, "loss": 4.28925895690918, "lr": 0.0002, "elapsed_sec": 29650.294713258743, "step_time_sec": 8.231886736000888, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3567, "loss": 4.246706962585449, "lr": 0.0002, "elapsed_sec": 29658.852959394455, "step_time_sec": 8.558074224973097, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3568, "loss": 4.454339027404785, "lr": 0.0002, "elapsed_sec": 29667.082911014557, "step_time_sec": 8.229757016990334, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3569, "loss": 4.318968296051025, "lr": 0.0002, "elapsed_sec": 29675.314628124237, "step_time_sec": 8.231562262051739, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3570, "loss": 4.424945831298828, "lr": 0.0002, "elapsed_sec": 29683.545729875565, "step_time_sec": 8.230894539039582, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3571, "loss": 4.333166599273682, "lr": 0.0002, "elapsed_sec": 29691.77629494667, "step_time_sec": 8.230396129074506, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3572, "loss": 4.291715145111084, "lr": 0.0002, "elapsed_sec": 29700.006556034088, "step_time_sec": 8.230109935044311, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3573, "loss": 4.255983352661133, "lr": 0.0002, "elapsed_sec": 29708.237746715546, "step_time_sec": 8.231057827011682, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3574, "loss": 4.506494998931885, "lr": 0.0002, "elapsed_sec": 29716.470376491547, "step_time_sec": 8.232485763030127, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3575, "loss": 4.385729789733887, "lr": 0.0002, "elapsed_sec": 29724.700734615326, "step_time_sec": 8.23016703990288, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3576, "loss": 4.4345197677612305, "lr": 0.0002, "elapsed_sec": 29732.931682109833, "step_time_sec": 8.230753982090391, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3577, "loss": 4.382013320922852, "lr": 0.0002, "elapsed_sec": 29741.164164543152, "step_time_sec": 8.232341180904768, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3578, "loss": 4.29502534866333, "lr": 0.0002, "elapsed_sec": 29749.39487838745, "step_time_sec": 8.230560548021458, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3579, "loss": 4.382268905639648, "lr": 0.0002, "elapsed_sec": 29757.626394033432, "step_time_sec": 8.231401975033805, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3580, "loss": 4.557867527008057, "lr": 0.0002, "elapsed_sec": 29765.855013608932, "step_time_sec": 8.22841623192653, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3581, "loss": 4.327199459075928, "lr": 0.0002, "elapsed_sec": 29774.08694911003, "step_time_sec": 8.231784043950029, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3582, "loss": 4.408513069152832, "lr": 0.0002, "elapsed_sec": 29782.31775689125, "step_time_sec": 8.230667322990485, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3583, "loss": 4.487490653991699, "lr": 0.0002, "elapsed_sec": 29790.877439022064, "step_time_sec": 8.559559928020462, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3584, "loss": 4.4095845222473145, "lr": 0.0002, "elapsed_sec": 29799.109162330627, "step_time_sec": 8.231518972082995, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3585, "loss": 4.337475776672363, "lr": 0.0002, "elapsed_sec": 29807.339597702026, "step_time_sec": 8.230276233982295, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3586, "loss": 4.391106605529785, "lr": 0.0002, "elapsed_sec": 29815.571437358856, "step_time_sec": 8.231652262969874, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3587, "loss": 4.481678009033203, "lr": 0.0002, "elapsed_sec": 29823.80415081978, "step_time_sec": 8.232582529075444, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3588, "loss": 4.428464412689209, "lr": 0.0002, "elapsed_sec": 29832.036331892014, "step_time_sec": 8.231982136028819, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3589, "loss": 4.345834255218506, "lr": 0.0002, "elapsed_sec": 29840.268698453903, "step_time_sec": 8.232263642945327, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3590, "loss": 4.310297966003418, "lr": 0.0002, "elapsed_sec": 29848.50053334236, "step_time_sec": 8.231650145025924, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3591, "loss": 4.431910037994385, "lr": 0.0002, "elapsed_sec": 29856.72957277298, "step_time_sec": 8.228914177976549, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3592, "loss": 4.426026821136475, "lr": 0.0002, "elapsed_sec": 29864.959705114365, "step_time_sec": 8.229934011003934, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3593, "loss": 4.333749771118164, "lr": 0.0002, "elapsed_sec": 29873.190638065338, "step_time_sec": 8.23079510498792, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3594, "loss": 4.280147552490234, "lr": 0.0002, "elapsed_sec": 29881.421765565872, "step_time_sec": 8.230952684069052, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3595, "loss": 4.389430046081543, "lr": 0.0002, "elapsed_sec": 29889.651263713837, "step_time_sec": 8.229331295005977, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3596, "loss": 4.311544418334961, "lr": 0.0002, "elapsed_sec": 29897.881464719772, "step_time_sec": 8.230025901109912, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3597, "loss": 4.3346781730651855, "lr": 0.0002, "elapsed_sec": 29906.112776994705, "step_time_sec": 8.231137054041028, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3598, "loss": 4.453455924987793, "lr": 0.0002, "elapsed_sec": 29914.344766378403, "step_time_sec": 8.231816930929199, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3599, "loss": 4.424271583557129, "lr": 0.0002, "elapsed_sec": 29922.577129125595, "step_time_sec": 8.232168136979453, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3600, "loss": 4.370329856872559, "lr": 0.0002, "elapsed_sec": 29930.809387922287, "step_time_sec": 8.232146788970567, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3601, "loss": 4.35621452331543, "lr": 0.0002, "elapsed_sec": 29939.040065050125, "step_time_sec": 8.23052139999345, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3602, "loss": 4.375487327575684, "lr": 0.0002, "elapsed_sec": 29947.27072119713, "step_time_sec": 8.230488976929337, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3603, "loss": 4.28093957901001, "lr": 0.0002, "elapsed_sec": 29955.502872228622, "step_time_sec": 8.232034116052091, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3604, "loss": 4.369831085205078, "lr": 0.0002, "elapsed_sec": 29963.735132932663, "step_time_sec": 8.23209953296464, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3605, "loss": 4.232161045074463, "lr": 0.0002, "elapsed_sec": 29971.96717262268, "step_time_sec": 8.231853029923514, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3606, "loss": 4.566591262817383, "lr": 0.0002, "elapsed_sec": 29980.199582338333, "step_time_sec": 8.232235885923728, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3607, "loss": 4.363700866699219, "lr": 0.0002, "elapsed_sec": 29988.43056368828, "step_time_sec": 8.230861170915887, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3608, "loss": 4.325138092041016, "lr": 0.0002, "elapsed_sec": 29996.661962985992, "step_time_sec": 8.231240370078012, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3609, "loss": 4.346066951751709, "lr": 0.0002, "elapsed_sec": 30004.89325261116, "step_time_sec": 8.23107483703643, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3610, "loss": 4.303188800811768, "lr": 0.0002, "elapsed_sec": 30013.12472844124, "step_time_sec": 8.231360525940545, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3611, "loss": 4.316436290740967, "lr": 0.0002, "elapsed_sec": 30021.35621738434, "step_time_sec": 8.231295029981993, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3612, "loss": 4.400259017944336, "lr": 0.0002, "elapsed_sec": 30029.588705301285, "step_time_sec": 8.232331596082076, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3613, "loss": 4.22941255569458, "lr": 0.0002, "elapsed_sec": 30037.821171045303, "step_time_sec": 8.232285583973862, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3614, "loss": 4.3090128898620605, "lr": 0.0002, "elapsed_sec": 30046.054250717163, "step_time_sec": 8.232950677978806, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3615, "loss": 4.300355434417725, "lr": 0.0002, "elapsed_sec": 30054.28604030609, "step_time_sec": 8.231680885073729, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3616, "loss": 4.321260452270508, "lr": 0.0002, "elapsed_sec": 30062.516979694366, "step_time_sec": 8.230762050952762, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3617, "loss": 4.420942306518555, "lr": 0.0002, "elapsed_sec": 30070.74719953537, "step_time_sec": 8.230048372992314, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3618, "loss": 4.423195838928223, "lr": 0.0002, "elapsed_sec": 30078.979719400406, "step_time_sec": 8.232383819995448, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3619, "loss": 4.389720916748047, "lr": 0.0002, "elapsed_sec": 30087.211082220078, "step_time_sec": 8.231141390977427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3620, "loss": 4.3400774002075195, "lr": 0.0002, "elapsed_sec": 30095.442799329758, "step_time_sec": 8.231571743031964, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3621, "loss": 4.38229513168335, "lr": 0.0002, "elapsed_sec": 30103.67517066002, "step_time_sec": 8.23222220200114, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3622, "loss": 4.250494480133057, "lr": 0.0002, "elapsed_sec": 30111.906662225723, "step_time_sec": 8.231318756006658, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3623, "loss": 4.40867805480957, "lr": 0.0002, "elapsed_sec": 30120.13880133629, "step_time_sec": 8.231933339964598, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3624, "loss": 4.198604583740234, "lr": 0.0002, "elapsed_sec": 30128.370985031128, "step_time_sec": 8.232068654033355, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3625, "loss": 4.397555828094482, "lr": 0.0002, "elapsed_sec": 30136.603311538696, "step_time_sec": 8.232201809994876, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3626, "loss": 4.406981468200684, "lr": 0.0002, "elapsed_sec": 30144.834187746048, "step_time_sec": 8.230662284069695, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3627, "loss": 4.291843891143799, "lr": 0.0002, "elapsed_sec": 30153.066432237625, "step_time_sec": 8.232112240977585, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3628, "loss": 4.272375583648682, "lr": 0.0002, "elapsed_sec": 30161.298310279846, "step_time_sec": 8.231672546942718, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3629, "loss": 4.393533229827881, "lr": 0.0002, "elapsed_sec": 30169.529433488846, "step_time_sec": 8.230990372016095, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3630, "loss": 4.324523448944092, "lr": 0.0002, "elapsed_sec": 30177.760068416595, "step_time_sec": 8.230504130013287, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3631, "loss": 4.348246097564697, "lr": 0.0002, "elapsed_sec": 30185.991428613663, "step_time_sec": 8.231183483963832, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3632, "loss": 4.397109508514404, "lr": 0.0002, "elapsed_sec": 30194.22249531746, "step_time_sec": 8.230901002069004, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3633, "loss": 4.301020622253418, "lr": 0.0002, "elapsed_sec": 30202.454594373703, "step_time_sec": 8.231906407978386, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3634, "loss": 4.2878923416137695, "lr": 0.0002, "elapsed_sec": 30210.686578035355, "step_time_sec": 8.231849881005473, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3635, "loss": 4.3978376388549805, "lr": 0.0002, "elapsed_sec": 30218.919767141342, "step_time_sec": 8.233026631991379, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3636, "loss": 4.230461120605469, "lr": 0.0002, "elapsed_sec": 30227.476741075516, "step_time_sec": 8.556786515051499, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3637, "loss": 4.333133220672607, "lr": 0.0002, "elapsed_sec": 30235.70772600174, "step_time_sec": 8.230851004947908, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3638, "loss": 4.250816345214844, "lr": 0.0002, "elapsed_sec": 30243.937878370285, "step_time_sec": 8.229962373967282, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3639, "loss": 4.238544940948486, "lr": 0.0002, "elapsed_sec": 30252.352885246277, "step_time_sec": 8.414855348062702, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3640, "loss": 4.381183624267578, "lr": 0.0002, "elapsed_sec": 30260.894859552383, "step_time_sec": 8.54183263995219, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3641, "loss": 4.493958950042725, "lr": 0.0002, "elapsed_sec": 30269.45170521736, "step_time_sec": 8.556728685973212, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3642, "loss": 4.328390121459961, "lr": 0.0002, "elapsed_sec": 30277.684216976166, "step_time_sec": 8.232308592065237, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3643, "loss": 4.391475200653076, "lr": 0.0002, "elapsed_sec": 30285.91568493843, "step_time_sec": 8.231338648009114, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3644, "loss": 4.539215087890625, "lr": 0.0002, "elapsed_sec": 30294.310519218445, "step_time_sec": 8.394694710965268, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3645, "loss": 4.165042400360107, "lr": 0.0002, "elapsed_sec": 30302.703637838364, "step_time_sec": 8.392910355934873, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3646, "loss": 4.386484622955322, "lr": 0.0002, "elapsed_sec": 30311.20184803009, "step_time_sec": 8.498072394053452, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3647, "loss": 4.368122577667236, "lr": 0.0002, "elapsed_sec": 30319.435445308685, "step_time_sec": 8.233418088057078, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3648, "loss": 4.388525009155273, "lr": 0.0002, "elapsed_sec": 30327.666874170303, "step_time_sec": 8.231260726926848, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3649, "loss": 4.155283451080322, "lr": 0.0002, "elapsed_sec": 30336.223866462708, "step_time_sec": 8.55685693398118, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3650, "loss": 4.346869468688965, "lr": 0.0002, "elapsed_sec": 30344.492230653763, "step_time_sec": 8.268209154950455, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3651, "loss": 4.40453577041626, "lr": 0.0002, "elapsed_sec": 30353.013013601303, "step_time_sec": 8.52057838102337, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3652, "loss": 4.4388251304626465, "lr": 0.0002, "elapsed_sec": 30361.244856595993, "step_time_sec": 8.231649163062684, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3653, "loss": 4.361173629760742, "lr": 0.0002, "elapsed_sec": 30369.805725574493, "step_time_sec": 8.560719558037817, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3654, "loss": 4.290287017822266, "lr": 0.0002, "elapsed_sec": 30378.25482773781, "step_time_sec": 8.448969784076326, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3655, "loss": 4.308467864990234, "lr": 0.0002, "elapsed_sec": 30386.54475426674, "step_time_sec": 8.289768873015419, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3656, "loss": 4.289902210235596, "lr": 0.0002, "elapsed_sec": 30395.10166144371, "step_time_sec": 8.556693925987929, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3657, "loss": 4.40156364440918, "lr": 0.0002, "elapsed_sec": 30403.333831071854, "step_time_sec": 8.23201021400746, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3658, "loss": 4.385124206542969, "lr": 0.0002, "elapsed_sec": 30411.893782377243, "step_time_sec": 8.559800110990182, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3659, "loss": 4.406643867492676, "lr": 0.0002, "elapsed_sec": 30420.16707968712, "step_time_sec": 8.273188516963273, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3660, "loss": 4.3591413497924805, "lr": 0.0002, "elapsed_sec": 30428.65946674347, "step_time_sec": 8.492171859019436, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3661, "loss": 4.353904724121094, "lr": 0.0002, "elapsed_sec": 30437.21945500374, "step_time_sec": 8.559799686074257, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3662, "loss": 4.45654821395874, "lr": 0.0002, "elapsed_sec": 30445.485231876373, "step_time_sec": 8.265718189068139, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3663, "loss": 4.400012493133545, "lr": 0.0002, "elapsed_sec": 30454.01262140274, "step_time_sec": 8.527152619091794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3664, "loss": 4.368051528930664, "lr": 0.0002, "elapsed_sec": 30462.24309706688, "step_time_sec": 8.230257421033457, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3665, "loss": 4.475909233093262, "lr": 0.0002, "elapsed_sec": 30470.473447322845, "step_time_sec": 8.230208470020443, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3666, "loss": 4.384555339813232, "lr": 0.0002, "elapsed_sec": 30478.704489707947, "step_time_sec": 8.230880726943724, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3667, "loss": 4.324977874755859, "lr": 0.0002, "elapsed_sec": 30486.93511247635, "step_time_sec": 8.230400072061457, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3668, "loss": 4.336788177490234, "lr": 0.0002, "elapsed_sec": 30495.1657204628, "step_time_sec": 8.230442365980707, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3669, "loss": 4.4179277420043945, "lr": 0.0002, "elapsed_sec": 30503.396883249283, "step_time_sec": 8.231011848081835, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3670, "loss": 4.6216254234313965, "lr": 0.0002, "elapsed_sec": 30511.627645254135, "step_time_sec": 8.230630228063092, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3671, "loss": 4.370880603790283, "lr": 0.0002, "elapsed_sec": 30519.858492851257, "step_time_sec": 8.230687981005758, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3672, "loss": 4.410046100616455, "lr": 0.0002, "elapsed_sec": 30528.08824443817, "step_time_sec": 8.22953727596905, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3673, "loss": 4.362456798553467, "lr": 0.0002, "elapsed_sec": 30536.318860292435, "step_time_sec": 8.230524934013374, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3674, "loss": 4.459511756896973, "lr": 0.0002, "elapsed_sec": 30544.55176091194, "step_time_sec": 8.232719274936244, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3675, "loss": 4.342826843261719, "lr": 0.0002, "elapsed_sec": 30552.783475875854, "step_time_sec": 8.231508392025717, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3676, "loss": 4.341835021972656, "lr": 0.0002, "elapsed_sec": 30561.016808986664, "step_time_sec": 8.233180096023716, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3677, "loss": 4.252288341522217, "lr": 0.0002, "elapsed_sec": 30569.248682022095, "step_time_sec": 8.231679418007843, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3678, "loss": 4.273694038391113, "lr": 0.0002, "elapsed_sec": 30577.481626987457, "step_time_sec": 8.232825796003453, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3679, "loss": 4.317773818969727, "lr": 0.0002, "elapsed_sec": 30585.71305513382, "step_time_sec": 8.23127820994705, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3680, "loss": 4.2671613693237305, "lr": 0.0002, "elapsed_sec": 30593.94622850418, "step_time_sec": 8.232958865934052, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3681, "loss": 4.290622711181641, "lr": 0.0002, "elapsed_sec": 30602.17813706398, "step_time_sec": 8.231746481964365, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3682, "loss": 4.319314956665039, "lr": 0.0002, "elapsed_sec": 30610.41091775894, "step_time_sec": 8.232608296093531, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3683, "loss": 4.359030246734619, "lr": 0.0002, "elapsed_sec": 30618.64356827736, "step_time_sec": 8.23254590400029, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3684, "loss": 4.335131645202637, "lr": 0.0002, "elapsed_sec": 30626.877325057983, "step_time_sec": 8.233589594019577, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3685, "loss": 4.359796047210693, "lr": 0.0002, "elapsed_sec": 30635.11022877693, "step_time_sec": 8.232729854993522, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3686, "loss": 4.21575403213501, "lr": 0.0002, "elapsed_sec": 30643.344045639038, "step_time_sec": 8.233688072068617, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3687, "loss": 4.192582607269287, "lr": 0.0002, "elapsed_sec": 30651.577874422073, "step_time_sec": 8.233669608016498, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3688, "loss": 4.39650821685791, "lr": 0.0002, "elapsed_sec": 30659.810546398163, "step_time_sec": 8.232547716004774, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3689, "loss": 4.280862808227539, "lr": 0.0002, "elapsed_sec": 30668.042508125305, "step_time_sec": 8.231773474020883, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3690, "loss": 4.327223300933838, "lr": 0.0002, "elapsed_sec": 30676.27548623085, "step_time_sec": 8.232767495093867, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3691, "loss": 4.452199935913086, "lr": 0.0002, "elapsed_sec": 30684.507855653763, "step_time_sec": 8.232202248997055, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3692, "loss": 4.211335182189941, "lr": 0.0002, "elapsed_sec": 30692.738654851913, "step_time_sec": 8.230599884991534, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3693, "loss": 4.355599880218506, "lr": 0.0002, "elapsed_sec": 30700.970583438873, "step_time_sec": 8.231807967997156, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3694, "loss": 4.222733974456787, "lr": 0.0002, "elapsed_sec": 30709.200700998306, "step_time_sec": 8.22996519098524, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3695, "loss": 4.4838786125183105, "lr": 0.0002, "elapsed_sec": 30717.432574033737, "step_time_sec": 8.231689242995344, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3696, "loss": 4.307257652282715, "lr": 0.0002, "elapsed_sec": 30725.6641497612, "step_time_sec": 8.231418229988776, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3697, "loss": 4.3344879150390625, "lr": 0.0002, "elapsed_sec": 30733.896201848984, "step_time_sec": 8.231920119025744, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3698, "loss": 4.317476272583008, "lr": 0.0002, "elapsed_sec": 30742.127289772034, "step_time_sec": 8.230901912902482, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3699, "loss": 4.341614723205566, "lr": 0.0002, "elapsed_sec": 30750.357853412628, "step_time_sec": 8.23046076099854, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3700, "loss": 4.3863525390625, "lr": 0.0002, "elapsed_sec": 30758.588705301285, "step_time_sec": 8.230615515029058, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3701, "loss": 4.395769119262695, "lr": 0.0002, "elapsed_sec": 30766.819834709167, "step_time_sec": 8.231011210940778, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3702, "loss": 4.35446834564209, "lr": 0.0002, "elapsed_sec": 30775.05276465416, "step_time_sec": 8.2327287979424, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3703, "loss": 4.348256587982178, "lr": 0.0002, "elapsed_sec": 30783.284079551697, "step_time_sec": 8.231185768963769, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3704, "loss": 4.231194496154785, "lr": 0.0002, "elapsed_sec": 30791.515986680984, "step_time_sec": 8.23171059798915, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3705, "loss": 4.3840789794921875, "lr": 0.0002, "elapsed_sec": 30799.747358322144, "step_time_sec": 8.231247441959567, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3706, "loss": 4.344690322875977, "lr": 0.0002, "elapsed_sec": 30807.979483366013, "step_time_sec": 8.231963778031059, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3707, "loss": 4.30032205581665, "lr": 0.0002, "elapsed_sec": 30816.211315870285, "step_time_sec": 8.23169610206969, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3708, "loss": 4.3107194900512695, "lr": 0.0002, "elapsed_sec": 30824.442356348038, "step_time_sec": 8.230826286016963, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3709, "loss": 4.434398174285889, "lr": 0.0002, "elapsed_sec": 30832.674011707306, "step_time_sec": 8.23150083294604, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3710, "loss": 4.349682807922363, "lr": 0.0002, "elapsed_sec": 30840.905663728714, "step_time_sec": 8.231485264957882, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3711, "loss": 4.282641410827637, "lr": 0.0002, "elapsed_sec": 30849.13691329956, "step_time_sec": 8.231068920926191, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3712, "loss": 4.238544940948486, "lr": 0.0002, "elapsed_sec": 30857.36797451973, "step_time_sec": 8.230939673958346, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3713, "loss": 4.356337547302246, "lr": 0.0002, "elapsed_sec": 30865.599499702454, "step_time_sec": 8.231302311993204, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3714, "loss": 4.33939266204834, "lr": 0.0002, "elapsed_sec": 30873.830296754837, "step_time_sec": 8.230620043003, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3715, "loss": 4.409566879272461, "lr": 0.0002, "elapsed_sec": 30882.06172990799, "step_time_sec": 8.231302305939607, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3716, "loss": 4.166172981262207, "lr": 0.0002, "elapsed_sec": 30890.292221307755, "step_time_sec": 8.230332886916585, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3717, "loss": 4.396103382110596, "lr": 0.0002, "elapsed_sec": 30898.522885084152, "step_time_sec": 8.230520133045502, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3718, "loss": 4.393064975738525, "lr": 0.0002, "elapsed_sec": 30906.753225564957, "step_time_sec": 8.23022776201833, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3719, "loss": 4.289558410644531, "lr": 0.0002, "elapsed_sec": 30914.98370742798, "step_time_sec": 8.230290204985067, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3720, "loss": 4.3877973556518555, "lr": 0.0002, "elapsed_sec": 30923.21520924568, "step_time_sec": 8.231346862041391, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3721, "loss": 4.296817302703857, "lr": 0.0002, "elapsed_sec": 30931.44521331787, "step_time_sec": 8.229857133002952, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3722, "loss": 4.547384262084961, "lr": 0.0002, "elapsed_sec": 30939.67696785927, "step_time_sec": 8.23159228998702, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3723, "loss": 4.38734245300293, "lr": 0.0002, "elapsed_sec": 30947.908473968506, "step_time_sec": 8.23132391704712, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3724, "loss": 4.33601188659668, "lr": 0.0002, "elapsed_sec": 30956.140733718872, "step_time_sec": 8.232078055036254, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3725, "loss": 4.497491836547852, "lr": 0.0002, "elapsed_sec": 30964.373461008072, "step_time_sec": 8.232556091039442, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3726, "loss": 4.275764465332031, "lr": 0.0002, "elapsed_sec": 30972.606336593628, "step_time_sec": 8.232775641954504, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3727, "loss": 4.397895812988281, "lr": 0.0002, "elapsed_sec": 30980.839005231857, "step_time_sec": 8.23246377601754, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3728, "loss": 4.412827968597412, "lr": 0.0002, "elapsed_sec": 30989.07063150406, "step_time_sec": 8.231398431002162, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3729, "loss": 4.44332218170166, "lr": 0.0002, "elapsed_sec": 30997.30358839035, "step_time_sec": 8.232855293084867, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3730, "loss": 4.306421756744385, "lr": 0.0002, "elapsed_sec": 31005.535624742508, "step_time_sec": 8.231865230947733, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3731, "loss": 4.310703754425049, "lr": 0.0002, "elapsed_sec": 31013.768505096436, "step_time_sec": 8.232695971964858, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3732, "loss": 4.199986934661865, "lr": 0.0002, "elapsed_sec": 31021.998612880707, "step_time_sec": 8.229942072066478, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3733, "loss": 4.345512866973877, "lr": 0.0002, "elapsed_sec": 31030.230429410934, "step_time_sec": 8.231670748908073, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3734, "loss": 4.4190521240234375, "lr": 0.0002, "elapsed_sec": 31038.463456869125, "step_time_sec": 8.232907374040224, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3735, "loss": 4.2076334953308105, "lr": 0.0002, "elapsed_sec": 31046.69596529007, "step_time_sec": 8.232332665007561, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3736, "loss": 4.395307540893555, "lr": 0.0002, "elapsed_sec": 31055.000438928604, "step_time_sec": 8.24205188988708, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3737, "loss": 4.303370475769043, "lr": 0.0002, "elapsed_sec": 31063.232050180435, "step_time_sec": 8.231449608923867, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3738, "loss": 4.248187065124512, "lr": 0.0002, "elapsed_sec": 31071.463908195496, "step_time_sec": 8.231700242962688, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3739, "loss": 4.384119510650635, "lr": 0.0002, "elapsed_sec": 31079.694539308548, "step_time_sec": 8.230487086926587, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3740, "loss": 4.374110698699951, "lr": 0.0002, "elapsed_sec": 31087.927075386047, "step_time_sec": 8.232364811003208, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3741, "loss": 4.200296401977539, "lr": 0.0002, "elapsed_sec": 31096.15940594673, "step_time_sec": 8.232158552040346, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3742, "loss": 4.319217681884766, "lr": 0.0002, "elapsed_sec": 31104.392842531204, "step_time_sec": 8.233280522981659, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3743, "loss": 4.301770210266113, "lr": 0.0002, "elapsed_sec": 31112.625786066055, "step_time_sec": 8.23276258003898, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3744, "loss": 4.301020622253418, "lr": 0.0002, "elapsed_sec": 31120.856251955032, "step_time_sec": 8.230337037937716, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3745, "loss": 4.261225700378418, "lr": 0.0002, "elapsed_sec": 31129.088579416275, "step_time_sec": 8.232172321062535, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3746, "loss": 4.299419403076172, "lr": 0.0002, "elapsed_sec": 31137.319882392883, "step_time_sec": 8.231113433022983, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3747, "loss": 4.382871150970459, "lr": 0.0002, "elapsed_sec": 31145.551505565643, "step_time_sec": 8.231478257104754, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3748, "loss": 4.235910415649414, "lr": 0.0002, "elapsed_sec": 31153.781566143036, "step_time_sec": 8.229928120039403, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3749, "loss": 4.239006519317627, "lr": 0.0002, "elapsed_sec": 31162.01330924034, "step_time_sec": 8.231540139997378, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3750, "loss": 4.3919477462768555, "lr": 0.0002, "elapsed_sec": 31170.243840932846, "step_time_sec": 8.230382511974312, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3751, "loss": 4.272585391998291, "lr": 0.0002, "elapsed_sec": 31178.47398352623, "step_time_sec": 8.229971694992855, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3752, "loss": 4.379637241363525, "lr": 0.0002, "elapsed_sec": 31186.70520710945, "step_time_sec": 8.231083435937762, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3753, "loss": 4.323855400085449, "lr": 0.0002, "elapsed_sec": 31194.934344291687, "step_time_sec": 8.228957457002252, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3754, "loss": 4.284846782684326, "lr": 0.0002, "elapsed_sec": 31203.167857646942, "step_time_sec": 8.233321039006114, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3755, "loss": 4.317812442779541, "lr": 0.0002, "elapsed_sec": 31211.399975061417, "step_time_sec": 8.23201121808961, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3756, "loss": 4.37839412689209, "lr": 0.0002, "elapsed_sec": 31219.633654117584, "step_time_sec": 8.233489531092346, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3757, "loss": 4.337089538574219, "lr": 0.0002, "elapsed_sec": 31227.86424970627, "step_time_sec": 8.230418098974042, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3758, "loss": 4.3045783042907715, "lr": 0.0002, "elapsed_sec": 31236.09463763237, "step_time_sec": 8.230232077999972, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3759, "loss": 4.341947078704834, "lr": 0.0002, "elapsed_sec": 31244.32639694214, "step_time_sec": 8.231611385010183, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3760, "loss": 4.3628997802734375, "lr": 0.0002, "elapsed_sec": 31252.557671785355, "step_time_sec": 8.231129797990434, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3761, "loss": 4.184247016906738, "lr": 0.0002, "elapsed_sec": 31260.788358449936, "step_time_sec": 8.230559255927801, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3762, "loss": 4.149169921875, "lr": 0.0002, "elapsed_sec": 31269.02049612999, "step_time_sec": 8.231982211000286, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3763, "loss": 4.324045658111572, "lr": 0.0002, "elapsed_sec": 31277.25346636772, "step_time_sec": 8.232766986009665, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3764, "loss": 4.5251641273498535, "lr": 0.0002, "elapsed_sec": 31285.48275923729, "step_time_sec": 8.229156689019874, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3765, "loss": 4.304664134979248, "lr": 0.0002, "elapsed_sec": 31293.71458363533, "step_time_sec": 8.231639373931102, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3766, "loss": 4.357307434082031, "lr": 0.0002, "elapsed_sec": 31301.945912837982, "step_time_sec": 8.231128984945826, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3767, "loss": 4.349873065948486, "lr": 0.0002, "elapsed_sec": 31310.17827272415, "step_time_sec": 8.232251013047062, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3768, "loss": 4.3110575675964355, "lr": 0.0002, "elapsed_sec": 31318.41102361679, "step_time_sec": 8.23257030104287, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3769, "loss": 4.311573505401611, "lr": 0.0002, "elapsed_sec": 31326.643297433853, "step_time_sec": 8.232116188039072, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3770, "loss": 4.327720642089844, "lr": 0.0002, "elapsed_sec": 31334.87693786621, "step_time_sec": 8.233470592997037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3771, "loss": 4.420047283172607, "lr": 0.0002, "elapsed_sec": 31343.108276844025, "step_time_sec": 8.231147860991769, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3772, "loss": 4.274677276611328, "lr": 0.0002, "elapsed_sec": 31351.340764522552, "step_time_sec": 8.2323669239413, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3773, "loss": 4.219516754150391, "lr": 0.0002, "elapsed_sec": 31359.571089029312, "step_time_sec": 8.230123254936188, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3774, "loss": 4.17676305770874, "lr": 0.0002, "elapsed_sec": 31367.801309108734, "step_time_sec": 8.230062606045976, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3775, "loss": 4.2958984375, "lr": 0.0002, "elapsed_sec": 31376.03243613243, "step_time_sec": 8.230964655056596, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3776, "loss": 4.362560272216797, "lr": 0.0002, "elapsed_sec": 31384.262571573257, "step_time_sec": 8.22998891491443, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3777, "loss": 4.293008804321289, "lr": 0.0002, "elapsed_sec": 31392.492371082306, "step_time_sec": 8.229624278028496, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3778, "loss": 4.391413688659668, "lr": 0.0002, "elapsed_sec": 31400.72400212288, "step_time_sec": 8.231476229964755, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3779, "loss": 4.260414123535156, "lr": 0.0002, "elapsed_sec": 31408.956698179245, "step_time_sec": 8.232523052021861, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3780, "loss": 4.43612813949585, "lr": 0.0002, "elapsed_sec": 31417.188783168793, "step_time_sec": 8.231986874947324, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3781, "loss": 4.563995838165283, "lr": 0.0002, "elapsed_sec": 31425.420348882675, "step_time_sec": 8.23137326806318, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3782, "loss": 4.35543155670166, "lr": 0.0002, "elapsed_sec": 31433.651205778122, "step_time_sec": 8.230683682020754, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3783, "loss": 4.363354206085205, "lr": 0.0002, "elapsed_sec": 31441.883669614792, "step_time_sec": 8.232338704052381, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3784, "loss": 4.236871242523193, "lr": 0.0002, "elapsed_sec": 31450.11480641365, "step_time_sec": 8.230973006109707, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3785, "loss": 4.229630947113037, "lr": 0.0002, "elapsed_sec": 31458.347311258316, "step_time_sec": 8.232305868063122, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3786, "loss": 4.5061869621276855, "lr": 0.0002, "elapsed_sec": 31466.5790643692, "step_time_sec": 8.231640588026494, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3787, "loss": 4.3035807609558105, "lr": 0.0002, "elapsed_sec": 31474.810950040817, "step_time_sec": 8.231671622954309, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3788, "loss": 4.287727355957031, "lr": 0.0002, "elapsed_sec": 31483.044063806534, "step_time_sec": 8.23292850307189, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3789, "loss": 4.208687782287598, "lr": 0.0002, "elapsed_sec": 31491.27635574341, "step_time_sec": 8.232150199939497, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3790, "loss": 4.280169486999512, "lr": 0.0002, "elapsed_sec": 31499.50833415985, "step_time_sec": 8.231786031043157, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3791, "loss": 4.2262115478515625, "lr": 0.0002, "elapsed_sec": 31507.740245580673, "step_time_sec": 8.231807658914477, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3792, "loss": 4.466879367828369, "lr": 0.0002, "elapsed_sec": 31515.97318148613, "step_time_sec": 8.23274303006474, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3793, "loss": 4.260326385498047, "lr": 0.0002, "elapsed_sec": 31524.20550107956, "step_time_sec": 8.23214810306672, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3794, "loss": 4.420652866363525, "lr": 0.0002, "elapsed_sec": 31532.43843150139, "step_time_sec": 8.23275228694547, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3795, "loss": 4.193563938140869, "lr": 0.0002, "elapsed_sec": 31540.66901254654, "step_time_sec": 8.230442145955749, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3796, "loss": 4.285314083099365, "lr": 0.0002, "elapsed_sec": 31548.899547100067, "step_time_sec": 8.230400032945909, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3797, "loss": 4.188035011291504, "lr": 0.0002, "elapsed_sec": 31557.12895131111, "step_time_sec": 8.229190076002851, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3798, "loss": 4.378669738769531, "lr": 0.0002, "elapsed_sec": 31565.35848736763, "step_time_sec": 8.22943408798892, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3799, "loss": 4.2317280769348145, "lr": 0.0002, "elapsed_sec": 31573.58968281746, "step_time_sec": 8.231030564988032, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3800, "loss": 4.2723565101623535, "lr": 0.0002, "elapsed_sec": 31581.822063207626, "step_time_sec": 8.2321980149718, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3801, "loss": 4.380197048187256, "lr": 0.0002, "elapsed_sec": 31590.054775953293, "step_time_sec": 8.232528618071228, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3802, "loss": 4.310886859893799, "lr": 0.0002, "elapsed_sec": 31598.286388397217, "step_time_sec": 8.231446153018624, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3803, "loss": 4.23431921005249, "lr": 0.0002, "elapsed_sec": 31606.519212961197, "step_time_sec": 8.23270879802294, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3804, "loss": 4.326918601989746, "lr": 0.0002, "elapsed_sec": 31614.750258922577, "step_time_sec": 8.230856147943996, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3805, "loss": 4.358035564422607, "lr": 0.0002, "elapsed_sec": 31622.981719732285, "step_time_sec": 8.231298241997138, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3806, "loss": 4.220424175262451, "lr": 0.0002, "elapsed_sec": 31631.213653087616, "step_time_sec": 8.231758694979362, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3807, "loss": 4.348308086395264, "lr": 0.0002, "elapsed_sec": 31639.44740319252, "step_time_sec": 8.233579072984867, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3808, "loss": 4.364212989807129, "lr": 0.0002, "elapsed_sec": 31647.678215503693, "step_time_sec": 8.230671224067919, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3809, "loss": 4.458616256713867, "lr": 0.0002, "elapsed_sec": 31655.909351825714, "step_time_sec": 8.23099019494839, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3810, "loss": 4.3452887535095215, "lr": 0.0002, "elapsed_sec": 31664.140284061432, "step_time_sec": 8.230728573049419, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3811, "loss": 4.297701358795166, "lr": 0.0002, "elapsed_sec": 31672.37227654457, "step_time_sec": 8.23186848894693, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3812, "loss": 4.335720062255859, "lr": 0.0002, "elapsed_sec": 31680.60423564911, "step_time_sec": 8.231777766020969, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3813, "loss": 4.262918472290039, "lr": 0.0002, "elapsed_sec": 31688.837170362473, "step_time_sec": 8.232807049062103, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3814, "loss": 4.322876453399658, "lr": 0.0002, "elapsed_sec": 31697.070553302765, "step_time_sec": 8.233254128019325, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3815, "loss": 4.173427104949951, "lr": 0.0002, "elapsed_sec": 31705.301884889603, "step_time_sec": 8.231164555996656, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3816, "loss": 4.283087253570557, "lr": 0.0002, "elapsed_sec": 31713.534152507782, "step_time_sec": 8.232072563027032, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3817, "loss": 4.378073215484619, "lr": 0.0002, "elapsed_sec": 31721.76541352272, "step_time_sec": 8.231130040017888, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3818, "loss": 4.301743030548096, "lr": 0.0002, "elapsed_sec": 31729.994850873947, "step_time_sec": 8.229262573062442, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3819, "loss": 4.161161422729492, "lr": 0.0002, "elapsed_sec": 31738.227952480316, "step_time_sec": 8.232950514997356, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3820, "loss": 4.149143695831299, "lr": 0.0002, "elapsed_sec": 31746.459849834442, "step_time_sec": 8.231776539003477, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3821, "loss": 4.229341506958008, "lr": 0.0002, "elapsed_sec": 31754.693077087402, "step_time_sec": 8.23302342405077, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3822, "loss": 4.319754600524902, "lr": 0.0002, "elapsed_sec": 31762.923314094543, "step_time_sec": 8.23008413205389, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3823, "loss": 4.184565544128418, "lr": 0.0002, "elapsed_sec": 31771.155879974365, "step_time_sec": 8.23242608399596, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3824, "loss": 4.344513893127441, "lr": 0.0002, "elapsed_sec": 31779.386471271515, "step_time_sec": 8.23041237401776, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3825, "loss": 4.228661060333252, "lr": 0.0002, "elapsed_sec": 31787.61903500557, "step_time_sec": 8.232402435038239, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3826, "loss": 4.32811975479126, "lr": 0.0002, "elapsed_sec": 31795.849916934967, "step_time_sec": 8.230740407016128, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3827, "loss": 4.369149684906006, "lr": 0.0002, "elapsed_sec": 31804.081669807434, "step_time_sec": 8.231587070971727, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3828, "loss": 4.1750969886779785, "lr": 0.0002, "elapsed_sec": 31812.313224315643, "step_time_sec": 8.231366937980056, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3829, "loss": 4.386231899261475, "lr": 0.0002, "elapsed_sec": 31820.54508304596, "step_time_sec": 8.231780436006375, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3830, "loss": 4.229065895080566, "lr": 0.0002, "elapsed_sec": 31828.777650356293, "step_time_sec": 8.232338231988251, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3831, "loss": 4.284616947174072, "lr": 0.0002, "elapsed_sec": 31837.00905418396, "step_time_sec": 8.231260676984675, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3832, "loss": 4.295931816101074, "lr": 0.0002, "elapsed_sec": 31845.24153470993, "step_time_sec": 8.232311649015173, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3833, "loss": 4.463324546813965, "lr": 0.0002, "elapsed_sec": 31853.47366666794, "step_time_sec": 8.231971159926616, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3834, "loss": 4.306450366973877, "lr": 0.0002, "elapsed_sec": 31861.706208467484, "step_time_sec": 8.232390835997649, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3835, "loss": 4.285749912261963, "lr": 0.0002, "elapsed_sec": 31869.938470840454, "step_time_sec": 8.2320865560323, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3836, "loss": 4.180942535400391, "lr": 0.0002, "elapsed_sec": 31878.171535491943, "step_time_sec": 8.23288731102366, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3837, "loss": 4.3898725509643555, "lr": 0.0002, "elapsed_sec": 31886.403688669205, "step_time_sec": 8.232039193972014, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3838, "loss": 4.295004367828369, "lr": 0.0002, "elapsed_sec": 31894.63759613037, "step_time_sec": 8.233676000032574, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3839, "loss": 4.210633754730225, "lr": 0.0002, "elapsed_sec": 31902.868300914764, "step_time_sec": 8.230598205002025, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3840, "loss": 4.2374114990234375, "lr": 0.0002, "elapsed_sec": 31911.101642131805, "step_time_sec": 8.23314570391085, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3841, "loss": 4.134252548217773, "lr": 0.0002, "elapsed_sec": 31919.331695079803, "step_time_sec": 8.229907053057104, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3842, "loss": 4.338742733001709, "lr": 0.0002, "elapsed_sec": 31927.56433033943, "step_time_sec": 8.23244747100398, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3843, "loss": 4.370997905731201, "lr": 0.0002, "elapsed_sec": 31935.79633951187, "step_time_sec": 8.23190370190423, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3844, "loss": 4.212208271026611, "lr": 0.0002, "elapsed_sec": 31944.027268648148, "step_time_sec": 8.230781776015647, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3845, "loss": 4.260286331176758, "lr": 0.0002, "elapsed_sec": 31952.262459754944, "step_time_sec": 8.235006658011116, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3846, "loss": 4.198212623596191, "lr": 0.0002, "elapsed_sec": 31960.49399447441, "step_time_sec": 8.231397455907427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3847, "loss": 4.2113142013549805, "lr": 0.0002, "elapsed_sec": 31968.72709941864, "step_time_sec": 8.232924112002365, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3848, "loss": 4.317789554595947, "lr": 0.0002, "elapsed_sec": 31976.95826482773, "step_time_sec": 8.231009103008546, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3849, "loss": 3.978966474533081, "lr": 0.0002, "elapsed_sec": 31985.19087934494, "step_time_sec": 8.232421836000867, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3850, "loss": 4.407773971557617, "lr": 0.0002, "elapsed_sec": 31993.422513961792, "step_time_sec": 8.231476042070426, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3851, "loss": 4.263838291168213, "lr": 0.0002, "elapsed_sec": 32001.65629196167, "step_time_sec": 8.233613969990984, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3852, "loss": 4.324657917022705, "lr": 0.0002, "elapsed_sec": 32009.888496398926, "step_time_sec": 8.232031172025017, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3853, "loss": 4.2346577644348145, "lr": 0.0002, "elapsed_sec": 32018.121153831482, "step_time_sec": 8.232488767011091, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3854, "loss": 4.237102031707764, "lr": 0.0002, "elapsed_sec": 32026.35351061821, "step_time_sec": 8.232275575981475, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3855, "loss": 4.390290260314941, "lr": 0.0002, "elapsed_sec": 32034.58683037758, "step_time_sec": 8.233090488938615, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3856, "loss": 4.275707244873047, "lr": 0.0002, "elapsed_sec": 32042.81914138794, "step_time_sec": 8.23214600502979, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3857, "loss": 4.252288341522217, "lr": 0.0002, "elapsed_sec": 32051.05124926567, "step_time_sec": 8.231969402986579, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3858, "loss": 4.2259697914123535, "lr": 0.0002, "elapsed_sec": 32059.283483743668, "step_time_sec": 8.232098206062801, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3859, "loss": 4.275235176086426, "lr": 0.0002, "elapsed_sec": 32067.514309883118, "step_time_sec": 8.230601832037792, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3860, "loss": 4.3237996101379395, "lr": 0.0002, "elapsed_sec": 32075.74661874771, "step_time_sec": 8.232205886975862, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3861, "loss": 4.239235877990723, "lr": 0.0002, "elapsed_sec": 32083.977961540222, "step_time_sec": 8.231169938924722, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3862, "loss": 4.265195369720459, "lr": 0.0002, "elapsed_sec": 32092.210358142853, "step_time_sec": 8.232238791068085, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3863, "loss": 4.111939907073975, "lr": 0.0002, "elapsed_sec": 32100.44184589386, "step_time_sec": 8.23130658804439, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3864, "loss": 4.226391315460205, "lr": 0.0002, "elapsed_sec": 32108.67487335205, "step_time_sec": 8.232903141062707, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3865, "loss": 4.390605926513672, "lr": 0.0002, "elapsed_sec": 32116.90740442276, "step_time_sec": 8.232314264052548, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3866, "loss": 4.262502193450928, "lr": 0.0002, "elapsed_sec": 32125.140347003937, "step_time_sec": 8.23276283708401, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3867, "loss": 4.206676483154297, "lr": 0.0002, "elapsed_sec": 32133.371628284454, "step_time_sec": 8.231177325942554, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3868, "loss": 4.208867073059082, "lr": 0.0002, "elapsed_sec": 32141.603853225708, "step_time_sec": 8.23201388400048, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3869, "loss": 4.28013277053833, "lr": 0.0002, "elapsed_sec": 32149.836252212524, "step_time_sec": 8.232252275105566, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3870, "loss": 4.286380767822266, "lr": 0.0002, "elapsed_sec": 32158.0678358078, "step_time_sec": 8.231447516009212, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3871, "loss": 4.363669395446777, "lr": 0.0002, "elapsed_sec": 32166.298096895218, "step_time_sec": 8.230142735992558, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3872, "loss": 4.266783714294434, "lr": 0.0002, "elapsed_sec": 32174.530901670456, "step_time_sec": 8.232594265020452, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3873, "loss": 4.320171356201172, "lr": 0.0002, "elapsed_sec": 32182.763315677643, "step_time_sec": 8.232254207017832, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3874, "loss": 4.220532417297363, "lr": 0.0002, "elapsed_sec": 32190.995010614395, "step_time_sec": 8.231556883081794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3875, "loss": 4.360326766967773, "lr": 0.0002, "elapsed_sec": 32199.227605819702, "step_time_sec": 8.232449078001082, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3876, "loss": 4.059722423553467, "lr": 0.0002, "elapsed_sec": 32207.459993124008, "step_time_sec": 8.232182865031064, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3877, "loss": 4.090085983276367, "lr": 0.0002, "elapsed_sec": 32215.692129850388, "step_time_sec": 8.231989669031464, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3878, "loss": 4.287569999694824, "lr": 0.0002, "elapsed_sec": 32223.924590349197, "step_time_sec": 8.232266488019377, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3879, "loss": 4.344931125640869, "lr": 0.0002, "elapsed_sec": 32232.15805721283, "step_time_sec": 8.233301503933035, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3880, "loss": 4.194520473480225, "lr": 0.0002, "elapsed_sec": 32240.38793206215, "step_time_sec": 8.229723607073538, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3881, "loss": 4.104952812194824, "lr": 0.0002, "elapsed_sec": 32248.62259197235, "step_time_sec": 8.234489505062811, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3882, "loss": 4.2880096435546875, "lr": 0.0002, "elapsed_sec": 32256.856134176254, "step_time_sec": 8.233421289012767, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3883, "loss": 4.22550106048584, "lr": 0.0002, "elapsed_sec": 32265.089706897736, "step_time_sec": 8.233369053923525, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3884, "loss": 4.3607306480407715, "lr": 0.0002, "elapsed_sec": 32273.322725772858, "step_time_sec": 8.232883500983007, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3885, "loss": 4.28170108795166, "lr": 0.0002, "elapsed_sec": 32281.55600643158, "step_time_sec": 8.233098800992593, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3886, "loss": 4.31209135055542, "lr": 0.0002, "elapsed_sec": 32289.788474559784, "step_time_sec": 8.232336058048531, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3887, "loss": 4.136880397796631, "lr": 0.0002, "elapsed_sec": 32298.020088672638, "step_time_sec": 8.231401206925511, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3888, "loss": 4.253072738647461, "lr": 0.0002, "elapsed_sec": 32306.253101348877, "step_time_sec": 8.232909236103296, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3889, "loss": 4.20249080657959, "lr": 0.0002, "elapsed_sec": 32314.485248327255, "step_time_sec": 8.231931793037802, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3890, "loss": 4.29530143737793, "lr": 0.0002, "elapsed_sec": 32322.71731042862, "step_time_sec": 8.231884915963747, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3891, "loss": 4.290690898895264, "lr": 0.0002, "elapsed_sec": 32330.949414730072, "step_time_sec": 8.231977034010924, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3892, "loss": 4.302656173706055, "lr": 0.0002, "elapsed_sec": 32339.182849407196, "step_time_sec": 8.233231832971796, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3893, "loss": 4.277725696563721, "lr": 0.0002, "elapsed_sec": 32347.41453909874, "step_time_sec": 8.231530562974513, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3894, "loss": 4.2945966720581055, "lr": 0.0002, "elapsed_sec": 32355.648716449738, "step_time_sec": 8.234056122018956, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3895, "loss": 4.27380895614624, "lr": 0.0002, "elapsed_sec": 32363.880113363266, "step_time_sec": 8.23122851899825, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3896, "loss": 4.206611633300781, "lr": 0.0002, "elapsed_sec": 32372.112567186356, "step_time_sec": 8.232259753975086, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3897, "loss": 4.387523174285889, "lr": 0.0002, "elapsed_sec": 32380.34472632408, "step_time_sec": 8.231967616011389, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3898, "loss": 4.342737197875977, "lr": 0.0002, "elapsed_sec": 32388.57700920105, "step_time_sec": 8.23213164799381, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3899, "loss": 4.285547256469727, "lr": 0.0002, "elapsed_sec": 32396.81006884575, "step_time_sec": 8.232930884929374, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3900, "loss": 4.2902302742004395, "lr": 0.0002, "elapsed_sec": 32405.044780254364, "step_time_sec": 8.234527237014845, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3901, "loss": 4.226855754852295, "lr": 0.0002, "elapsed_sec": 32413.28085589409, "step_time_sec": 8.235895988997072, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3902, "loss": 4.2887115478515625, "lr": 0.0002, "elapsed_sec": 32421.5164642334, "step_time_sec": 8.235422392026521, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3903, "loss": 4.4617109298706055, "lr": 0.0002, "elapsed_sec": 32429.747950553894, "step_time_sec": 8.23131879302673, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3904, "loss": 4.319272994995117, "lr": 0.0002, "elapsed_sec": 32437.97926425934, "step_time_sec": 8.231171363033354, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3905, "loss": 4.191202640533447, "lr": 0.0002, "elapsed_sec": 32446.21156001091, "step_time_sec": 8.232208020985126, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3906, "loss": 4.279592514038086, "lr": 0.0002, "elapsed_sec": 32454.443495988846, "step_time_sec": 8.231683396967128, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3907, "loss": 4.177465915679932, "lr": 0.0002, "elapsed_sec": 32462.67550110817, "step_time_sec": 8.231857649050653, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3908, "loss": 4.129162788391113, "lr": 0.0002, "elapsed_sec": 32470.907706022263, "step_time_sec": 8.232052172068506, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3909, "loss": 4.102784633636475, "lr": 0.0002, "elapsed_sec": 32479.141429424286, "step_time_sec": 8.23357053392101, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3910, "loss": 4.357762336730957, "lr": 0.0002, "elapsed_sec": 32487.37683081627, "step_time_sec": 8.235205137985758, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3911, "loss": 4.268552303314209, "lr": 0.0002, "elapsed_sec": 32495.613372802734, "step_time_sec": 8.23638961208053, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3912, "loss": 4.168773651123047, "lr": 0.0002, "elapsed_sec": 32503.843994617462, "step_time_sec": 8.230494908988476, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3913, "loss": 4.174071788787842, "lr": 0.0002, "elapsed_sec": 32512.07603788376, "step_time_sec": 8.231839732965454, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3914, "loss": 4.272578239440918, "lr": 0.0002, "elapsed_sec": 32520.30754852295, "step_time_sec": 8.231351780006662, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3915, "loss": 4.206847667694092, "lr": 0.0002, "elapsed_sec": 32528.54094338417, "step_time_sec": 8.233221285045147, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3916, "loss": 4.252470970153809, "lr": 0.0002, "elapsed_sec": 32536.77410721779, "step_time_sec": 8.233049905044027, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3917, "loss": 4.345106601715088, "lr": 0.0002, "elapsed_sec": 32545.00786972046, "step_time_sec": 8.233548121061176, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3918, "loss": 4.330766201019287, "lr": 0.0002, "elapsed_sec": 32553.239806175232, "step_time_sec": 8.231769138015807, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3919, "loss": 4.158252716064453, "lr": 0.0002, "elapsed_sec": 32561.47247672081, "step_time_sec": 8.232542640995234, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3920, "loss": 4.34836483001709, "lr": 0.0002, "elapsed_sec": 32569.70484972, "step_time_sec": 8.232194766984321, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3921, "loss": 4.373547554016113, "lr": 0.0002, "elapsed_sec": 32577.937335968018, "step_time_sec": 8.232340593007393, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3922, "loss": 4.2768425941467285, "lr": 0.0002, "elapsed_sec": 32586.17196726799, "step_time_sec": 8.234479432925582, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3923, "loss": 4.163631916046143, "lr": 0.0002, "elapsed_sec": 32594.40475511551, "step_time_sec": 8.232615230954252, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3924, "loss": 4.478577136993408, "lr": 0.0002, "elapsed_sec": 32602.63825559616, "step_time_sec": 8.233340293983929, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3925, "loss": 4.395518779754639, "lr": 0.0002, "elapsed_sec": 32610.871212244034, "step_time_sec": 8.232815183931962, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3926, "loss": 4.445562362670898, "lr": 0.0002, "elapsed_sec": 32619.102778196335, "step_time_sec": 8.231387357111089, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3927, "loss": 4.1209845542907715, "lr": 0.0002, "elapsed_sec": 32627.334659814835, "step_time_sec": 8.231674936949275, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3928, "loss": 4.314448356628418, "lr": 0.0002, "elapsed_sec": 32635.566858530045, "step_time_sec": 8.232055981992744, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3929, "loss": 4.257870674133301, "lr": 0.0002, "elapsed_sec": 32643.79864168167, "step_time_sec": 8.23165308800526, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3930, "loss": 4.203645706176758, "lr": 0.0002, "elapsed_sec": 32652.030382156372, "step_time_sec": 8.23155502404552, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3931, "loss": 4.300198554992676, "lr": 0.0002, "elapsed_sec": 32660.263920545578, "step_time_sec": 8.233390383073129, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3932, "loss": 4.418941497802734, "lr": 0.0002, "elapsed_sec": 32668.49685406685, "step_time_sec": 8.232782191014849, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3933, "loss": 4.329725742340088, "lr": 0.0002, "elapsed_sec": 32676.728753566742, "step_time_sec": 8.231797950109467, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3934, "loss": 4.285494327545166, "lr": 0.0002, "elapsed_sec": 32684.963799476624, "step_time_sec": 8.23485339502804, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3935, "loss": 4.271912574768066, "lr": 0.0002, "elapsed_sec": 32693.196707725525, "step_time_sec": 8.232705774949864, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3936, "loss": 4.222797393798828, "lr": 0.0002, "elapsed_sec": 32701.429998636246, "step_time_sec": 8.233140237047337, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3937, "loss": 4.292397975921631, "lr": 0.0002, "elapsed_sec": 32709.66056251526, "step_time_sec": 8.230426385998726, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3938, "loss": 4.245872497558594, "lr": 0.0002, "elapsed_sec": 32717.897376537323, "step_time_sec": 8.236662173992954, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3939, "loss": 4.299844264984131, "lr": 0.0002, "elapsed_sec": 32726.132318019867, "step_time_sec": 8.234788318048231, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3940, "loss": 4.215028762817383, "lr": 0.0002, "elapsed_sec": 32734.366718053818, "step_time_sec": 8.234207749948837, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3941, "loss": 4.192986011505127, "lr": 0.0002, "elapsed_sec": 32742.6007168293, "step_time_sec": 8.23390135599766, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3942, "loss": 4.254145622253418, "lr": 0.0002, "elapsed_sec": 32750.838063955307, "step_time_sec": 8.237130230059847, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3943, "loss": 4.193084716796875, "lr": 0.0002, "elapsed_sec": 32759.071652173996, "step_time_sec": 8.233413422014564, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3944, "loss": 4.482717037200928, "lr": 0.0002, "elapsed_sec": 32767.30542230606, "step_time_sec": 8.233605067012832, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3945, "loss": 4.263672351837158, "lr": 0.0002, "elapsed_sec": 32775.53927588463, "step_time_sec": 8.23373591399286, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3946, "loss": 4.340557098388672, "lr": 0.0002, "elapsed_sec": 32783.77802538872, "step_time_sec": 8.238592618028633, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3947, "loss": 4.232678413391113, "lr": 0.0002, "elapsed_sec": 32792.011639118195, "step_time_sec": 8.233424209058285, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3948, "loss": 4.288065433502197, "lr": 0.0002, "elapsed_sec": 32800.24408721924, "step_time_sec": 8.232268802006729, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3949, "loss": 4.218039512634277, "lr": 0.0002, "elapsed_sec": 32808.47769713402, "step_time_sec": 8.233447268023156, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3950, "loss": 4.287330627441406, "lr": 0.0002, "elapsed_sec": 32816.714129924774, "step_time_sec": 8.23633260303177, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3951, "loss": 4.4259538650512695, "lr": 0.0002, "elapsed_sec": 32824.94653201103, "step_time_sec": 8.232221549958922, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3952, "loss": 4.4119791984558105, "lr": 0.0002, "elapsed_sec": 32833.1788725853, "step_time_sec": 8.23218100098893, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3953, "loss": 4.182929992675781, "lr": 0.0002, "elapsed_sec": 32841.41369962692, "step_time_sec": 8.234680273919366, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3954, "loss": 4.252352714538574, "lr": 0.0002, "elapsed_sec": 32849.65105342865, "step_time_sec": 8.23715107794851, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3955, "loss": 4.2033610343933105, "lr": 0.0002, "elapsed_sec": 32857.88258457184, "step_time_sec": 8.231408925028518, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3956, "loss": 4.422245502471924, "lr": 0.0002, "elapsed_sec": 32866.12062191963, "step_time_sec": 8.237880570930429, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3957, "loss": 4.241769790649414, "lr": 0.0002, "elapsed_sec": 32874.353703022, "step_time_sec": 8.232894786982797, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3958, "loss": 4.393881320953369, "lr": 0.0002, "elapsed_sec": 32882.59208393097, "step_time_sec": 8.238188570947386, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3959, "loss": 4.257700443267822, "lr": 0.0002, "elapsed_sec": 32890.82754588127, "step_time_sec": 8.235323064029217, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3960, "loss": 4.3419108390808105, "lr": 0.0002, "elapsed_sec": 32899.0634458065, "step_time_sec": 8.235763950971887, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3961, "loss": 4.29240608215332, "lr": 0.0002, "elapsed_sec": 32907.29905080795, "step_time_sec": 8.235431983019225, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3962, "loss": 4.313394546508789, "lr": 0.0002, "elapsed_sec": 32915.533378362656, "step_time_sec": 8.234160221996717, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3963, "loss": 4.254045486450195, "lr": 0.0002, "elapsed_sec": 32923.765964984894, "step_time_sec": 8.232411480974406, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3964, "loss": 4.216768264770508, "lr": 0.0002, "elapsed_sec": 32931.99884271622, "step_time_sec": 8.232685465016402, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3965, "loss": 4.317829608917236, "lr": 0.0002, "elapsed_sec": 32940.2313272953, "step_time_sec": 8.232322215102613, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3966, "loss": 4.262599468231201, "lr": 0.0002, "elapsed_sec": 32948.46367096901, "step_time_sec": 8.232169201946817, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3967, "loss": 4.196369647979736, "lr": 0.0002, "elapsed_sec": 32956.69811940193, "step_time_sec": 8.234333944041282, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3968, "loss": 4.186678409576416, "lr": 0.0002, "elapsed_sec": 32964.93065094948, "step_time_sec": 8.232348336954601, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3969, "loss": 4.271485805511475, "lr": 0.0002, "elapsed_sec": 32973.16305184364, "step_time_sec": 8.23221441800706, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3970, "loss": 4.251561641693115, "lr": 0.0002, "elapsed_sec": 32981.399842739105, "step_time_sec": 8.236641871975735, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3971, "loss": 4.370481491088867, "lr": 0.0002, "elapsed_sec": 32989.63680887222, "step_time_sec": 8.23677769198548, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3972, "loss": 4.307857036590576, "lr": 0.0002, "elapsed_sec": 32997.8697617054, "step_time_sec": 8.232833753922023, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3973, "loss": 4.322100639343262, "lr": 0.0002, "elapsed_sec": 33006.10622167587, "step_time_sec": 8.236291947076097, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3974, "loss": 4.290164947509766, "lr": 0.0002, "elapsed_sec": 33014.34104156494, "step_time_sec": 8.234652662998997, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3975, "loss": 4.177071571350098, "lr": 0.0002, "elapsed_sec": 33022.57889389992, "step_time_sec": 8.237686570035294, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3976, "loss": 4.305195331573486, "lr": 0.0002, "elapsed_sec": 33030.81711244583, "step_time_sec": 8.238081375020556, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3977, "loss": 4.266956329345703, "lr": 0.0002, "elapsed_sec": 33039.04968261719, "step_time_sec": 8.232392739038914, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3978, "loss": 4.321513652801514, "lr": 0.0002, "elapsed_sec": 33047.28581881523, "step_time_sec": 8.235964469029568, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3979, "loss": 4.240926742553711, "lr": 0.0002, "elapsed_sec": 33055.52326273918, "step_time_sec": 8.23733701708261, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3980, "loss": 4.306832313537598, "lr": 0.0002, "elapsed_sec": 33063.7581717968, "step_time_sec": 8.234692253987305, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3981, "loss": 4.402628421783447, "lr": 0.0002, "elapsed_sec": 33071.99085044861, "step_time_sec": 8.232531010988168, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3982, "loss": 4.204723834991455, "lr": 0.0002, "elapsed_sec": 33080.22263836861, "step_time_sec": 8.231642264989205, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3983, "loss": 4.235792636871338, "lr": 0.0002, "elapsed_sec": 33088.45995044708, "step_time_sec": 8.23713872290682, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3984, "loss": 4.1766037940979, "lr": 0.0002, "elapsed_sec": 33096.6922557354, "step_time_sec": 8.232184153050184, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3985, "loss": 4.160264492034912, "lr": 0.0002, "elapsed_sec": 33104.92487168312, "step_time_sec": 8.23239285708405, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3986, "loss": 4.414303302764893, "lr": 0.0002, "elapsed_sec": 33113.15648674965, "step_time_sec": 8.231478799018078, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3987, "loss": 4.292331695556641, "lr": 0.0002, "elapsed_sec": 33121.39519619942, "step_time_sec": 8.238564776955172, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3988, "loss": 4.337551593780518, "lr": 0.0002, "elapsed_sec": 33129.6324300766, "step_time_sec": 8.237027434981428, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3989, "loss": 4.22496223449707, "lr": 0.0002, "elapsed_sec": 33137.87060570717, "step_time_sec": 8.23801947303582, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3990, "loss": 4.315516471862793, "lr": 0.0002, "elapsed_sec": 33146.102177381516, "step_time_sec": 8.231436345959082, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3991, "loss": 4.30466890335083, "lr": 0.0002, "elapsed_sec": 33154.33491611481, "step_time_sec": 8.232596518006176, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3992, "loss": 4.350588321685791, "lr": 0.0002, "elapsed_sec": 33162.56664586067, "step_time_sec": 8.231509319972247, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3993, "loss": 4.050795555114746, "lr": 0.0002, "elapsed_sec": 33170.80417585373, "step_time_sec": 8.237363078980707, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3994, "loss": 4.228816509246826, "lr": 0.0002, "elapsed_sec": 33179.039920568466, "step_time_sec": 8.235590946977027, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3995, "loss": 4.184492111206055, "lr": 0.0002, "elapsed_sec": 33187.27724099159, "step_time_sec": 8.237147774896584, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3996, "loss": 4.28781795501709, "lr": 0.0002, "elapsed_sec": 33195.51476120949, "step_time_sec": 8.237430126988329, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3997, "loss": 4.225176811218262, "lr": 0.0002, "elapsed_sec": 33203.752452135086, "step_time_sec": 8.237513666972518, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3998, "loss": 4.276324272155762, "lr": 0.0002, "elapsed_sec": 33211.990580797195, "step_time_sec": 8.23792461596895, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3999, "loss": 4.487155437469482, "lr": 0.0002, "elapsed_sec": 33220.222331762314, "step_time_sec": 8.231577254016884, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4000, "loss": 4.321393966674805, "lr": 0.0002, "elapsed_sec": 33228.455527067184, "step_time_sec": 56.06848807295319, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": true, "validation_ran": true, "probe_ran": true, "probe_elapsed_sec": 0.9815248120576143, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4001, "loss": 4.3377532958984375, "lr": 0.0002, "elapsed_sec": 33284.5358338356, "step_time_sec": 8.244786225026473, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4002, "loss": 4.21795654296875, "lr": 0.0002, "elapsed_sec": 33292.76671433449, "step_time_sec": 8.23068553803023, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4003, "loss": 4.398975372314453, "lr": 0.0002, "elapsed_sec": 33300.99987959862, "step_time_sec": 8.23299708799459, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4004, "loss": 4.365635871887207, "lr": 0.0002, "elapsed_sec": 33309.23130488396, "step_time_sec": 8.23125715192873, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4005, "loss": 4.4066643714904785, "lr": 0.0002, "elapsed_sec": 33317.46436357498, "step_time_sec": 8.232925893971696, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4006, "loss": 4.278335094451904, "lr": 0.0002, "elapsed_sec": 33325.696466207504, "step_time_sec": 8.23193899798207, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4007, "loss": 4.231847286224365, "lr": 0.0002, "elapsed_sec": 33333.92867875099, "step_time_sec": 8.232004682999104, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4008, "loss": 4.3529229164123535, "lr": 0.0002, "elapsed_sec": 33342.161194086075, "step_time_sec": 8.232385357958265, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4009, "loss": 4.566204071044922, "lr": 0.0002, "elapsed_sec": 33350.39466667175, "step_time_sec": 8.233337354962714, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4010, "loss": 4.235172271728516, "lr": 0.0002, "elapsed_sec": 33358.62704086304, "step_time_sec": 8.23217830597423, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4011, "loss": 4.4752912521362305, "lr": 0.0002, "elapsed_sec": 33366.8590734005, "step_time_sec": 8.231891082017682, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4012, "loss": 4.311892509460449, "lr": 0.0002, "elapsed_sec": 33375.092052698135, "step_time_sec": 8.232760402956046, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4013, "loss": 4.3274030685424805, "lr": 0.0002, "elapsed_sec": 33383.330731868744, "step_time_sec": 8.2385580369737, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4014, "loss": 4.293509483337402, "lr": 0.0002, "elapsed_sec": 33391.56997203827, "step_time_sec": 8.239078703918494, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4015, "loss": 4.341063499450684, "lr": 0.0002, "elapsed_sec": 33399.80772948265, "step_time_sec": 8.237577231950127, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4016, "loss": 4.270970344543457, "lr": 0.0002, "elapsed_sec": 33408.039511203766, "step_time_sec": 8.231601075036451, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4017, "loss": 4.384822368621826, "lr": 0.0002, "elapsed_sec": 33416.27277135849, "step_time_sec": 8.233150594984181, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4018, "loss": 4.315184593200684, "lr": 0.0002, "elapsed_sec": 33424.50593328476, "step_time_sec": 8.232956963009201, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4019, "loss": 4.319252014160156, "lr": 0.0002, "elapsed_sec": 33432.7394490242, "step_time_sec": 8.233319208957255, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4020, "loss": 4.1784515380859375, "lr": 0.0002, "elapsed_sec": 33440.97732281685, "step_time_sec": 8.237706993939355, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4021, "loss": 4.360888481140137, "lr": 0.0002, "elapsed_sec": 33449.21640443802, "step_time_sec": 8.238971011014655, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4022, "loss": 4.337378025054932, "lr": 0.0002, "elapsed_sec": 33457.452536821365, "step_time_sec": 8.235904103028588, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4023, "loss": 4.260566234588623, "lr": 0.0002, "elapsed_sec": 33465.69099020958, "step_time_sec": 8.238294653012417, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4024, "loss": 4.216698169708252, "lr": 0.0002, "elapsed_sec": 33473.92921042442, "step_time_sec": 8.237985165906139, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4025, "loss": 4.316434383392334, "lr": 0.0002, "elapsed_sec": 33482.16728377342, "step_time_sec": 8.237913673045114, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4026, "loss": 4.40480899810791, "lr": 0.0002, "elapsed_sec": 33490.405391693115, "step_time_sec": 8.237983271945268, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4027, "loss": 4.291334629058838, "lr": 0.0002, "elapsed_sec": 33498.63791060448, "step_time_sec": 8.232346979086287, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4028, "loss": 4.296834468841553, "lr": 0.0002, "elapsed_sec": 33506.87115979195, "step_time_sec": 8.233077573007904, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4029, "loss": 4.272634506225586, "lr": 0.0002, "elapsed_sec": 33515.10301733017, "step_time_sec": 8.231631791917607, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4030, "loss": 4.241986274719238, "lr": 0.0002, "elapsed_sec": 33523.33598399162, "step_time_sec": 8.232828756910749, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4031, "loss": 4.178284645080566, "lr": 0.0002, "elapsed_sec": 33531.56809306145, "step_time_sec": 8.231924119987525, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4032, "loss": 4.134247303009033, "lr": 0.0002, "elapsed_sec": 33539.80013465881, "step_time_sec": 8.231880576000549, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4033, "loss": 4.223697662353516, "lr": 0.0002, "elapsed_sec": 33548.03241991997, "step_time_sec": 8.232146519934759, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4034, "loss": 4.295772075653076, "lr": 0.0002, "elapsed_sec": 33556.26489377022, "step_time_sec": 8.232340921997093, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4035, "loss": 4.189704895019531, "lr": 0.0002, "elapsed_sec": 33564.49780130386, "step_time_sec": 8.23269158101175, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4036, "loss": 4.309937477111816, "lr": 0.0002, "elapsed_sec": 33572.730249881744, "step_time_sec": 8.232276653987356, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4037, "loss": 4.185074806213379, "lr": 0.0002, "elapsed_sec": 33580.963394641876, "step_time_sec": 8.232950453064404, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4038, "loss": 4.287308216094971, "lr": 0.0002, "elapsed_sec": 33589.200135707855, "step_time_sec": 8.236622812924907, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4039, "loss": 4.271745204925537, "lr": 0.0002, "elapsed_sec": 33597.438642024994, "step_time_sec": 8.238337185001, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4040, "loss": 4.176998138427734, "lr": 0.0002, "elapsed_sec": 33605.677852630615, "step_time_sec": 8.239034416037612, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4041, "loss": 4.220877647399902, "lr": 0.0002, "elapsed_sec": 33613.910496234894, "step_time_sec": 8.23251098091714, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4042, "loss": 4.173821449279785, "lr": 0.0002, "elapsed_sec": 33622.14218521118, "step_time_sec": 8.231480309041217, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4043, "loss": 4.3983025550842285, "lr": 0.0002, "elapsed_sec": 33630.37531399727, "step_time_sec": 8.232982769957744, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4044, "loss": 4.401283264160156, "lr": 0.0002, "elapsed_sec": 33638.607342004776, "step_time_sec": 8.23188894696068, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4045, "loss": 4.173664093017578, "lr": 0.0002, "elapsed_sec": 33646.839711904526, "step_time_sec": 8.232251413050108, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4046, "loss": 4.283886909484863, "lr": 0.0002, "elapsed_sec": 33655.072548151016, "step_time_sec": 8.232603120035492, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4047, "loss": 4.358650207519531, "lr": 0.0002, "elapsed_sec": 33663.3105506897, "step_time_sec": 8.23785626993049, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4048, "loss": 4.259366989135742, "lr": 0.0002, "elapsed_sec": 33671.54798913002, "step_time_sec": 8.23727213696111, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4049, "loss": 4.304877758026123, "lr": 0.0002, "elapsed_sec": 33679.7863471508, "step_time_sec": 8.238211514079012, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4050, "loss": 4.291046619415283, "lr": 0.0002, "elapsed_sec": 33688.01890587807, "step_time_sec": 8.23241512500681, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4051, "loss": 4.253415107727051, "lr": 0.0002, "elapsed_sec": 33696.2558157444, "step_time_sec": 8.236793279997073, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4052, "loss": 4.3278889656066895, "lr": 0.0002, "elapsed_sec": 33704.49478507042, "step_time_sec": 8.238763735047542, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4053, "loss": 4.224570274353027, "lr": 0.0002, "elapsed_sec": 33712.731981277466, "step_time_sec": 8.237021281966008, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4054, "loss": 4.389617919921875, "lr": 0.0002, "elapsed_sec": 33720.971277713776, "step_time_sec": 8.239111509057693, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4055, "loss": 4.317081451416016, "lr": 0.0002, "elapsed_sec": 33729.20985484123, "step_time_sec": 8.238418327993713, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4056, "loss": 4.329215049743652, "lr": 0.0002, "elapsed_sec": 33748.91796708107, "step_time_sec": 19.70795160997659, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4057, "loss": 4.2251691818237305, "lr": 0.0002, "elapsed_sec": 33757.149331092834, "step_time_sec": 8.23121805198025, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4058, "loss": 4.246039867401123, "lr": 0.0002, "elapsed_sec": 33765.3816344738, "step_time_sec": 8.23210590693634, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4059, "loss": 4.382614612579346, "lr": 0.0002, "elapsed_sec": 33773.61560988426, "step_time_sec": 8.233863475034013, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4060, "loss": 4.231380939483643, "lr": 0.0002, "elapsed_sec": 33781.851049661636, "step_time_sec": 8.235291282995604, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4061, "loss": 4.194345474243164, "lr": 0.0002, "elapsed_sec": 33790.08716797829, "step_time_sec": 8.235958052915521, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4062, "loss": 4.245811939239502, "lr": 0.0002, "elapsed_sec": 33798.32540678978, "step_time_sec": 8.238069929997437, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4063, "loss": 4.308565616607666, "lr": 0.0002, "elapsed_sec": 33806.56260037422, "step_time_sec": 8.237048475071788, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4064, "loss": 4.284576416015625, "lr": 0.0002, "elapsed_sec": 33814.79955172539, "step_time_sec": 8.236722522997297, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4065, "loss": 4.373230934143066, "lr": 0.0002, "elapsed_sec": 33823.036684036255, "step_time_sec": 8.236965496093035, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4066, "loss": 4.326335430145264, "lr": 0.0002, "elapsed_sec": 33831.26833033562, "step_time_sec": 8.23149594804272, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4067, "loss": 4.441843509674072, "lr": 0.0002, "elapsed_sec": 33839.501573085785, "step_time_sec": 8.233100187033415, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4068, "loss": 4.1949944496154785, "lr": 0.0002, "elapsed_sec": 33847.73352241516, "step_time_sec": 8.231814057915471, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4069, "loss": 4.28256893157959, "lr": 0.0002, "elapsed_sec": 33855.9671690464, "step_time_sec": 8.233483138959855, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4070, "loss": 4.260140419006348, "lr": 0.0002, "elapsed_sec": 33864.20486783981, "step_time_sec": 8.237495107925497, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4071, "loss": 4.123480796813965, "lr": 0.0002, "elapsed_sec": 33872.44489240646, "step_time_sec": 8.23987428098917, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4072, "loss": 4.368625640869141, "lr": 0.0002, "elapsed_sec": 33880.68288564682, "step_time_sec": 8.237848195014521, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4073, "loss": 4.29653263092041, "lr": 0.0002, "elapsed_sec": 33888.921022892, "step_time_sec": 8.237990495981649, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4074, "loss": 4.2123942375183105, "lr": 0.0002, "elapsed_sec": 33897.157920360565, "step_time_sec": 8.236711088917218, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4075, "loss": 4.2802414894104, "lr": 0.0002, "elapsed_sec": 33905.39621639252, "step_time_sec": 8.238106298958883, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4076, "loss": 4.225265026092529, "lr": 0.0002, "elapsed_sec": 33913.63129067421, "step_time_sec": 8.234961012960412, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4077, "loss": 4.307445526123047, "lr": 0.0002, "elapsed_sec": 33921.86889004707, "step_time_sec": 8.237400060985237, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4078, "loss": 4.2814154624938965, "lr": 0.0002, "elapsed_sec": 33930.108689546585, "step_time_sec": 8.239617650047876, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4079, "loss": 4.122759819030762, "lr": 0.0002, "elapsed_sec": 33938.34906458855, "step_time_sec": 8.240227765054442, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4080, "loss": 4.218970775604248, "lr": 0.0002, "elapsed_sec": 33946.58556675911, "step_time_sec": 8.236381409922615, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4081, "loss": 4.3538384437561035, "lr": 0.0002, "elapsed_sec": 33954.82213258743, "step_time_sec": 8.236349362996407, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4082, "loss": 4.109810829162598, "lr": 0.0002, "elapsed_sec": 33963.058863162994, "step_time_sec": 8.236567558022216, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4083, "loss": 4.254650592803955, "lr": 0.0002, "elapsed_sec": 33971.297105550766, "step_time_sec": 8.238083313917741, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4084, "loss": 4.3275628089904785, "lr": 0.0002, "elapsed_sec": 33979.53092455864, "step_time_sec": 8.233649107976817, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4085, "loss": 4.270400047302246, "lr": 0.0002, "elapsed_sec": 33987.763244628906, "step_time_sec": 8.232171738985926, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4086, "loss": 4.256400108337402, "lr": 0.0002, "elapsed_sec": 33996.00105524063, "step_time_sec": 8.237659550039098, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4087, "loss": 4.3739914894104, "lr": 0.0002, "elapsed_sec": 34004.238906145096, "step_time_sec": 8.237667080946267, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4088, "loss": 4.3195414543151855, "lr": 0.0002, "elapsed_sec": 34012.47774004936, "step_time_sec": 8.238688495010138, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4089, "loss": 4.197033882141113, "lr": 0.0002, "elapsed_sec": 34020.7145152092, "step_time_sec": 8.23663766705431, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4090, "loss": 4.317953586578369, "lr": 0.0002, "elapsed_sec": 34028.95322179794, "step_time_sec": 8.238478422979824, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4091, "loss": 4.148891448974609, "lr": 0.0002, "elapsed_sec": 34037.189807891846, "step_time_sec": 8.236428110045381, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4092, "loss": 4.104363441467285, "lr": 0.0002, "elapsed_sec": 34045.42830324173, "step_time_sec": 8.238310002023354, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4093, "loss": 4.085171222686768, "lr": 0.0002, "elapsed_sec": 34053.66492342949, "step_time_sec": 8.236468181014061, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4094, "loss": 4.140453338623047, "lr": 0.0002, "elapsed_sec": 34061.90288925171, "step_time_sec": 8.237813072046265, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4095, "loss": 4.3417205810546875, "lr": 0.0002, "elapsed_sec": 34070.140989780426, "step_time_sec": 8.237925078021362, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4096, "loss": 4.195154190063477, "lr": 0.0002, "elapsed_sec": 34078.37815403938, "step_time_sec": 8.237050435971469, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4097, "loss": 4.269741535186768, "lr": 0.0002, "elapsed_sec": 34086.6169219017, "step_time_sec": 8.238630583975464, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4098, "loss": 4.40084171295166, "lr": 0.0002, "elapsed_sec": 34094.85393691063, "step_time_sec": 8.236795429023914, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4099, "loss": 4.286020278930664, "lr": 0.0002, "elapsed_sec": 34103.09379315376, "step_time_sec": 8.239694054005668, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4100, "loss": 4.300240516662598, "lr": 0.0002, "elapsed_sec": 34111.332615852356, "step_time_sec": 8.238724475027993, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4101, "loss": 4.230967998504639, "lr": 0.0002, "elapsed_sec": 34119.564767837524, "step_time_sec": 8.231930656009354, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4102, "loss": 4.139490604400635, "lr": 0.0002, "elapsed_sec": 34127.79615855217, "step_time_sec": 8.231220735935494, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4103, "loss": 4.374843597412109, "lr": 0.0002, "elapsed_sec": 34136.028864622116, "step_time_sec": 8.232544646947645, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4104, "loss": 4.355344295501709, "lr": 0.0002, "elapsed_sec": 34144.260668992996, "step_time_sec": 8.231681458069943, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4105, "loss": 4.178894519805908, "lr": 0.0002, "elapsed_sec": 34152.493059158325, "step_time_sec": 8.232209403067827, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4106, "loss": 4.493013858795166, "lr": 0.0002, "elapsed_sec": 34160.72559762001, "step_time_sec": 8.232379838009365, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4107, "loss": 4.278100967407227, "lr": 0.0002, "elapsed_sec": 34168.959285736084, "step_time_sec": 8.233504622010514, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4108, "loss": 4.228912353515625, "lr": 0.0002, "elapsed_sec": 34177.19112086296, "step_time_sec": 8.231689857086167, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4109, "loss": 4.355241775512695, "lr": 0.0002, "elapsed_sec": 34185.42332839966, "step_time_sec": 8.23200941097457, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4110, "loss": 4.262130260467529, "lr": 0.0002, "elapsed_sec": 34193.66034293175, "step_time_sec": 8.236844348954037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4111, "loss": 4.403822898864746, "lr": 0.0002, "elapsed_sec": 34201.89875006676, "step_time_sec": 8.238272662973031, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4112, "loss": 4.30178165435791, "lr": 0.0002, "elapsed_sec": 34210.13561916351, "step_time_sec": 8.236689781071618, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4113, "loss": 4.3288702964782715, "lr": 0.0002, "elapsed_sec": 34218.37513375282, "step_time_sec": 8.239375636912882, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4114, "loss": 4.182949542999268, "lr": 0.0002, "elapsed_sec": 34226.612735271454, "step_time_sec": 8.237474964931607, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4115, "loss": 4.224637985229492, "lr": 0.0002, "elapsed_sec": 34234.850222587585, "step_time_sec": 8.237256691092625, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4116, "loss": 4.1411213874816895, "lr": 0.0002, "elapsed_sec": 34243.084172964096, "step_time_sec": 8.233809093013406, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4117, "loss": 4.265749931335449, "lr": 0.0002, "elapsed_sec": 34251.32225394249, "step_time_sec": 8.237913589924574, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4118, "loss": 4.2320027351379395, "lr": 0.0002, "elapsed_sec": 34259.56131839752, "step_time_sec": 8.238902009907179, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4119, "loss": 4.288942337036133, "lr": 0.0002, "elapsed_sec": 34267.79936337471, "step_time_sec": 8.237887889961712, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4120, "loss": 4.267353057861328, "lr": 0.0002, "elapsed_sec": 34276.03936004639, "step_time_sec": 8.239827045938, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4121, "loss": 4.233428955078125, "lr": 0.0002, "elapsed_sec": 34284.27909040451, "step_time_sec": 8.239569216966629, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4122, "loss": 4.3616862297058105, "lr": 0.0002, "elapsed_sec": 34292.51787734032, "step_time_sec": 8.238649940001778, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4123, "loss": 4.164259910583496, "lr": 0.0002, "elapsed_sec": 34300.758246183395, "step_time_sec": 8.240182758076116, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4124, "loss": 4.237726211547852, "lr": 0.0002, "elapsed_sec": 34308.997602939606, "step_time_sec": 8.23923353292048, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4125, "loss": 4.233513355255127, "lr": 0.0002, "elapsed_sec": 34317.235223293304, "step_time_sec": 8.237434374983422, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4126, "loss": 4.138772964477539, "lr": 0.0002, "elapsed_sec": 34325.473014354706, "step_time_sec": 8.237605037051253, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4127, "loss": 4.2836198806762695, "lr": 0.0002, "elapsed_sec": 34333.70443868637, "step_time_sec": 8.231313107069582, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4128, "loss": 4.132162570953369, "lr": 0.0002, "elapsed_sec": 34341.937678813934, "step_time_sec": 8.233022160013206, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4129, "loss": 4.261144638061523, "lr": 0.0002, "elapsed_sec": 34350.16880750656, "step_time_sec": 8.230959974927828, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4130, "loss": 4.39418888092041, "lr": 0.0002, "elapsed_sec": 34358.40767931938, "step_time_sec": 8.238723924034275, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4131, "loss": 4.285389423370361, "lr": 0.0002, "elapsed_sec": 34366.64531970024, "step_time_sec": 8.23748719098512, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4132, "loss": 4.357826232910156, "lr": 0.0002, "elapsed_sec": 34374.88489508629, "step_time_sec": 8.2393604159588, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4133, "loss": 4.27202033996582, "lr": 0.0002, "elapsed_sec": 34383.121995687485, "step_time_sec": 8.23694253701251, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4134, "loss": 4.268932342529297, "lr": 0.0002, "elapsed_sec": 34391.35514974594, "step_time_sec": 8.23298855801113, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4135, "loss": 4.254486083984375, "lr": 0.0002, "elapsed_sec": 34399.58848237991, "step_time_sec": 8.233182575902902, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4136, "loss": 4.334771633148193, "lr": 0.0002, "elapsed_sec": 34407.82104730606, "step_time_sec": 8.232385348994285, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4137, "loss": 4.454463005065918, "lr": 0.0002, "elapsed_sec": 34416.05929040909, "step_time_sec": 8.238077020971105, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4138, "loss": 4.109420299530029, "lr": 0.0002, "elapsed_sec": 34424.29848027229, "step_time_sec": 8.239044705987908, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4139, "loss": 4.248810768127441, "lr": 0.0002, "elapsed_sec": 34432.535506010056, "step_time_sec": 8.236866257968359, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4140, "loss": 4.353295803070068, "lr": 0.0002, "elapsed_sec": 34440.77295565605, "step_time_sec": 8.237274420098402, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4141, "loss": 4.2135725021362305, "lr": 0.0002, "elapsed_sec": 34449.01222252846, "step_time_sec": 8.23907913803123, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4142, "loss": 4.296557903289795, "lr": 0.0002, "elapsed_sec": 34457.24873971939, "step_time_sec": 8.236347199999727, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4143, "loss": 4.1549973487854, "lr": 0.0002, "elapsed_sec": 34465.48731613159, "step_time_sec": 8.238409910933115, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4144, "loss": 4.199618816375732, "lr": 0.0002, "elapsed_sec": 34473.72529935837, "step_time_sec": 8.237825476098806, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4145, "loss": 4.3132243156433105, "lr": 0.0002, "elapsed_sec": 34481.96285057068, "step_time_sec": 8.237400576006621, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4146, "loss": 4.222167491912842, "lr": 0.0002, "elapsed_sec": 34490.19962954521, "step_time_sec": 8.236618643044494, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4147, "loss": 4.466310501098633, "lr": 0.0002, "elapsed_sec": 34498.43146657944, "step_time_sec": 8.231694804970175, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4148, "loss": 4.168558120727539, "lr": 0.0002, "elapsed_sec": 34506.66487789154, "step_time_sec": 8.233287056908011, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4149, "loss": 4.319684028625488, "lr": 0.0002, "elapsed_sec": 34514.89801669121, "step_time_sec": 8.232916660024785, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4150, "loss": 4.065457820892334, "lr": 0.0002, "elapsed_sec": 34523.13754630089, "step_time_sec": 8.23941897705663, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4151, "loss": 4.232524394989014, "lr": 0.0002, "elapsed_sec": 34531.374428749084, "step_time_sec": 8.236670282902196, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4152, "loss": 4.273317813873291, "lr": 0.0002, "elapsed_sec": 34539.61342096329, "step_time_sec": 8.238851065048948, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4153, "loss": 4.264802932739258, "lr": 0.0002, "elapsed_sec": 34547.85148906708, "step_time_sec": 8.237847247044556, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4154, "loss": 4.1798272132873535, "lr": 0.0002, "elapsed_sec": 34556.09006643295, "step_time_sec": 8.238399637979455, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4155, "loss": 4.21885347366333, "lr": 0.0002, "elapsed_sec": 34564.328909635544, "step_time_sec": 8.23870839504525, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4156, "loss": 4.276443958282471, "lr": 0.0002, "elapsed_sec": 34572.56686735153, "step_time_sec": 8.237808489007875, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4157, "loss": 4.371874809265137, "lr": 0.0002, "elapsed_sec": 34580.80168914795, "step_time_sec": 8.23466295003891, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4158, "loss": 4.278094291687012, "lr": 0.0002, "elapsed_sec": 34589.03845119476, "step_time_sec": 8.236563417012803, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4159, "loss": 4.1661481857299805, "lr": 0.0002, "elapsed_sec": 34597.27702116966, "step_time_sec": 8.238390239072032, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4160, "loss": 4.227484703063965, "lr": 0.0002, "elapsed_sec": 34605.513751506805, "step_time_sec": 8.236585709964857, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4161, "loss": 4.207951068878174, "lr": 0.0002, "elapsed_sec": 34613.7518427372, "step_time_sec": 8.237931906012818, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4162, "loss": 4.2322211265563965, "lr": 0.0002, "elapsed_sec": 34621.99163150787, "step_time_sec": 8.23962016007863, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4163, "loss": 4.175563335418701, "lr": 0.0002, "elapsed_sec": 34630.22974038124, "step_time_sec": 8.237923830049112, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4164, "loss": 4.152891159057617, "lr": 0.0002, "elapsed_sec": 34638.46307301521, "step_time_sec": 8.233208750956692, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4165, "loss": 4.141213893890381, "lr": 0.0002, "elapsed_sec": 34646.699961185455, "step_time_sec": 8.236725648050196, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4166, "loss": 4.33009147644043, "lr": 0.0002, "elapsed_sec": 34654.93824362755, "step_time_sec": 8.238069067010656, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4167, "loss": 4.129702568054199, "lr": 0.0002, "elapsed_sec": 34663.17624115944, "step_time_sec": 8.237838221946731, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4168, "loss": 4.280572414398193, "lr": 0.0002, "elapsed_sec": 34671.415549993515, "step_time_sec": 8.239153563976288, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4169, "loss": 4.133970737457275, "lr": 0.0002, "elapsed_sec": 34679.655269384384, "step_time_sec": 8.239598284009844, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4170, "loss": 4.203865051269531, "lr": 0.0002, "elapsed_sec": 34687.89308786392, "step_time_sec": 8.237661893013865, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4171, "loss": 4.201096534729004, "lr": 0.0002, "elapsed_sec": 34696.13312268257, "step_time_sec": 8.239858490996994, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4172, "loss": 4.165910243988037, "lr": 0.0002, "elapsed_sec": 34704.371908426285, "step_time_sec": 8.238597784074955, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4173, "loss": 4.193349361419678, "lr": 0.0002, "elapsed_sec": 34712.6107070446, "step_time_sec": 8.238640198018402, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4174, "loss": 4.302778720855713, "lr": 0.0002, "elapsed_sec": 34720.84922862053, "step_time_sec": 8.238327736966312, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4175, "loss": 4.081832408905029, "lr": 0.0002, "elapsed_sec": 34729.08172225952, "step_time_sec": 8.232362234964967, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4176, "loss": 4.337916851043701, "lr": 0.0002, "elapsed_sec": 34737.31521201134, "step_time_sec": 8.233292126911692, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4177, "loss": 4.296930313110352, "lr": 0.0002, "elapsed_sec": 34745.54749441147, "step_time_sec": 8.232120772008784, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4178, "loss": 4.242758274078369, "lr": 0.0002, "elapsed_sec": 34753.78609204292, "step_time_sec": 8.238476756028831, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4179, "loss": 4.295257568359375, "lr": 0.0002, "elapsed_sec": 34762.02259540558, "step_time_sec": 8.236313030007295, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4180, "loss": 4.031978607177734, "lr": 0.0002, "elapsed_sec": 34770.26102089882, "step_time_sec": 8.238280947087333, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4181, "loss": 4.139175891876221, "lr": 0.0002, "elapsed_sec": 34778.49849820137, "step_time_sec": 8.237297960091382, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4182, "loss": 4.23964786529541, "lr": 0.0002, "elapsed_sec": 34786.73109507561, "step_time_sec": 8.232472906005569, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4183, "loss": 4.216531276702881, "lr": 0.0002, "elapsed_sec": 34794.96324086189, "step_time_sec": 8.231930136913434, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4184, "loss": 4.278514862060547, "lr": 0.0002, "elapsed_sec": 34803.19632411003, "step_time_sec": 8.232928420067765, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4185, "loss": 4.242458820343018, "lr": 0.0002, "elapsed_sec": 34811.43437933922, "step_time_sec": 8.23789492296055, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4186, "loss": 4.226989269256592, "lr": 0.0002, "elapsed_sec": 34819.67428898811, "step_time_sec": 8.239772911998443, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4187, "loss": 4.121819019317627, "lr": 0.0002, "elapsed_sec": 34827.912073135376, "step_time_sec": 8.237624553963542, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4188, "loss": 4.12571382522583, "lr": 0.0002, "elapsed_sec": 34836.14476060867, "step_time_sec": 8.23250563396141, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4189, "loss": 4.1758527755737305, "lr": 0.0002, "elapsed_sec": 34844.378412246704, "step_time_sec": 8.233478066977113, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4190, "loss": 4.3299784660339355, "lr": 0.0002, "elapsed_sec": 34852.60953307152, "step_time_sec": 8.230948952957988, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4191, "loss": 4.217464447021484, "lr": 0.0002, "elapsed_sec": 34860.848593235016, "step_time_sec": 8.23889808007516, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4192, "loss": 4.315417289733887, "lr": 0.0002, "elapsed_sec": 34869.08583021164, "step_time_sec": 8.237076853052713, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4193, "loss": 4.123600959777832, "lr": 0.0002, "elapsed_sec": 34877.32631587982, "step_time_sec": 8.24033608706668, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4194, "loss": 4.097240447998047, "lr": 0.0002, "elapsed_sec": 34885.56447553635, "step_time_sec": 8.237990070017986, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4195, "loss": 4.231419563293457, "lr": 0.0002, "elapsed_sec": 34893.802951812744, "step_time_sec": 8.23828745400533, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4196, "loss": 4.2098774909973145, "lr": 0.0002, "elapsed_sec": 34902.04145669937, "step_time_sec": 8.238346095080487, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4197, "loss": 4.400824069976807, "lr": 0.0002, "elapsed_sec": 34910.27992081642, "step_time_sec": 8.238317131064832, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4198, "loss": 4.086031913757324, "lr": 0.0002, "elapsed_sec": 34918.51841545105, "step_time_sec": 8.238286759937182, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4199, "loss": 4.222039699554443, "lr": 0.0002, "elapsed_sec": 34926.75595211983, "step_time_sec": 8.237410919042304, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4200, "loss": 4.215821743011475, "lr": 0.0002, "elapsed_sec": 34934.995958805084, "step_time_sec": 8.239782572025433, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4201, "loss": 4.22055196762085, "lr": 0.0002, "elapsed_sec": 34943.23608446121, "step_time_sec": 8.239973460091278, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4202, "loss": 4.215322017669678, "lr": 0.0002, "elapsed_sec": 34951.46789383888, "step_time_sec": 8.231719229952432, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4203, "loss": 4.279864311218262, "lr": 0.0002, "elapsed_sec": 34959.70058012009, "step_time_sec": 8.232466649031267, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4204, "loss": 3.9878792762756348, "lr": 0.0002, "elapsed_sec": 34967.93277692795, "step_time_sec": 8.232059402042069, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4205, "loss": 4.1341142654418945, "lr": 0.0002, "elapsed_sec": 34976.170509815216, "step_time_sec": 8.237545322976075, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4206, "loss": 4.243988513946533, "lr": 0.0002, "elapsed_sec": 34984.407829999924, "step_time_sec": 8.237190366024151, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4207, "loss": 4.240276336669922, "lr": 0.0002, "elapsed_sec": 34992.64485645294, "step_time_sec": 8.236828077002428, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4208, "loss": 4.1673054695129395, "lr": 0.0002, "elapsed_sec": 35000.8832924366, "step_time_sec": 8.238307568011805, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4209, "loss": 4.2215962409973145, "lr": 0.0002, "elapsed_sec": 35009.12211561203, "step_time_sec": 8.23864203307312, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4210, "loss": 4.187723636627197, "lr": 0.0002, "elapsed_sec": 35017.36082601547, "step_time_sec": 8.238545599975623, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4211, "loss": 4.1579508781433105, "lr": 0.0002, "elapsed_sec": 35025.598146915436, "step_time_sec": 8.237173282075673, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4212, "loss": 4.153714179992676, "lr": 0.0002, "elapsed_sec": 35033.835698843, "step_time_sec": 8.237377380020916, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4213, "loss": 4.099534034729004, "lr": 0.0002, "elapsed_sec": 35042.07276773453, "step_time_sec": 8.236910072038881, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4214, "loss": 4.24243688583374, "lr": 0.0002, "elapsed_sec": 35050.312144994736, "step_time_sec": 8.239212811924517, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4215, "loss": 4.186648845672607, "lr": 0.0002, "elapsed_sec": 35058.551230192184, "step_time_sec": 8.238977261004038, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4216, "loss": 4.120812892913818, "lr": 0.0002, "elapsed_sec": 35066.79024195671, "step_time_sec": 8.238856040057726, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4217, "loss": 4.232240200042725, "lr": 0.0002, "elapsed_sec": 35075.02981185913, "step_time_sec": 8.239351241034456, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4218, "loss": 4.163890361785889, "lr": 0.0002, "elapsed_sec": 35083.26728892326, "step_time_sec": 8.237303039990366, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4219, "loss": 4.27099609375, "lr": 0.0002, "elapsed_sec": 35091.50679850578, "step_time_sec": 8.239348469069228, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4220, "loss": 4.090818405151367, "lr": 0.0002, "elapsed_sec": 35099.73870396614, "step_time_sec": 8.231781824026257, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4221, "loss": 4.124814510345459, "lr": 0.0002, "elapsed_sec": 35107.97140431404, "step_time_sec": 8.2325701310765, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4222, "loss": 4.211115837097168, "lr": 0.0002, "elapsed_sec": 35116.20371341705, "step_time_sec": 8.23209935799241, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4223, "loss": 4.164063453674316, "lr": 0.0002, "elapsed_sec": 35124.43789601326, "step_time_sec": 8.234005524078384, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4224, "loss": 4.170238494873047, "lr": 0.0002, "elapsed_sec": 35132.66944050789, "step_time_sec": 8.231364844017662, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4225, "loss": 4.023270606994629, "lr": 0.0002, "elapsed_sec": 35140.902792453766, "step_time_sec": 8.23320282495115, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4226, "loss": 4.300549507141113, "lr": 0.0002, "elapsed_sec": 35149.1400475502, "step_time_sec": 8.237101324019022, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4227, "loss": 4.253047943115234, "lr": 0.0002, "elapsed_sec": 35157.3791308403, "step_time_sec": 8.238935084897093, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4228, "loss": 4.245334625244141, "lr": 0.0002, "elapsed_sec": 35165.61732387543, "step_time_sec": 8.23801587393973, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4229, "loss": 4.392899990081787, "lr": 0.0002, "elapsed_sec": 35173.850095033646, "step_time_sec": 8.232659762958065, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4230, "loss": 4.276659965515137, "lr": 0.0002, "elapsed_sec": 35182.08257126808, "step_time_sec": 8.23228892707266, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4231, "loss": 4.1970319747924805, "lr": 0.0002, "elapsed_sec": 35190.315171957016, "step_time_sec": 8.232425458962098, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4232, "loss": 4.192067623138428, "lr": 0.0002, "elapsed_sec": 35198.55181336403, "step_time_sec": 8.236471988959238, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4233, "loss": 4.1079535484313965, "lr": 0.0002, "elapsed_sec": 35206.79072523117, "step_time_sec": 8.238813086063601, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4234, "loss": 4.25039529800415, "lr": 0.0002, "elapsed_sec": 35215.02989578247, "step_time_sec": 8.238973589031957, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4235, "loss": 4.203714847564697, "lr": 0.0002, "elapsed_sec": 35223.261478185654, "step_time_sec": 8.231441859039478, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4236, "loss": 4.171274662017822, "lr": 0.0002, "elapsed_sec": 35231.49455332756, "step_time_sec": 8.232915060012601, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4237, "loss": 4.210160732269287, "lr": 0.0002, "elapsed_sec": 35239.728684425354, "step_time_sec": 8.234063146985136, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4238, "loss": 4.256021499633789, "lr": 0.0002, "elapsed_sec": 35247.96866083145, "step_time_sec": 8.239692656090483, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4239, "loss": 4.273581504821777, "lr": 0.0002, "elapsed_sec": 35256.205414772034, "step_time_sec": 8.236644493066706, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4240, "loss": 4.063835144042969, "lr": 0.0002, "elapsed_sec": 35264.44391345978, "step_time_sec": 8.23821946803946, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4241, "loss": 4.029870510101318, "lr": 0.0002, "elapsed_sec": 35272.68155121803, "step_time_sec": 8.237444545025937, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4242, "loss": 4.248944282531738, "lr": 0.0002, "elapsed_sec": 35280.920439720154, "step_time_sec": 8.238745096023194, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4243, "loss": 4.076876163482666, "lr": 0.0002, "elapsed_sec": 35289.15774321556, "step_time_sec": 8.237140426994301, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4244, "loss": 4.159182548522949, "lr": 0.0002, "elapsed_sec": 35297.395231723785, "step_time_sec": 8.237332809017971, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4245, "loss": 4.195458889007568, "lr": 0.0002, "elapsed_sec": 35305.628860235214, "step_time_sec": 8.233447927981615, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4246, "loss": 4.050971984863281, "lr": 0.0002, "elapsed_sec": 35313.86653494835, "step_time_sec": 8.237502228002995, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4247, "loss": 3.9839022159576416, "lr": 0.0002, "elapsed_sec": 35322.10530757904, "step_time_sec": 8.238594835973345, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4248, "loss": 4.095301151275635, "lr": 0.0002, "elapsed_sec": 35330.342952251434, "step_time_sec": 8.2374787859153, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4249, "loss": 4.210585594177246, "lr": 0.0002, "elapsed_sec": 35338.58203840256, "step_time_sec": 8.238921009935439, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4250, "loss": 4.135420799255371, "lr": 0.0002, "elapsed_sec": 35346.82092285156, "step_time_sec": 8.238767076982185, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4251, "loss": 4.220116138458252, "lr": 0.0002, "elapsed_sec": 35355.05907559395, "step_time_sec": 8.237951374030672, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4252, "loss": 4.188188076019287, "lr": 0.0002, "elapsed_sec": 35363.291568517685, "step_time_sec": 8.23236080002971, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4253, "loss": 4.150836944580078, "lr": 0.0002, "elapsed_sec": 35371.52828288078, "step_time_sec": 8.23652141704224, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4254, "loss": 4.22931432723999, "lr": 0.0002, "elapsed_sec": 35379.767847299576, "step_time_sec": 8.239400762016885, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4255, "loss": 4.452173709869385, "lr": 0.0002, "elapsed_sec": 35388.006251096725, "step_time_sec": 8.238259518053383, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4256, "loss": 4.1760358810424805, "lr": 0.0002, "elapsed_sec": 35396.24451446533, "step_time_sec": 8.238163904985413, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4257, "loss": 4.247791767120361, "lr": 0.0002, "elapsed_sec": 35404.48395252228, "step_time_sec": 8.239203689037822, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4258, "loss": 4.091419696807861, "lr": 0.0002, "elapsed_sec": 35412.72243189812, "step_time_sec": 8.238310517044738, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4259, "loss": 4.243398189544678, "lr": 0.0002, "elapsed_sec": 35420.96314096451, "step_time_sec": 8.240541261970066, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4260, "loss": 3.9905569553375244, "lr": 0.0002, "elapsed_sec": 35429.20192718506, "step_time_sec": 8.238638121983968, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4261, "loss": 4.0952959060668945, "lr": 0.0002, "elapsed_sec": 35437.4408018589, "step_time_sec": 8.238702416070737, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4262, "loss": 4.2389326095581055, "lr": 0.0002, "elapsed_sec": 35445.67912817001, "step_time_sec": 8.238161024055444, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4263, "loss": 4.10849142074585, "lr": 0.0002, "elapsed_sec": 35453.91183304787, "step_time_sec": 8.232549726963043, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4264, "loss": 4.160950183868408, "lr": 0.0002, "elapsed_sec": 35462.15181136131, "step_time_sec": 8.239829847007059, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4265, "loss": 4.1471381187438965, "lr": 0.0002, "elapsed_sec": 35470.390521764755, "step_time_sec": 8.238520342041738, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4266, "loss": 4.276170253753662, "lr": 0.0002, "elapsed_sec": 35478.62994194031, "step_time_sec": 8.239305353956297, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4267, "loss": 4.2725419998168945, "lr": 0.0002, "elapsed_sec": 35486.868137836456, "step_time_sec": 8.23802639101632, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4268, "loss": 4.114505767822266, "lr": 0.0002, "elapsed_sec": 35495.10667467117, "step_time_sec": 8.238351706997491, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4269, "loss": 4.276519298553467, "lr": 0.0002, "elapsed_sec": 35503.34518313408, "step_time_sec": 8.238354214001447, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4270, "loss": 4.267933368682861, "lr": 0.0002, "elapsed_sec": 35511.58286881447, "step_time_sec": 8.237472424050793, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4271, "loss": 4.089602470397949, "lr": 0.0002, "elapsed_sec": 35519.82234954834, "step_time_sec": 8.239332825061865, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4272, "loss": 4.0906853675842285, "lr": 0.0002, "elapsed_sec": 35528.060516119, "step_time_sec": 8.238071647007018, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4273, "loss": 4.200492858886719, "lr": 0.0002, "elapsed_sec": 35536.30069756508, "step_time_sec": 8.240012496011332, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4274, "loss": 4.08300256729126, "lr": 0.0002, "elapsed_sec": 35544.5395026207, "step_time_sec": 8.238621862023138, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4275, "loss": 4.0578227043151855, "lr": 0.0002, "elapsed_sec": 35552.77811193466, "step_time_sec": 8.23839736590162, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4276, "loss": 4.1517767906188965, "lr": 0.0002, "elapsed_sec": 35561.017781972885, "step_time_sec": 8.239508013008162, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4277, "loss": 4.1333699226379395, "lr": 0.0002, "elapsed_sec": 35569.257329940796, "step_time_sec": 8.239353126031347, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4278, "loss": 4.229036808013916, "lr": 0.0002, "elapsed_sec": 35577.49540042877, "step_time_sec": 8.237899923929945, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4279, "loss": 4.070662498474121, "lr": 0.0002, "elapsed_sec": 35585.72831940651, "step_time_sec": 8.232773694908246, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4280, "loss": 3.9938089847564697, "lr": 0.0002, "elapsed_sec": 35593.96094274521, "step_time_sec": 8.232475341996178, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4281, "loss": 4.205127716064453, "lr": 0.0002, "elapsed_sec": 35602.1944231987, "step_time_sec": 8.233312234980986, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4282, "loss": 4.161639213562012, "lr": 0.0002, "elapsed_sec": 35610.43217206001, "step_time_sec": 8.237560646957718, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4283, "loss": 4.1052069664001465, "lr": 0.0002, "elapsed_sec": 35618.671994924545, "step_time_sec": 8.239657610072754, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4284, "loss": 4.172520160675049, "lr": 0.0002, "elapsed_sec": 35626.90985631943, "step_time_sec": 8.237670292030089, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4285, "loss": 4.338489532470703, "lr": 0.0002, "elapsed_sec": 35635.14793729782, "step_time_sec": 8.237921869964339, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4286, "loss": 4.255366802215576, "lr": 0.0002, "elapsed_sec": 35643.38659453392, "step_time_sec": 8.238521156948991, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4287, "loss": 4.2762885093688965, "lr": 0.0002, "elapsed_sec": 35651.62424278259, "step_time_sec": 8.237435807008296, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4288, "loss": 4.333683013916016, "lr": 0.0002, "elapsed_sec": 35659.86308979988, "step_time_sec": 8.238694707048126, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4289, "loss": 4.305037498474121, "lr": 0.0002, "elapsed_sec": 35668.102390527725, "step_time_sec": 8.239144537015818, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4290, "loss": 4.283905982971191, "lr": 0.0002, "elapsed_sec": 35676.34229660034, "step_time_sec": 8.239790449035354, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4291, "loss": 4.125494003295898, "lr": 0.0002, "elapsed_sec": 35684.581706523895, "step_time_sec": 8.239230063045397, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4292, "loss": 4.3356757164001465, "lr": 0.0002, "elapsed_sec": 35692.81319761276, "step_time_sec": 8.231269731069915, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4293, "loss": 4.2229790687561035, "lr": 0.0002, "elapsed_sec": 35701.046686172485, "step_time_sec": 8.233316414989531, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4294, "loss": 4.126786231994629, "lr": 0.0002, "elapsed_sec": 35709.279433488846, "step_time_sec": 8.232590432977304, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4295, "loss": 4.160880088806152, "lr": 0.0002, "elapsed_sec": 35717.51806926727, "step_time_sec": 8.238453704980202, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4296, "loss": 4.2952070236206055, "lr": 0.0002, "elapsed_sec": 35725.75567531586, "step_time_sec": 8.237447464023717, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4297, "loss": 4.098113059997559, "lr": 0.0002, "elapsed_sec": 35733.99409389496, "step_time_sec": 8.238264685031027, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4298, "loss": 4.027859687805176, "lr": 0.0002, "elapsed_sec": 35742.23269844055, "step_time_sec": 8.238449149997905, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4299, "loss": 4.272815704345703, "lr": 0.0002, "elapsed_sec": 35750.47207427025, "step_time_sec": 8.239228046964854, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4300, "loss": 4.0981597900390625, "lr": 0.0002, "elapsed_sec": 35758.711252212524, "step_time_sec": 8.239022949943319, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4301, "loss": 4.233047962188721, "lr": 0.0002, "elapsed_sec": 35766.949216365814, "step_time_sec": 8.237823675968684, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4302, "loss": 4.242016315460205, "lr": 0.0002, "elapsed_sec": 35775.18831348419, "step_time_sec": 8.238876850926317, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4303, "loss": 4.191473484039307, "lr": 0.0002, "elapsed_sec": 35783.420452833176, "step_time_sec": 8.232023264048621, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4304, "loss": 4.221047878265381, "lr": 0.0002, "elapsed_sec": 35791.65408062935, "step_time_sec": 8.233462115051225, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4305, "loss": 4.293889999389648, "lr": 0.0002, "elapsed_sec": 35799.88609981537, "step_time_sec": 8.231810476980172, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4306, "loss": 4.138814926147461, "lr": 0.0002, "elapsed_sec": 35808.12494516373, "step_time_sec": 8.238675864064135, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4307, "loss": 4.196054935455322, "lr": 0.0002, "elapsed_sec": 35816.363877773285, "step_time_sec": 8.238845588988625, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4308, "loss": 4.037459850311279, "lr": 0.0002, "elapsed_sec": 35824.60268115997, "step_time_sec": 8.238594055059366, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4309, "loss": 4.21457576751709, "lr": 0.0002, "elapsed_sec": 35832.84006047249, "step_time_sec": 8.237245346070267, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4310, "loss": 4.293305397033691, "lr": 0.0002, "elapsed_sec": 35841.078342199326, "step_time_sec": 8.238074889988638, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4311, "loss": 4.263105392456055, "lr": 0.0002, "elapsed_sec": 35849.31660819054, "step_time_sec": 8.238124581985176, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4312, "loss": 4.214753150939941, "lr": 0.0002, "elapsed_sec": 35857.55559825897, "step_time_sec": 8.238831881899387, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4313, "loss": 4.113504886627197, "lr": 0.0002, "elapsed_sec": 35865.79427862167, "step_time_sec": 8.238471224089153, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4314, "loss": 4.143804550170898, "lr": 0.0002, "elapsed_sec": 35874.03347802162, "step_time_sec": 8.239049113006331, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4315, "loss": 4.248356819152832, "lr": 0.0002, "elapsed_sec": 35882.27134895325, "step_time_sec": 8.237693857052363, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4316, "loss": 4.408604145050049, "lr": 0.0002, "elapsed_sec": 35890.5096821785, "step_time_sec": 8.238199956947938, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4317, "loss": 4.194693565368652, "lr": 0.0002, "elapsed_sec": 35898.74816274643, "step_time_sec": 8.238355887006037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4318, "loss": 4.2072882652282715, "lr": 0.0002, "elapsed_sec": 35906.98771595955, "step_time_sec": 8.239312760997564, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4319, "loss": 4.166452884674072, "lr": 0.0002, "elapsed_sec": 35915.22708964348, "step_time_sec": 8.239224233082496, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4320, "loss": 4.096378326416016, "lr": 0.0002, "elapsed_sec": 35923.458781957626, "step_time_sec": 8.23156070697587, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4321, "loss": 4.245582103729248, "lr": 0.0002, "elapsed_sec": 35931.69206929207, "step_time_sec": 8.233098716009408, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4322, "loss": 4.221848487854004, "lr": 0.0002, "elapsed_sec": 35939.9248046875, "step_time_sec": 8.232544144033454, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4323, "loss": 4.279162406921387, "lr": 0.0002, "elapsed_sec": 35948.15815734863, "step_time_sec": 8.23325691302307, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4324, "loss": 4.17483377456665, "lr": 0.0002, "elapsed_sec": 35956.392063617706, "step_time_sec": 8.233744655968621, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4325, "loss": 4.267733097076416, "lr": 0.0002, "elapsed_sec": 35964.626240730286, "step_time_sec": 8.233946608030237, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4326, "loss": 4.1487603187561035, "lr": 0.0002, "elapsed_sec": 35972.859288454056, "step_time_sec": 8.232877923059277, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4327, "loss": 4.232209205627441, "lr": 0.0002, "elapsed_sec": 35981.09255051613, "step_time_sec": 8.233107211999595, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4328, "loss": 4.132627487182617, "lr": 0.0002, "elapsed_sec": 35989.32493567467, "step_time_sec": 8.232278316980228, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4329, "loss": 4.074333190917969, "lr": 0.0002, "elapsed_sec": 35997.56290984154, "step_time_sec": 8.237837487948127, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4330, "loss": 4.213912010192871, "lr": 0.0002, "elapsed_sec": 36005.80076169968, "step_time_sec": 8.237667476991192, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4331, "loss": 4.080029010772705, "lr": 0.0002, "elapsed_sec": 36014.040942430496, "step_time_sec": 8.239978626952507, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4332, "loss": 4.225863456726074, "lr": 0.0002, "elapsed_sec": 36022.279958963394, "step_time_sec": 8.238858793047257, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4333, "loss": 4.144627094268799, "lr": 0.0002, "elapsed_sec": 36030.51391553879, "step_time_sec": 8.233817750005983, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4334, "loss": 4.197020530700684, "lr": 0.0002, "elapsed_sec": 36038.747202157974, "step_time_sec": 8.233137044939213, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4335, "loss": 4.22036075592041, "lr": 0.0002, "elapsed_sec": 36046.98038840294, "step_time_sec": 8.232968854019418, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4336, "loss": 4.429856777191162, "lr": 0.0002, "elapsed_sec": 36055.21802067757, "step_time_sec": 8.237487104022875, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4337, "loss": 4.188218116760254, "lr": 0.0002, "elapsed_sec": 36063.45709013939, "step_time_sec": 8.238914603018202, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4338, "loss": 4.35130500793457, "lr": 0.0002, "elapsed_sec": 36071.69607067108, "step_time_sec": 8.238811880000867, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4339, "loss": 4.344934463500977, "lr": 0.0002, "elapsed_sec": 36079.933299064636, "step_time_sec": 8.237030811025761, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4340, "loss": 4.20473051071167, "lr": 0.0002, "elapsed_sec": 36088.17246890068, "step_time_sec": 8.239023498026654, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4341, "loss": 4.2644829750061035, "lr": 0.0002, "elapsed_sec": 36096.41097736359, "step_time_sec": 8.238398211891763, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4342, "loss": 4.279411792755127, "lr": 0.0002, "elapsed_sec": 36104.65038108826, "step_time_sec": 8.23920937802177, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4343, "loss": 4.19563102722168, "lr": 0.0002, "elapsed_sec": 36112.8892724514, "step_time_sec": 8.238709785975516, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4344, "loss": 4.014268398284912, "lr": 0.0002, "elapsed_sec": 36121.127057790756, "step_time_sec": 8.237575520994142, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4345, "loss": 4.4225592613220215, "lr": 0.0002, "elapsed_sec": 36129.36617541313, "step_time_sec": 8.23897466994822, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4346, "loss": 4.15346622467041, "lr": 0.0002, "elapsed_sec": 36137.60482263565, "step_time_sec": 8.238481804961339, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4347, "loss": 4.169974327087402, "lr": 0.0002, "elapsed_sec": 36145.843992471695, "step_time_sec": 8.239010218996555, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4348, "loss": 3.9458022117614746, "lr": 0.0002, "elapsed_sec": 36154.08235454559, "step_time_sec": 8.238200944033451, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4349, "loss": 4.2326555252075195, "lr": 0.0002, "elapsed_sec": 36162.31867790222, "step_time_sec": 8.236160427914001, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4350, "loss": 4.1095194816589355, "lr": 0.0002, "elapsed_sec": 36170.55378603935, "step_time_sec": 8.23498590302188, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4351, "loss": 4.350014686584473, "lr": 0.0002, "elapsed_sec": 36178.79120492935, "step_time_sec": 8.237250801059417, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4352, "loss": 4.238202095031738, "lr": 0.0002, "elapsed_sec": 36187.029153585434, "step_time_sec": 8.237751535023563, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4353, "loss": 4.172548294067383, "lr": 0.0002, "elapsed_sec": 36195.26805138588, "step_time_sec": 8.23874118598178, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4354, "loss": 4.312662124633789, "lr": 0.0002, "elapsed_sec": 36203.50617456436, "step_time_sec": 8.237958580954, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4355, "loss": 4.295468330383301, "lr": 0.0002, "elapsed_sec": 36211.74484419823, "step_time_sec": 8.238539024954662, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4356, "loss": 4.288863658905029, "lr": 0.0002, "elapsed_sec": 36220.05794548988, "step_time_sec": 8.245740441023372, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4357, "loss": 4.435678958892822, "lr": 0.0002, "elapsed_sec": 36228.29627251625, "step_time_sec": 8.238184208981693, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4358, "loss": 4.045337677001953, "lr": 0.0002, "elapsed_sec": 36236.535299539566, "step_time_sec": 8.238901898032054, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4359, "loss": 4.268290042877197, "lr": 0.0002, "elapsed_sec": 36244.77304196358, "step_time_sec": 8.237557530053891, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4360, "loss": 4.072965621948242, "lr": 0.0002, "elapsed_sec": 36253.010796546936, "step_time_sec": 8.237577999010682, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4361, "loss": 4.236793041229248, "lr": 0.0002, "elapsed_sec": 36261.24555134773, "step_time_sec": 8.234577558003366, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4362, "loss": 4.213023662567139, "lr": 0.0002, "elapsed_sec": 36269.48294329643, "step_time_sec": 8.237208631937392, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4363, "loss": 4.111668109893799, "lr": 0.0002, "elapsed_sec": 36277.72200345993, "step_time_sec": 8.238909263978712, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4364, "loss": 4.144235610961914, "lr": 0.0002, "elapsed_sec": 36285.95938539505, "step_time_sec": 8.237204964039847, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4365, "loss": 4.130001068115234, "lr": 0.0002, "elapsed_sec": 36294.19842290878, "step_time_sec": 8.238907144987024, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4366, "loss": 4.019574165344238, "lr": 0.0002, "elapsed_sec": 36302.436749219894, "step_time_sec": 8.238109836005606, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4367, "loss": 4.124683856964111, "lr": 0.0002, "elapsed_sec": 36310.6751909256, "step_time_sec": 8.23834157595411, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4368, "loss": 4.248991012573242, "lr": 0.0002, "elapsed_sec": 36318.91616439819, "step_time_sec": 8.240791246062145, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4369, "loss": 4.193465709686279, "lr": 0.0002, "elapsed_sec": 36327.15420269966, "step_time_sec": 8.237816547974944, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4370, "loss": 4.264767646789551, "lr": 0.0002, "elapsed_sec": 36335.39199090004, "step_time_sec": 8.23764518799726, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4371, "loss": 4.17282772064209, "lr": 0.0002, "elapsed_sec": 36343.630875349045, "step_time_sec": 8.23874800093472, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4372, "loss": 4.175219535827637, "lr": 0.0002, "elapsed_sec": 36351.86269760132, "step_time_sec": 8.231651089969091, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4373, "loss": 4.163656711578369, "lr": 0.0002, "elapsed_sec": 36360.09612131119, "step_time_sec": 8.233235197025351, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4374, "loss": 4.320730686187744, "lr": 0.0002, "elapsed_sec": 36368.32826447487, "step_time_sec": 8.232028569909744, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4375, "loss": 4.169099807739258, "lr": 0.0002, "elapsed_sec": 36376.567695856094, "step_time_sec": 8.23928388603963, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4376, "loss": 4.223737716674805, "lr": 0.0002, "elapsed_sec": 36384.80450463295, "step_time_sec": 8.23659989295993, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4377, "loss": 4.095751762390137, "lr": 0.0002, "elapsed_sec": 36393.04436802864, "step_time_sec": 8.23970004695002, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4378, "loss": 4.1126484870910645, "lr": 0.0002, "elapsed_sec": 36401.2809946537, "step_time_sec": 8.236422969028354, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4379, "loss": 4.296238899230957, "lr": 0.0002, "elapsed_sec": 36409.51463222504, "step_time_sec": 8.233508566976525, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4380, "loss": 4.184413433074951, "lr": 0.0002, "elapsed_sec": 36417.74683356285, "step_time_sec": 8.232046792982146, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4381, "loss": 4.259632110595703, "lr": 0.0002, "elapsed_sec": 36425.97961783409, "step_time_sec": 8.232643357012421, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4382, "loss": 4.250765323638916, "lr": 0.0002, "elapsed_sec": 36434.21749663353, "step_time_sec": 8.237732556997798, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4383, "loss": 4.05950927734375, "lr": 0.0002, "elapsed_sec": 36442.45700550079, "step_time_sec": 8.239288330078125, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4384, "loss": 4.201176643371582, "lr": 0.0002, "elapsed_sec": 36450.69517493248, "step_time_sec": 8.238031568005681, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4385, "loss": 4.291750431060791, "lr": 0.0002, "elapsed_sec": 36458.93180680275, "step_time_sec": 8.236505965935066, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4386, "loss": 4.320202827453613, "lr": 0.0002, "elapsed_sec": 36467.17202377319, "step_time_sec": 8.239995097974315, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4387, "loss": 4.243524551391602, "lr": 0.0002, "elapsed_sec": 36475.41061544418, "step_time_sec": 8.2384513260331, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4388, "loss": 4.1603899002075195, "lr": 0.0002, "elapsed_sec": 36483.64773392677, "step_time_sec": 8.23695518495515, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4389, "loss": 4.232344150543213, "lr": 0.0002, "elapsed_sec": 36491.887249946594, "step_time_sec": 8.239365739980713, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4390, "loss": 4.19340181350708, "lr": 0.0002, "elapsed_sec": 36500.12553405762, "step_time_sec": 8.238101333961822, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4391, "loss": 4.171411037445068, "lr": 0.0002, "elapsed_sec": 36508.36467409134, "step_time_sec": 8.238984889001586, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4392, "loss": 4.1520490646362305, "lr": 0.0002, "elapsed_sec": 36516.60262274742, "step_time_sec": 8.23785699706059, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4393, "loss": 4.251656532287598, "lr": 0.0002, "elapsed_sec": 36524.84195923805, "step_time_sec": 8.239124491927214, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4394, "loss": 4.226690769195557, "lr": 0.0002, "elapsed_sec": 36533.08004260063, "step_time_sec": 8.237879892927594, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4395, "loss": 4.269175052642822, "lr": 0.0002, "elapsed_sec": 36541.31811261177, "step_time_sec": 8.237897951970808, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4396, "loss": 4.231863021850586, "lr": 0.0002, "elapsed_sec": 36549.55618214607, "step_time_sec": 8.237892196979374, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4397, "loss": 4.396976470947266, "lr": 0.0002, "elapsed_sec": 36557.79529809952, "step_time_sec": 8.238971640006639, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4398, "loss": 4.156890392303467, "lr": 0.0002, "elapsed_sec": 36566.03540587425, "step_time_sec": 8.23995667102281, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4399, "loss": 4.1842875480651855, "lr": 0.0002, "elapsed_sec": 36574.27498579025, "step_time_sec": 8.239405420026742, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4400, "loss": 4.322281837463379, "lr": 0.0002, "elapsed_sec": 36582.513113737106, "step_time_sec": 8.23794897296466, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4401, "loss": 4.238060474395752, "lr": 0.0002, "elapsed_sec": 36590.75246620178, "step_time_sec": 8.239203843986616, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4402, "loss": 4.289977073669434, "lr": 0.0002, "elapsed_sec": 36598.99046134949, "step_time_sec": 8.237806940102018, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4403, "loss": 4.399914264678955, "lr": 0.0002, "elapsed_sec": 36607.22921347618, "step_time_sec": 8.238583478960209, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4404, "loss": 4.159686088562012, "lr": 0.0002, "elapsed_sec": 36615.46979475021, "step_time_sec": 8.240457523963414, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4405, "loss": 4.177679538726807, "lr": 0.0002, "elapsed_sec": 36623.70759057999, "step_time_sec": 8.237642076914199, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4406, "loss": 4.312222480773926, "lr": 0.0002, "elapsed_sec": 36631.94673705101, "step_time_sec": 8.23894103395287, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4407, "loss": 4.2648024559021, "lr": 0.0002, "elapsed_sec": 36640.18481040001, "step_time_sec": 8.23789490701165, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4408, "loss": 4.144003391265869, "lr": 0.0002, "elapsed_sec": 36648.42566370964, "step_time_sec": 8.240696961991489, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4409, "loss": 4.225218772888184, "lr": 0.0002, "elapsed_sec": 36656.66545152664, "step_time_sec": 8.239679254940711, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4410, "loss": 4.215823173522949, "lr": 0.0002, "elapsed_sec": 36664.90404391289, "step_time_sec": 8.238362638978288, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4411, "loss": 4.256727695465088, "lr": 0.0002, "elapsed_sec": 36673.14368343353, "step_time_sec": 8.239516586996615, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4412, "loss": 4.202453136444092, "lr": 0.0002, "elapsed_sec": 36681.38386464119, "step_time_sec": 8.240050824941136, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4413, "loss": 4.1996684074401855, "lr": 0.0002, "elapsed_sec": 36689.62242627144, "step_time_sec": 8.238409177050926, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4414, "loss": 4.2412428855896, "lr": 0.0002, "elapsed_sec": 36697.86129164696, "step_time_sec": 8.238660017028451, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4415, "loss": 4.2433905601501465, "lr": 0.0002, "elapsed_sec": 36706.0996465683, "step_time_sec": 8.238236750010401, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4416, "loss": 4.204154968261719, "lr": 0.0002, "elapsed_sec": 36714.33844828606, "step_time_sec": 8.238615936017595, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4417, "loss": 4.200295448303223, "lr": 0.0002, "elapsed_sec": 36722.57694673538, "step_time_sec": 8.238275986979716, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4418, "loss": 4.119863510131836, "lr": 0.0002, "elapsed_sec": 36730.81606578827, "step_time_sec": 8.239017854910344, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4419, "loss": 4.171496391296387, "lr": 0.0002, "elapsed_sec": 36739.0571501255, "step_time_sec": 8.240919169969857, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4420, "loss": 4.203119277954102, "lr": 0.0002, "elapsed_sec": 36747.319466114044, "step_time_sec": 8.262129348004237, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4421, "loss": 4.116614818572998, "lr": 0.0002, "elapsed_sec": 36755.55992913246, "step_time_sec": 8.240293629001826, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4422, "loss": 4.203859806060791, "lr": 0.0002, "elapsed_sec": 36763.79933953285, "step_time_sec": 8.239249462960288, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4423, "loss": 4.031872749328613, "lr": 0.0002, "elapsed_sec": 36772.03860783577, "step_time_sec": 8.239149569999427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4424, "loss": 4.086124420166016, "lr": 0.0002, "elapsed_sec": 36780.27825164795, "step_time_sec": 8.239367678994313, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4425, "loss": 4.19700288772583, "lr": 0.0002, "elapsed_sec": 36788.518335819244, "step_time_sec": 8.2399636649061, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4426, "loss": 4.429899215698242, "lr": 0.0002, "elapsed_sec": 36796.75882554054, "step_time_sec": 8.240321036078967, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4427, "loss": 4.230428218841553, "lr": 0.0002, "elapsed_sec": 36804.99688696861, "step_time_sec": 8.237849861034192, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4428, "loss": 4.129528522491455, "lr": 0.0002, "elapsed_sec": 36813.23666214943, "step_time_sec": 8.239639673032798, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4429, "loss": 4.320734977722168, "lr": 0.0002, "elapsed_sec": 36821.47609567642, "step_time_sec": 8.23926601104904, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4430, "loss": 4.3243608474731445, "lr": 0.0002, "elapsed_sec": 36829.71519112587, "step_time_sec": 8.238912547007203, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4431, "loss": 4.1354594230651855, "lr": 0.0002, "elapsed_sec": 36837.953224658966, "step_time_sec": 8.237860545981675, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4432, "loss": 4.209555625915527, "lr": 0.0002, "elapsed_sec": 36846.19076657295, "step_time_sec": 8.23742656398099, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4433, "loss": 4.1760783195495605, "lr": 0.0002, "elapsed_sec": 36854.43053126335, "step_time_sec": 8.239612270961516, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4434, "loss": 4.212662696838379, "lr": 0.0002, "elapsed_sec": 36862.66945052147, "step_time_sec": 8.23870122001972, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4435, "loss": 4.171741962432861, "lr": 0.0002, "elapsed_sec": 36870.90836644173, "step_time_sec": 8.23875464394223, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4436, "loss": 4.162563323974609, "lr": 0.0002, "elapsed_sec": 36879.146597862244, "step_time_sec": 8.238110955921002, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4437, "loss": 4.355331897735596, "lr": 0.0002, "elapsed_sec": 36887.38399243355, "step_time_sec": 8.237206736928783, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4438, "loss": 4.215048313140869, "lr": 0.0002, "elapsed_sec": 36895.622176885605, "step_time_sec": 8.238051195046864, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4439, "loss": 4.343101978302002, "lr": 0.0002, "elapsed_sec": 36903.86041331291, "step_time_sec": 8.238038216950372, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4440, "loss": 4.28181266784668, "lr": 0.0002, "elapsed_sec": 36912.100044727325, "step_time_sec": 8.23944593896158, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4441, "loss": 4.19566535949707, "lr": 0.0002, "elapsed_sec": 36920.340098142624, "step_time_sec": 8.239905255031772, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4442, "loss": 4.120377063751221, "lr": 0.0002, "elapsed_sec": 36928.57881522179, "step_time_sec": 8.238558443030342, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4443, "loss": 4.246312141418457, "lr": 0.0002, "elapsed_sec": 36936.81907534599, "step_time_sec": 8.240128093981184, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4444, "loss": 4.221486568450928, "lr": 0.0002, "elapsed_sec": 36945.05673336983, "step_time_sec": 8.237442132085562, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4445, "loss": 4.223291873931885, "lr": 0.0002, "elapsed_sec": 36953.29534006119, "step_time_sec": 8.238464890047908, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4446, "loss": 4.203988552093506, "lr": 0.0002, "elapsed_sec": 36961.533968925476, "step_time_sec": 8.238491429947317, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4447, "loss": 4.261205673217773, "lr": 0.0002, "elapsed_sec": 36969.772236824036, "step_time_sec": 8.23811572091654, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4448, "loss": 4.329805374145508, "lr": 0.0002, "elapsed_sec": 36978.01134610176, "step_time_sec": 8.238891327986494, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4449, "loss": 4.179314613342285, "lr": 0.0002, "elapsed_sec": 36986.24967765808, "step_time_sec": 8.238243788015097, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4450, "loss": 4.166324138641357, "lr": 0.0002, "elapsed_sec": 36994.488604307175, "step_time_sec": 8.238747844006866, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4451, "loss": 4.098905563354492, "lr": 0.0002, "elapsed_sec": 37002.727729320526, "step_time_sec": 8.238925814977847, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4452, "loss": 4.169383525848389, "lr": 0.0002, "elapsed_sec": 37010.96488404274, "step_time_sec": 8.236978985951282, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4453, "loss": 4.14601993560791, "lr": 0.0002, "elapsed_sec": 37019.203777074814, "step_time_sec": 8.238732739002444, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4454, "loss": 4.261724948883057, "lr": 0.0002, "elapsed_sec": 37027.442551374435, "step_time_sec": 8.238618984003551, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4455, "loss": 4.113875389099121, "lr": 0.0002, "elapsed_sec": 37035.681327581406, "step_time_sec": 8.238658495014533, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4456, "loss": 4.137791156768799, "lr": 0.0002, "elapsed_sec": 37043.92009162903, "step_time_sec": 8.23858649295289, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4457, "loss": 4.2236647605896, "lr": 0.0002, "elapsed_sec": 37052.15779209137, "step_time_sec": 8.237484431010671, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4458, "loss": 4.240055561065674, "lr": 0.0002, "elapsed_sec": 37060.395223617554, "step_time_sec": 8.237349225906655, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4459, "loss": 4.141507625579834, "lr": 0.0002, "elapsed_sec": 37068.63449692726, "step_time_sec": 8.239099361002445, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4460, "loss": 4.231062412261963, "lr": 0.0002, "elapsed_sec": 37076.8733420372, "step_time_sec": 8.238706205040216, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4461, "loss": 4.141073226928711, "lr": 0.0002, "elapsed_sec": 37085.11316227913, "step_time_sec": 8.239598631975241, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4462, "loss": 4.166920185089111, "lr": 0.0002, "elapsed_sec": 37093.35017442703, "step_time_sec": 8.236858560005203, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4463, "loss": 4.063992500305176, "lr": 0.0002, "elapsed_sec": 37101.59004998207, "step_time_sec": 8.2397481049411, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4464, "loss": 4.200520038604736, "lr": 0.0002, "elapsed_sec": 37109.828347206116, "step_time_sec": 8.238113002036698, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4465, "loss": 4.122169494628906, "lr": 0.0002, "elapsed_sec": 37118.06774806976, "step_time_sec": 8.239218308008276, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4466, "loss": 4.260088920593262, "lr": 0.0002, "elapsed_sec": 37126.307074546814, "step_time_sec": 8.239219090086408, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4467, "loss": 4.221467018127441, "lr": 0.0002, "elapsed_sec": 37134.54458737373, "step_time_sec": 8.237319741980173, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4468, "loss": 4.0965895652771, "lr": 0.0002, "elapsed_sec": 37142.78304767609, "step_time_sec": 8.238268445944414, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4469, "loss": 4.29210901260376, "lr": 0.0002, "elapsed_sec": 37151.02149963379, "step_time_sec": 8.238284558989108, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4470, "loss": 4.210205554962158, "lr": 0.0002, "elapsed_sec": 37159.26111483574, "step_time_sec": 8.239474091911688, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4471, "loss": 4.112438678741455, "lr": 0.0002, "elapsed_sec": 37167.5010304451, "step_time_sec": 8.239755083923228, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4472, "loss": 4.119547367095947, "lr": 0.0002, "elapsed_sec": 37175.73870110512, "step_time_sec": 8.237526195007376, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4473, "loss": 3.993128538131714, "lr": 0.0002, "elapsed_sec": 37183.977722644806, "step_time_sec": 8.238855987088755, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4474, "loss": 4.203884124755859, "lr": 0.0002, "elapsed_sec": 37192.21583247185, "step_time_sec": 8.237896186998114, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4475, "loss": 4.144719123840332, "lr": 0.0002, "elapsed_sec": 37200.455414533615, "step_time_sec": 8.239451322937384, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4476, "loss": 4.195065975189209, "lr": 0.0002, "elapsed_sec": 37208.69386386871, "step_time_sec": 8.238267223001458, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4477, "loss": 4.1497955322265625, "lr": 0.0002, "elapsed_sec": 37216.92629122734, "step_time_sec": 8.232275619055144, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4478, "loss": 4.27049446105957, "lr": 0.0002, "elapsed_sec": 37225.158759355545, "step_time_sec": 8.232288837898523, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4479, "loss": 4.177828788757324, "lr": 0.0002, "elapsed_sec": 37233.39065694809, "step_time_sec": 8.231758035952225, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4480, "loss": 4.218564987182617, "lr": 0.0002, "elapsed_sec": 37241.628470659256, "step_time_sec": 8.23768886202015, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4481, "loss": 4.257343769073486, "lr": 0.0002, "elapsed_sec": 37249.866379499435, "step_time_sec": 8.237747863051482, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4482, "loss": 4.0376458168029785, "lr": 0.0002, "elapsed_sec": 37258.10363936424, "step_time_sec": 8.237058989005163, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4483, "loss": 3.9884562492370605, "lr": 0.0002, "elapsed_sec": 37266.34055662155, "step_time_sec": 8.236813626019284, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4484, "loss": 4.172069549560547, "lr": 0.0002, "elapsed_sec": 37274.58269619942, "step_time_sec": 8.241941842017695, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4485, "loss": 4.283120155334473, "lr": 0.0002, "elapsed_sec": 37282.82100868225, "step_time_sec": 8.238092961953953, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4486, "loss": 4.186913967132568, "lr": 0.0002, "elapsed_sec": 37291.05942797661, "step_time_sec": 8.238275988958776, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4487, "loss": 4.134679317474365, "lr": 0.0002, "elapsed_sec": 37299.29944014549, "step_time_sec": 8.239814106957056, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4488, "loss": 4.264252662658691, "lr": 0.0002, "elapsed_sec": 37307.536836862564, "step_time_sec": 8.23722756805364, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4489, "loss": 4.162650108337402, "lr": 0.0002, "elapsed_sec": 37315.775958776474, "step_time_sec": 8.23896617803257, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4490, "loss": 4.10653829574585, "lr": 0.0002, "elapsed_sec": 37324.01515221596, "step_time_sec": 8.239052159013227, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4491, "loss": 4.133395195007324, "lr": 0.0002, "elapsed_sec": 37332.252185583115, "step_time_sec": 8.236848797998391, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4492, "loss": 4.158099174499512, "lr": 0.0002, "elapsed_sec": 37340.491438150406, "step_time_sec": 8.239134311908856, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4493, "loss": 4.21279764175415, "lr": 0.0002, "elapsed_sec": 37348.729055166245, "step_time_sec": 8.2374160339823, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4494, "loss": 4.234971523284912, "lr": 0.0002, "elapsed_sec": 37356.96843981743, "step_time_sec": 8.239271593978629, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4495, "loss": 4.263383388519287, "lr": 0.0002, "elapsed_sec": 37365.20843195915, "step_time_sec": 8.239768441068009, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4496, "loss": 4.183688640594482, "lr": 0.0002, "elapsed_sec": 37373.44685816765, "step_time_sec": 8.238257772056386, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4497, "loss": 4.039827346801758, "lr": 0.0002, "elapsed_sec": 37381.686974048615, "step_time_sec": 8.240023448946886, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4498, "loss": 4.192317962646484, "lr": 0.0002, "elapsed_sec": 37389.92565917969, "step_time_sec": 8.23852676700335, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4499, "loss": 4.219346046447754, "lr": 0.0002, "elapsed_sec": 37398.16588616371, "step_time_sec": 8.240018227952532, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4500, "loss": 4.182009696960449, "lr": 0.0002, "elapsed_sec": 37406.40582585335, "step_time_sec": 24.160866376943886, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": true, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4501, "loss": 4.278177738189697, "lr": 0.0002, "elapsed_sec": 37430.57583737373, "step_time_sec": 8.248453418025747, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4502, "loss": 4.146446704864502, "lr": 0.0002, "elapsed_sec": 37438.807658195496, "step_time_sec": 8.231665916973725, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4503, "loss": 4.444203853607178, "lr": 0.0002, "elapsed_sec": 37447.042108774185, "step_time_sec": 8.2343206750229, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4504, "loss": 4.070265769958496, "lr": 0.0002, "elapsed_sec": 37455.27529120445, "step_time_sec": 8.23297414300032, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4505, "loss": 4.290860176086426, "lr": 0.0002, "elapsed_sec": 37463.508420705795, "step_time_sec": 8.232958727981895, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4506, "loss": 4.233179569244385, "lr": 0.0002, "elapsed_sec": 37471.74047398567, "step_time_sec": 8.231947927968577, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4507, "loss": 4.27384090423584, "lr": 0.0002, "elapsed_sec": 37479.978714466095, "step_time_sec": 8.23806988808792, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4508, "loss": 4.026594161987305, "lr": 0.0002, "elapsed_sec": 37488.216829776764, "step_time_sec": 8.23790894402191, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4509, "loss": 4.095904350280762, "lr": 0.0002, "elapsed_sec": 37496.45650219917, "step_time_sec": 8.23957723705098, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4510, "loss": 4.166849136352539, "lr": 0.0002, "elapsed_sec": 37504.69689178467, "step_time_sec": 8.240149782970548, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4511, "loss": 4.136377811431885, "lr": 0.0002, "elapsed_sec": 37512.93402481079, "step_time_sec": 8.236999206012115, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4512, "loss": 4.258389949798584, "lr": 0.0002, "elapsed_sec": 37521.17398238182, "step_time_sec": 8.239736168994568, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4513, "loss": 4.171840190887451, "lr": 0.0002, "elapsed_sec": 37529.41256427765, "step_time_sec": 8.238455303013325, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4514, "loss": 4.185736179351807, "lr": 0.0002, "elapsed_sec": 37537.65083837509, "step_time_sec": 8.238142536953092, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4515, "loss": 4.169240951538086, "lr": 0.0002, "elapsed_sec": 37545.89050459862, "step_time_sec": 8.239488102030009, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4516, "loss": 4.219455242156982, "lr": 0.0002, "elapsed_sec": 37554.128759384155, "step_time_sec": 8.238079670933075, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4517, "loss": 4.189693927764893, "lr": 0.0002, "elapsed_sec": 37562.3679625988, "step_time_sec": 8.239019476925023, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4518, "loss": 4.056425094604492, "lr": 0.0002, "elapsed_sec": 37570.60642504692, "step_time_sec": 8.238315245951526, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4519, "loss": 4.143102645874023, "lr": 0.0002, "elapsed_sec": 37578.84541487694, "step_time_sec": 8.238843113998882, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4520, "loss": 4.254786491394043, "lr": 0.0002, "elapsed_sec": 37587.083360672, "step_time_sec": 8.237722103018314, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4521, "loss": 4.131209373474121, "lr": 0.0002, "elapsed_sec": 37595.321084976196, "step_time_sec": 8.237543110968545, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4522, "loss": 4.136390686035156, "lr": 0.0002, "elapsed_sec": 37603.56126785278, "step_time_sec": 8.24001142301131, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4523, "loss": 4.117629051208496, "lr": 0.0002, "elapsed_sec": 37611.801021814346, "step_time_sec": 8.239661274012178, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4524, "loss": 4.155912399291992, "lr": 0.0002, "elapsed_sec": 37620.041546821594, "step_time_sec": 8.24036909500137, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4525, "loss": 3.954573154449463, "lr": 0.0002, "elapsed_sec": 37628.28084516525, "step_time_sec": 8.23911019496154, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4526, "loss": 4.12672233581543, "lr": 0.0002, "elapsed_sec": 37636.51925086975, "step_time_sec": 8.238279098994099, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4527, "loss": 4.208263397216797, "lr": 0.0002, "elapsed_sec": 37644.759650468826, "step_time_sec": 8.24019081890583, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4528, "loss": 4.035861015319824, "lr": 0.0002, "elapsed_sec": 37652.99759888649, "step_time_sec": 8.237779721966945, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4529, "loss": 3.9528558254241943, "lr": 0.0002, "elapsed_sec": 37661.23749470711, "step_time_sec": 8.239713251008652, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4530, "loss": 4.186199188232422, "lr": 0.0002, "elapsed_sec": 37669.4774646759, "step_time_sec": 8.239802046911791, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4531, "loss": 4.205166339874268, "lr": 0.0002, "elapsed_sec": 37677.71401667595, "step_time_sec": 8.236418743035756, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4532, "loss": 4.254093170166016, "lr": 0.0002, "elapsed_sec": 37685.95274877548, "step_time_sec": 8.238562186947092, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4533, "loss": 4.377471446990967, "lr": 0.0002, "elapsed_sec": 37694.191541433334, "step_time_sec": 8.238625827943906, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4534, "loss": 4.204338073730469, "lr": 0.0002, "elapsed_sec": 37702.4297003746, "step_time_sec": 8.238041052012704, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4535, "loss": 4.159305572509766, "lr": 0.0002, "elapsed_sec": 37710.67022562027, "step_time_sec": 8.240313039976172, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4536, "loss": 3.984607696533203, "lr": 0.0002, "elapsed_sec": 37718.90867304802, "step_time_sec": 8.238285079947673, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4537, "loss": 4.193096160888672, "lr": 0.0002, "elapsed_sec": 37727.14838862419, "step_time_sec": 8.239540240028873, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4538, "loss": 4.117556095123291, "lr": 0.0002, "elapsed_sec": 37735.387221336365, "step_time_sec": 8.238639571936801, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4539, "loss": 4.343083381652832, "lr": 0.0002, "elapsed_sec": 37743.62550783157, "step_time_sec": 8.238123761955649, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4540, "loss": 4.239068508148193, "lr": 0.0002, "elapsed_sec": 37751.863948345184, "step_time_sec": 8.238307669060305, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4541, "loss": 4.100168228149414, "lr": 0.0002, "elapsed_sec": 37760.101793050766, "step_time_sec": 8.237688921042718, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4542, "loss": 4.116884708404541, "lr": 0.0002, "elapsed_sec": 37768.34166145325, "step_time_sec": 8.239703965955414, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4543, "loss": 4.155554294586182, "lr": 0.0002, "elapsed_sec": 37776.581008434296, "step_time_sec": 8.239183994941413, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4544, "loss": 4.148594856262207, "lr": 0.0002, "elapsed_sec": 37784.81962609291, "step_time_sec": 8.23843431507703, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4545, "loss": 4.2612152099609375, "lr": 0.0002, "elapsed_sec": 37793.0579533577, "step_time_sec": 8.238141965004615, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4546, "loss": 4.3667497634887695, "lr": 0.0002, "elapsed_sec": 37801.29450035095, "step_time_sec": 8.236396232037805, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4547, "loss": 4.016406536102295, "lr": 0.0002, "elapsed_sec": 37809.53217339516, "step_time_sec": 8.237495351000689, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4548, "loss": 4.177494525909424, "lr": 0.0002, "elapsed_sec": 37817.77047753334, "step_time_sec": 8.238146179006435, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4549, "loss": 4.044159412384033, "lr": 0.0002, "elapsed_sec": 37826.00970745087, "step_time_sec": 8.239075461984612, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4550, "loss": 4.087081432342529, "lr": 0.0002, "elapsed_sec": 37834.248361587524, "step_time_sec": 8.23850008205045, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4551, "loss": 4.076793193817139, "lr": 0.0002, "elapsed_sec": 37842.48607778549, "step_time_sec": 8.237559082917869, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4552, "loss": 3.804229497909546, "lr": 0.0002, "elapsed_sec": 37850.725821733475, "step_time_sec": 8.239554345025681, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4553, "loss": 4.2554192543029785, "lr": 0.0002, "elapsed_sec": 37858.963690042496, "step_time_sec": 8.237698602955788, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4554, "loss": 3.8925271034240723, "lr": 0.0002, "elapsed_sec": 37867.20273780823, "step_time_sec": 8.238931880914606, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4555, "loss": 4.079890727996826, "lr": 0.0002, "elapsed_sec": 37875.44007945061, "step_time_sec": 8.237201490905136, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4556, "loss": 4.146846771240234, "lr": 0.0002, "elapsed_sec": 37883.677978515625, "step_time_sec": 8.237734992988408, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4557, "loss": 4.2529473304748535, "lr": 0.0002, "elapsed_sec": 37891.9165956974, "step_time_sec": 8.238416469073854, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4558, "loss": 4.172493934631348, "lr": 0.0002, "elapsed_sec": 37900.15496087074, "step_time_sec": 8.238243765081279, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4559, "loss": 4.212314128875732, "lr": 0.0002, "elapsed_sec": 37908.39386701584, "step_time_sec": 8.238739269087091, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4560, "loss": 4.127017498016357, "lr": 0.0002, "elapsed_sec": 37916.63324165344, "step_time_sec": 8.23920456902124, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4561, "loss": 4.2863006591796875, "lr": 0.0002, "elapsed_sec": 37924.870673418045, "step_time_sec": 8.23722387792077, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4562, "loss": 3.9834537506103516, "lr": 0.0002, "elapsed_sec": 37933.11089015007, "step_time_sec": 8.240131658036262, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4563, "loss": 4.2282490730285645, "lr": 0.0002, "elapsed_sec": 37941.3516805172, "step_time_sec": 8.240565613028593, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4564, "loss": 4.138115882873535, "lr": 0.0002, "elapsed_sec": 37949.590013980865, "step_time_sec": 8.23819824506063, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4565, "loss": 3.9945993423461914, "lr": 0.0002, "elapsed_sec": 37957.82883667946, "step_time_sec": 8.238627478014678, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4566, "loss": 4.08632230758667, "lr": 0.0002, "elapsed_sec": 37966.06674838066, "step_time_sec": 8.237762490054592, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4567, "loss": 4.161208629608154, "lr": 0.0002, "elapsed_sec": 37974.30678462982, "step_time_sec": 8.239934182027355, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4568, "loss": 4.223702907562256, "lr": 0.0002, "elapsed_sec": 37982.546322345734, "step_time_sec": 8.239347449038178, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4569, "loss": 4.1165313720703125, "lr": 0.0002, "elapsed_sec": 37990.78558444977, "step_time_sec": 8.239065343979746, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4570, "loss": 4.063323974609375, "lr": 0.0002, "elapsed_sec": 37999.025490522385, "step_time_sec": 8.239767327904701, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4571, "loss": 4.167749881744385, "lr": 0.0002, "elapsed_sec": 38007.263897657394, "step_time_sec": 8.238209739909507, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4572, "loss": 4.223121166229248, "lr": 0.0002, "elapsed_sec": 38015.503381729126, "step_time_sec": 8.239348453003913, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4573, "loss": 4.152512550354004, "lr": 0.0002, "elapsed_sec": 38023.7426469326, "step_time_sec": 8.239142318023369, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4574, "loss": 4.0032639503479, "lr": 0.0002, "elapsed_sec": 38031.98148393631, "step_time_sec": 8.238604141050018, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4575, "loss": 4.115846157073975, "lr": 0.0002, "elapsed_sec": 38040.22101569176, "step_time_sec": 8.239383267937228, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4576, "loss": 4.1774373054504395, "lr": 0.0002, "elapsed_sec": 38048.46009039879, "step_time_sec": 8.238899627001956, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4577, "loss": 4.022233963012695, "lr": 0.0002, "elapsed_sec": 38056.70047545433, "step_time_sec": 8.240257707075216, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4578, "loss": 4.185888290405273, "lr": 0.0002, "elapsed_sec": 38064.94060230255, "step_time_sec": 8.239938559941947, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4579, "loss": 4.261570930480957, "lr": 0.0002, "elapsed_sec": 38073.178492069244, "step_time_sec": 8.237721206038259, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4580, "loss": 4.137690544128418, "lr": 0.0002, "elapsed_sec": 38081.41914272308, "step_time_sec": 8.240479853004217, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4581, "loss": 4.377776145935059, "lr": 0.0002, "elapsed_sec": 38089.658853292465, "step_time_sec": 8.239525421988219, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4582, "loss": 4.214595317840576, "lr": 0.0002, "elapsed_sec": 38097.89791536331, "step_time_sec": 8.238899399060756, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4583, "loss": 4.107464790344238, "lr": 0.0002, "elapsed_sec": 38106.137382507324, "step_time_sec": 8.23929215199314, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4584, "loss": 4.15262508392334, "lr": 0.0002, "elapsed_sec": 38114.374442100525, "step_time_sec": 8.236933114007115, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4585, "loss": 4.2828779220581055, "lr": 0.0002, "elapsed_sec": 38122.61436891556, "step_time_sec": 8.23975878499914, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4586, "loss": 4.149560928344727, "lr": 0.0002, "elapsed_sec": 38130.85270690918, "step_time_sec": 8.238185281981714, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4587, "loss": 4.250027656555176, "lr": 0.0002, "elapsed_sec": 38139.09110379219, "step_time_sec": 8.238235961995088, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4588, "loss": 4.142847537994385, "lr": 0.0002, "elapsed_sec": 38147.33068752289, "step_time_sec": 8.239429254084826, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4589, "loss": 4.053590297698975, "lr": 0.0002, "elapsed_sec": 38155.56848692894, "step_time_sec": 8.237650856026448, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4590, "loss": 4.0577473640441895, "lr": 0.0002, "elapsed_sec": 38163.80761051178, "step_time_sec": 8.238932688022032, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4591, "loss": 3.989232063293457, "lr": 0.0002, "elapsed_sec": 38172.04613637924, "step_time_sec": 8.238367417943664, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4592, "loss": 4.069944858551025, "lr": 0.0002, "elapsed_sec": 38180.28512072563, "step_time_sec": 8.238857351010665, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4593, "loss": 4.189377784729004, "lr": 0.0002, "elapsed_sec": 38188.52427768707, "step_time_sec": 8.238974564010277, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4594, "loss": 4.131505012512207, "lr": 0.0002, "elapsed_sec": 38196.76304316521, "step_time_sec": 8.238639905001037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4595, "loss": 4.0796589851379395, "lr": 0.0002, "elapsed_sec": 38205.002391815186, "step_time_sec": 8.239144438062795, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4596, "loss": 4.0889434814453125, "lr": 0.0002, "elapsed_sec": 38213.241112709045, "step_time_sec": 8.238546523964033, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4597, "loss": 4.1937384605407715, "lr": 0.0002, "elapsed_sec": 38221.4788980484, "step_time_sec": 8.237643448985182, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4598, "loss": 4.232490539550781, "lr": 0.0002, "elapsed_sec": 38229.71681666374, "step_time_sec": 8.237774257082492, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4599, "loss": 4.1483564376831055, "lr": 0.0002, "elapsed_sec": 38237.95454645157, "step_time_sec": 8.23756991699338, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4600, "loss": 4.126089572906494, "lr": 0.0002, "elapsed_sec": 38246.19447302818, "step_time_sec": 8.239753450034186, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4601, "loss": 4.272608757019043, "lr": 0.0002, "elapsed_sec": 38254.43304204941, "step_time_sec": 8.238421767950058, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4602, "loss": 4.08730411529541, "lr": 0.0002, "elapsed_sec": 38262.67053294182, "step_time_sec": 8.237329320982099, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4603, "loss": 3.950087547302246, "lr": 0.0002, "elapsed_sec": 38270.910042762756, "step_time_sec": 8.239315603976138, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4604, "loss": 4.1010942459106445, "lr": 0.0002, "elapsed_sec": 38279.14769363403, "step_time_sec": 8.237478184979409, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4605, "loss": 4.20667028427124, "lr": 0.0002, "elapsed_sec": 38287.38656044006, "step_time_sec": 8.23873097600881, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4606, "loss": 4.032876014709473, "lr": 0.0002, "elapsed_sec": 38295.62629175186, "step_time_sec": 8.239597542094998, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4607, "loss": 4.2115654945373535, "lr": 0.0002, "elapsed_sec": 38303.86536669731, "step_time_sec": 8.238912650034763, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4608, "loss": 4.2118940353393555, "lr": 0.0002, "elapsed_sec": 38312.104841947556, "step_time_sec": 8.239298482076265, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4609, "loss": 4.128664016723633, "lr": 0.0002, "elapsed_sec": 38320.34360361099, "step_time_sec": 8.238594174967147, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4610, "loss": 4.144681930541992, "lr": 0.0002, "elapsed_sec": 38328.58390712738, "step_time_sec": 8.240158454980701, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4611, "loss": 4.125495910644531, "lr": 0.0002, "elapsed_sec": 38336.82266306877, "step_time_sec": 8.238584981998429, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4612, "loss": 4.174413681030273, "lr": 0.0002, "elapsed_sec": 38345.0613360405, "step_time_sec": 8.238479719962925, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4613, "loss": 4.1528825759887695, "lr": 0.0002, "elapsed_sec": 38353.301975011826, "step_time_sec": 8.240466999006458, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4614, "loss": 4.261250019073486, "lr": 0.0002, "elapsed_sec": 38361.542682886124, "step_time_sec": 8.240544043947011, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4615, "loss": 4.1526665687561035, "lr": 0.0002, "elapsed_sec": 38369.78226971626, "step_time_sec": 8.239446350955404, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4616, "loss": 4.187887191772461, "lr": 0.0002, "elapsed_sec": 38378.02167510986, "step_time_sec": 8.239216715912335, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4617, "loss": 4.205005168914795, "lr": 0.0002, "elapsed_sec": 38386.2606985569, "step_time_sec": 8.238947916892357, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4618, "loss": 4.154027462005615, "lr": 0.0002, "elapsed_sec": 38394.49943614006, "step_time_sec": 8.238471660064533, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4619, "loss": 4.1102986335754395, "lr": 0.0002, "elapsed_sec": 38402.73657274246, "step_time_sec": 8.236996869905852, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4620, "loss": 4.181642055511475, "lr": 0.0002, "elapsed_sec": 38410.97396969795, "step_time_sec": 8.237175934016705, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4621, "loss": 4.274555683135986, "lr": 0.0002, "elapsed_sec": 38419.21460771561, "step_time_sec": 8.240490140975453, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4622, "loss": 4.116032123565674, "lr": 0.0002, "elapsed_sec": 38427.45233821869, "step_time_sec": 8.237561783986166, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4623, "loss": 4.200381755828857, "lr": 0.0002, "elapsed_sec": 38435.692638874054, "step_time_sec": 8.240163660026155, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4624, "loss": 3.9470043182373047, "lr": 0.0002, "elapsed_sec": 38443.93286204338, "step_time_sec": 8.240085705998354, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4625, "loss": 4.050996780395508, "lr": 0.0002, "elapsed_sec": 38452.17044520378, "step_time_sec": 8.237348995986395, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4626, "loss": 3.9535224437713623, "lr": 0.0002, "elapsed_sec": 38460.40974569321, "step_time_sec": 8.23919009999372, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4627, "loss": 4.300438404083252, "lr": 0.0002, "elapsed_sec": 38468.6474711895, "step_time_sec": 8.237543444964103, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4628, "loss": 4.265542507171631, "lr": 0.0002, "elapsed_sec": 38476.8870446682, "step_time_sec": 8.239442504942417, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4629, "loss": 4.134401798248291, "lr": 0.0002, "elapsed_sec": 38485.12630486488, "step_time_sec": 8.23903491999954, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4630, "loss": 4.297188758850098, "lr": 0.0002, "elapsed_sec": 38493.36525845528, "step_time_sec": 8.238820691942237, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4631, "loss": 4.244501113891602, "lr": 0.0002, "elapsed_sec": 38501.60401082039, "step_time_sec": 8.238593087997288, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4632, "loss": 4.333240985870361, "lr": 0.0002, "elapsed_sec": 38509.842416763306, "step_time_sec": 8.238227540045045, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4633, "loss": 4.097887992858887, "lr": 0.0002, "elapsed_sec": 38518.08188199997, "step_time_sec": 8.239312662975863, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4634, "loss": 4.286348342895508, "lr": 0.0002, "elapsed_sec": 38526.320907115936, "step_time_sec": 8.23892299795989, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4635, "loss": 4.181149482727051, "lr": 0.0002, "elapsed_sec": 38534.55827283859, "step_time_sec": 8.237134953029454, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4636, "loss": 4.010741233825684, "lr": 0.0002, "elapsed_sec": 38542.79688477516, "step_time_sec": 8.238473134930246, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4637, "loss": 4.140244483947754, "lr": 0.0002, "elapsed_sec": 38551.03421497345, "step_time_sec": 8.23716089001391, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4638, "loss": 4.273528575897217, "lr": 0.0002, "elapsed_sec": 38559.27408862114, "step_time_sec": 8.239685058942996, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4639, "loss": 4.122528553009033, "lr": 0.0002, "elapsed_sec": 38567.512352228165, "step_time_sec": 8.23810774507001, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4640, "loss": 4.143361568450928, "lr": 0.0002, "elapsed_sec": 38575.7506043911, "step_time_sec": 8.23814069002401, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4641, "loss": 4.129969596862793, "lr": 0.0002, "elapsed_sec": 38583.99027132988, "step_time_sec": 8.239461747929454, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4642, "loss": 4.162743091583252, "lr": 0.0002, "elapsed_sec": 38592.22753238678, "step_time_sec": 8.237080838996917, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4643, "loss": 4.195800304412842, "lr": 0.0002, "elapsed_sec": 38600.4671792984, "step_time_sec": 8.239545757998712, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4644, "loss": 4.234490394592285, "lr": 0.0002, "elapsed_sec": 38608.7060046196, "step_time_sec": 8.238681914983317, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4645, "loss": 4.29621696472168, "lr": 0.0002, "elapsed_sec": 38616.9445104599, "step_time_sec": 8.238304055994377, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4646, "loss": 4.243565082550049, "lr": 0.0002, "elapsed_sec": 38625.184101104736, "step_time_sec": 8.239424510044046, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4647, "loss": 4.110744476318359, "lr": 0.0002, "elapsed_sec": 38633.4238550663, "step_time_sec": 8.239582234062254, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4648, "loss": 4.1611528396606445, "lr": 0.0002, "elapsed_sec": 38641.66239333153, "step_time_sec": 8.238390853977762, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4649, "loss": 4.077610015869141, "lr": 0.0002, "elapsed_sec": 38649.9018433094, "step_time_sec": 8.239273731946014, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4650, "loss": 4.094393730163574, "lr": 0.0002, "elapsed_sec": 38658.13982009888, "step_time_sec": 8.237804061966017, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4651, "loss": 3.9902658462524414, "lr": 0.0002, "elapsed_sec": 38666.38029909134, "step_time_sec": 8.24038857605774, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4652, "loss": 4.047671318054199, "lr": 0.0002, "elapsed_sec": 38674.61925530434, "step_time_sec": 8.23872008593753, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4653, "loss": 4.106886386871338, "lr": 0.0002, "elapsed_sec": 38682.85884356499, "step_time_sec": 8.239455361035652, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4654, "loss": 4.025763511657715, "lr": 0.0002, "elapsed_sec": 38691.098665714264, "step_time_sec": 8.239685475011356, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4655, "loss": 4.06766414642334, "lr": 0.0002, "elapsed_sec": 38699.3365149498, "step_time_sec": 8.237682570004836, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4656, "loss": 4.100513935089111, "lr": 0.0002, "elapsed_sec": 38707.57577705383, "step_time_sec": 8.239052272983827, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4657, "loss": 4.0253472328186035, "lr": 0.0002, "elapsed_sec": 38715.81496787071, "step_time_sec": 8.239076521946117, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4658, "loss": 4.143789768218994, "lr": 0.0002, "elapsed_sec": 38724.053945064545, "step_time_sec": 8.238830678979866, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4659, "loss": 4.312814235687256, "lr": 0.0002, "elapsed_sec": 38732.29380130768, "step_time_sec": 8.239652179996483, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4660, "loss": 4.130011081695557, "lr": 0.0002, "elapsed_sec": 38740.53198361397, "step_time_sec": 8.238045012927614, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4661, "loss": 4.260742664337158, "lr": 0.0002, "elapsed_sec": 38748.7725083828, "step_time_sec": 8.240378096932545, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4662, "loss": 4.130903244018555, "lr": 0.0002, "elapsed_sec": 38757.012335538864, "step_time_sec": 8.239616860984825, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4663, "loss": 4.221081256866455, "lr": 0.0002, "elapsed_sec": 38765.250849723816, "step_time_sec": 8.23835220199544, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4664, "loss": 4.067434787750244, "lr": 0.0002, "elapsed_sec": 38773.491612911224, "step_time_sec": 8.240655835019425, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4665, "loss": 4.100475311279297, "lr": 0.0002, "elapsed_sec": 38781.731424331665, "step_time_sec": 8.239604313042946, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4666, "loss": 4.121437072753906, "lr": 0.0002, "elapsed_sec": 38789.971187114716, "step_time_sec": 8.239555436070077, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4667, "loss": 4.013936996459961, "lr": 0.0002, "elapsed_sec": 38798.21056723595, "step_time_sec": 8.239235665067099, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4668, "loss": 4.171989440917969, "lr": 0.0002, "elapsed_sec": 38806.449016332626, "step_time_sec": 8.238309574080631, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4669, "loss": 4.110549449920654, "lr": 0.0002, "elapsed_sec": 38814.68900823593, "step_time_sec": 8.239771626074798, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4670, "loss": 4.102273464202881, "lr": 0.0002, "elapsed_sec": 38822.92964339256, "step_time_sec": 8.240516375983134, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4671, "loss": 3.9988768100738525, "lr": 0.0002, "elapsed_sec": 38831.1675992012, "step_time_sec": 8.237818935071118, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4672, "loss": 4.255617618560791, "lr": 0.0002, "elapsed_sec": 38839.408972263336, "step_time_sec": 8.241164486971684, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4673, "loss": 4.066312789916992, "lr": 0.0002, "elapsed_sec": 38847.64636683464, "step_time_sec": 8.237240461981855, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4674, "loss": 4.100693225860596, "lr": 0.0002, "elapsed_sec": 38855.88589930534, "step_time_sec": 8.23940614797175, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4675, "loss": 4.312236309051514, "lr": 0.0002, "elapsed_sec": 38864.1260137558, "step_time_sec": 8.239922929904424, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4676, "loss": 4.083249568939209, "lr": 0.0002, "elapsed_sec": 38872.36380457878, "step_time_sec": 8.237632066942751, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4677, "loss": 4.2776312828063965, "lr": 0.0002, "elapsed_sec": 38880.60288333893, "step_time_sec": 8.238918544026092, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4678, "loss": 4.02595329284668, "lr": 0.0002, "elapsed_sec": 38888.8415415287, "step_time_sec": 8.23846786096692, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4679, "loss": 4.012117385864258, "lr": 0.0002, "elapsed_sec": 38897.08105993271, "step_time_sec": 8.239394741947763, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4680, "loss": 4.044582843780518, "lr": 0.0002, "elapsed_sec": 38905.32029390335, "step_time_sec": 8.239046138944104, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4681, "loss": 4.24574089050293, "lr": 0.0002, "elapsed_sec": 38913.55916452408, "step_time_sec": 8.238679764908738, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4682, "loss": 4.072991371154785, "lr": 0.0002, "elapsed_sec": 38921.79896569252, "step_time_sec": 8.239719112985767, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4683, "loss": 4.081079483032227, "lr": 0.0002, "elapsed_sec": 38930.038544654846, "step_time_sec": 8.239346242975444, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4684, "loss": 4.038358688354492, "lr": 0.0002, "elapsed_sec": 38938.27804327011, "step_time_sec": 8.239371357951313, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4685, "loss": 3.906062126159668, "lr": 0.0002, "elapsed_sec": 38946.516827106476, "step_time_sec": 8.238659070106223, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4686, "loss": 4.275180339813232, "lr": 0.0002, "elapsed_sec": 38954.74995803833, "step_time_sec": 8.232896807021461, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4687, "loss": 4.161892414093018, "lr": 0.0002, "elapsed_sec": 38962.989132881165, "step_time_sec": 8.239055886981077, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4688, "loss": 4.143716335296631, "lr": 0.0002, "elapsed_sec": 38971.22596335411, "step_time_sec": 8.236685898969881, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4689, "loss": 4.054052352905273, "lr": 0.0002, "elapsed_sec": 38979.465779066086, "step_time_sec": 8.239603010006249, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4690, "loss": 4.199673652648926, "lr": 0.0002, "elapsed_sec": 38987.699191093445, "step_time_sec": 8.23323893302586, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4691, "loss": 4.2078776359558105, "lr": 0.0002, "elapsed_sec": 38995.93222975731, "step_time_sec": 8.23291716305539, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4692, "loss": 4.177518367767334, "lr": 0.0002, "elapsed_sec": 39004.1656498909, "step_time_sec": 8.233252760022879, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4693, "loss": 4.128676414489746, "lr": 0.0002, "elapsed_sec": 39012.404925346375, "step_time_sec": 8.239112220006064, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4694, "loss": 4.176625728607178, "lr": 0.0002, "elapsed_sec": 39020.64409685135, "step_time_sec": 8.239031646982767, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4695, "loss": 4.137102127075195, "lr": 0.0002, "elapsed_sec": 39028.88247227669, "step_time_sec": 8.238185534020886, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4696, "loss": 4.1973114013671875, "lr": 0.0002, "elapsed_sec": 39037.122797966, "step_time_sec": 8.24018259102013, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4697, "loss": 4.149167537689209, "lr": 0.0002, "elapsed_sec": 39045.36143732071, "step_time_sec": 8.23846681707073, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4698, "loss": 4.26624059677124, "lr": 0.0002, "elapsed_sec": 39053.600623607635, "step_time_sec": 8.239032723009586, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4699, "loss": 4.102248668670654, "lr": 0.0002, "elapsed_sec": 39061.83959007263, "step_time_sec": 8.238805397064425, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4700, "loss": 4.117250442504883, "lr": 0.0002, "elapsed_sec": 39070.077905893326, "step_time_sec": 8.238090327009559, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4701, "loss": 4.215727806091309, "lr": 0.0002, "elapsed_sec": 39078.31918478012, "step_time_sec": 8.241144258063287, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4702, "loss": 4.136692047119141, "lr": 0.0002, "elapsed_sec": 39086.557810783386, "step_time_sec": 8.238521838909946, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4703, "loss": 4.0837225914001465, "lr": 0.0002, "elapsed_sec": 39094.79606604576, "step_time_sec": 8.238042440963909, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4704, "loss": 4.220212936401367, "lr": 0.0002, "elapsed_sec": 39103.03516745567, "step_time_sec": 8.238943279022351, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4705, "loss": 4.051661491394043, "lr": 0.0002, "elapsed_sec": 39111.272874593735, "step_time_sec": 8.23757325601764, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4706, "loss": 4.1526994705200195, "lr": 0.0002, "elapsed_sec": 39119.51176071167, "step_time_sec": 8.238685494055972, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4707, "loss": 4.060310363769531, "lr": 0.0002, "elapsed_sec": 39127.75087714195, "step_time_sec": 8.238975814078003, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4708, "loss": 4.093542575836182, "lr": 0.0002, "elapsed_sec": 39135.990943193436, "step_time_sec": 8.239979037083685, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4709, "loss": 4.25946569442749, "lr": 0.0002, "elapsed_sec": 39144.23033666611, "step_time_sec": 8.239183401921764, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4710, "loss": 4.12413215637207, "lr": 0.0002, "elapsed_sec": 39152.46937394142, "step_time_sec": 8.238900972995907, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4711, "loss": 4.152604103088379, "lr": 0.0002, "elapsed_sec": 39160.70953822136, "step_time_sec": 8.240001350990497, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4712, "loss": 4.326601982116699, "lr": 0.0002, "elapsed_sec": 39168.949622154236, "step_time_sec": 8.239893800928257, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4713, "loss": 4.111496925354004, "lr": 0.0002, "elapsed_sec": 39177.18736243248, "step_time_sec": 8.23750737903174, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4714, "loss": 4.177214622497559, "lr": 0.0002, "elapsed_sec": 39185.42773485184, "step_time_sec": 8.240216037956998, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4715, "loss": 4.086289882659912, "lr": 0.0002, "elapsed_sec": 39193.66600394249, "step_time_sec": 8.238163414062, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4716, "loss": 4.01261568069458, "lr": 0.0002, "elapsed_sec": 39201.90519499779, "step_time_sec": 8.238965988042764, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4717, "loss": 4.024293422698975, "lr": 0.0002, "elapsed_sec": 39210.14513206482, "step_time_sec": 8.239764130092226, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4718, "loss": 4.1021728515625, "lr": 0.0002, "elapsed_sec": 39218.38354253769, "step_time_sec": 8.238294602022506, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4719, "loss": 4.161799430847168, "lr": 0.0002, "elapsed_sec": 39226.62375974655, "step_time_sec": 8.24008182506077, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4720, "loss": 4.142172336578369, "lr": 0.0002, "elapsed_sec": 39234.86202764511, "step_time_sec": 8.238060061004944, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4721, "loss": 4.004161357879639, "lr": 0.0002, "elapsed_sec": 39243.09992814064, "step_time_sec": 8.237717484938912, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4722, "loss": 4.126598358154297, "lr": 0.0002, "elapsed_sec": 39251.33887386322, "step_time_sec": 8.23884976201225, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4723, "loss": 4.064079284667969, "lr": 0.0002, "elapsed_sec": 39259.57786679268, "step_time_sec": 8.238795051933266, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4724, "loss": 4.016458511352539, "lr": 0.0002, "elapsed_sec": 39267.816192388535, "step_time_sec": 8.238169874995947, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4725, "loss": 4.140739440917969, "lr": 0.0002, "elapsed_sec": 39276.055314302444, "step_time_sec": 8.238964522024617, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4726, "loss": 4.215104579925537, "lr": 0.0002, "elapsed_sec": 39284.2932240963, "step_time_sec": 8.237789130071178, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4727, "loss": 4.073436737060547, "lr": 0.0002, "elapsed_sec": 39292.532962560654, "step_time_sec": 8.239600404980592, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4728, "loss": 4.153374671936035, "lr": 0.0002, "elapsed_sec": 39300.769501686096, "step_time_sec": 8.236341448966414, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4729, "loss": 4.1798295974731445, "lr": 0.0002, "elapsed_sec": 39309.00947570801, "step_time_sec": 8.239827466080897, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4730, "loss": 4.025701999664307, "lr": 0.0002, "elapsed_sec": 39317.24898838997, "step_time_sec": 8.239348030998372, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4731, "loss": 4.129063129425049, "lr": 0.0002, "elapsed_sec": 39325.48769068718, "step_time_sec": 8.23854711605236, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4732, "loss": 4.127139091491699, "lr": 0.0002, "elapsed_sec": 39333.72606086731, "step_time_sec": 8.238198499078862, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4733, "loss": 4.1742262840271, "lr": 0.0002, "elapsed_sec": 39341.9638004303, "step_time_sec": 8.237582099973224, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4734, "loss": 4.132920742034912, "lr": 0.0002, "elapsed_sec": 39350.20389795303, "step_time_sec": 8.239935911959037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4735, "loss": 4.178291320800781, "lr": 0.0002, "elapsed_sec": 39358.44391131401, "step_time_sec": 8.239890825003386, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4736, "loss": 4.209036827087402, "lr": 0.0002, "elapsed_sec": 39366.683055877686, "step_time_sec": 8.239017508924007, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4737, "loss": 4.15705680847168, "lr": 0.0002, "elapsed_sec": 39374.92447185516, "step_time_sec": 8.241207695915364, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4738, "loss": 4.145126819610596, "lr": 0.0002, "elapsed_sec": 39383.16308379173, "step_time_sec": 8.238450948963873, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4739, "loss": 4.068809509277344, "lr": 0.0002, "elapsed_sec": 39391.40039753914, "step_time_sec": 8.237171254935674, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4740, "loss": 4.416853427886963, "lr": 0.0002, "elapsed_sec": 39399.639397382736, "step_time_sec": 8.238839503959753, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4741, "loss": 4.1415300369262695, "lr": 0.0002, "elapsed_sec": 39407.87723636627, "step_time_sec": 8.237645654007792, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4742, "loss": 4.189897537231445, "lr": 0.0002, "elapsed_sec": 39416.11696600914, "step_time_sec": 8.239596721017733, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4743, "loss": 4.147607803344727, "lr": 0.0002, "elapsed_sec": 39424.35443854332, "step_time_sec": 8.23731025098823, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4744, "loss": 4.144862651824951, "lr": 0.0002, "elapsed_sec": 39432.595610141754, "step_time_sec": 8.24098942510318, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4745, "loss": 4.103348731994629, "lr": 0.0002, "elapsed_sec": 39440.83531522751, "step_time_sec": 8.239553404971957, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4746, "loss": 4.206355571746826, "lr": 0.0002, "elapsed_sec": 39449.073744535446, "step_time_sec": 8.23824308195617, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4747, "loss": 4.221984386444092, "lr": 0.0002, "elapsed_sec": 39457.31284022331, "step_time_sec": 8.238962024915963, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4748, "loss": 4.318146705627441, "lr": 0.0002, "elapsed_sec": 39465.551403045654, "step_time_sec": 8.238404246978462, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4749, "loss": 4.177560806274414, "lr": 0.0002, "elapsed_sec": 39473.79061818123, "step_time_sec": 8.239039306063205, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4750, "loss": 4.017252445220947, "lr": 0.0002, "elapsed_sec": 39482.03124189377, "step_time_sec": 8.240450387005694, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4751, "loss": 4.093981742858887, "lr": 0.0002, "elapsed_sec": 39490.26803064346, "step_time_sec": 8.23662935406901, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4752, "loss": 4.066623210906982, "lr": 0.0002, "elapsed_sec": 39498.50782608986, "step_time_sec": 8.239678896963596, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4753, "loss": 4.1737589836120605, "lr": 0.0002, "elapsed_sec": 39506.746185064316, "step_time_sec": 8.238199413055554, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4754, "loss": 4.144886493682861, "lr": 0.0002, "elapsed_sec": 39514.984657764435, "step_time_sec": 8.238251900998875, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4755, "loss": 4.207074165344238, "lr": 0.0002, "elapsed_sec": 39523.22374296188, "step_time_sec": 8.238953132997267, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4756, "loss": 4.228336334228516, "lr": 0.0002, "elapsed_sec": 39531.462369441986, "step_time_sec": 8.238465858041309, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4757, "loss": 4.2269606590271, "lr": 0.0002, "elapsed_sec": 39539.702476501465, "step_time_sec": 8.239908115938306, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4758, "loss": 4.142115116119385, "lr": 0.0002, "elapsed_sec": 39547.94159960747, "step_time_sec": 8.23897709499579, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4759, "loss": 4.215454578399658, "lr": 0.0002, "elapsed_sec": 39556.18058204651, "step_time_sec": 8.238797478028573, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4760, "loss": 4.0829386711120605, "lr": 0.0002, "elapsed_sec": 39564.419573545456, "step_time_sec": 8.238849696004763, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4761, "loss": 4.074687957763672, "lr": 0.0002, "elapsed_sec": 39572.65791964531, "step_time_sec": 8.238145519047976, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4762, "loss": 4.136775970458984, "lr": 0.0002, "elapsed_sec": 39580.89781045914, "step_time_sec": 8.239755015005358, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4763, "loss": 4.1510162353515625, "lr": 0.0002, "elapsed_sec": 39589.136414289474, "step_time_sec": 8.238496973994188, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4764, "loss": 4.084309101104736, "lr": 0.0002, "elapsed_sec": 39597.37471842766, "step_time_sec": 8.238078227033839, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4765, "loss": 4.237222671508789, "lr": 0.0002, "elapsed_sec": 39605.61293077469, "step_time_sec": 8.238073363085277, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4766, "loss": 4.171922206878662, "lr": 0.0002, "elapsed_sec": 39613.84995555878, "step_time_sec": 8.236899468000047, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4767, "loss": 4.092893123626709, "lr": 0.0002, "elapsed_sec": 39622.0893638134, "step_time_sec": 8.239215527893975, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4768, "loss": 4.302736282348633, "lr": 0.0002, "elapsed_sec": 39630.32838535309, "step_time_sec": 8.238863034057431, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4769, "loss": 4.042953014373779, "lr": 0.0002, "elapsed_sec": 39638.567308187485, "step_time_sec": 8.238809097092599, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4770, "loss": 4.081009864807129, "lr": 0.0002, "elapsed_sec": 39646.80776333809, "step_time_sec": 8.240314599010162, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4771, "loss": 4.111154079437256, "lr": 0.0002, "elapsed_sec": 39655.0451836586, "step_time_sec": 8.237210953957401, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4772, "loss": 4.088853359222412, "lr": 0.0002, "elapsed_sec": 39663.28332352638, "step_time_sec": 8.237979213008657, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4773, "loss": 4.140563011169434, "lr": 0.0002, "elapsed_sec": 39671.52216243744, "step_time_sec": 8.238734747981653, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4774, "loss": 4.040669918060303, "lr": 0.0002, "elapsed_sec": 39679.7605407238, "step_time_sec": 8.23821980599314, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4775, "loss": 4.071056842803955, "lr": 0.0002, "elapsed_sec": 39688.000125169754, "step_time_sec": 8.239367536036298, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4776, "loss": 4.124977111816406, "lr": 0.0002, "elapsed_sec": 39696.23869276047, "step_time_sec": 8.238403782015666, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4777, "loss": 4.137086868286133, "lr": 0.0002, "elapsed_sec": 39704.47814464569, "step_time_sec": 8.239299052977003, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4778, "loss": 4.024371147155762, "lr": 0.0002, "elapsed_sec": 39712.71585536003, "step_time_sec": 8.237554942956194, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4779, "loss": 4.132758617401123, "lr": 0.0002, "elapsed_sec": 39720.95347070694, "step_time_sec": 8.237459865980782, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4780, "loss": 4.0973286628723145, "lr": 0.0002, "elapsed_sec": 39729.192749500275, "step_time_sec": 8.239100483013317, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4781, "loss": 4.2141337394714355, "lr": 0.0002, "elapsed_sec": 39737.43034672737, "step_time_sec": 8.237440712982789, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4782, "loss": 4.171318531036377, "lr": 0.0002, "elapsed_sec": 39745.6707508564, "step_time_sec": 8.240251656039618, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4783, "loss": 4.307850360870361, "lr": 0.0002, "elapsed_sec": 39753.90973162651, "step_time_sec": 8.238864970975555, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4784, "loss": 4.239909648895264, "lr": 0.0002, "elapsed_sec": 39762.14790439606, "step_time_sec": 8.237964855972677, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4785, "loss": 4.06162691116333, "lr": 0.0002, "elapsed_sec": 39770.38767719269, "step_time_sec": 8.239628138020635, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4786, "loss": 4.076796531677246, "lr": 0.0002, "elapsed_sec": 39778.62546348572, "step_time_sec": 8.23763009707909, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4787, "loss": 4.1636857986450195, "lr": 0.0002, "elapsed_sec": 39786.86444783211, "step_time_sec": 8.23884879401885, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4788, "loss": 4.1406168937683105, "lr": 0.0002, "elapsed_sec": 39795.103663921356, "step_time_sec": 8.238995600957423, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4789, "loss": 4.152181148529053, "lr": 0.0002, "elapsed_sec": 39803.34206676483, "step_time_sec": 8.238245096057653, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4790, "loss": 4.021103858947754, "lr": 0.0002, "elapsed_sec": 39811.58041763306, "step_time_sec": 8.238234953023493, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4791, "loss": 4.2251200675964355, "lr": 0.0002, "elapsed_sec": 39819.817080020905, "step_time_sec": 8.236467229900882, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4792, "loss": 4.230233669281006, "lr": 0.0002, "elapsed_sec": 39828.05537867546, "step_time_sec": 8.23814550996758, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4793, "loss": 4.250064373016357, "lr": 0.0002, "elapsed_sec": 39836.294452905655, "step_time_sec": 8.23894312698394, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4794, "loss": 4.190449237823486, "lr": 0.0002, "elapsed_sec": 39844.53332281113, "step_time_sec": 8.238688008044846, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4795, "loss": 4.107606410980225, "lr": 0.0002, "elapsed_sec": 39852.77315402031, "step_time_sec": 8.239610578049906, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4796, "loss": 4.091568470001221, "lr": 0.0002, "elapsed_sec": 39861.010714292526, "step_time_sec": 8.237413077964447, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4797, "loss": 4.272489547729492, "lr": 0.0002, "elapsed_sec": 39869.25052046776, "step_time_sec": 8.239659154089168, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4798, "loss": 4.178097248077393, "lr": 0.0002, "elapsed_sec": 39877.48983645439, "step_time_sec": 8.239138504955918, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4799, "loss": 4.1077117919921875, "lr": 0.0002, "elapsed_sec": 39885.7277264595, "step_time_sec": 8.237750180065632, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4800, "loss": 4.0762619972229, "lr": 0.0002, "elapsed_sec": 39893.96754050255, "step_time_sec": 8.239680749946274, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4801, "loss": 4.183941841125488, "lr": 0.0002, "elapsed_sec": 39902.20518684387, "step_time_sec": 8.237457836046815, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4802, "loss": 4.261552333831787, "lr": 0.0002, "elapsed_sec": 39910.444558382034, "step_time_sec": 8.23921321099624, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4803, "loss": 4.110013484954834, "lr": 0.0002, "elapsed_sec": 39918.68344449997, "step_time_sec": 8.238755678059533, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4804, "loss": 4.155966758728027, "lr": 0.0002, "elapsed_sec": 39926.921696424484, "step_time_sec": 8.238079301081598, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4805, "loss": 4.24713659286499, "lr": 0.0002, "elapsed_sec": 39935.16065144539, "step_time_sec": 8.238744769012555, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4806, "loss": 4.160298824310303, "lr": 0.0002, "elapsed_sec": 39943.39889907837, "step_time_sec": 8.238063666969538, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4807, "loss": 4.057831287384033, "lr": 0.0002, "elapsed_sec": 39951.638916015625, "step_time_sec": 8.239854653947987, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4808, "loss": 4.074643135070801, "lr": 0.0002, "elapsed_sec": 39959.87821769714, "step_time_sec": 8.23913627397269, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4809, "loss": 4.1235480308532715, "lr": 0.0002, "elapsed_sec": 39968.11715078354, "step_time_sec": 8.238790187053382, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4810, "loss": 4.092194080352783, "lr": 0.0002, "elapsed_sec": 39976.35813689232, "step_time_sec": 8.240850626025349, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4811, "loss": 4.253403663635254, "lr": 0.0002, "elapsed_sec": 39984.597618579865, "step_time_sec": 8.23930235602893, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4812, "loss": 4.352220058441162, "lr": 0.0002, "elapsed_sec": 39992.83767604828, "step_time_sec": 8.239853242994286, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4813, "loss": 4.155423641204834, "lr": 0.0002, "elapsed_sec": 40001.07608914375, "step_time_sec": 8.238253801944666, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4814, "loss": 4.179509162902832, "lr": 0.0002, "elapsed_sec": 40009.31436705589, "step_time_sec": 8.238137703971006, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4815, "loss": 4.104786396026611, "lr": 0.0002, "elapsed_sec": 40017.55464363098, "step_time_sec": 8.240116694010794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4816, "loss": 4.167576789855957, "lr": 0.0002, "elapsed_sec": 40025.793061733246, "step_time_sec": 8.23826428304892, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4817, "loss": 4.160863876342773, "lr": 0.0002, "elapsed_sec": 40034.03326368332, "step_time_sec": 8.240068585961126, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4818, "loss": 4.229764938354492, "lr": 0.0002, "elapsed_sec": 40042.274152994156, "step_time_sec": 8.240719706052914, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4819, "loss": 4.079870223999023, "lr": 0.0002, "elapsed_sec": 40050.512018203735, "step_time_sec": 8.237701745005324, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4820, "loss": 4.122951030731201, "lr": 0.0002, "elapsed_sec": 40058.75152802467, "step_time_sec": 8.239364502951503, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4821, "loss": 4.050198554992676, "lr": 0.0002, "elapsed_sec": 40066.990570783615, "step_time_sec": 8.238879117998295, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4822, "loss": 4.2000226974487305, "lr": 0.0002, "elapsed_sec": 40075.228348493576, "step_time_sec": 8.237560170935467, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4823, "loss": 4.156618595123291, "lr": 0.0002, "elapsed_sec": 40083.46704912186, "step_time_sec": 8.238567794091068, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4824, "loss": 4.147660732269287, "lr": 0.0002, "elapsed_sec": 40091.70608186722, "step_time_sec": 8.238893119967543, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4825, "loss": 4.119850158691406, "lr": 0.0002, "elapsed_sec": 40099.94722890854, "step_time_sec": 8.240988361998461, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4826, "loss": 4.130701541900635, "lr": 0.0002, "elapsed_sec": 40108.18709516525, "step_time_sec": 8.23970150295645, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4827, "loss": 4.25770378112793, "lr": 0.0002, "elapsed_sec": 40116.42476797104, "step_time_sec": 8.23756439902354, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4828, "loss": 4.154445648193359, "lr": 0.0002, "elapsed_sec": 40124.66442108154, "step_time_sec": 8.239463525009342, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4829, "loss": 4.099498271942139, "lr": 0.0002, "elapsed_sec": 40132.90287518501, "step_time_sec": 8.23823852406349, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4830, "loss": 4.131127834320068, "lr": 0.0002, "elapsed_sec": 40141.14147353172, "step_time_sec": 8.238445762079209, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4831, "loss": 4.097785472869873, "lr": 0.0002, "elapsed_sec": 40149.380469083786, "step_time_sec": 8.238806849927641, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4832, "loss": 4.15742826461792, "lr": 0.0002, "elapsed_sec": 40157.61874461174, "step_time_sec": 8.23812731495127, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4833, "loss": 4.101774215698242, "lr": 0.0002, "elapsed_sec": 40165.858001708984, "step_time_sec": 8.239073539036326, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4834, "loss": 3.973407030105591, "lr": 0.0002, "elapsed_sec": 40174.09694290161, "step_time_sec": 8.238829332054593, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4835, "loss": 4.117198467254639, "lr": 0.0002, "elapsed_sec": 40182.33641576767, "step_time_sec": 8.239297105930746, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4836, "loss": 4.192171096801758, "lr": 0.0002, "elapsed_sec": 40190.57566046715, "step_time_sec": 8.239076254074462, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4837, "loss": 4.270814418792725, "lr": 0.0002, "elapsed_sec": 40198.81348490715, "step_time_sec": 8.237664104090072, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4838, "loss": 4.137204170227051, "lr": 0.0002, "elapsed_sec": 40207.05242609978, "step_time_sec": 8.238796792109497, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4839, "loss": 4.088387966156006, "lr": 0.0002, "elapsed_sec": 40215.29194664955, "step_time_sec": 8.239365836023353, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4840, "loss": 4.263343334197998, "lr": 0.0002, "elapsed_sec": 40223.52997159958, "step_time_sec": 8.23782068991568, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4841, "loss": 4.092051982879639, "lr": 0.0002, "elapsed_sec": 40231.76966905594, "step_time_sec": 8.239543460891582, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4842, "loss": 4.169785022735596, "lr": 0.0002, "elapsed_sec": 40240.0071849823, "step_time_sec": 8.237361045088619, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4843, "loss": 4.122716903686523, "lr": 0.0002, "elapsed_sec": 40248.24730825424, "step_time_sec": 8.239933940931223, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4844, "loss": 4.0946526527404785, "lr": 0.0002, "elapsed_sec": 40256.48600053787, "step_time_sec": 8.238597288960591, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4845, "loss": 4.158722877502441, "lr": 0.0002, "elapsed_sec": 40264.72401571274, "step_time_sec": 8.237845786032267, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4846, "loss": 4.246831893920898, "lr": 0.0002, "elapsed_sec": 40272.9628367424, "step_time_sec": 8.238616626011208, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4847, "loss": 4.21773624420166, "lr": 0.0002, "elapsed_sec": 40281.201605796814, "step_time_sec": 8.238646381068975, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4848, "loss": 4.186227321624756, "lr": 0.0002, "elapsed_sec": 40289.43993258476, "step_time_sec": 8.238166357041337, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4849, "loss": 4.092898845672607, "lr": 0.0002, "elapsed_sec": 40297.678183078766, "step_time_sec": 8.238008113927208, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4850, "loss": 4.157634258270264, "lr": 0.0002, "elapsed_sec": 40305.917625427246, "step_time_sec": 8.23935102194082, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4851, "loss": 4.162813663482666, "lr": 0.0002, "elapsed_sec": 40314.15739560127, "step_time_sec": 8.23959480097983, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4852, "loss": 4.083255767822266, "lr": 0.0002, "elapsed_sec": 40322.39571595192, "step_time_sec": 8.238164770998992, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4853, "loss": 3.9933786392211914, "lr": 0.0002, "elapsed_sec": 40330.63644909859, "step_time_sec": 8.24061087903101, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4854, "loss": 4.397839546203613, "lr": 0.0002, "elapsed_sec": 40338.876425266266, "step_time_sec": 8.239783272962086, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4855, "loss": 4.294090747833252, "lr": 0.0002, "elapsed_sec": 40347.1174325943, "step_time_sec": 8.240823069936596, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4856, "loss": 4.220835208892822, "lr": 0.0002, "elapsed_sec": 40355.36085939407, "step_time_sec": 8.243262692936696, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4857, "loss": 4.194004058837891, "lr": 0.0002, "elapsed_sec": 40363.60060381889, "step_time_sec": 8.239572049002163, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4858, "loss": 4.121707439422607, "lr": 0.0002, "elapsed_sec": 40371.840017318726, "step_time_sec": 8.239216325921007, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4859, "loss": 4.096020698547363, "lr": 0.0002, "elapsed_sec": 40380.078986644745, "step_time_sec": 8.238776971935295, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4860, "loss": 4.174028396606445, "lr": 0.0002, "elapsed_sec": 40388.31749176979, "step_time_sec": 8.238389678066596, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4861, "loss": 4.215226650238037, "lr": 0.0002, "elapsed_sec": 40396.557005643845, "step_time_sec": 8.239353899960406, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4862, "loss": 4.238656997680664, "lr": 0.0002, "elapsed_sec": 40404.79570770264, "step_time_sec": 8.238524771062657, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4863, "loss": 4.0650835037231445, "lr": 0.0002, "elapsed_sec": 40413.03360795975, "step_time_sec": 8.237718140007928, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4864, "loss": 4.1736249923706055, "lr": 0.0002, "elapsed_sec": 40421.27298808098, "step_time_sec": 8.239225036930293, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4865, "loss": 3.9969446659088135, "lr": 0.0002, "elapsed_sec": 40429.5098259449, "step_time_sec": 8.236696509993635, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4866, "loss": 4.152768135070801, "lr": 0.0002, "elapsed_sec": 40437.74936056137, "step_time_sec": 8.239359945058823, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4867, "loss": 4.187843322753906, "lr": 0.0002, "elapsed_sec": 40457.37921190262, "step_time_sec": 19.62971558305435, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4868, "loss": 4.071170806884766, "lr": 0.0002, "elapsed_sec": 40465.610901117325, "step_time_sec": 8.231510345940478, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4869, "loss": 4.157163619995117, "lr": 0.0002, "elapsed_sec": 40473.84396696091, "step_time_sec": 8.23287235898897, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4870, "loss": 4.152864933013916, "lr": 0.0002, "elapsed_sec": 40482.07509684563, "step_time_sec": 8.230960014974698, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4871, "loss": 4.029043197631836, "lr": 0.0002, "elapsed_sec": 40490.30770254135, "step_time_sec": 8.232458021957427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4872, "loss": 4.084499359130859, "lr": 0.0002, "elapsed_sec": 40498.53972363472, "step_time_sec": 8.231902354047634, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4873, "loss": 4.085766315460205, "lr": 0.0002, "elapsed_sec": 40506.77329874039, "step_time_sec": 8.233436002978124, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4874, "loss": 4.096930027008057, "lr": 0.0002, "elapsed_sec": 40515.013590574265, "step_time_sec": 8.240070412983187, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4875, "loss": 4.182590007781982, "lr": 0.0002, "elapsed_sec": 40523.25260066986, "step_time_sec": 8.23884991905652, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4876, "loss": 4.125461101531982, "lr": 0.0002, "elapsed_sec": 40531.49098134041, "step_time_sec": 8.238183805020526, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4877, "loss": 4.054988861083984, "lr": 0.0002, "elapsed_sec": 40539.730766773224, "step_time_sec": 8.239622116088867, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4878, "loss": 4.085845470428467, "lr": 0.0002, "elapsed_sec": 40547.96874785423, "step_time_sec": 8.23784444492776, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4879, "loss": 3.973134994506836, "lr": 0.0002, "elapsed_sec": 40556.205912828445, "step_time_sec": 8.236997288069688, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4880, "loss": 4.136908054351807, "lr": 0.0002, "elapsed_sec": 40564.44596409798, "step_time_sec": 8.239917653030716, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4881, "loss": 4.255107879638672, "lr": 0.0002, "elapsed_sec": 40572.68359398842, "step_time_sec": 8.23745549004525, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4882, "loss": 4.003203392028809, "lr": 0.0002, "elapsed_sec": 40580.923847436905, "step_time_sec": 8.240027918014675, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4883, "loss": 4.041574478149414, "lr": 0.0002, "elapsed_sec": 40589.16213583946, "step_time_sec": 8.238161831046455, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4884, "loss": 4.003581523895264, "lr": 0.0002, "elapsed_sec": 40597.40056705475, "step_time_sec": 8.238298989948817, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4885, "loss": 4.291770935058594, "lr": 0.0002, "elapsed_sec": 40605.63852930069, "step_time_sec": 8.237757130991668, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4886, "loss": 4.10371732711792, "lr": 0.0002, "elapsed_sec": 40613.876735925674, "step_time_sec": 8.238069572020322, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4887, "loss": 4.126156806945801, "lr": 0.0002, "elapsed_sec": 40622.116220235825, "step_time_sec": 8.239295150036924, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4888, "loss": 4.105935096740723, "lr": 0.0002, "elapsed_sec": 40630.35477375984, "step_time_sec": 8.238402171991765, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4889, "loss": 4.164562225341797, "lr": 0.0002, "elapsed_sec": 40638.59354305267, "step_time_sec": 8.238635273999535, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4890, "loss": 3.9893853664398193, "lr": 0.0002, "elapsed_sec": 40646.83193206787, "step_time_sec": 8.238233062904328, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4891, "loss": 4.162720680236816, "lr": 0.0002, "elapsed_sec": 40655.06965661049, "step_time_sec": 8.237530892016366, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4892, "loss": 4.114948749542236, "lr": 0.0002, "elapsed_sec": 40663.31002449989, "step_time_sec": 8.240183491958305, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4893, "loss": 4.019961833953857, "lr": 0.0002, "elapsed_sec": 40671.54920578003, "step_time_sec": 8.239029043004848, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4894, "loss": 4.130122184753418, "lr": 0.0002, "elapsed_sec": 40679.78833270073, "step_time_sec": 8.2389913800871, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4895, "loss": 4.102042198181152, "lr": 0.0002, "elapsed_sec": 40688.02683019638, "step_time_sec": 8.23837698600255, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4896, "loss": 4.031798362731934, "lr": 0.0002, "elapsed_sec": 40696.26467394829, "step_time_sec": 8.237672788091004, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4897, "loss": 4.209552764892578, "lr": 0.0002, "elapsed_sec": 40704.504950761795, "step_time_sec": 8.240107523975894, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4898, "loss": 4.0513410568237305, "lr": 0.0002, "elapsed_sec": 40712.744848012924, "step_time_sec": 8.23970483802259, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4899, "loss": 4.185558795928955, "lr": 0.0002, "elapsed_sec": 40720.98171329498, "step_time_sec": 8.236670581973158, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4900, "loss": 4.157683372497559, "lr": 0.0002, "elapsed_sec": 40729.22056770325, "step_time_sec": 8.238734898041002, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4901, "loss": 4.214667320251465, "lr": 0.0002, "elapsed_sec": 40737.458522081375, "step_time_sec": 8.237790773971938, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4902, "loss": 4.1318864822387695, "lr": 0.0002, "elapsed_sec": 40745.69676566124, "step_time_sec": 8.238038228009827, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4903, "loss": 4.2968902587890625, "lr": 0.0002, "elapsed_sec": 40753.93587350845, "step_time_sec": 8.238946510944515, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4904, "loss": 4.133375644683838, "lr": 0.0002, "elapsed_sec": 40762.17496943474, "step_time_sec": 8.238898272975348, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4905, "loss": 4.268742561340332, "lr": 0.0002, "elapsed_sec": 40770.41541099548, "step_time_sec": 8.240315970033407, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4906, "loss": 4.171965599060059, "lr": 0.0002, "elapsed_sec": 40778.65302538872, "step_time_sec": 8.237421074998565, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4907, "loss": 4.050008296966553, "lr": 0.0002, "elapsed_sec": 40786.891580581665, "step_time_sec": 8.238427271950059, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4908, "loss": 4.161492824554443, "lr": 0.0002, "elapsed_sec": 40795.13037085533, "step_time_sec": 8.238587275962345, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4909, "loss": 3.9416792392730713, "lr": 0.0002, "elapsed_sec": 40803.36803674698, "step_time_sec": 8.237517275032587, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4910, "loss": 4.021021366119385, "lr": 0.0002, "elapsed_sec": 40811.60710787773, "step_time_sec": 8.23889063403476, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4911, "loss": 4.041302680969238, "lr": 0.0002, "elapsed_sec": 40819.84564328194, "step_time_sec": 8.238408387056552, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4912, "loss": 4.00541353225708, "lr": 0.0002, "elapsed_sec": 40828.08603692055, "step_time_sec": 8.240196125931107, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4913, "loss": 4.163133144378662, "lr": 0.0002, "elapsed_sec": 40836.32634425163, "step_time_sec": 8.240217071026564, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4914, "loss": 4.071058750152588, "lr": 0.0002, "elapsed_sec": 40844.56464624405, "step_time_sec": 8.23808653501328, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4915, "loss": 4.163447856903076, "lr": 0.0002, "elapsed_sec": 40852.80439400673, "step_time_sec": 8.239583645015955, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4916, "loss": 4.07961893081665, "lr": 0.0002, "elapsed_sec": 40861.04167842865, "step_time_sec": 8.237133490969427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4917, "loss": 4.122663497924805, "lr": 0.0002, "elapsed_sec": 40869.280290842056, "step_time_sec": 8.238416114007123, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4918, "loss": 4.170470237731934, "lr": 0.0002, "elapsed_sec": 40877.51971459389, "step_time_sec": 8.239280998008326, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4919, "loss": 4.112881183624268, "lr": 0.0002, "elapsed_sec": 40885.75863599777, "step_time_sec": 8.238719507004134, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4920, "loss": 4.107306003570557, "lr": 0.0002, "elapsed_sec": 40893.99833536148, "step_time_sec": 8.239547144039534, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4921, "loss": 4.020818710327148, "lr": 0.0002, "elapsed_sec": 40902.23608660698, "step_time_sec": 8.237581979017705, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4922, "loss": 4.070854187011719, "lr": 0.0002, "elapsed_sec": 40910.47587656975, "step_time_sec": 8.239653998054564, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4923, "loss": 4.045253753662109, "lr": 0.0002, "elapsed_sec": 40918.71495985985, "step_time_sec": 8.23889048199635, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4924, "loss": 4.005764961242676, "lr": 0.0002, "elapsed_sec": 40926.95311546326, "step_time_sec": 8.238020405056886, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4925, "loss": 4.064863681793213, "lr": 0.0002, "elapsed_sec": 40935.19176077843, "step_time_sec": 8.23842948989477, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4926, "loss": 4.13576078414917, "lr": 0.0002, "elapsed_sec": 40943.43058300018, "step_time_sec": 8.238657571026124, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4927, "loss": 3.9339771270751953, "lr": 0.0002, "elapsed_sec": 40951.670439481735, "step_time_sec": 8.239689337904565, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4928, "loss": 4.142961025238037, "lr": 0.0002, "elapsed_sec": 40959.9098341465, "step_time_sec": 8.239237417001277, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4929, "loss": 4.28857421875, "lr": 0.0002, "elapsed_sec": 40968.14798259735, "step_time_sec": 8.237961348961107, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4930, "loss": 4.00868034362793, "lr": 0.0002, "elapsed_sec": 40976.3878223896, "step_time_sec": 8.239745738916099, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4931, "loss": 4.101114749908447, "lr": 0.0002, "elapsed_sec": 40984.62756276131, "step_time_sec": 8.239592605968937, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4932, "loss": 4.057075500488281, "lr": 0.0002, "elapsed_sec": 40992.86698913574, "step_time_sec": 8.239221827010624, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4933, "loss": 4.059626579284668, "lr": 0.0002, "elapsed_sec": 41001.10610628128, "step_time_sec": 8.238903561024927, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4934, "loss": 4.301113128662109, "lr": 0.0002, "elapsed_sec": 41009.343223810196, "step_time_sec": 8.23697146307677, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4935, "loss": 3.976165771484375, "lr": 0.0002, "elapsed_sec": 41017.58317875862, "step_time_sec": 8.23981322790496, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4936, "loss": 4.134791374206543, "lr": 0.0002, "elapsed_sec": 41025.82026934624, "step_time_sec": 8.236891307984479, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4937, "loss": 4.369020462036133, "lr": 0.0002, "elapsed_sec": 41034.059159994125, "step_time_sec": 8.238780276034959, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4938, "loss": 4.242422580718994, "lr": 0.0002, "elapsed_sec": 41042.297550201416, "step_time_sec": 8.23816871503368, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4939, "loss": 4.244493007659912, "lr": 0.0002, "elapsed_sec": 41050.53599500656, "step_time_sec": 8.238302509998903, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4940, "loss": 4.104984760284424, "lr": 0.0002, "elapsed_sec": 41058.77754664421, "step_time_sec": 8.24143588298466, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4941, "loss": 4.071835041046143, "lr": 0.0002, "elapsed_sec": 41067.01704335213, "step_time_sec": 8.239319882006384, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4942, "loss": 4.080965042114258, "lr": 0.0002, "elapsed_sec": 41075.25614309311, "step_time_sec": 8.238898670999333, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4943, "loss": 4.159749507904053, "lr": 0.0002, "elapsed_sec": 41083.49661421776, "step_time_sec": 8.240315877948888, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4944, "loss": 4.144870281219482, "lr": 0.0002, "elapsed_sec": 41091.73431181908, "step_time_sec": 8.237520494963974, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4945, "loss": 4.10481595993042, "lr": 0.0002, "elapsed_sec": 41099.97431445122, "step_time_sec": 8.23987320996821, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4946, "loss": 4.083935260772705, "lr": 0.0002, "elapsed_sec": 41108.21249628067, "step_time_sec": 8.23801592097152, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4947, "loss": 4.242053985595703, "lr": 0.0002, "elapsed_sec": 41116.45088100433, "step_time_sec": 8.23827725101728, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4948, "loss": 4.078409671783447, "lr": 0.0002, "elapsed_sec": 41124.69108629227, "step_time_sec": 8.239998358068988, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4949, "loss": 4.078609943389893, "lr": 0.0002, "elapsed_sec": 41132.928785800934, "step_time_sec": 8.237567203002982, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4950, "loss": 4.077216625213623, "lr": 0.0002, "elapsed_sec": 41141.16802883148, "step_time_sec": 8.239054000005126, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4951, "loss": 4.059115409851074, "lr": 0.0002, "elapsed_sec": 41149.40781092644, "step_time_sec": 8.239615186932497, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4952, "loss": 4.112122535705566, "lr": 0.0002, "elapsed_sec": 41157.646161317825, "step_time_sec": 8.238255738979205, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4953, "loss": 4.141870498657227, "lr": 0.0002, "elapsed_sec": 41165.88460731506, "step_time_sec": 8.238219500984997, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4954, "loss": 4.02409029006958, "lr": 0.0002, "elapsed_sec": 41174.123014211655, "step_time_sec": 8.23825523396954, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4955, "loss": 4.144510746002197, "lr": 0.0002, "elapsed_sec": 41182.36155343056, "step_time_sec": 8.238374528940767, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4956, "loss": 4.129667282104492, "lr": 0.0002, "elapsed_sec": 41190.60159778595, "step_time_sec": 8.239881492918357, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4957, "loss": 4.117086887359619, "lr": 0.0002, "elapsed_sec": 41198.83896136284, "step_time_sec": 8.237185870995745, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4958, "loss": 4.066431045532227, "lr": 0.0002, "elapsed_sec": 41207.077848911285, "step_time_sec": 8.238740767003037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4959, "loss": 4.1589484214782715, "lr": 0.0002, "elapsed_sec": 41215.31785130501, "step_time_sec": 8.239844401949085, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4960, "loss": 4.069546222686768, "lr": 0.0002, "elapsed_sec": 41223.557361125946, "step_time_sec": 8.239338532905094, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4961, "loss": 3.8735413551330566, "lr": 0.0002, "elapsed_sec": 41231.796292066574, "step_time_sec": 8.23878059303388, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4962, "loss": 4.3230156898498535, "lr": 0.0002, "elapsed_sec": 41240.035800457, "step_time_sec": 8.2393510149559, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4963, "loss": 4.114287376403809, "lr": 0.0002, "elapsed_sec": 41248.27585148811, "step_time_sec": 8.23988415801432, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4964, "loss": 4.214768886566162, "lr": 0.0002, "elapsed_sec": 41256.51601243019, "step_time_sec": 8.240041288081557, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4965, "loss": 4.019986152648926, "lr": 0.0002, "elapsed_sec": 41264.75472831726, "step_time_sec": 8.238530062022619, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4966, "loss": 4.12523889541626, "lr": 0.0002, "elapsed_sec": 41272.9931306839, "step_time_sec": 8.238271953072399, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4967, "loss": 3.779630661010742, "lr": 0.0002, "elapsed_sec": 41281.23181557655, "step_time_sec": 8.238495658966713, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4968, "loss": 4.2662811279296875, "lr": 0.0002, "elapsed_sec": 41289.47114825249, "step_time_sec": 8.239204008947127, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4969, "loss": 4.26709508895874, "lr": 0.0002, "elapsed_sec": 41297.70944952965, "step_time_sec": 8.238136133062653, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4970, "loss": 4.172757148742676, "lr": 0.0002, "elapsed_sec": 41305.94779467583, "step_time_sec": 8.238133111037314, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4971, "loss": 4.010554790496826, "lr": 0.0002, "elapsed_sec": 41314.18607997894, "step_time_sec": 8.238131153048016, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4972, "loss": 4.115826606750488, "lr": 0.0002, "elapsed_sec": 41322.423475027084, "step_time_sec": 8.237262613954954, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4973, "loss": 3.889843225479126, "lr": 0.0002, "elapsed_sec": 41330.66209053993, "step_time_sec": 8.238411582889967, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4974, "loss": 4.14697265625, "lr": 0.0002, "elapsed_sec": 41338.8993139267, "step_time_sec": 8.237078772042878, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4975, "loss": 4.089334011077881, "lr": 0.0002, "elapsed_sec": 41347.13743376732, "step_time_sec": 8.237953797914088, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4976, "loss": 4.070104122161865, "lr": 0.0002, "elapsed_sec": 41355.37764644623, "step_time_sec": 8.240084397024475, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4977, "loss": 4.168833255767822, "lr": 0.0002, "elapsed_sec": 41363.61605453491, "step_time_sec": 8.238196127000265, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4978, "loss": 4.07090425491333, "lr": 0.0002, "elapsed_sec": 41371.856935977936, "step_time_sec": 8.240706709912047, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4979, "loss": 4.094298362731934, "lr": 0.0002, "elapsed_sec": 41380.09639668465, "step_time_sec": 8.239332242053933, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4980, "loss": 4.074748516082764, "lr": 0.0002, "elapsed_sec": 41388.33524823189, "step_time_sec": 8.238671925966628, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4981, "loss": 4.020994663238525, "lr": 0.0002, "elapsed_sec": 41396.639462947845, "step_time_sec": 8.241164661943913, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4982, "loss": 4.097323894500732, "lr": 0.0002, "elapsed_sec": 41404.87808060646, "step_time_sec": 8.23842941597104, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4983, "loss": 4.096621513366699, "lr": 0.0002, "elapsed_sec": 41413.11521077156, "step_time_sec": 8.236972014070489, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4984, "loss": 4.048238754272461, "lr": 0.0002, "elapsed_sec": 41421.354206323624, "step_time_sec": 8.238797605969012, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4985, "loss": 4.11186408996582, "lr": 0.0002, "elapsed_sec": 41429.59345078468, "step_time_sec": 8.239065097062849, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4986, "loss": 4.103038311004639, "lr": 0.0002, "elapsed_sec": 41437.832121133804, "step_time_sec": 8.238531754934229, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4987, "loss": 4.065207481384277, "lr": 0.0002, "elapsed_sec": 41446.07123708725, "step_time_sec": 8.238916043075733, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4988, "loss": 4.189300060272217, "lr": 0.0002, "elapsed_sec": 41454.30893993378, "step_time_sec": 8.237561955931596, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4989, "loss": 3.9479634761810303, "lr": 0.0002, "elapsed_sec": 41462.547902822495, "step_time_sec": 8.238767003058456, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4990, "loss": 4.085784912109375, "lr": 0.0002, "elapsed_sec": 41470.787197589874, "step_time_sec": 8.239173948997632, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4991, "loss": 4.149892330169678, "lr": 0.0002, "elapsed_sec": 41479.02627182007, "step_time_sec": 8.238912006025203, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4992, "loss": 4.021501541137695, "lr": 0.0002, "elapsed_sec": 41487.26467370987, "step_time_sec": 8.238205250934698, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4993, "loss": 4.058623313903809, "lr": 0.0002, "elapsed_sec": 41495.50344777107, "step_time_sec": 8.238668162026443, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4994, "loss": 4.2566142082214355, "lr": 0.0002, "elapsed_sec": 41503.74406456947, "step_time_sec": 8.240408212994225, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4995, "loss": 4.100237846374512, "lr": 0.0002, "elapsed_sec": 41511.984166145325, "step_time_sec": 8.239905674010515, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4996, "loss": 3.9336414337158203, "lr": 0.0002, "elapsed_sec": 41520.22298026085, "step_time_sec": 8.238646275014617, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4997, "loss": 4.028706073760986, "lr": 0.0002, "elapsed_sec": 41528.463502168655, "step_time_sec": 8.240403022966348, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4998, "loss": 4.02070426940918, "lr": 0.0002, "elapsed_sec": 41536.704127788544, "step_time_sec": 8.240452485042624, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4999, "loss": 4.006834506988525, "lr": 0.0002, "elapsed_sec": 41544.94434142113, "step_time_sec": 8.240057446993887, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5000, "loss": 4.0556640625, "lr": 0.0002, "elapsed_sec": 41553.18463110924, "step_time_sec": 50.394446288002655, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": true, "validation_ran": true, "probe_ran": true, "probe_elapsed_sec": 0.977522014058195, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5001, "loss": 4.080062389373779, "lr": 0.0002, "elapsed_sec": 41603.58448600769, "step_time_sec": 8.245397972059436, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5002, "loss": 3.9436299800872803, "lr": 0.0002, "elapsed_sec": 41611.82300424576, "step_time_sec": 8.23836056701839, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5003, "loss": 3.9554855823516846, "lr": 0.0002, "elapsed_sec": 41620.06085419655, "step_time_sec": 8.237689792062156, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5004, "loss": 4.445052146911621, "lr": 0.0002, "elapsed_sec": 41628.29971122742, "step_time_sec": 8.238732076948509, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5005, "loss": 4.036650657653809, "lr": 0.0002, "elapsed_sec": 41636.53867769241, "step_time_sec": 8.238820650032721, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5006, "loss": 4.115616321563721, "lr": 0.0002, "elapsed_sec": 41644.77790951729, "step_time_sec": 8.239054810954258, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5007, "loss": 4.09035062789917, "lr": 0.0002, "elapsed_sec": 41653.018042087555, "step_time_sec": 8.23997794999741, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5008, "loss": 4.188896179199219, "lr": 0.0002, "elapsed_sec": 41661.256319999695, "step_time_sec": 8.238065212033689, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5009, "loss": 3.92336368560791, "lr": 0.0002, "elapsed_sec": 41669.49676680565, "step_time_sec": 8.24028983595781, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5010, "loss": 4.113170146942139, "lr": 0.0002, "elapsed_sec": 41677.73685979843, "step_time_sec": 8.239932374912314, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5011, "loss": 4.0227227210998535, "lr": 0.0002, "elapsed_sec": 41685.97383785248, "step_time_sec": 8.236814246047288, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5012, "loss": 4.134769439697266, "lr": 0.0002, "elapsed_sec": 41694.21278500557, "step_time_sec": 8.238790074014105, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5013, "loss": 4.198122024536133, "lr": 0.0002, "elapsed_sec": 41702.450603961945, "step_time_sec": 8.237679232959636, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5014, "loss": 4.00008487701416, "lr": 0.0002, "elapsed_sec": 41710.69074201584, "step_time_sec": 8.239989707013592, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5015, "loss": 4.205422878265381, "lr": 0.0002, "elapsed_sec": 41718.93117785454, "step_time_sec": 8.240296724019572, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5016, "loss": 4.022375106811523, "lr": 0.0002, "elapsed_sec": 41727.16940784454, "step_time_sec": 8.238004715996794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5017, "loss": 4.2121405601501465, "lr": 0.0002, "elapsed_sec": 41735.40827178955, "step_time_sec": 8.2387287550373, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5018, "loss": 4.070863246917725, "lr": 0.0002, "elapsed_sec": 41743.64722132683, "step_time_sec": 8.238795096985996, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5019, "loss": 4.1110687255859375, "lr": 0.0002, "elapsed_sec": 41751.88612985611, "step_time_sec": 8.238729058066383, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5020, "loss": 4.031642436981201, "lr": 0.0002, "elapsed_sec": 41760.12545657158, "step_time_sec": 8.239141257945448, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5021, "loss": 4.08297872543335, "lr": 0.0002, "elapsed_sec": 41768.364775419235, "step_time_sec": 8.23920133896172, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5022, "loss": 4.155036449432373, "lr": 0.0002, "elapsed_sec": 41776.604007959366, "step_time_sec": 8.239072744967416, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5023, "loss": 4.06643533706665, "lr": 0.0002, "elapsed_sec": 41784.843540906906, "step_time_sec": 8.239363197935745, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5024, "loss": 4.243485450744629, "lr": 0.0002, "elapsed_sec": 41793.08322238922, "step_time_sec": 8.239482783013955, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5025, "loss": 4.044867992401123, "lr": 0.0002, "elapsed_sec": 41801.32175517082, "step_time_sec": 8.238373438012786, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5026, "loss": 4.033174514770508, "lr": 0.0002, "elapsed_sec": 41809.55985426903, "step_time_sec": 8.23795081791468, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5027, "loss": 4.070858001708984, "lr": 0.0002, "elapsed_sec": 41817.79823541641, "step_time_sec": 8.238209806964733, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5028, "loss": 4.057438373565674, "lr": 0.0002, "elapsed_sec": 41826.0373647213, "step_time_sec": 8.238943916046992, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5029, "loss": 3.8384323120117188, "lr": 0.0002, "elapsed_sec": 41834.27757334709, "step_time_sec": 8.240069845109247, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5030, "loss": 4.163053035736084, "lr": 0.0002, "elapsed_sec": 41842.516942977905, "step_time_sec": 8.239209122024477, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5031, "loss": 4.189707279205322, "lr": 0.0002, "elapsed_sec": 41850.754383802414, "step_time_sec": 8.237285192939453, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5032, "loss": 4.143815517425537, "lr": 0.0002, "elapsed_sec": 41858.9934694767, "step_time_sec": 8.238868024083786, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5033, "loss": 3.9822587966918945, "lr": 0.0002, "elapsed_sec": 41867.23269248009, "step_time_sec": 8.239085093024187, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5034, "loss": 4.084002494812012, "lr": 0.0002, "elapsed_sec": 41875.47207021713, "step_time_sec": 8.239214664092287, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5035, "loss": 3.859417676925659, "lr": 0.0002, "elapsed_sec": 41883.712288856506, "step_time_sec": 8.240045677055605, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5036, "loss": 4.057746410369873, "lr": 0.0002, "elapsed_sec": 41891.950372219086, "step_time_sec": 8.237916799960658, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5037, "loss": 4.010528564453125, "lr": 0.0002, "elapsed_sec": 41900.19156932831, "step_time_sec": 8.241045043920167, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5038, "loss": 4.215525150299072, "lr": 0.0002, "elapsed_sec": 41908.43126440048, "step_time_sec": 8.23956407001242, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5039, "loss": 4.072420597076416, "lr": 0.0002, "elapsed_sec": 41916.66918230057, "step_time_sec": 8.237775343004614, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5040, "loss": 4.045332431793213, "lr": 0.0002, "elapsed_sec": 41924.90727877617, "step_time_sec": 8.237878367071971, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5041, "loss": 4.060039520263672, "lr": 0.0002, "elapsed_sec": 41933.14492535591, "step_time_sec": 8.237502194941044, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5042, "loss": 4.107755184173584, "lr": 0.0002, "elapsed_sec": 41941.384823322296, "step_time_sec": 8.23973892396316, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5043, "loss": 4.1024980545043945, "lr": 0.0002, "elapsed_sec": 41949.6240541935, "step_time_sec": 8.239091743016616, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5044, "loss": 4.037164688110352, "lr": 0.0002, "elapsed_sec": 41957.862651109695, "step_time_sec": 8.238411145051941, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5045, "loss": 4.101559162139893, "lr": 0.0002, "elapsed_sec": 41966.10298895836, "step_time_sec": 8.240132248960435, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5046, "loss": 4.122997760772705, "lr": 0.0002, "elapsed_sec": 41974.341935396194, "step_time_sec": 8.238806609995663, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5047, "loss": 4.154376983642578, "lr": 0.0002, "elapsed_sec": 41982.58152484894, "step_time_sec": 8.239420852973126, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5048, "loss": 4.032959938049316, "lr": 0.0002, "elapsed_sec": 41990.82021713257, "step_time_sec": 8.238561747944914, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5049, "loss": 4.164361000061035, "lr": 0.0002, "elapsed_sec": 41999.05860996246, "step_time_sec": 8.238210551906377, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5050, "loss": 4.015750885009766, "lr": 0.0002, "elapsed_sec": 42007.29829454422, "step_time_sec": 8.239513457985595, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5051, "loss": 4.169455051422119, "lr": 0.0002, "elapsed_sec": 42015.537843465805, "step_time_sec": 8.239443322992884, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5052, "loss": 4.125247478485107, "lr": 0.0002, "elapsed_sec": 42023.77701044083, "step_time_sec": 8.239000339061022, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5053, "loss": 4.106062412261963, "lr": 0.0002, "elapsed_sec": 42032.0152015686, "step_time_sec": 8.237968553090468, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5054, "loss": 4.082470417022705, "lr": 0.0002, "elapsed_sec": 42040.25233435631, "step_time_sec": 8.23698833596427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5055, "loss": 3.985858201980591, "lr": 0.0002, "elapsed_sec": 42048.49134397507, "step_time_sec": 8.238874739967287, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5056, "loss": 3.936324119567871, "lr": 0.0002, "elapsed_sec": 42056.73089790344, "step_time_sec": 8.239412627997808, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5057, "loss": 4.054917812347412, "lr": 0.0002, "elapsed_sec": 42064.970000743866, "step_time_sec": 8.238938330090605, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5058, "loss": 4.1032514572143555, "lr": 0.0002, "elapsed_sec": 42073.21059155464, "step_time_sec": 8.240430283010937, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5059, "loss": 3.961907148361206, "lr": 0.0002, "elapsed_sec": 42081.449682712555, "step_time_sec": 8.23890090500936, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5060, "loss": 4.010202884674072, "lr": 0.0002, "elapsed_sec": 42089.6906785965, "step_time_sec": 8.240872093010694, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5061, "loss": 3.9467694759368896, "lr": 0.0002, "elapsed_sec": 42097.93079972267, "step_time_sec": 8.239956552046351, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5062, "loss": 4.070496559143066, "lr": 0.0002, "elapsed_sec": 42106.16861176491, "step_time_sec": 8.237607916002162, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5063, "loss": 4.099966526031494, "lr": 0.0002, "elapsed_sec": 42114.40930390358, "step_time_sec": 8.240505808033049, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5064, "loss": 4.1086835861206055, "lr": 0.0002, "elapsed_sec": 42122.64730763435, "step_time_sec": 8.237864698050544, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5065, "loss": 3.976149082183838, "lr": 0.0002, "elapsed_sec": 42130.88559341431, "step_time_sec": 8.238152986974455, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5066, "loss": 3.8981480598449707, "lr": 0.0002, "elapsed_sec": 42139.12525629997, "step_time_sec": 8.239525812095962, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5067, "loss": 4.111557483673096, "lr": 0.0002, "elapsed_sec": 42147.36389899254, "step_time_sec": 8.23842352302745, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5068, "loss": 4.062744140625, "lr": 0.0002, "elapsed_sec": 42155.60314130783, "step_time_sec": 8.239099947037175, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5069, "loss": 4.211150646209717, "lr": 0.0002, "elapsed_sec": 42163.84036111832, "step_time_sec": 8.23708357498981, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5070, "loss": 4.07243537902832, "lr": 0.0002, "elapsed_sec": 42172.08031272888, "step_time_sec": 8.239749726955779, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5071, "loss": 4.051743507385254, "lr": 0.0002, "elapsed_sec": 42180.31956076622, "step_time_sec": 8.239116101991385, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5072, "loss": 4.0629472732543945, "lr": 0.0002, "elapsed_sec": 42188.55667257309, "step_time_sec": 8.236936124041677, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5073, "loss": 4.061192512512207, "lr": 0.0002, "elapsed_sec": 42196.79554057121, "step_time_sec": 8.238749623997137, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5074, "loss": 3.9828128814697266, "lr": 0.0002, "elapsed_sec": 42205.03373527527, "step_time_sec": 8.237973550101742, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5075, "loss": 4.0780863761901855, "lr": 0.0002, "elapsed_sec": 42213.27225232124, "step_time_sec": 8.238333246903494, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5076, "loss": 4.113883018493652, "lr": 0.0002, "elapsed_sec": 42221.51238155365, "step_time_sec": 8.239962992956862, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5077, "loss": 3.892606496810913, "lr": 0.0002, "elapsed_sec": 42229.74858570099, "step_time_sec": 8.236048287013546, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5078, "loss": 4.173596382141113, "lr": 0.0002, "elapsed_sec": 42237.98713135719, "step_time_sec": 8.238401121925563, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5079, "loss": 4.144282341003418, "lr": 0.0002, "elapsed_sec": 42246.225690841675, "step_time_sec": 8.238358015893027, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5080, "loss": 4.153972148895264, "lr": 0.0002, "elapsed_sec": 42254.465997457504, "step_time_sec": 8.240144210052676, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5081, "loss": 4.124938011169434, "lr": 0.0002, "elapsed_sec": 42262.70426273346, "step_time_sec": 8.238140937988646, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5082, "loss": 4.015096187591553, "lr": 0.0002, "elapsed_sec": 42270.943075180054, "step_time_sec": 8.238656520028599, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5083, "loss": 4.0181450843811035, "lr": 0.0002, "elapsed_sec": 42279.18358826637, "step_time_sec": 8.240326022030786, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5084, "loss": 4.226128101348877, "lr": 0.0002, "elapsed_sec": 42287.421464920044, "step_time_sec": 8.237706581945531, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5085, "loss": 4.06558895111084, "lr": 0.0002, "elapsed_sec": 42295.66147971153, "step_time_sec": 8.239835210959427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5086, "loss": 4.069235324859619, "lr": 0.0002, "elapsed_sec": 42303.90097475052, "step_time_sec": 8.239363768021576, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5087, "loss": 4.073005676269531, "lr": 0.0002, "elapsed_sec": 42312.13924384117, "step_time_sec": 8.238088620011695, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5088, "loss": 4.141012668609619, "lr": 0.0002, "elapsed_sec": 42320.380202293396, "step_time_sec": 8.240825503016822, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5089, "loss": 4.106104373931885, "lr": 0.0002, "elapsed_sec": 42328.619550943375, "step_time_sec": 8.239161561010405, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5090, "loss": 4.130704879760742, "lr": 0.0002, "elapsed_sec": 42336.85870838165, "step_time_sec": 8.239032885991037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5091, "loss": 3.85766863822937, "lr": 0.0002, "elapsed_sec": 42345.098623752594, "step_time_sec": 8.239718046970665, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5092, "loss": 4.034947395324707, "lr": 0.0002, "elapsed_sec": 42353.33706855774, "step_time_sec": 8.238280676072463, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5093, "loss": 4.084237575531006, "lr": 0.0002, "elapsed_sec": 42361.57745027542, "step_time_sec": 8.240226249909028, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5094, "loss": 4.122691631317139, "lr": 0.0002, "elapsed_sec": 42369.816093206406, "step_time_sec": 8.238471593009308, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5095, "loss": 4.08256196975708, "lr": 0.0002, "elapsed_sec": 42378.05390357971, "step_time_sec": 8.237689841073006, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5096, "loss": 3.959321975708008, "lr": 0.0002, "elapsed_sec": 42386.29388356209, "step_time_sec": 8.239817457972094, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5097, "loss": 4.0357255935668945, "lr": 0.0002, "elapsed_sec": 42394.53376197815, "step_time_sec": 8.23964285000693, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5098, "loss": 4.0403642654418945, "lr": 0.0002, "elapsed_sec": 42402.77303504944, "step_time_sec": 8.239174237009138, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5099, "loss": 3.9843780994415283, "lr": 0.0002, "elapsed_sec": 42411.01260972023, "step_time_sec": 8.239402231993154, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5100, "loss": 4.088181972503662, "lr": 0.0002, "elapsed_sec": 42419.2492723465, "step_time_sec": 8.236475712968968, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5101, "loss": 4.127853870391846, "lr": 0.0002, "elapsed_sec": 42427.48761296272, "step_time_sec": 8.238208951079287, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5102, "loss": 4.019680500030518, "lr": 0.0002, "elapsed_sec": 42435.72798013687, "step_time_sec": 8.240231344010681, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5103, "loss": 4.057219982147217, "lr": 0.0002, "elapsed_sec": 42443.9667699337, "step_time_sec": 8.238612574990839, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5104, "loss": 4.144941806793213, "lr": 0.0002, "elapsed_sec": 42452.207134246826, "step_time_sec": 8.24015597102698, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5105, "loss": 4.073936462402344, "lr": 0.0002, "elapsed_sec": 42460.44604349136, "step_time_sec": 8.238754875026643, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5106, "loss": 4.346707344055176, "lr": 0.0002, "elapsed_sec": 42468.68524098396, "step_time_sec": 8.239056588034146, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5107, "loss": 4.038227081298828, "lr": 0.0002, "elapsed_sec": 42476.92342925072, "step_time_sec": 8.238025747006759, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5108, "loss": 3.9353268146514893, "lr": 0.0002, "elapsed_sec": 42485.162294626236, "step_time_sec": 8.238642289070413, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5109, "loss": 4.21855354309082, "lr": 0.0002, "elapsed_sec": 42493.400933265686, "step_time_sec": 8.238502438063733, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5110, "loss": 4.093448638916016, "lr": 0.0002, "elapsed_sec": 42501.63997912407, "step_time_sec": 8.238914065994322, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5111, "loss": 4.03218936920166, "lr": 0.0002, "elapsed_sec": 42509.881536483765, "step_time_sec": 8.241415679920465, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5112, "loss": 4.031274318695068, "lr": 0.0002, "elapsed_sec": 42518.12109065056, "step_time_sec": 8.239339626976289, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5113, "loss": 4.0858378410339355, "lr": 0.0002, "elapsed_sec": 42526.358776807785, "step_time_sec": 8.237584298010916, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5114, "loss": 4.204731464385986, "lr": 0.0002, "elapsed_sec": 42534.59754014015, "step_time_sec": 8.2385304820491, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5115, "loss": 4.174278259277344, "lr": 0.0002, "elapsed_sec": 42542.8353202343, "step_time_sec": 8.237607470015064, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5116, "loss": 3.9126510620117188, "lr": 0.0002, "elapsed_sec": 42551.075340270996, "step_time_sec": 8.23989004897885, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5117, "loss": 3.9743497371673584, "lr": 0.0002, "elapsed_sec": 42559.314994096756, "step_time_sec": 8.239478311967105, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5118, "loss": 4.192272663116455, "lr": 0.0002, "elapsed_sec": 42567.552307128906, "step_time_sec": 8.237109696958214, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5119, "loss": 4.144796848297119, "lr": 0.0002, "elapsed_sec": 42575.79080796242, "step_time_sec": 8.2383451720234, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5120, "loss": 3.988069772720337, "lr": 0.0002, "elapsed_sec": 42584.029598236084, "step_time_sec": 8.238673005020246, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5121, "loss": 4.029634952545166, "lr": 0.0002, "elapsed_sec": 42592.26854920387, "step_time_sec": 8.238744913018309, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5122, "loss": 3.970003128051758, "lr": 0.0002, "elapsed_sec": 42600.507296562195, "step_time_sec": 8.238559618010186, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5123, "loss": 4.140105724334717, "lr": 0.0002, "elapsed_sec": 42608.74553966522, "step_time_sec": 8.238088887068443, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5124, "loss": 4.283989429473877, "lr": 0.0002, "elapsed_sec": 42616.98572230339, "step_time_sec": 8.24003268906381, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5125, "loss": 3.975667953491211, "lr": 0.0002, "elapsed_sec": 42625.22474551201, "step_time_sec": 8.238899920950644, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5126, "loss": 3.982072591781616, "lr": 0.0002, "elapsed_sec": 42633.463883161545, "step_time_sec": 8.238933590939268, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5127, "loss": 4.06020450592041, "lr": 0.0002, "elapsed_sec": 42641.70309758186, "step_time_sec": 8.239041232969612, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5128, "loss": 4.14274263381958, "lr": 0.0002, "elapsed_sec": 42649.940490722656, "step_time_sec": 8.237230891012587, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5129, "loss": 4.056638240814209, "lr": 0.0002, "elapsed_sec": 42658.180344343185, "step_time_sec": 8.239704618928954, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5130, "loss": 4.0962934494018555, "lr": 0.0002, "elapsed_sec": 42666.419100522995, "step_time_sec": 8.238638748996891, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5131, "loss": 4.201206684112549, "lr": 0.0002, "elapsed_sec": 42674.65901136398, "step_time_sec": 8.239694564021192, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5132, "loss": 4.0904083251953125, "lr": 0.0002, "elapsed_sec": 42682.8992664814, "step_time_sec": 8.240091822925024, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5133, "loss": 4.162088394165039, "lr": 0.0002, "elapsed_sec": 42691.1381251812, "step_time_sec": 8.238735366030596, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5134, "loss": 4.089338779449463, "lr": 0.0002, "elapsed_sec": 42699.37823176384, "step_time_sec": 8.239894292899407, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5135, "loss": 4.0951128005981445, "lr": 0.0002, "elapsed_sec": 42707.61757516861, "step_time_sec": 8.23915899603162, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5136, "loss": 4.2199931144714355, "lr": 0.0002, "elapsed_sec": 42715.85592842102, "step_time_sec": 8.238199252053164, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5137, "loss": 4.124657154083252, "lr": 0.0002, "elapsed_sec": 42724.095881700516, "step_time_sec": 8.239835973945446, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5138, "loss": 3.9146697521209717, "lr": 0.0002, "elapsed_sec": 42732.333045482635, "step_time_sec": 8.237024928908795, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5139, "loss": 4.045979976654053, "lr": 0.0002, "elapsed_sec": 42740.57486200333, "step_time_sec": 8.241624696995132, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5140, "loss": 4.101470947265625, "lr": 0.0002, "elapsed_sec": 42748.82799530029, "step_time_sec": 8.252990695997141, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5141, "loss": 4.01272439956665, "lr": 0.0002, "elapsed_sec": 42757.06713104248, "step_time_sec": 8.238974142004736, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5142, "loss": 4.227344989776611, "lr": 0.0002, "elapsed_sec": 42765.30636882782, "step_time_sec": 8.239065833040513, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5143, "loss": 4.160346984863281, "lr": 0.0002, "elapsed_sec": 42773.54723358154, "step_time_sec": 8.2406692589866, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5144, "loss": 4.363208770751953, "lr": 0.0002, "elapsed_sec": 42781.78547167778, "step_time_sec": 8.238107731915079, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5145, "loss": 3.907888412475586, "lr": 0.0002, "elapsed_sec": 42790.02544260025, "step_time_sec": 8.239775035064667, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5146, "loss": 4.074304580688477, "lr": 0.0002, "elapsed_sec": 42798.263966560364, "step_time_sec": 8.238366530044004, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5147, "loss": 3.985077142715454, "lr": 0.0002, "elapsed_sec": 42806.50209569931, "step_time_sec": 8.238019526004791, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5148, "loss": 4.136843681335449, "lr": 0.0002, "elapsed_sec": 42814.74144792557, "step_time_sec": 8.239119916921481, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5149, "loss": 4.196480751037598, "lr": 0.0002, "elapsed_sec": 42822.9800798893, "step_time_sec": 8.238478233921342, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5150, "loss": 4.2599897384643555, "lr": 0.0002, "elapsed_sec": 42831.22020196915, "step_time_sec": 8.239931375952438, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5151, "loss": 4.128847599029541, "lr": 0.0002, "elapsed_sec": 42839.46020793915, "step_time_sec": 8.23986686207354, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5152, "loss": 4.300471782684326, "lr": 0.0002, "elapsed_sec": 42847.69826412201, "step_time_sec": 8.237909534014761, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5153, "loss": 4.225473880767822, "lr": 0.0002, "elapsed_sec": 42855.93886065483, "step_time_sec": 8.240412654005922, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5154, "loss": 3.900116205215454, "lr": 0.0002, "elapsed_sec": 42864.178035497665, "step_time_sec": 8.239044155925512, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5155, "loss": 4.066451549530029, "lr": 0.0002, "elapsed_sec": 42872.41661834717, "step_time_sec": 8.23843761102762, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5156, "loss": 4.153931140899658, "lr": 0.0002, "elapsed_sec": 42880.65619921684, "step_time_sec": 8.239371331990696, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5157, "loss": 4.110726356506348, "lr": 0.0002, "elapsed_sec": 42888.89405012131, "step_time_sec": 8.237677021999843, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5158, "loss": 4.0128912925720215, "lr": 0.0002, "elapsed_sec": 42897.133617162704, "step_time_sec": 8.239357881015167, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5159, "loss": 4.0948381423950195, "lr": 0.0002, "elapsed_sec": 42905.373797893524, "step_time_sec": 8.24006229301449, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5160, "loss": 4.012397289276123, "lr": 0.0002, "elapsed_sec": 42913.61191558838, "step_time_sec": 8.237950679962523, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5161, "loss": 4.097811698913574, "lr": 0.0002, "elapsed_sec": 42921.851053237915, "step_time_sec": 8.238926918013021, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5162, "loss": 4.175906181335449, "lr": 0.0002, "elapsed_sec": 42930.088708639145, "step_time_sec": 8.237500144983642, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5163, "loss": 4.003879547119141, "lr": 0.0002, "elapsed_sec": 42938.327733039856, "step_time_sec": 8.238872151938267, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5164, "loss": 3.93979549407959, "lr": 0.0002, "elapsed_sec": 42946.5666680336, "step_time_sec": 8.238818426034413, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5165, "loss": 4.224118232727051, "lr": 0.0002, "elapsed_sec": 42954.80483388901, "step_time_sec": 8.237954829004593, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5166, "loss": 4.01954984664917, "lr": 0.0002, "elapsed_sec": 42963.04473233223, "step_time_sec": 8.239725823979825, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5167, "loss": 3.9692673683166504, "lr": 0.0002, "elapsed_sec": 42971.28329753876, "step_time_sec": 8.238394650048576, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5168, "loss": 4.056615829467773, "lr": 0.0002, "elapsed_sec": 42979.524225234985, "step_time_sec": 8.240762197063304, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5169, "loss": 4.023412704467773, "lr": 0.0002, "elapsed_sec": 42987.76485824585, "step_time_sec": 8.240485243033618, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5170, "loss": 4.120045185089111, "lr": 0.0002, "elapsed_sec": 42996.00276231766, "step_time_sec": 8.23772578802891, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5171, "loss": 3.9828085899353027, "lr": 0.0002, "elapsed_sec": 43004.2420694828, "step_time_sec": 8.2391725089401, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5172, "loss": 4.099973201751709, "lr": 0.0002, "elapsed_sec": 43012.48131489754, "step_time_sec": 8.239091873983853, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5173, "loss": 3.9982857704162598, "lr": 0.0002, "elapsed_sec": 43020.72104501724, "step_time_sec": 8.239553220104426, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5174, "loss": 4.240679740905762, "lr": 0.0002, "elapsed_sec": 43028.960641384125, "step_time_sec": 8.23940896091517, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5175, "loss": 4.117981433868408, "lr": 0.0002, "elapsed_sec": 43037.20051455498, "step_time_sec": 8.23970069992356, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5176, "loss": 4.182404518127441, "lr": 0.0002, "elapsed_sec": 43045.44076061249, "step_time_sec": 8.240145123098046, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5177, "loss": 4.223796367645264, "lr": 0.0002, "elapsed_sec": 43053.68066287041, "step_time_sec": 8.239707696950063, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5178, "loss": 4.206144332885742, "lr": 0.0002, "elapsed_sec": 43061.918800115585, "step_time_sec": 8.237967488006689, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5179, "loss": 3.9912807941436768, "lr": 0.0002, "elapsed_sec": 43070.158544540405, "step_time_sec": 8.239576902007684, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5180, "loss": 4.23325252532959, "lr": 0.0002, "elapsed_sec": 43078.39823889732, "step_time_sec": 8.23952865996398, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5181, "loss": 4.251077175140381, "lr": 0.0002, "elapsed_sec": 43086.637761592865, "step_time_sec": 8.239411839051172, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5182, "loss": 4.10121488571167, "lr": 0.0002, "elapsed_sec": 43094.87676811218, "step_time_sec": 8.2387895709835, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5183, "loss": 4.077738285064697, "lr": 0.0002, "elapsed_sec": 43103.114253759384, "step_time_sec": 8.237315453006886, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5184, "loss": 4.133181095123291, "lr": 0.0002, "elapsed_sec": 43111.35469985008, "step_time_sec": 8.240295402007177, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5185, "loss": 4.175108909606934, "lr": 0.0002, "elapsed_sec": 43119.59437251091, "step_time_sec": 8.239538998925127, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5186, "loss": 4.1591291427612305, "lr": 0.0002, "elapsed_sec": 43127.83493614197, "step_time_sec": 8.240383981028572, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5187, "loss": 4.08061408996582, "lr": 0.0002, "elapsed_sec": 43136.07547068596, "step_time_sec": 8.240334268077277, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5188, "loss": 4.1205644607543945, "lr": 0.0002, "elapsed_sec": 43144.31394481659, "step_time_sec": 8.2383723850362, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5189, "loss": 4.105857849121094, "lr": 0.0002, "elapsed_sec": 43152.554250240326, "step_time_sec": 8.240130293997936, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5190, "loss": 4.0663981437683105, "lr": 0.0002, "elapsed_sec": 43160.79347586632, "step_time_sec": 8.239063888904639, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5191, "loss": 4.2214837074279785, "lr": 0.0002, "elapsed_sec": 43169.03195619583, "step_time_sec": 8.238278409000486, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5192, "loss": 4.034345626831055, "lr": 0.0002, "elapsed_sec": 43177.271465301514, "step_time_sec": 8.239333149977028, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5193, "loss": 4.279489994049072, "lr": 0.0002, "elapsed_sec": 43185.50956821442, "step_time_sec": 8.23795333201997, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5194, "loss": 4.138822078704834, "lr": 0.0002, "elapsed_sec": 43193.74932527542, "step_time_sec": 8.239588112104684, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5195, "loss": 4.195322513580322, "lr": 0.0002, "elapsed_sec": 43201.989292144775, "step_time_sec": 8.239862181944773, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5196, "loss": 4.062926769256592, "lr": 0.0002, "elapsed_sec": 43210.227437496185, "step_time_sec": 8.237938628997654, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5197, "loss": 4.089693546295166, "lr": 0.0002, "elapsed_sec": 43218.465661525726, "step_time_sec": 8.238118393928744, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5198, "loss": 4.123053073883057, "lr": 0.0002, "elapsed_sec": 43226.703376054764, "step_time_sec": 8.237519141985103, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5199, "loss": 4.085000514984131, "lr": 0.0002, "elapsed_sec": 43234.94261312485, "step_time_sec": 8.239033610094339, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5200, "loss": 4.033352851867676, "lr": 0.0002, "elapsed_sec": 43243.18255996704, "step_time_sec": 8.239797960966825, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5201, "loss": 4.092674255371094, "lr": 0.0002, "elapsed_sec": 43251.420736551285, "step_time_sec": 8.237998239928856, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5202, "loss": 4.108627796173096, "lr": 0.0002, "elapsed_sec": 43259.661099910736, "step_time_sec": 8.240225118002854, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5203, "loss": 4.094766616821289, "lr": 0.0002, "elapsed_sec": 43267.89993715286, "step_time_sec": 8.238644652999938, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5204, "loss": 4.118463516235352, "lr": 0.0002, "elapsed_sec": 43276.13987779617, "step_time_sec": 8.23979384091217, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5205, "loss": 4.087412357330322, "lr": 0.0002, "elapsed_sec": 43284.38025140762, "step_time_sec": 8.24022388504818, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5206, "loss": 3.975250482559204, "lr": 0.0002, "elapsed_sec": 43292.61882734299, "step_time_sec": 8.238408660981804, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5207, "loss": 4.166294574737549, "lr": 0.0002, "elapsed_sec": 43300.85953474045, "step_time_sec": 8.240547113935463, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5208, "loss": 4.031696319580078, "lr": 0.0002, "elapsed_sec": 43309.099433898926, "step_time_sec": 8.239709606976248, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5209, "loss": 4.1815876960754395, "lr": 0.0002, "elapsed_sec": 43317.33738446236, "step_time_sec": 8.237765728030354, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5210, "loss": 4.040833950042725, "lr": 0.0002, "elapsed_sec": 43325.575293540955, "step_time_sec": 8.237745356978849, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5211, "loss": 4.1795759201049805, "lr": 0.0002, "elapsed_sec": 43333.813947439194, "step_time_sec": 8.238548620021902, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5212, "loss": 4.105316162109375, "lr": 0.0002, "elapsed_sec": 43342.054617881775, "step_time_sec": 8.240496719023213, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5213, "loss": 4.0299458503723145, "lr": 0.0002, "elapsed_sec": 43350.293731451035, "step_time_sec": 8.238934252993204, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5214, "loss": 4.020839691162109, "lr": 0.0002, "elapsed_sec": 43358.53127217293, "step_time_sec": 8.237357428995892, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5215, "loss": 4.028293132781982, "lr": 0.0002, "elapsed_sec": 43366.77067089081, "step_time_sec": 8.239302308065817, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5216, "loss": 4.137703895568848, "lr": 0.0002, "elapsed_sec": 43375.00926017761, "step_time_sec": 8.238380632945336, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5217, "loss": 4.037228107452393, "lr": 0.0002, "elapsed_sec": 43383.24961876869, "step_time_sec": 8.24021854007151, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5218, "loss": 4.125363826751709, "lr": 0.0002, "elapsed_sec": 43391.48924303055, "step_time_sec": 8.239412832888775, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5219, "loss": 4.118188858032227, "lr": 0.0002, "elapsed_sec": 43399.72744560242, "step_time_sec": 8.238066461053677, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5220, "loss": 3.9693965911865234, "lr": 0.0002, "elapsed_sec": 43407.967480421066, "step_time_sec": 8.239845306961797, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5221, "loss": 4.092645168304443, "lr": 0.0002, "elapsed_sec": 43416.20609784126, "step_time_sec": 8.238463607034646, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5222, "loss": 4.336493968963623, "lr": 0.0002, "elapsed_sec": 43424.44540929794, "step_time_sec": 8.239177512004972, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5223, "loss": 4.190122127532959, "lr": 0.0002, "elapsed_sec": 43432.684306144714, "step_time_sec": 8.23868876695633, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5224, "loss": 4.084224700927734, "lr": 0.0002, "elapsed_sec": 43440.921216487885, "step_time_sec": 8.236769781098701, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5225, "loss": 4.0655517578125, "lr": 0.0002, "elapsed_sec": 43449.159749507904, "step_time_sec": 8.2383543739561, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5226, "loss": 4.25469970703125, "lr": 0.0002, "elapsed_sec": 43457.399705410004, "step_time_sec": 8.239839436952025, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5227, "loss": 4.133974552154541, "lr": 0.0002, "elapsed_sec": 43465.639212846756, "step_time_sec": 8.239314898033626, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5228, "loss": 4.004455089569092, "lr": 0.0002, "elapsed_sec": 43473.87775230408, "step_time_sec": 8.238365991972387, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5229, "loss": 4.148144245147705, "lr": 0.0002, "elapsed_sec": 43482.11411070824, "step_time_sec": 8.236207703012042, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5230, "loss": 4.0526323318481445, "lr": 0.0002, "elapsed_sec": 43490.35387992859, "step_time_sec": 8.239577161031775, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5231, "loss": 4.055722236633301, "lr": 0.0002, "elapsed_sec": 43498.59305500984, "step_time_sec": 8.239014700055122, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5232, "loss": 3.913745641708374, "lr": 0.0002, "elapsed_sec": 43506.83324122429, "step_time_sec": 8.240063891978934, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5233, "loss": 4.086653709411621, "lr": 0.0002, "elapsed_sec": 43515.07280731201, "step_time_sec": 8.239375168923289, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5234, "loss": 4.078609466552734, "lr": 0.0002, "elapsed_sec": 43523.31146979332, "step_time_sec": 8.238488348084502, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5235, "loss": 4.122514724731445, "lr": 0.0002, "elapsed_sec": 43531.553221940994, "step_time_sec": 8.241580898989923, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5236, "loss": 4.1172308921813965, "lr": 0.0002, "elapsed_sec": 43539.79264497757, "step_time_sec": 8.239321789937094, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5237, "loss": 4.095926761627197, "lr": 0.0002, "elapsed_sec": 43548.03236699104, "step_time_sec": 8.239538128022105, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5238, "loss": 4.139796257019043, "lr": 0.0002, "elapsed_sec": 43556.271280288696, "step_time_sec": 8.238771777017973, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5239, "loss": 4.112030029296875, "lr": 0.0002, "elapsed_sec": 43564.51107263565, "step_time_sec": 8.239670222974382, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5240, "loss": 4.177435874938965, "lr": 0.0002, "elapsed_sec": 43572.75086283684, "step_time_sec": 8.239599370979704, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5241, "loss": 4.062669277191162, "lr": 0.0002, "elapsed_sec": 43580.990622758865, "step_time_sec": 8.239593786071055, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5242, "loss": 4.085263729095459, "lr": 0.0002, "elapsed_sec": 43589.229328632355, "step_time_sec": 8.238555704941973, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5243, "loss": 3.960771083831787, "lr": 0.0002, "elapsed_sec": 43597.46997666359, "step_time_sec": 8.240475450991653, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5244, "loss": 4.037290573120117, "lr": 0.0002, "elapsed_sec": 43605.70920586586, "step_time_sec": 8.239088641945273, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5245, "loss": 4.005635738372803, "lr": 0.0002, "elapsed_sec": 43613.9474298954, "step_time_sec": 8.238042163895443, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5246, "loss": 4.163434982299805, "lr": 0.0002, "elapsed_sec": 43622.18797111511, "step_time_sec": 8.240360322990455, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5247, "loss": 4.165866374969482, "lr": 0.0002, "elapsed_sec": 43630.42720365524, "step_time_sec": 8.239078214974143, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5248, "loss": 4.034741401672363, "lr": 0.0002, "elapsed_sec": 43638.66808056831, "step_time_sec": 8.240757227991708, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5249, "loss": 4.156164646148682, "lr": 0.0002, "elapsed_sec": 43646.90857243538, "step_time_sec": 8.240329816006124, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5250, "loss": 4.038532257080078, "lr": 0.0002, "elapsed_sec": 43655.14683485031, "step_time_sec": 8.238049396080896, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5251, "loss": 4.033046722412109, "lr": 0.0002, "elapsed_sec": 43663.38605427742, "step_time_sec": 8.239063488086686, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5252, "loss": 4.088461399078369, "lr": 0.0002, "elapsed_sec": 43671.6248960495, "step_time_sec": 8.238677185028791, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5253, "loss": 4.1329803466796875, "lr": 0.0002, "elapsed_sec": 43679.86328625679, "step_time_sec": 8.238256894052029, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5254, "loss": 4.081340312957764, "lr": 0.0002, "elapsed_sec": 43688.103697776794, "step_time_sec": 8.240239543025382, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5255, "loss": 4.021458148956299, "lr": 0.0002, "elapsed_sec": 43696.34185433388, "step_time_sec": 8.238035873044282, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5256, "loss": 4.1577887535095215, "lr": 0.0002, "elapsed_sec": 43704.582130908966, "step_time_sec": 8.240045296028256, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5257, "loss": 3.9895708560943604, "lr": 0.0002, "elapsed_sec": 43712.82130050659, "step_time_sec": 8.239027210045606, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5258, "loss": 4.048677921295166, "lr": 0.0002, "elapsed_sec": 43721.059119701385, "step_time_sec": 8.237622206099331, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5259, "loss": 4.1406779289245605, "lr": 0.0002, "elapsed_sec": 43729.297656059265, "step_time_sec": 8.23845560499467, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5260, "loss": 4.034667491912842, "lr": 0.0002, "elapsed_sec": 43737.53542923927, "step_time_sec": 8.23753859498538, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5261, "loss": 4.127345561981201, "lr": 0.0002, "elapsed_sec": 43745.77551007271, "step_time_sec": 8.239939799997956, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5262, "loss": 4.113903999328613, "lr": 0.0002, "elapsed_sec": 43754.013627529144, "step_time_sec": 8.237959482008591, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5263, "loss": 4.10443115234375, "lr": 0.0002, "elapsed_sec": 43762.25141406059, "step_time_sec": 8.23762768600136, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5264, "loss": 4.139532089233398, "lr": 0.0002, "elapsed_sec": 43770.491322517395, "step_time_sec": 8.239738216041587, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5265, "loss": 4.054778575897217, "lr": 0.0002, "elapsed_sec": 43778.729674100876, "step_time_sec": 8.238207219052128, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5266, "loss": 4.076364994049072, "lr": 0.0002, "elapsed_sec": 43786.968160390854, "step_time_sec": 8.238346808007918, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5267, "loss": 4.046377182006836, "lr": 0.0002, "elapsed_sec": 43795.20776176453, "step_time_sec": 8.239419001969509, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5268, "loss": 3.9758126735687256, "lr": 0.0002, "elapsed_sec": 43803.44515156746, "step_time_sec": 8.237190069979988, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5269, "loss": 3.951555013656616, "lr": 0.0002, "elapsed_sec": 43811.685499191284, "step_time_sec": 8.240174905047752, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5270, "loss": 3.995651960372925, "lr": 0.0002, "elapsed_sec": 43819.92366719246, "step_time_sec": 8.238037300063297, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5271, "loss": 4.023589611053467, "lr": 0.0002, "elapsed_sec": 43828.164511203766, "step_time_sec": 8.240664203069173, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5272, "loss": 3.9489967823028564, "lr": 0.0002, "elapsed_sec": 43836.404465675354, "step_time_sec": 8.239833715022542, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5273, "loss": 4.209513187408447, "lr": 0.0002, "elapsed_sec": 43844.640803813934, "step_time_sec": 8.236113295075484, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5274, "loss": 3.9962892532348633, "lr": 0.0002, "elapsed_sec": 43852.880130290985, "step_time_sec": 8.239172484027222, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5275, "loss": 4.1808600425720215, "lr": 0.0002, "elapsed_sec": 43861.11796069145, "step_time_sec": 8.237673338036984, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5276, "loss": 4.208982467651367, "lr": 0.0002, "elapsed_sec": 43869.35830402374, "step_time_sec": 8.240199978929013, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5277, "loss": 4.064759731292725, "lr": 0.0002, "elapsed_sec": 43877.59817695618, "step_time_sec": 8.239707911969163, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5278, "loss": 4.097937107086182, "lr": 0.0002, "elapsed_sec": 43885.83598613739, "step_time_sec": 8.237621778971516, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5279, "loss": 3.9669501781463623, "lr": 0.0002, "elapsed_sec": 43894.075850486755, "step_time_sec": 8.239721646998078, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5280, "loss": 4.181426048278809, "lr": 0.0002, "elapsed_sec": 43902.31517410278, "step_time_sec": 8.239184072008356, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5281, "loss": 4.124014377593994, "lr": 0.0002, "elapsed_sec": 43910.55534386635, "step_time_sec": 8.23999323497992, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5282, "loss": 3.9417104721069336, "lr": 0.0002, "elapsed_sec": 43918.794392585754, "step_time_sec": 8.238910009968095, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5283, "loss": 4.021225452423096, "lr": 0.0002, "elapsed_sec": 43927.033989191055, "step_time_sec": 8.23939248395618, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5284, "loss": 4.092995643615723, "lr": 0.0002, "elapsed_sec": 43935.27547836304, "step_time_sec": 8.24139085097704, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5285, "loss": 4.142068862915039, "lr": 0.0002, "elapsed_sec": 43943.52985882759, "step_time_sec": 8.254246809985489, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5286, "loss": 4.029745578765869, "lr": 0.0002, "elapsed_sec": 43951.77818751335, "step_time_sec": 8.2480799210025, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5287, "loss": 4.232324600219727, "lr": 0.0002, "elapsed_sec": 43960.01591563225, "step_time_sec": 8.237561000045389, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5288, "loss": 3.9758553504943848, "lr": 0.0002, "elapsed_sec": 43968.25542664528, "step_time_sec": 8.239309481927194, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5289, "loss": 4.167261123657227, "lr": 0.0002, "elapsed_sec": 43976.49292016029, "step_time_sec": 8.237389997928403, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5290, "loss": 3.972219705581665, "lr": 0.0002, "elapsed_sec": 43984.731954574585, "step_time_sec": 8.23885409499053, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5291, "loss": 4.117619514465332, "lr": 0.0002, "elapsed_sec": 43992.971398830414, "step_time_sec": 8.239229978993535, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5292, "loss": 4.149857521057129, "lr": 0.0002, "elapsed_sec": 44001.20956683159, "step_time_sec": 8.23803855094593, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5293, "loss": 4.030140399932861, "lr": 0.0002, "elapsed_sec": 44009.44911527634, "step_time_sec": 8.239399824058637, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5294, "loss": 4.071619987487793, "lr": 0.0002, "elapsed_sec": 44017.68937253952, "step_time_sec": 8.24009890097659, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5295, "loss": 4.130918979644775, "lr": 0.0002, "elapsed_sec": 44025.92847895622, "step_time_sec": 8.238954707980156, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5296, "loss": 4.171492576599121, "lr": 0.0002, "elapsed_sec": 44034.16830396652, "step_time_sec": 8.23962169198785, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5297, "loss": 4.083106994628906, "lr": 0.0002, "elapsed_sec": 44042.40593481064, "step_time_sec": 8.237498719012365, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5298, "loss": 4.065842628479004, "lr": 0.0002, "elapsed_sec": 44050.64683032036, "step_time_sec": 8.240715899970382, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5299, "loss": 4.189378261566162, "lr": 0.0002, "elapsed_sec": 44058.88792562485, "step_time_sec": 8.240956310997717, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5300, "loss": 4.206831455230713, "lr": 0.0002, "elapsed_sec": 44067.12692952156, "step_time_sec": 8.238768477924168, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5301, "loss": 4.2605695724487305, "lr": 0.0002, "elapsed_sec": 44075.36720752716, "step_time_sec": 8.240184949943796, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5302, "loss": 4.226275444030762, "lr": 0.0002, "elapsed_sec": 44083.60680627823, "step_time_sec": 8.23942542693112, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5303, "loss": 4.093939781188965, "lr": 0.0002, "elapsed_sec": 44091.84669446945, "step_time_sec": 8.239684492000379, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5304, "loss": 4.0412211418151855, "lr": 0.0002, "elapsed_sec": 44100.085087537766, "step_time_sec": 8.23824437498115, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5305, "loss": 4.10746955871582, "lr": 0.0002, "elapsed_sec": 44108.32253623009, "step_time_sec": 8.237271798076108, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5306, "loss": 4.414701461791992, "lr": 0.0002, "elapsed_sec": 44116.56231236458, "step_time_sec": 8.23966089298483, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5307, "loss": 4.192601680755615, "lr": 0.0002, "elapsed_sec": 44124.801966905594, "step_time_sec": 8.239443155005574, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5308, "loss": 3.932079553604126, "lr": 0.0002, "elapsed_sec": 44133.0399415493, "step_time_sec": 8.237816049018875, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5309, "loss": 4.189888954162598, "lr": 0.0002, "elapsed_sec": 44141.27895927429, "step_time_sec": 8.238850140944123, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5310, "loss": 4.194020748138428, "lr": 0.0002, "elapsed_sec": 44149.51578330994, "step_time_sec": 8.236675065942109, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5311, "loss": 4.031322002410889, "lr": 0.0002, "elapsed_sec": 44157.75460147858, "step_time_sec": 8.238707622047514, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5312, "loss": 4.123570442199707, "lr": 0.0002, "elapsed_sec": 44165.99279284477, "step_time_sec": 8.23802739707753, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5313, "loss": 4.1516499519348145, "lr": 0.0002, "elapsed_sec": 44174.232541799545, "step_time_sec": 8.239545623073354, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5314, "loss": 4.037506103515625, "lr": 0.0002, "elapsed_sec": 44182.4716694355, "step_time_sec": 8.238982715061866, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5315, "loss": 4.0054473876953125, "lr": 0.0002, "elapsed_sec": 44190.71121096611, "step_time_sec": 8.239377329940908, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5316, "loss": 4.115409851074219, "lr": 0.0002, "elapsed_sec": 44198.95124030113, "step_time_sec": 8.239850557991304, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5317, "loss": 4.053591251373291, "lr": 0.0002, "elapsed_sec": 44207.19173026085, "step_time_sec": 8.240319742937572, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5318, "loss": 4.192033290863037, "lr": 0.0002, "elapsed_sec": 44215.43048095703, "step_time_sec": 8.238606888917275, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5319, "loss": 4.136270999908447, "lr": 0.0002, "elapsed_sec": 44223.67059087753, "step_time_sec": 8.239930790034123, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5320, "loss": 3.933671474456787, "lr": 0.0002, "elapsed_sec": 44231.910393714905, "step_time_sec": 8.239633652963676, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5321, "loss": 4.1756911277771, "lr": 0.0002, "elapsed_sec": 44240.150151729584, "step_time_sec": 8.239585297065787, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5322, "loss": 4.097448825836182, "lr": 0.0002, "elapsed_sec": 44248.38921761513, "step_time_sec": 8.238883984973654, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5323, "loss": 4.106073379516602, "lr": 0.0002, "elapsed_sec": 44256.62686085701, "step_time_sec": 8.237531335093081, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5324, "loss": 4.113275527954102, "lr": 0.0002, "elapsed_sec": 44264.8660197258, "step_time_sec": 8.238941813004203, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5325, "loss": 4.03726863861084, "lr": 0.0002, "elapsed_sec": 44273.10611701012, "step_time_sec": 8.239929045084864, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5326, "loss": 3.9439289569854736, "lr": 0.0002, "elapsed_sec": 44281.345186948776, "step_time_sec": 8.238914633053355, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5327, "loss": 4.101567268371582, "lr": 0.0002, "elapsed_sec": 44289.58457374573, "step_time_sec": 8.23927211901173, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5328, "loss": 4.034478187561035, "lr": 0.0002, "elapsed_sec": 44297.82320332527, "step_time_sec": 8.2384580289945, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5329, "loss": 4.2356696128845215, "lr": 0.0002, "elapsed_sec": 44306.06410455704, "step_time_sec": 8.24070106097497, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5330, "loss": 3.8971755504608154, "lr": 0.0002, "elapsed_sec": 44314.30458045006, "step_time_sec": 8.240350781008601, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5331, "loss": 4.124226093292236, "lr": 0.0002, "elapsed_sec": 44322.54325056076, "step_time_sec": 8.238491709926166, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5332, "loss": 4.0053019523620605, "lr": 0.0002, "elapsed_sec": 44330.7832968235, "step_time_sec": 8.239850685931742, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5333, "loss": 4.168674468994141, "lr": 0.0002, "elapsed_sec": 44339.02091932297, "step_time_sec": 8.23746413399931, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5334, "loss": 4.09586238861084, "lr": 0.0002, "elapsed_sec": 44347.26149439812, "step_time_sec": 8.240383828990161, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5335, "loss": 4.070459842681885, "lr": 0.0002, "elapsed_sec": 44355.5037279129, "step_time_sec": 8.242122191004455, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5336, "loss": 4.02841854095459, "lr": 0.0002, "elapsed_sec": 44363.742797613144, "step_time_sec": 8.238903317018412, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5337, "loss": 4.1569294929504395, "lr": 0.0002, "elapsed_sec": 44371.98481750488, "step_time_sec": 8.24182831007056, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5338, "loss": 4.047674655914307, "lr": 0.0002, "elapsed_sec": 44380.22453999519, "step_time_sec": 8.239595826016739, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5339, "loss": 4.083645820617676, "lr": 0.0002, "elapsed_sec": 44388.46342039108, "step_time_sec": 8.238683104049414, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5340, "loss": 3.861229419708252, "lr": 0.0002, "elapsed_sec": 44396.70378136635, "step_time_sec": 8.240233737044036, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5341, "loss": 4.0082855224609375, "lr": 0.0002, "elapsed_sec": 44404.94277000427, "step_time_sec": 8.238775226986036, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5342, "loss": 4.055927276611328, "lr": 0.0002, "elapsed_sec": 44413.18221640587, "step_time_sec": 8.239307343028486, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5343, "loss": 4.0909743309021, "lr": 0.0002, "elapsed_sec": 44421.42185688019, "step_time_sec": 8.239439693978056, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5344, "loss": 4.080451011657715, "lr": 0.0002, "elapsed_sec": 44429.66000413895, "step_time_sec": 8.237986971042119, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5345, "loss": 4.030267238616943, "lr": 0.0002, "elapsed_sec": 44437.899663448334, "step_time_sec": 8.239518829039298, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5346, "loss": 4.058223247528076, "lr": 0.0002, "elapsed_sec": 44446.13962602615, "step_time_sec": 8.239825999946333, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5347, "loss": 3.9865243434906006, "lr": 0.0002, "elapsed_sec": 44454.37883234024, "step_time_sec": 8.23902715195436, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5348, "loss": 3.907322883605957, "lr": 0.0002, "elapsed_sec": 44462.618517160416, "step_time_sec": 8.239511627936736, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5349, "loss": 3.984901189804077, "lr": 0.0002, "elapsed_sec": 44470.857387304306, "step_time_sec": 8.238724340917543, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5350, "loss": 4.023837566375732, "lr": 0.0002, "elapsed_sec": 44479.09709882736, "step_time_sec": 8.239559283014387, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5351, "loss": 4.04196834564209, "lr": 0.0002, "elapsed_sec": 44487.33795142174, "step_time_sec": 8.24066790600773, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5352, "loss": 3.9772825241088867, "lr": 0.0002, "elapsed_sec": 44495.575091362, "step_time_sec": 8.237011813907884, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5353, "loss": 4.070279598236084, "lr": 0.0002, "elapsed_sec": 44503.815697193146, "step_time_sec": 8.240428789984435, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5354, "loss": 4.129555702209473, "lr": 0.0002, "elapsed_sec": 44512.05484557152, "step_time_sec": 8.238970475969836, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5355, "loss": 4.118356227874756, "lr": 0.0002, "elapsed_sec": 44520.29457259178, "step_time_sec": 8.239591943915002, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5356, "loss": 3.9620304107666016, "lr": 0.0002, "elapsed_sec": 44528.53435730934, "step_time_sec": 8.239616150967777, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5357, "loss": 4.04769229888916, "lr": 0.0002, "elapsed_sec": 44536.771730184555, "step_time_sec": 8.237246342003345, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5358, "loss": 3.805560350418091, "lr": 0.0002, "elapsed_sec": 44545.01073861122, "step_time_sec": 8.238820704980753, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5359, "loss": 4.074831008911133, "lr": 0.0002, "elapsed_sec": 44553.25018572807, "step_time_sec": 8.239260431961156, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5360, "loss": 4.139689922332764, "lr": 0.0002, "elapsed_sec": 44561.48987054825, "step_time_sec": 8.239532189094462, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5361, "loss": 4.08780574798584, "lr": 0.0002, "elapsed_sec": 44569.73026943207, "step_time_sec": 8.240230755065568, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5362, "loss": 4.109202861785889, "lr": 0.0002, "elapsed_sec": 44577.96964049339, "step_time_sec": 8.23922655894421, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5363, "loss": 4.0221123695373535, "lr": 0.0002, "elapsed_sec": 44586.21042966843, "step_time_sec": 8.24057780194562, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5364, "loss": 3.8566324710845947, "lr": 0.0002, "elapsed_sec": 44594.449893713, "step_time_sec": 8.239299301058054, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5365, "loss": 4.108381748199463, "lr": 0.0002, "elapsed_sec": 44602.68819332123, "step_time_sec": 8.238212001044303, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5366, "loss": 4.031296253204346, "lr": 0.0002, "elapsed_sec": 44610.92750668526, "step_time_sec": 8.239124980987981, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5367, "loss": 4.121464252471924, "lr": 0.0002, "elapsed_sec": 44619.166283369064, "step_time_sec": 8.238563040969893, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5368, "loss": 4.133921146392822, "lr": 0.0002, "elapsed_sec": 44627.404837846756, "step_time_sec": 8.238391957012936, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5369, "loss": 4.00117826461792, "lr": 0.0002, "elapsed_sec": 44635.64403247833, "step_time_sec": 8.23907352797687, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5370, "loss": 4.079786777496338, "lr": 0.0002, "elapsed_sec": 44643.88164997101, "step_time_sec": 8.237430828041397, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5371, "loss": 3.883486032485962, "lr": 0.0002, "elapsed_sec": 44652.12153816223, "step_time_sec": 8.23971324309241, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5372, "loss": 4.148911952972412, "lr": 0.0002, "elapsed_sec": 44660.36095905304, "step_time_sec": 8.23925813799724, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5373, "loss": 3.991183042526245, "lr": 0.0002, "elapsed_sec": 44668.60080528259, "step_time_sec": 8.239685332053341, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5374, "loss": 4.101474285125732, "lr": 0.0002, "elapsed_sec": 44676.84007024765, "step_time_sec": 8.239167295978405, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5375, "loss": 3.957733154296875, "lr": 0.0002, "elapsed_sec": 44685.078129053116, "step_time_sec": 8.237825624994002, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5376, "loss": 3.9885165691375732, "lr": 0.0002, "elapsed_sec": 44693.31989169121, "step_time_sec": 8.241618764935993, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5377, "loss": 4.005331516265869, "lr": 0.0002, "elapsed_sec": 44701.55995965004, "step_time_sec": 8.239897851948626, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5378, "loss": 3.987309694290161, "lr": 0.0002, "elapsed_sec": 44709.799332380295, "step_time_sec": 8.239202655036934, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5379, "loss": 4.165535926818848, "lr": 0.0002, "elapsed_sec": 44718.04179286957, "step_time_sec": 8.242337297997437, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5380, "loss": 4.1523051261901855, "lr": 0.0002, "elapsed_sec": 44726.28136634827, "step_time_sec": 8.239375230041333, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5381, "loss": 4.18119478225708, "lr": 0.0002, "elapsed_sec": 44734.51945781708, "step_time_sec": 8.237920225947164, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5382, "loss": 4.155643463134766, "lr": 0.0002, "elapsed_sec": 44742.75771379471, "step_time_sec": 8.238104294985533, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5383, "loss": 3.970078468322754, "lr": 0.0002, "elapsed_sec": 44750.99625277519, "step_time_sec": 8.238391375984065, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5384, "loss": 4.08048152923584, "lr": 0.0002, "elapsed_sec": 44759.2350461483, "step_time_sec": 8.23857341404073, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5385, "loss": 4.110383033752441, "lr": 0.0002, "elapsed_sec": 44767.473846673965, "step_time_sec": 8.238696484011598, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5386, "loss": 3.9800519943237305, "lr": 0.0002, "elapsed_sec": 44775.71282982826, "step_time_sec": 8.238820706028491, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5387, "loss": 3.9467711448669434, "lr": 0.0002, "elapsed_sec": 44783.95269846916, "step_time_sec": 8.239730024011806, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5388, "loss": 3.9355363845825195, "lr": 0.0002, "elapsed_sec": 44792.18991136551, "step_time_sec": 8.237002374022268, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5389, "loss": 4.026029586791992, "lr": 0.0002, "elapsed_sec": 44800.42937755585, "step_time_sec": 8.23931751807686, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5390, "loss": 3.9859137535095215, "lr": 0.0002, "elapsed_sec": 44808.66692829132, "step_time_sec": 8.237408132990822, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5391, "loss": 3.963496446609497, "lr": 0.0002, "elapsed_sec": 44816.90617775917, "step_time_sec": 8.239106449997053, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5392, "loss": 3.9724576473236084, "lr": 0.0002, "elapsed_sec": 44825.14752101898, "step_time_sec": 8.241133474046364, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5393, "loss": 3.9638943672180176, "lr": 0.0002, "elapsed_sec": 44833.385900735855, "step_time_sec": 8.238217614009045, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5394, "loss": 4.027091026306152, "lr": 0.0002, "elapsed_sec": 44841.627363204956, "step_time_sec": 8.241267240955494, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5395, "loss": 4.063674449920654, "lr": 0.0002, "elapsed_sec": 44849.86642742157, "step_time_sec": 8.23892098700162, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5396, "loss": 3.9687068462371826, "lr": 0.0002, "elapsed_sec": 44858.105194330215, "step_time_sec": 8.238617222988978, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5397, "loss": 3.9243905544281006, "lr": 0.0002, "elapsed_sec": 44866.34538984299, "step_time_sec": 8.24008422496263, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5398, "loss": 3.9739885330200195, "lr": 0.0002, "elapsed_sec": 44874.58501315117, "step_time_sec": 8.23938139306847, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5399, "loss": 3.969254732131958, "lr": 0.0002, "elapsed_sec": 44882.82443809509, "step_time_sec": 8.239285124000162, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5400, "loss": 4.157851696014404, "lr": 0.0002, "elapsed_sec": 44891.06389951706, "step_time_sec": 8.239271494094282, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5401, "loss": 4.031073570251465, "lr": 0.0002, "elapsed_sec": 44899.30283713341, "step_time_sec": 8.238784468965605, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5402, "loss": 4.081763744354248, "lr": 0.0002, "elapsed_sec": 44907.542927503586, "step_time_sec": 8.239895135047846, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5403, "loss": 4.04936408996582, "lr": 0.0002, "elapsed_sec": 44915.78232717514, "step_time_sec": 8.239297794993035, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5404, "loss": 4.131897926330566, "lr": 0.0002, "elapsed_sec": 44924.02211141586, "step_time_sec": 8.239602375077084, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5405, "loss": 4.042471885681152, "lr": 0.0002, "elapsed_sec": 44932.26146173477, "step_time_sec": 8.239205572986975, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5406, "loss": 3.9405691623687744, "lr": 0.0002, "elapsed_sec": 44940.50004696846, "step_time_sec": 8.238414366962388, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5407, "loss": 3.938556671142578, "lr": 0.0002, "elapsed_sec": 44948.73870801926, "step_time_sec": 8.23852750507649, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5408, "loss": 4.027815818786621, "lr": 0.0002, "elapsed_sec": 44956.97882795334, "step_time_sec": 8.239959568018094, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5409, "loss": 4.026549816131592, "lr": 0.0002, "elapsed_sec": 44965.21635222435, "step_time_sec": 8.237325894995593, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5410, "loss": 4.04214334487915, "lr": 0.0002, "elapsed_sec": 44973.4560008049, "step_time_sec": 8.2394952919567, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5411, "loss": 4.092801094055176, "lr": 0.0002, "elapsed_sec": 44981.69418334961, "step_time_sec": 8.23801441094838, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5412, "loss": 4.017883777618408, "lr": 0.0002, "elapsed_sec": 44989.93380880356, "step_time_sec": 8.239484281977639, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5413, "loss": 3.8778913021087646, "lr": 0.0002, "elapsed_sec": 44998.17362117767, "step_time_sec": 8.23965791799128, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5414, "loss": 4.160486221313477, "lr": 0.0002, "elapsed_sec": 45006.41091918945, "step_time_sec": 8.23716646793764, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5415, "loss": 4.092925548553467, "lr": 0.0002, "elapsed_sec": 45014.65119934082, "step_time_sec": 8.240036709001288, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5416, "loss": 4.099578380584717, "lr": 0.0002, "elapsed_sec": 45022.889877557755, "step_time_sec": 8.238530935021117, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5417, "loss": 4.069387912750244, "lr": 0.0002, "elapsed_sec": 45031.12916254997, "step_time_sec": 8.239120604004711, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5418, "loss": 4.0549468994140625, "lr": 0.0002, "elapsed_sec": 45039.368037462234, "step_time_sec": 8.238726894953288, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5419, "loss": 4.066713333129883, "lr": 0.0002, "elapsed_sec": 45047.60637831688, "step_time_sec": 8.23817860102281, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5420, "loss": 4.079292297363281, "lr": 0.0002, "elapsed_sec": 45055.845497369766, "step_time_sec": 8.23898523603566, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5421, "loss": 4.16606330871582, "lr": 0.0002, "elapsed_sec": 45064.08398795128, "step_time_sec": 8.238293385948054, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5422, "loss": 4.020812511444092, "lr": 0.0002, "elapsed_sec": 45072.3231215477, "step_time_sec": 8.23897649301216, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5423, "loss": 4.063438415527344, "lr": 0.0002, "elapsed_sec": 45080.56393432617, "step_time_sec": 8.240662515978329, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5424, "loss": 3.9699904918670654, "lr": 0.0002, "elapsed_sec": 45088.80183672905, "step_time_sec": 8.237714049988426, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5425, "loss": 4.013505458831787, "lr": 0.0002, "elapsed_sec": 45097.042073726654, "step_time_sec": 8.240090242004953, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5426, "loss": 4.095826148986816, "lr": 0.0002, "elapsed_sec": 45105.28070759773, "step_time_sec": 8.238470197073184, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5427, "loss": 3.921520709991455, "lr": 0.0002, "elapsed_sec": 45113.520091056824, "step_time_sec": 8.239248555037193, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5428, "loss": 3.9245853424072266, "lr": 0.0002, "elapsed_sec": 45121.75856757164, "step_time_sec": 8.23827158100903, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5429, "loss": 3.9293675422668457, "lr": 0.0002, "elapsed_sec": 45129.99714398384, "step_time_sec": 8.238482607994229, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5430, "loss": 4.0662922859191895, "lr": 0.0002, "elapsed_sec": 45138.23727989197, "step_time_sec": 8.239987349952571, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5431, "loss": 4.038564205169678, "lr": 0.0002, "elapsed_sec": 45146.47575926781, "step_time_sec": 8.238323991070502, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5432, "loss": 4.0539703369140625, "lr": 0.0002, "elapsed_sec": 45154.728747844696, "step_time_sec": 8.252796196029522, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5433, "loss": 3.985055685043335, "lr": 0.0002, "elapsed_sec": 45162.96870589256, "step_time_sec": 8.239785579033196, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5434, "loss": 4.103540897369385, "lr": 0.0002, "elapsed_sec": 45171.208846092224, "step_time_sec": 8.239967608940788, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5435, "loss": 3.871166229248047, "lr": 0.0002, "elapsed_sec": 45179.44630908966, "step_time_sec": 8.237303172936663, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5436, "loss": 3.9825618267059326, "lr": 0.0002, "elapsed_sec": 45187.68697762489, "step_time_sec": 8.240486721973866, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5437, "loss": 3.95595645904541, "lr": 0.0002, "elapsed_sec": 45195.9267373085, "step_time_sec": 8.23962767899502, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5438, "loss": 4.0904974937438965, "lr": 0.0002, "elapsed_sec": 45204.164226293564, "step_time_sec": 8.237334906007163, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5439, "loss": 4.116065979003906, "lr": 0.0002, "elapsed_sec": 45212.40399122238, "step_time_sec": 8.239588198950514, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5440, "loss": 3.8951783180236816, "lr": 0.0002, "elapsed_sec": 45220.64087176323, "step_time_sec": 8.236718797008507, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5441, "loss": 3.8878166675567627, "lr": 0.0002, "elapsed_sec": 45228.87976861, "step_time_sec": 8.238760164007545, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5442, "loss": 3.994711399078369, "lr": 0.0002, "elapsed_sec": 45237.11890268326, "step_time_sec": 8.238921212032437, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5443, "loss": 4.072843074798584, "lr": 0.0002, "elapsed_sec": 45245.35885691643, "step_time_sec": 8.239800547016785, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5444, "loss": 4.0237884521484375, "lr": 0.0002, "elapsed_sec": 45253.59849214554, "step_time_sec": 8.239510560058989, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5445, "loss": 4.150521278381348, "lr": 0.0002, "elapsed_sec": 45261.83777141571, "step_time_sec": 8.239095575991087, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5446, "loss": 3.996567964553833, "lr": 0.0002, "elapsed_sec": 45270.07818841934, "step_time_sec": 8.240211277035996, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5447, "loss": 3.904219388961792, "lr": 0.0002, "elapsed_sec": 45278.31807112694, "step_time_sec": 8.239698690013029, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5448, "loss": 3.97991681098938, "lr": 0.0002, "elapsed_sec": 45286.55694651604, "step_time_sec": 8.238760062027723, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5449, "loss": 3.9984774589538574, "lr": 0.0002, "elapsed_sec": 45294.796022892, "step_time_sec": 8.238862755941227, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5450, "loss": 4.090061187744141, "lr": 0.0002, "elapsed_sec": 45303.035173892975, "step_time_sec": 8.238977289991453, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5451, "loss": 4.04928731918335, "lr": 0.0002, "elapsed_sec": 45311.27420544624, "step_time_sec": 8.238876893068664, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5452, "loss": 3.8506736755371094, "lr": 0.0002, "elapsed_sec": 45319.5139029026, "step_time_sec": 8.239560978021473, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5453, "loss": 4.1736884117126465, "lr": 0.0002, "elapsed_sec": 45327.75264501572, "step_time_sec": 8.238568710978143, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5454, "loss": 4.057085037231445, "lr": 0.0002, "elapsed_sec": 45335.992307424545, "step_time_sec": 8.239513622014783, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5455, "loss": 4.010211944580078, "lr": 0.0002, "elapsed_sec": 45344.23089027405, "step_time_sec": 8.238382560899481, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5456, "loss": 3.8482794761657715, "lr": 0.0002, "elapsed_sec": 45352.470168828964, "step_time_sec": 8.239118981990032, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5457, "loss": 3.9917707443237305, "lr": 0.0002, "elapsed_sec": 45360.709562301636, "step_time_sec": 8.239264968084171, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5458, "loss": 4.043748378753662, "lr": 0.0002, "elapsed_sec": 45368.94706749916, "step_time_sec": 8.237362730898894, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5459, "loss": 3.958087921142578, "lr": 0.0002, "elapsed_sec": 45377.18549513817, "step_time_sec": 8.23821174108889, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5460, "loss": 4.0503411293029785, "lr": 0.0002, "elapsed_sec": 45385.42398953438, "step_time_sec": 8.238362835953012, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5461, "loss": 3.920405149459839, "lr": 0.0002, "elapsed_sec": 45393.662685871124, "step_time_sec": 8.238510543014854, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5462, "loss": 4.149733543395996, "lr": 0.0002, "elapsed_sec": 45401.902391433716, "step_time_sec": 8.239536006003618, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5463, "loss": 4.062972545623779, "lr": 0.0002, "elapsed_sec": 45410.141154527664, "step_time_sec": 8.238622438977472, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5464, "loss": 3.9636735916137695, "lr": 0.0002, "elapsed_sec": 45418.381622076035, "step_time_sec": 8.240284629981034, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5465, "loss": 4.025906562805176, "lr": 0.0002, "elapsed_sec": 45426.62031579018, "step_time_sec": 8.238612229004502, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5466, "loss": 4.279775619506836, "lr": 0.0002, "elapsed_sec": 45434.85937023163, "step_time_sec": 8.238862216006964, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5467, "loss": 3.7988240718841553, "lr": 0.0002, "elapsed_sec": 45443.10004281998, "step_time_sec": 8.240494368015788, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5468, "loss": 4.082206726074219, "lr": 0.0002, "elapsed_sec": 45451.338705301285, "step_time_sec": 8.238515826058574, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5469, "loss": 4.172452449798584, "lr": 0.0002, "elapsed_sec": 45459.57792901993, "step_time_sec": 8.239032199955545, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5470, "loss": 4.011116027832031, "lr": 0.0002, "elapsed_sec": 45467.81574463844, "step_time_sec": 8.237679512938485, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5471, "loss": 4.235109329223633, "lr": 0.0002, "elapsed_sec": 45476.054099559784, "step_time_sec": 8.238197684986517, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5472, "loss": 4.058620929718018, "lr": 0.0002, "elapsed_sec": 45484.293259859085, "step_time_sec": 8.238993918057531, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5473, "loss": 4.099123477935791, "lr": 0.0002, "elapsed_sec": 45492.53227686882, "step_time_sec": 8.238832860952243, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5474, "loss": 4.148177146911621, "lr": 0.0002, "elapsed_sec": 45500.77198147774, "step_time_sec": 8.23955625994131, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5475, "loss": 4.180672645568848, "lr": 0.0002, "elapsed_sec": 45509.0113325119, "step_time_sec": 8.2391658059787, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5476, "loss": 4.2185444831848145, "lr": 0.0002, "elapsed_sec": 45517.24871850014, "step_time_sec": 8.237238534959033, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5477, "loss": 4.17595911026001, "lr": 0.0002, "elapsed_sec": 45525.48958277702, "step_time_sec": 8.240706886979751, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5478, "loss": 4.0590410232543945, "lr": 0.0002, "elapsed_sec": 45533.72911000252, "step_time_sec": 8.239387259935029, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5479, "loss": 4.174250602722168, "lr": 0.0002, "elapsed_sec": 45541.96829175949, "step_time_sec": 8.239000846981071, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5480, "loss": 3.974794626235962, "lr": 0.0002, "elapsed_sec": 45550.20782470703, "step_time_sec": 8.239399145939387, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5481, "loss": 4.097290992736816, "lr": 0.0002, "elapsed_sec": 45558.44520688057, "step_time_sec": 8.237247452954762, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5482, "loss": 3.9143478870391846, "lr": 0.0002, "elapsed_sec": 45566.684265613556, "step_time_sec": 8.238875978044234, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5483, "loss": 4.043920993804932, "lr": 0.0002, "elapsed_sec": 45574.92297196388, "step_time_sec": 8.238523312960751, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5484, "loss": 3.9962875843048096, "lr": 0.0002, "elapsed_sec": 45583.16286802292, "step_time_sec": 8.239703392027877, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5485, "loss": 3.9467601776123047, "lr": 0.0002, "elapsed_sec": 45591.40368747711, "step_time_sec": 8.24068452604115, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5486, "loss": 4.105154514312744, "lr": 0.0002, "elapsed_sec": 45599.6425139904, "step_time_sec": 8.238643327029422, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5487, "loss": 3.965259075164795, "lr": 0.0002, "elapsed_sec": 45607.883496046066, "step_time_sec": 8.240828400012106, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5488, "loss": 4.065288066864014, "lr": 0.0002, "elapsed_sec": 45616.12402391434, "step_time_sec": 8.240376719972119, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5489, "loss": 4.020224094390869, "lr": 0.0002, "elapsed_sec": 45624.36121106148, "step_time_sec": 8.237000623019412, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5490, "loss": 3.996628522872925, "lr": 0.0002, "elapsed_sec": 45632.59935545921, "step_time_sec": 8.23799598007463, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5491, "loss": 4.004002571105957, "lr": 0.0002, "elapsed_sec": 45640.83937549591, "step_time_sec": 8.239871341036633, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5492, "loss": 4.304018497467041, "lr": 0.0002, "elapsed_sec": 45649.07945394516, "step_time_sec": 8.239917191909626, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5493, "loss": 4.032181262969971, "lr": 0.0002, "elapsed_sec": 45657.31849718094, "step_time_sec": 8.23880824400112, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5494, "loss": 4.122825622558594, "lr": 0.0002, "elapsed_sec": 45665.55702662468, "step_time_sec": 8.238372970954515, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5495, "loss": 3.965743064880371, "lr": 0.0002, "elapsed_sec": 45673.79852771759, "step_time_sec": 8.241402885993011, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5496, "loss": 3.9722542762756348, "lr": 0.0002, "elapsed_sec": 45682.038682222366, "step_time_sec": 8.239957434940152, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5497, "loss": 3.968958616256714, "lr": 0.0002, "elapsed_sec": 45690.27662110329, "step_time_sec": 8.237788965925574, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5498, "loss": 4.020539283752441, "lr": 0.0002, "elapsed_sec": 45698.51543164253, "step_time_sec": 8.238620714982972, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5499, "loss": 4.0608086585998535, "lr": 0.0002, "elapsed_sec": 45706.75419354439, "step_time_sec": 8.238644920056686, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5500, "loss": 3.9552528858184814, "lr": 0.0002, "elapsed_sec": 45714.99473285675, "step_time_sec": 35.60790728905704, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": true, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5501, "loss": 3.998615026473999, "lr": 0.0002, "elapsed_sec": 45750.61497783661, "step_time_sec": 8.252188420039602, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5502, "loss": 4.140445232391357, "lr": 0.0002, "elapsed_sec": 45758.846509218216, "step_time_sec": 8.231338083045557, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5503, "loss": 3.957082748413086, "lr": 0.0002, "elapsed_sec": 45767.07924365997, "step_time_sec": 8.232565460028127, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5504, "loss": 4.0156660079956055, "lr": 0.0002, "elapsed_sec": 45775.31113934517, "step_time_sec": 8.231737975031137, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5505, "loss": 4.023518085479736, "lr": 0.0002, "elapsed_sec": 45783.54229712486, "step_time_sec": 8.230954712955281, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5506, "loss": 4.075667858123779, "lr": 0.0002, "elapsed_sec": 45791.78067827225, "step_time_sec": 8.238266065949574, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5507, "loss": 3.979501247406006, "lr": 0.0002, "elapsed_sec": 45800.014213085175, "step_time_sec": 8.233336576027796, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5508, "loss": 4.176025390625, "lr": 0.0002, "elapsed_sec": 45808.2481136322, "step_time_sec": 8.233691087923944, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5509, "loss": 4.041301727294922, "lr": 0.0002, "elapsed_sec": 45816.47971820831, "step_time_sec": 8.231526653980836, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5510, "loss": 4.100955486297607, "lr": 0.0002, "elapsed_sec": 45824.72005939484, "step_time_sec": 8.240095869987272, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5511, "loss": 4.1662139892578125, "lr": 0.0002, "elapsed_sec": 45832.95915007591, "step_time_sec": 8.238952872925438, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5512, "loss": 4.088325023651123, "lr": 0.0002, "elapsed_sec": 45841.19809746742, "step_time_sec": 8.238796498044394, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5513, "loss": 3.835521936416626, "lr": 0.0002, "elapsed_sec": 45849.437271595, "step_time_sec": 8.239006569958292, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5514, "loss": 3.9270172119140625, "lr": 0.0002, "elapsed_sec": 45857.67577958107, "step_time_sec": 8.238355136942118, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5515, "loss": 4.149894714355469, "lr": 0.0002, "elapsed_sec": 45865.91569471359, "step_time_sec": 8.239759101998061, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5516, "loss": 3.986727714538574, "lr": 0.0002, "elapsed_sec": 45874.15427184105, "step_time_sec": 8.238437067950144, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5517, "loss": 4.044897556304932, "lr": 0.0002, "elapsed_sec": 45882.392837285995, "step_time_sec": 8.2383917379193, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5518, "loss": 4.035121440887451, "lr": 0.0002, "elapsed_sec": 45890.63255953789, "step_time_sec": 8.239548587007448, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5519, "loss": 4.126046657562256, "lr": 0.0002, "elapsed_sec": 45898.87228393555, "step_time_sec": 8.239578866981901, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5520, "loss": 4.070061206817627, "lr": 0.0002, "elapsed_sec": 45907.11129426956, "step_time_sec": 8.238852648995817, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5521, "loss": 3.9554409980773926, "lr": 0.0002, "elapsed_sec": 45915.351891994476, "step_time_sec": 8.240466599003412, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5522, "loss": 4.143190383911133, "lr": 0.0002, "elapsed_sec": 45923.58930134773, "step_time_sec": 8.237175931921229, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5523, "loss": 4.040602207183838, "lr": 0.0002, "elapsed_sec": 45931.82827997208, "step_time_sec": 8.238862443948165, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5524, "loss": 4.044198989868164, "lr": 0.0002, "elapsed_sec": 45940.068219423294, "step_time_sec": 8.239748970023356, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5525, "loss": 4.059913635253906, "lr": 0.0002, "elapsed_sec": 45948.30755710602, "step_time_sec": 8.239182249992155, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5526, "loss": 4.125090599060059, "lr": 0.0002, "elapsed_sec": 45956.547876119614, "step_time_sec": 8.240217252052389, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5527, "loss": 3.9745523929595947, "lr": 0.0002, "elapsed_sec": 45964.78652381897, "step_time_sec": 8.238405629992485, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5528, "loss": 4.140078544616699, "lr": 0.0002, "elapsed_sec": 45973.0269036293, "step_time_sec": 8.240237532998435, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5529, "loss": 4.080377578735352, "lr": 0.0002, "elapsed_sec": 45981.26629734039, "step_time_sec": 8.239221452968195, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5530, "loss": 4.422500133514404, "lr": 0.0002, "elapsed_sec": 45989.50448536873, "step_time_sec": 8.238019287935458, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5531, "loss": 4.217076301574707, "lr": 0.0002, "elapsed_sec": 45997.74314713478, "step_time_sec": 8.238514552940615, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5532, "loss": 4.1433868408203125, "lr": 0.0002, "elapsed_sec": 46005.98199224472, "step_time_sec": 8.238669051905163, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5533, "loss": 4.046621799468994, "lr": 0.0002, "elapsed_sec": 46014.2214307785, "step_time_sec": 8.23930612904951, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5534, "loss": 4.043701171875, "lr": 0.0002, "elapsed_sec": 46022.460414886475, "step_time_sec": 8.238828656030819, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5535, "loss": 4.031062602996826, "lr": 0.0002, "elapsed_sec": 46030.699271678925, "step_time_sec": 8.23862325400114, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5536, "loss": 4.081558704376221, "lr": 0.0002, "elapsed_sec": 46038.94158196449, "step_time_sec": 8.242225533002056, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5537, "loss": 4.087198734283447, "lr": 0.0002, "elapsed_sec": 46047.18193459511, "step_time_sec": 8.240137708024122, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5538, "loss": 4.049866676330566, "lr": 0.0002, "elapsed_sec": 46055.420784235, "step_time_sec": 8.238714187056758, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5539, "loss": 4.111293315887451, "lr": 0.0002, "elapsed_sec": 46063.661608457565, "step_time_sec": 8.240681202965789, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5540, "loss": 3.994001865386963, "lr": 0.0002, "elapsed_sec": 46071.900163173676, "step_time_sec": 8.238383015966974, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5541, "loss": 4.03822135925293, "lr": 0.0002, "elapsed_sec": 46080.13913178444, "step_time_sec": 8.238774267025292, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5542, "loss": 4.024017810821533, "lr": 0.0002, "elapsed_sec": 46088.37795114517, "step_time_sec": 8.238704464980401, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5543, "loss": 4.041519641876221, "lr": 0.0002, "elapsed_sec": 46096.61709237099, "step_time_sec": 8.239005357027054, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5544, "loss": 4.1070237159729, "lr": 0.0002, "elapsed_sec": 46104.85806846619, "step_time_sec": 8.240806597983465, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5545, "loss": 4.093332290649414, "lr": 0.0002, "elapsed_sec": 46113.09760904312, "step_time_sec": 8.239349849987775, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5546, "loss": 4.083255290985107, "lr": 0.0002, "elapsed_sec": 46121.336817502975, "step_time_sec": 8.239091368042864, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5547, "loss": 4.083189964294434, "lr": 0.0002, "elapsed_sec": 46129.57654309273, "step_time_sec": 8.239497614908032, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5548, "loss": 4.149317741394043, "lr": 0.0002, "elapsed_sec": 46137.81536889076, "step_time_sec": 8.238675324013457, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5549, "loss": 4.158195495605469, "lr": 0.0002, "elapsed_sec": 46146.0554022789, "step_time_sec": 8.239847584045492, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5550, "loss": 4.152517795562744, "lr": 0.0002, "elapsed_sec": 46154.29552054405, "step_time_sec": 8.239983991952613, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5551, "loss": 4.129289627075195, "lr": 0.0002, "elapsed_sec": 46162.53466105461, "step_time_sec": 8.23896349593997, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5552, "loss": 4.064225673675537, "lr": 0.0002, "elapsed_sec": 46170.77420425415, "step_time_sec": 8.239426414947957, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5553, "loss": 4.021513938903809, "lr": 0.0002, "elapsed_sec": 46179.013247966766, "step_time_sec": 8.238830008078367, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5554, "loss": 4.082821369171143, "lr": 0.0002, "elapsed_sec": 46187.25324654579, "step_time_sec": 8.239838903071359, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5555, "loss": 4.029101848602295, "lr": 0.0002, "elapsed_sec": 46195.49323439598, "step_time_sec": 8.239865040988661, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5556, "loss": 4.066998481750488, "lr": 0.0002, "elapsed_sec": 46203.73195147514, "step_time_sec": 8.238521844963543, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5557, "loss": 4.090279579162598, "lr": 0.0002, "elapsed_sec": 46211.9725818634, "step_time_sec": 8.240486984024756, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5558, "loss": 4.036337852478027, "lr": 0.0002, "elapsed_sec": 46220.21244263649, "step_time_sec": 8.239686260931194, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5559, "loss": 4.119172096252441, "lr": 0.0002, "elapsed_sec": 46228.45164966583, "step_time_sec": 8.239046829985455, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5560, "loss": 4.048248291015625, "lr": 0.0002, "elapsed_sec": 46236.69133710861, "step_time_sec": 8.239554346073419, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5561, "loss": 4.11829948425293, "lr": 0.0002, "elapsed_sec": 46244.93069410324, "step_time_sec": 8.239146536099724, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5562, "loss": 3.864717721939087, "lr": 0.0002, "elapsed_sec": 46253.17231178284, "step_time_sec": 8.241500917938538, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5563, "loss": 4.1376471519470215, "lr": 0.0002, "elapsed_sec": 46261.412658929825, "step_time_sec": 8.240173460100777, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5564, "loss": 4.015682697296143, "lr": 0.0002, "elapsed_sec": 46269.65161371231, "step_time_sec": 8.238774569006637, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5565, "loss": 4.201727867126465, "lr": 0.0002, "elapsed_sec": 46277.89181160927, "step_time_sec": 8.240050764055923, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5566, "loss": 4.07440185546875, "lr": 0.0002, "elapsed_sec": 46286.129905462265, "step_time_sec": 8.237923637032509, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5567, "loss": 4.126514434814453, "lr": 0.0002, "elapsed_sec": 46294.36856985092, "step_time_sec": 8.238515403005295, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5568, "loss": 4.223759174346924, "lr": 0.0002, "elapsed_sec": 46302.608312129974, "step_time_sec": 8.23957623494789, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5569, "loss": 3.976172685623169, "lr": 0.0002, "elapsed_sec": 46310.84518313408, "step_time_sec": 8.236687854980119, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5570, "loss": 4.095874309539795, "lr": 0.0002, "elapsed_sec": 46319.08594942093, "step_time_sec": 8.24058109393809, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5571, "loss": 4.163056373596191, "lr": 0.0002, "elapsed_sec": 46327.32534766197, "step_time_sec": 8.239260741975158, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5572, "loss": 4.055965423583984, "lr": 0.0002, "elapsed_sec": 46335.56561899185, "step_time_sec": 8.240145049989223, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5573, "loss": 3.8948380947113037, "lr": 0.0002, "elapsed_sec": 46343.80586504936, "step_time_sec": 8.240078701986931, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5574, "loss": 3.9948179721832275, "lr": 0.0002, "elapsed_sec": 46352.045632362366, "step_time_sec": 8.239643960958347, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5575, "loss": 4.159143447875977, "lr": 0.0002, "elapsed_sec": 46360.28601717949, "step_time_sec": 8.240188053110614, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5576, "loss": 4.0790629386901855, "lr": 0.0002, "elapsed_sec": 46368.52657985687, "step_time_sec": 8.240399798029102, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5577, "loss": 4.112653732299805, "lr": 0.0002, "elapsed_sec": 46376.764133930206, "step_time_sec": 8.23739615897648, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5578, "loss": 4.076621055603027, "lr": 0.0002, "elapsed_sec": 46385.00310230255, "step_time_sec": 8.238757424987853, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5579, "loss": 4.31248140335083, "lr": 0.0002, "elapsed_sec": 46393.24358820915, "step_time_sec": 8.240329394000582, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5580, "loss": 4.147904396057129, "lr": 0.0002, "elapsed_sec": 46401.48312783241, "step_time_sec": 8.239378939033486, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5581, "loss": 4.009223461151123, "lr": 0.0002, "elapsed_sec": 46409.72357439995, "step_time_sec": 8.240320194046944, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5582, "loss": 4.045321464538574, "lr": 0.0002, "elapsed_sec": 46417.962553977966, "step_time_sec": 8.23878361703828, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5583, "loss": 4.010269641876221, "lr": 0.0002, "elapsed_sec": 46426.20333480835, "step_time_sec": 8.240589480032213, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5584, "loss": 4.089986801147461, "lr": 0.0002, "elapsed_sec": 46434.44294023514, "step_time_sec": 8.239465284976177, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5585, "loss": 4.122764587402344, "lr": 0.0002, "elapsed_sec": 46442.6805460453, "step_time_sec": 8.237414164002985, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5586, "loss": 4.055458068847656, "lr": 0.0002, "elapsed_sec": 46450.91971182823, "step_time_sec": 8.2390109350672, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5587, "loss": 4.0973711013793945, "lr": 0.0002, "elapsed_sec": 46459.15836763382, "step_time_sec": 8.238500612904318, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5588, "loss": 4.154129981994629, "lr": 0.0002, "elapsed_sec": 46467.396538734436, "step_time_sec": 8.238037012983114, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5589, "loss": 4.02371883392334, "lr": 0.0002, "elapsed_sec": 46475.63492226601, "step_time_sec": 8.23823059303686, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5590, "loss": 3.9875664710998535, "lr": 0.0002, "elapsed_sec": 46483.87349200249, "step_time_sec": 8.238398222019896, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5591, "loss": 3.962825059890747, "lr": 0.0002, "elapsed_sec": 46492.11517572403, "step_time_sec": 8.241508187027648, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5592, "loss": 4.141443252563477, "lr": 0.0002, "elapsed_sec": 46500.35400438309, "step_time_sec": 8.238650497980416, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5593, "loss": 4.07545280456543, "lr": 0.0002, "elapsed_sec": 46508.59246015549, "step_time_sec": 8.238280151039362, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5594, "loss": 4.123881816864014, "lr": 0.0002, "elapsed_sec": 46516.83137178421, "step_time_sec": 8.238795128068887, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5595, "loss": 4.0530619621276855, "lr": 0.0002, "elapsed_sec": 46525.06966304779, "step_time_sec": 8.238063097000122, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5596, "loss": 4.267563343048096, "lr": 0.0002, "elapsed_sec": 46533.30920982361, "step_time_sec": 8.23939317499753, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5597, "loss": 4.126060485839844, "lr": 0.0002, "elapsed_sec": 46541.547781705856, "step_time_sec": 8.238433532998897, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5598, "loss": 4.049769401550293, "lr": 0.0002, "elapsed_sec": 46549.786976099014, "step_time_sec": 8.239023413974792, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5599, "loss": 3.9689314365386963, "lr": 0.0002, "elapsed_sec": 46558.026758909225, "step_time_sec": 8.239613606012426, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5600, "loss": 4.130436420440674, "lr": 0.0002, "elapsed_sec": 46566.263481616974, "step_time_sec": 8.236538575962186, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5601, "loss": 3.991415023803711, "lr": 0.0002, "elapsed_sec": 46574.51063513756, "step_time_sec": 8.240079297916964, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5602, "loss": 4.157336711883545, "lr": 0.0002, "elapsed_sec": 46582.74961566925, "step_time_sec": 8.238802558043972, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5603, "loss": 4.052137851715088, "lr": 0.0002, "elapsed_sec": 46590.988382816315, "step_time_sec": 8.238626179052517, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5604, "loss": 3.8727340698242188, "lr": 0.0002, "elapsed_sec": 46599.22909450531, "step_time_sec": 8.240503559005447, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5605, "loss": 4.045120716094971, "lr": 0.0002, "elapsed_sec": 46607.46803689003, "step_time_sec": 8.238793715951033, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5606, "loss": 4.026128768920898, "lr": 0.0002, "elapsed_sec": 46615.70701003075, "step_time_sec": 8.238807142945006, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5607, "loss": 3.9332733154296875, "lr": 0.0002, "elapsed_sec": 46623.94712471962, "step_time_sec": 8.239959942060523, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5608, "loss": 4.058091163635254, "lr": 0.0002, "elapsed_sec": 46632.186026096344, "step_time_sec": 8.238774332101457, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5609, "loss": 3.9925143718719482, "lr": 0.0002, "elapsed_sec": 46640.42692708969, "step_time_sec": 8.240733523038216, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5610, "loss": 4.063089847564697, "lr": 0.0002, "elapsed_sec": 46648.667535066605, "step_time_sec": 8.240414615953341, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5611, "loss": 4.028110504150391, "lr": 0.0002, "elapsed_sec": 46656.90650701523, "step_time_sec": 8.238828768953681, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5612, "loss": 3.996513605117798, "lr": 0.0002, "elapsed_sec": 46665.14772152901, "step_time_sec": 8.240987582015805, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5613, "loss": 4.033351898193359, "lr": 0.0002, "elapsed_sec": 46673.386763334274, "step_time_sec": 8.238884128979407, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5614, "loss": 3.9234089851379395, "lr": 0.0002, "elapsed_sec": 46681.625118255615, "step_time_sec": 8.238225967041217, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5615, "loss": 4.053791522979736, "lr": 0.0002, "elapsed_sec": 46689.86294865608, "step_time_sec": 8.237672566086985, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5616, "loss": 4.170106887817383, "lr": 0.0002, "elapsed_sec": 46698.10134959221, "step_time_sec": 8.238221989013255, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5617, "loss": 4.026155471801758, "lr": 0.0002, "elapsed_sec": 46706.3402235508, "step_time_sec": 8.23876002500765, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5618, "loss": 4.004689693450928, "lr": 0.0002, "elapsed_sec": 46714.580337285995, "step_time_sec": 8.239929831936024, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5619, "loss": 4.063261985778809, "lr": 0.0002, "elapsed_sec": 46722.81935381889, "step_time_sec": 8.238855903968215, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5620, "loss": 4.011825084686279, "lr": 0.0002, "elapsed_sec": 46731.059957265854, "step_time_sec": 8.24037512706127, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5621, "loss": 4.143438339233398, "lr": 0.0002, "elapsed_sec": 46739.298464775085, "step_time_sec": 8.238338645081967, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5622, "loss": 4.0790252685546875, "lr": 0.0002, "elapsed_sec": 46747.53797531128, "step_time_sec": 8.23932126397267, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5623, "loss": 4.093567848205566, "lr": 0.0002, "elapsed_sec": 46755.77745985985, "step_time_sec": 8.239367125905119, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5624, "loss": 3.9305832386016846, "lr": 0.0002, "elapsed_sec": 46764.01566648483, "step_time_sec": 8.238031995017081, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5625, "loss": 4.09433126449585, "lr": 0.0002, "elapsed_sec": 46772.25417280197, "step_time_sec": 8.238323061028495, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5626, "loss": 4.189207077026367, "lr": 0.0002, "elapsed_sec": 46780.49283003807, "step_time_sec": 8.238470221986063, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5627, "loss": 3.957343339920044, "lr": 0.0002, "elapsed_sec": 46788.73313713074, "step_time_sec": 8.240187005023472, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5628, "loss": 3.8741719722747803, "lr": 0.0002, "elapsed_sec": 46796.97359037399, "step_time_sec": 8.240306649007834, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5629, "loss": 4.07035493850708, "lr": 0.0002, "elapsed_sec": 46805.21244096756, "step_time_sec": 8.23864097893238, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5630, "loss": 4.012979507446289, "lr": 0.0002, "elapsed_sec": 46813.45281672478, "step_time_sec": 8.240208721952513, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5631, "loss": 4.034698486328125, "lr": 0.0002, "elapsed_sec": 46821.692425727844, "step_time_sec": 8.239448310108855, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5632, "loss": 4.127344131469727, "lr": 0.0002, "elapsed_sec": 46829.93103194237, "step_time_sec": 8.23844625800848, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5633, "loss": 4.16475248336792, "lr": 0.0002, "elapsed_sec": 46838.17021870613, "step_time_sec": 8.238997888052836, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5634, "loss": 4.017929553985596, "lr": 0.0002, "elapsed_sec": 46846.40862369537, "step_time_sec": 8.238317060982808, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5635, "loss": 3.993410587310791, "lr": 0.0002, "elapsed_sec": 46854.649364709854, "step_time_sec": 8.240491779986769, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5636, "loss": 4.025212287902832, "lr": 0.0002, "elapsed_sec": 46862.88880085945, "step_time_sec": 8.239291068050079, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5637, "loss": 4.031185150146484, "lr": 0.0002, "elapsed_sec": 46871.12830591202, "step_time_sec": 8.239353853045031, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5638, "loss": 4.033895015716553, "lr": 0.0002, "elapsed_sec": 46879.3694934845, "step_time_sec": 8.241034588078037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5639, "loss": 3.977091073989868, "lr": 0.0002, "elapsed_sec": 46887.610070705414, "step_time_sec": 8.240403722040355, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5640, "loss": 4.017411231994629, "lr": 0.0002, "elapsed_sec": 46895.84961247444, "step_time_sec": 8.239373911055736, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5641, "loss": 3.95761775970459, "lr": 0.0002, "elapsed_sec": 46904.0894510746, "step_time_sec": 8.239715146948583, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5642, "loss": 4.003139495849609, "lr": 0.0002, "elapsed_sec": 46912.32835102081, "step_time_sec": 8.238725778996013, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5643, "loss": 4.0293498039245605, "lr": 0.0002, "elapsed_sec": 46920.56848239899, "step_time_sec": 8.239969371003099, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5644, "loss": 4.053788661956787, "lr": 0.0002, "elapsed_sec": 46928.80764603615, "step_time_sec": 8.238990437006578, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5645, "loss": 4.113701820373535, "lr": 0.0002, "elapsed_sec": 46937.04545879364, "step_time_sec": 8.237626319052652, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5646, "loss": 4.0099077224731445, "lr": 0.0002, "elapsed_sec": 46945.28388595581, "step_time_sec": 8.238280005054548, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5647, "loss": 4.167938232421875, "lr": 0.0002, "elapsed_sec": 46953.523394584656, "step_time_sec": 8.23932087502908, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5648, "loss": 4.151117324829102, "lr": 0.0002, "elapsed_sec": 46961.763971328735, "step_time_sec": 8.240410532103851, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5649, "loss": 4.271321773529053, "lr": 0.0002, "elapsed_sec": 46970.00411653519, "step_time_sec": 8.240007413900457, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5650, "loss": 3.9361772537231445, "lr": 0.0002, "elapsed_sec": 46978.2427675724, "step_time_sec": 8.238458284991793, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5651, "loss": 3.8227107524871826, "lr": 0.0002, "elapsed_sec": 46986.48305153847, "step_time_sec": 8.240184399997815, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5652, "loss": 4.256772041320801, "lr": 0.0002, "elapsed_sec": 46994.72269296646, "step_time_sec": 8.239437383948825, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5653, "loss": 4.0736870765686035, "lr": 0.0002, "elapsed_sec": 47002.96128129959, "step_time_sec": 8.238423925009556, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5654, "loss": 4.191247940063477, "lr": 0.0002, "elapsed_sec": 47011.20035099983, "step_time_sec": 8.238892816007137, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5655, "loss": 4.178304672241211, "lr": 0.0002, "elapsed_sec": 47019.436868190765, "step_time_sec": 8.236354577937163, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5656, "loss": 4.003961086273193, "lr": 0.0002, "elapsed_sec": 47027.67629694939, "step_time_sec": 8.23931473097764, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5657, "loss": 3.9894933700561523, "lr": 0.0002, "elapsed_sec": 47035.914696455, "step_time_sec": 8.238194428035058, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5658, "loss": 4.1493659019470215, "lr": 0.0002, "elapsed_sec": 47044.15394330025, "step_time_sec": 8.239099455997348, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5659, "loss": 4.055155277252197, "lr": 0.0002, "elapsed_sec": 47052.394574165344, "step_time_sec": 8.240491622011177, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5660, "loss": 4.08793306350708, "lr": 0.0002, "elapsed_sec": 47060.63508605957, "step_time_sec": 8.240330016007647, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5661, "loss": 4.0566606521606445, "lr": 0.0002, "elapsed_sec": 47068.87556600571, "step_time_sec": 8.240269882022403, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5662, "loss": 4.057696342468262, "lr": 0.0002, "elapsed_sec": 47077.1147711277, "step_time_sec": 8.239050258998759, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5663, "loss": 4.241609573364258, "lr": 0.0002, "elapsed_sec": 47085.352363586426, "step_time_sec": 8.23743757000193, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5664, "loss": 3.8899085521698, "lr": 0.0002, "elapsed_sec": 47093.59115934372, "step_time_sec": 8.238635346991941, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5665, "loss": 4.091569900512695, "lr": 0.0002, "elapsed_sec": 47101.83066725731, "step_time_sec": 8.239325452013873, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5666, "loss": 3.9812169075012207, "lr": 0.0002, "elapsed_sec": 47110.0695810318, "step_time_sec": 8.23880803398788, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5667, "loss": 4.105301856994629, "lr": 0.0002, "elapsed_sec": 47118.30948019028, "step_time_sec": 8.239676296012476, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5668, "loss": 4.146763324737549, "lr": 0.0002, "elapsed_sec": 47126.548236846924, "step_time_sec": 8.238626971025951, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5669, "loss": 4.128457546234131, "lr": 0.0002, "elapsed_sec": 47134.788887023926, "step_time_sec": 8.240426103002392, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5670, "loss": 3.8580689430236816, "lr": 0.0002, "elapsed_sec": 47143.02726197243, "step_time_sec": 8.2382162300637, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5671, "loss": 4.010388374328613, "lr": 0.0002, "elapsed_sec": 47151.26708030701, "step_time_sec": 8.239657720085233, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5672, "loss": 4.164707660675049, "lr": 0.0002, "elapsed_sec": 47159.50615429878, "step_time_sec": 8.238909841980785, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5673, "loss": 4.051187992095947, "lr": 0.0002, "elapsed_sec": 47167.745234012604, "step_time_sec": 8.23895482905209, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5674, "loss": 4.017246723175049, "lr": 0.0002, "elapsed_sec": 47175.984587192535, "step_time_sec": 8.239208512939513, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5675, "loss": 4.066340446472168, "lr": 0.0002, "elapsed_sec": 47184.223290205, "step_time_sec": 8.238476763945073, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5676, "loss": 4.140827655792236, "lr": 0.0002, "elapsed_sec": 47192.46110486984, "step_time_sec": 8.23771418002434, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5677, "loss": 4.22730827331543, "lr": 0.0002, "elapsed_sec": 47200.69882822037, "step_time_sec": 8.237530141021125, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5678, "loss": 3.8520126342773438, "lr": 0.0002, "elapsed_sec": 47220.36032438278, "step_time_sec": 19.661377035896294, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5679, "loss": 4.189239025115967, "lr": 0.0002, "elapsed_sec": 47228.59265255928, "step_time_sec": 8.232155343052, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5680, "loss": 4.0740580558776855, "lr": 0.0002, "elapsed_sec": 47236.82702803612, "step_time_sec": 8.234260500059463, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5681, "loss": 4.14934778213501, "lr": 0.0002, "elapsed_sec": 47245.065286159515, "step_time_sec": 8.238060185918584, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5682, "loss": 4.109807014465332, "lr": 0.0002, "elapsed_sec": 47253.30212903023, "step_time_sec": 8.23667254706379, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5683, "loss": 4.120157241821289, "lr": 0.0002, "elapsed_sec": 47261.541180849075, "step_time_sec": 8.23889483208768, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5684, "loss": 3.987339496612549, "lr": 0.0002, "elapsed_sec": 47269.78075480461, "step_time_sec": 8.239395463024266, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5685, "loss": 4.080822944641113, "lr": 0.0002, "elapsed_sec": 47278.01989483833, "step_time_sec": 8.239002287038602, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5686, "loss": 4.0683698654174805, "lr": 0.0002, "elapsed_sec": 47286.26087927818, "step_time_sec": 8.240805504960008, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5687, "loss": 4.020198822021484, "lr": 0.0002, "elapsed_sec": 47294.499771118164, "step_time_sec": 8.238678972003981, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5688, "loss": 4.521420955657959, "lr": 0.0002, "elapsed_sec": 47302.740008831024, "step_time_sec": 8.24012278392911, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5689, "loss": 4.118460655212402, "lr": 0.0002, "elapsed_sec": 47310.98012495041, "step_time_sec": 8.239925728994422, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5690, "loss": 3.9476280212402344, "lr": 0.0002, "elapsed_sec": 47319.21853709221, "step_time_sec": 8.238233775016852, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5691, "loss": 4.000601291656494, "lr": 0.0002, "elapsed_sec": 47327.45745229721, "step_time_sec": 8.238782861968502, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5692, "loss": 3.975900411605835, "lr": 0.0002, "elapsed_sec": 47335.69649386406, "step_time_sec": 8.238863228005357, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5693, "loss": 4.223360061645508, "lr": 0.0002, "elapsed_sec": 47343.93549466133, "step_time_sec": 8.23885414400138, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5694, "loss": 4.0247039794921875, "lr": 0.0002, "elapsed_sec": 47352.174666404724, "step_time_sec": 8.238996428088285, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5695, "loss": 4.087759494781494, "lr": 0.0002, "elapsed_sec": 47360.41360783577, "step_time_sec": 8.23880510602612, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5696, "loss": 3.968649387359619, "lr": 0.0002, "elapsed_sec": 47368.653561115265, "step_time_sec": 8.239757172064856, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5697, "loss": 3.9952399730682373, "lr": 0.0002, "elapsed_sec": 47376.89261722565, "step_time_sec": 8.238939372007735, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5698, "loss": 4.003047943115234, "lr": 0.0002, "elapsed_sec": 47385.13067817688, "step_time_sec": 8.237909576972015, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5699, "loss": 4.045986175537109, "lr": 0.0002, "elapsed_sec": 47393.37059235573, "step_time_sec": 8.239690735004842, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5700, "loss": 4.035202980041504, "lr": 0.0002, "elapsed_sec": 47401.60919356346, "step_time_sec": 8.23847800202202, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5701, "loss": 4.044156551361084, "lr": 0.0002, "elapsed_sec": 47409.850850105286, "step_time_sec": 8.241453305003233, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5702, "loss": 4.0798797607421875, "lr": 0.0002, "elapsed_sec": 47418.091708660126, "step_time_sec": 8.24071747600101, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5703, "loss": 3.9598166942596436, "lr": 0.0002, "elapsed_sec": 47426.33084797859, "step_time_sec": 8.239001585054211, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5704, "loss": 4.009724140167236, "lr": 0.0002, "elapsed_sec": 47434.57196140289, "step_time_sec": 8.240885477978736, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5705, "loss": 3.9974541664123535, "lr": 0.0002, "elapsed_sec": 47442.81196117401, "step_time_sec": 8.239855628926307, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5706, "loss": 3.955488443374634, "lr": 0.0002, "elapsed_sec": 47451.0520427227, "step_time_sec": 8.23988420504611, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5707, "loss": 4.012871742248535, "lr": 0.0002, "elapsed_sec": 47459.291187524796, "step_time_sec": 8.23895430599805, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5708, "loss": 3.8884544372558594, "lr": 0.0002, "elapsed_sec": 47467.530806064606, "step_time_sec": 8.239494565990753, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5709, "loss": 3.9536664485931396, "lr": 0.0002, "elapsed_sec": 47475.7712533474, "step_time_sec": 8.240320103010163, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5710, "loss": 3.9556691646575928, "lr": 0.0002, "elapsed_sec": 47484.01012635231, "step_time_sec": 8.23871084395796, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5711, "loss": 4.075038909912109, "lr": 0.0002, "elapsed_sec": 47492.248922109604, "step_time_sec": 8.238590931985527, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5712, "loss": 4.055209159851074, "lr": 0.0002, "elapsed_sec": 47500.48842358589, "step_time_sec": 8.239347377908416, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5713, "loss": 4.057694911956787, "lr": 0.0002, "elapsed_sec": 47508.72757315636, "step_time_sec": 8.239025322953239, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5714, "loss": 4.117903709411621, "lr": 0.0002, "elapsed_sec": 47516.96705365181, "step_time_sec": 8.239306102041155, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5715, "loss": 3.930325508117676, "lr": 0.0002, "elapsed_sec": 47525.207057476044, "step_time_sec": 8.239783792058006, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5716, "loss": 4.0212626457214355, "lr": 0.0002, "elapsed_sec": 47533.44604182243, "step_time_sec": 8.238881646073423, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5717, "loss": 4.093756198883057, "lr": 0.0002, "elapsed_sec": 47541.687230825424, "step_time_sec": 8.241015572915785, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5718, "loss": 3.9671988487243652, "lr": 0.0002, "elapsed_sec": 47549.928560733795, "step_time_sec": 8.241195946000516, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5719, "loss": 4.009831428527832, "lr": 0.0002, "elapsed_sec": 47558.16679787636, "step_time_sec": 8.23798864998389, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5720, "loss": 3.9957287311553955, "lr": 0.0002, "elapsed_sec": 47566.40649986267, "step_time_sec": 8.23960192699451, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5721, "loss": 3.974339723587036, "lr": 0.0002, "elapsed_sec": 47574.64562869072, "step_time_sec": 8.238906042999588, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5722, "loss": 4.01398229598999, "lr": 0.0002, "elapsed_sec": 47582.88475227356, "step_time_sec": 8.238985687028617, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5723, "loss": 3.7682580947875977, "lr": 0.0002, "elapsed_sec": 47591.12400317192, "step_time_sec": 8.239051021984778, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5724, "loss": 3.9229300022125244, "lr": 0.0002, "elapsed_sec": 47599.36144566536, "step_time_sec": 8.23728453204967, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5725, "loss": 4.026584625244141, "lr": 0.0002, "elapsed_sec": 47607.601655483246, "step_time_sec": 8.24004258797504, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5726, "loss": 4.106549263000488, "lr": 0.0002, "elapsed_sec": 47615.83966374397, "step_time_sec": 8.237838158966042, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5727, "loss": 3.9698898792266846, "lr": 0.0002, "elapsed_sec": 47624.07880949974, "step_time_sec": 8.23899587104097, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5728, "loss": 4.018362522125244, "lr": 0.0002, "elapsed_sec": 47632.31822371483, "step_time_sec": 8.23924548306968, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5729, "loss": 4.005499362945557, "lr": 0.0002, "elapsed_sec": 47640.55554771423, "step_time_sec": 8.23716207407415, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5730, "loss": 4.008277416229248, "lr": 0.0002, "elapsed_sec": 47648.79405736923, "step_time_sec": 8.23834728100337, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5731, "loss": 4.085092544555664, "lr": 0.0002, "elapsed_sec": 47657.0321521759, "step_time_sec": 8.237981165060773, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5732, "loss": 3.927825450897217, "lr": 0.0002, "elapsed_sec": 47665.27122139931, "step_time_sec": 8.238854980096221, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5733, "loss": 4.034972190856934, "lr": 0.0002, "elapsed_sec": 47673.509303331375, "step_time_sec": 8.237949389964342, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5734, "loss": 3.963163137435913, "lr": 0.0002, "elapsed_sec": 47681.74797868729, "step_time_sec": 8.238513523014262, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5735, "loss": 4.112837314605713, "lr": 0.0002, "elapsed_sec": 47689.988797187805, "step_time_sec": 8.24066381691955, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5736, "loss": 4.1027607917785645, "lr": 0.0002, "elapsed_sec": 47698.229080200195, "step_time_sec": 8.24015030299779, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5737, "loss": 4.03216028213501, "lr": 0.0002, "elapsed_sec": 47706.46871089935, "step_time_sec": 8.23949010798242, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5738, "loss": 3.934995174407959, "lr": 0.0002, "elapsed_sec": 47714.7086956501, "step_time_sec": 8.239794165012427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5739, "loss": 3.9776504039764404, "lr": 0.0002, "elapsed_sec": 47722.94745659828, "step_time_sec": 8.238580912002362, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5740, "loss": 4.024539947509766, "lr": 0.0002, "elapsed_sec": 47731.187992572784, "step_time_sec": 8.240339770098217, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5741, "loss": 3.966431140899658, "lr": 0.0002, "elapsed_sec": 47739.42682552338, "step_time_sec": 8.238668828969821, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5742, "loss": 4.0380473136901855, "lr": 0.0002, "elapsed_sec": 47747.6650185585, "step_time_sec": 8.238070090999827, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5743, "loss": 4.018740177154541, "lr": 0.0002, "elapsed_sec": 47755.90493655205, "step_time_sec": 8.239734288072214, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5744, "loss": 4.0461931228637695, "lr": 0.0002, "elapsed_sec": 47764.14346265793, "step_time_sec": 8.23834772000555, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5745, "loss": 4.06404972076416, "lr": 0.0002, "elapsed_sec": 47772.38292527199, "step_time_sec": 8.239282140042633, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5746, "loss": 4.176819324493408, "lr": 0.0002, "elapsed_sec": 47780.62300682068, "step_time_sec": 8.239980798913166, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5747, "loss": 4.153160095214844, "lr": 0.0002, "elapsed_sec": 47788.86082315445, "step_time_sec": 8.237625736976042, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5748, "loss": 4.018514633178711, "lr": 0.0002, "elapsed_sec": 47797.100820064545, "step_time_sec": 8.239834522013552, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5749, "loss": 3.9363245964050293, "lr": 0.0002, "elapsed_sec": 47805.33987903595, "step_time_sec": 8.238898302079178, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5750, "loss": 4.11025333404541, "lr": 0.0002, "elapsed_sec": 47813.57880759239, "step_time_sec": 8.23875112296082, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5751, "loss": 3.9692249298095703, "lr": 0.0002, "elapsed_sec": 47821.81802606583, "step_time_sec": 8.23906887799967, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5752, "loss": 4.203830718994141, "lr": 0.0002, "elapsed_sec": 47830.05577015877, "step_time_sec": 8.237583387992345, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5753, "loss": 3.936164140701294, "lr": 0.0002, "elapsed_sec": 47838.29462337494, "step_time_sec": 8.238720530993305, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5754, "loss": 4.093998432159424, "lr": 0.0002, "elapsed_sec": 47846.53311395645, "step_time_sec": 8.238332863897085, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5755, "loss": 3.9840831756591797, "lr": 0.0002, "elapsed_sec": 47854.7728574276, "step_time_sec": 8.239607090945356, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5756, "loss": 3.9417061805725098, "lr": 0.0002, "elapsed_sec": 47863.01386928558, "step_time_sec": 8.240798256010748, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5757, "loss": 3.9281203746795654, "lr": 0.0002, "elapsed_sec": 47871.25202703476, "step_time_sec": 8.237959946971387, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5758, "loss": 4.085878849029541, "lr": 0.0002, "elapsed_sec": 47879.490888118744, "step_time_sec": 8.238696748041548, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5759, "loss": 4.002562999725342, "lr": 0.0002, "elapsed_sec": 47887.73030948639, "step_time_sec": 8.239299715962261, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5760, "loss": 4.02527379989624, "lr": 0.0002, "elapsed_sec": 47895.969984054565, "step_time_sec": 8.239515567896888, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5761, "loss": 4.073522567749023, "lr": 0.0002, "elapsed_sec": 47904.20944976807, "step_time_sec": 8.239291590056382, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5762, "loss": 3.936777353286743, "lr": 0.0002, "elapsed_sec": 47912.44913458824, "step_time_sec": 8.239492950029671, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5763, "loss": 3.9897146224975586, "lr": 0.0002, "elapsed_sec": 47920.68909788132, "step_time_sec": 8.239857297972776, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5764, "loss": 3.97238826751709, "lr": 0.0002, "elapsed_sec": 47928.9269015789, "step_time_sec": 8.237636432982981, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5765, "loss": 4.053814888000488, "lr": 0.0002, "elapsed_sec": 47937.165444374084, "step_time_sec": 8.238346169004217, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5766, "loss": 3.982919931411743, "lr": 0.0002, "elapsed_sec": 47945.40530180931, "step_time_sec": 8.239683381048962, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5767, "loss": 3.916414499282837, "lr": 0.0002, "elapsed_sec": 47953.64510130882, "step_time_sec": 8.239676263998263, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5768, "loss": 3.7418811321258545, "lr": 0.0002, "elapsed_sec": 47961.885267972946, "step_time_sec": 8.239957234938629, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5769, "loss": 3.9349725246429443, "lr": 0.0002, "elapsed_sec": 47970.125537872314, "step_time_sec": 8.240122347022407, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5770, "loss": 3.810802459716797, "lr": 0.0002, "elapsed_sec": 47978.36422920227, "step_time_sec": 8.238557218923233, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5771, "loss": 3.9412124156951904, "lr": 0.0002, "elapsed_sec": 47986.6043446064, "step_time_sec": 8.239953925949521, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5772, "loss": 4.148863315582275, "lr": 0.0002, "elapsed_sec": 47994.84431028366, "step_time_sec": 8.239749895059504, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5773, "loss": 4.0165934562683105, "lr": 0.0002, "elapsed_sec": 48003.08312010765, "step_time_sec": 8.238670768099837, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5774, "loss": 4.113156318664551, "lr": 0.0002, "elapsed_sec": 48011.323157310486, "step_time_sec": 8.239828272955492, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5775, "loss": 4.072091579437256, "lr": 0.0002, "elapsed_sec": 48019.563425302505, "step_time_sec": 8.2400967720896, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5776, "loss": 3.7562997341156006, "lr": 0.0002, "elapsed_sec": 48027.803272008896, "step_time_sec": 8.239727969048545, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5777, "loss": 3.9716644287109375, "lr": 0.0002, "elapsed_sec": 48036.0412709713, "step_time_sec": 8.237836083048023, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5778, "loss": 3.992217779159546, "lr": 0.0002, "elapsed_sec": 48044.27944660187, "step_time_sec": 8.237958865007386, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5779, "loss": 3.9981582164764404, "lr": 0.0002, "elapsed_sec": 48052.51760792732, "step_time_sec": 8.237988237058744, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5780, "loss": 3.842114210128784, "lr": 0.0002, "elapsed_sec": 48060.75628781319, "step_time_sec": 8.238511025090702, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5781, "loss": 3.99607515335083, "lr": 0.0002, "elapsed_sec": 48068.995913505554, "step_time_sec": 8.239455914008431, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5782, "loss": 4.095257759094238, "lr": 0.0002, "elapsed_sec": 48077.23581242561, "step_time_sec": 8.239748943946324, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5783, "loss": 4.011942386627197, "lr": 0.0002, "elapsed_sec": 48085.47377204895, "step_time_sec": 8.237772058928385, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5784, "loss": 3.845790386199951, "lr": 0.0002, "elapsed_sec": 48093.71387052536, "step_time_sec": 8.239945649984293, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5785, "loss": 4.115560054779053, "lr": 0.0002, "elapsed_sec": 48101.95230770111, "step_time_sec": 8.238269823021255, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5786, "loss": 3.885577440261841, "lr": 0.0002, "elapsed_sec": 48110.19104003906, "step_time_sec": 8.238579381955788, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5787, "loss": 4.072544097900391, "lr": 0.0002, "elapsed_sec": 48118.43071627617, "step_time_sec": 8.239537811023183, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5788, "loss": 3.9718167781829834, "lr": 0.0002, "elapsed_sec": 48126.66888666153, "step_time_sec": 8.238022431964055, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5789, "loss": 3.960326910018921, "lr": 0.0002, "elapsed_sec": 48134.910912275314, "step_time_sec": 8.24183632293716, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5790, "loss": 3.9777743816375732, "lr": 0.0002, "elapsed_sec": 48143.151379585266, "step_time_sec": 8.240313482936472, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5791, "loss": 3.9969441890716553, "lr": 0.0002, "elapsed_sec": 48151.389390945435, "step_time_sec": 8.237816454959102, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5792, "loss": 4.065148830413818, "lr": 0.0002, "elapsed_sec": 48159.62853693962, "step_time_sec": 8.238983802963048, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5793, "loss": 3.995748281478882, "lr": 0.0002, "elapsed_sec": 48167.86822915077, "step_time_sec": 8.239579336019233, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5794, "loss": 3.968423366546631, "lr": 0.0002, "elapsed_sec": 48176.10772037506, "step_time_sec": 8.23931596102193, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5795, "loss": 3.9529991149902344, "lr": 0.0002, "elapsed_sec": 48184.34654593468, "step_time_sec": 8.238650468993, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5796, "loss": 3.905478000640869, "lr": 0.0002, "elapsed_sec": 48192.58500647545, "step_time_sec": 8.238282090984285, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5797, "loss": 4.000119209289551, "lr": 0.0002, "elapsed_sec": 48200.824528217316, "step_time_sec": 8.239379837992601, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5798, "loss": 3.892656087875366, "lr": 0.0002, "elapsed_sec": 48209.06458902359, "step_time_sec": 8.239883002010174, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5799, "loss": 4.027870178222656, "lr": 0.0002, "elapsed_sec": 48217.30426263809, "step_time_sec": 8.23949716007337, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5800, "loss": 3.974804401397705, "lr": 0.0002, "elapsed_sec": 48225.542781353, "step_time_sec": 8.2383661539061, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5801, "loss": 4.004172325134277, "lr": 0.0002, "elapsed_sec": 48233.78104496002, "step_time_sec": 8.238110637059435, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5802, "loss": 3.985757827758789, "lr": 0.0002, "elapsed_sec": 48242.01940083504, "step_time_sec": 8.238204778986983, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5803, "loss": 3.9299144744873047, "lr": 0.0002, "elapsed_sec": 48250.25899100304, "step_time_sec": 8.239412951981649, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5804, "loss": 4.19440221786499, "lr": 0.0002, "elapsed_sec": 48258.49890208244, "step_time_sec": 8.23978860792704, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5805, "loss": 3.9331085681915283, "lr": 0.0002, "elapsed_sec": 48266.73828840256, "step_time_sec": 8.239265194046311, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5806, "loss": 4.0047430992126465, "lr": 0.0002, "elapsed_sec": 48274.975294589996, "step_time_sec": 8.236772235948592, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5807, "loss": 3.957913637161255, "lr": 0.0002, "elapsed_sec": 48283.213886499405, "step_time_sec": 8.238448026007973, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5808, "loss": 3.906815767288208, "lr": 0.0002, "elapsed_sec": 48291.452184677124, "step_time_sec": 8.238109502010047, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5809, "loss": 4.109234809875488, "lr": 0.0002, "elapsed_sec": 48299.691267967224, "step_time_sec": 8.238946500001475, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5810, "loss": 4.008090972900391, "lr": 0.0002, "elapsed_sec": 48307.93050670624, "step_time_sec": 8.239086899906397, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5811, "loss": 3.94046950340271, "lr": 0.0002, "elapsed_sec": 48316.16981816292, "step_time_sec": 8.239163252990693, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5812, "loss": 4.093693733215332, "lr": 0.0002, "elapsed_sec": 48324.41040349007, "step_time_sec": 8.240404835902154, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5813, "loss": 3.9084460735321045, "lr": 0.0002, "elapsed_sec": 48332.64939951897, "step_time_sec": 8.238835511961952, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5814, "loss": 3.9530797004699707, "lr": 0.0002, "elapsed_sec": 48340.88795518875, "step_time_sec": 8.238408792996779, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5815, "loss": 4.236993312835693, "lr": 0.0002, "elapsed_sec": 48349.12795090675, "step_time_sec": 8.239857585984282, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5816, "loss": 4.0409255027771, "lr": 0.0002, "elapsed_sec": 48357.36676287651, "step_time_sec": 8.238667724072002, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5817, "loss": 4.053347587585449, "lr": 0.0002, "elapsed_sec": 48365.60672092438, "step_time_sec": 8.239733574911952, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5818, "loss": 4.028824329376221, "lr": 0.0002, "elapsed_sec": 48373.845853328705, "step_time_sec": 8.23902271699626, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5819, "loss": 3.872572660446167, "lr": 0.0002, "elapsed_sec": 48382.084794044495, "step_time_sec": 8.238715057028458, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5820, "loss": 4.015152454376221, "lr": 0.0002, "elapsed_sec": 48390.32443380356, "step_time_sec": 8.239495210000314, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5821, "loss": 3.9784107208251953, "lr": 0.0002, "elapsed_sec": 48398.56342768669, "step_time_sec": 8.238861003075726, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5822, "loss": 4.055779457092285, "lr": 0.0002, "elapsed_sec": 48406.80323624611, "step_time_sec": 8.239698819001205, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5823, "loss": 3.9122424125671387, "lr": 0.0002, "elapsed_sec": 48415.04221868515, "step_time_sec": 8.238741139997728, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5824, "loss": 3.967982769012451, "lr": 0.0002, "elapsed_sec": 48423.27960515022, "step_time_sec": 8.23721268796362, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5825, "loss": 4.02655553817749, "lr": 0.0002, "elapsed_sec": 48431.518856048584, "step_time_sec": 8.239048883900978, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5826, "loss": 3.967017650604248, "lr": 0.0002, "elapsed_sec": 48439.75847363472, "step_time_sec": 8.239457587013021, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5827, "loss": 3.972970962524414, "lr": 0.0002, "elapsed_sec": 48447.997039318085, "step_time_sec": 8.238413847982883, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5828, "loss": 4.073890686035156, "lr": 0.0002, "elapsed_sec": 48456.236750125885, "step_time_sec": 8.23956526699476, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5829, "loss": 3.9435009956359863, "lr": 0.0002, "elapsed_sec": 48464.47399377823, "step_time_sec": 8.237087699002586, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5830, "loss": 4.023169040679932, "lr": 0.0002, "elapsed_sec": 48472.71373295784, "step_time_sec": 8.23956132994499, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5831, "loss": 3.922259569168091, "lr": 0.0002, "elapsed_sec": 48480.95228886604, "step_time_sec": 8.238391202990897, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5832, "loss": 3.991132974624634, "lr": 0.0002, "elapsed_sec": 48489.19145464897, "step_time_sec": 8.239026132971048, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5833, "loss": 3.9603023529052734, "lr": 0.0002, "elapsed_sec": 48497.43194437027, "step_time_sec": 8.24029554007575, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5834, "loss": 3.90846848487854, "lr": 0.0002, "elapsed_sec": 48505.670443058014, "step_time_sec": 8.238303104997613, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5835, "loss": 3.956707239151001, "lr": 0.0002, "elapsed_sec": 48513.91109967232, "step_time_sec": 8.240489906049334, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5836, "loss": 3.9986703395843506, "lr": 0.0002, "elapsed_sec": 48522.15154194832, "step_time_sec": 8.240286207059398, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5837, "loss": 4.005580425262451, "lr": 0.0002, "elapsed_sec": 48530.389503240585, "step_time_sec": 8.23780815990176, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5838, "loss": 3.916398525238037, "lr": 0.0002, "elapsed_sec": 48538.629385471344, "step_time_sec": 8.239725414081477, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5839, "loss": 4.044829368591309, "lr": 0.0002, "elapsed_sec": 48546.86826515198, "step_time_sec": 8.238756084931083, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5840, "loss": 3.8718976974487305, "lr": 0.0002, "elapsed_sec": 48555.10824751854, "step_time_sec": 8.239745329017751, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5841, "loss": 3.891162633895874, "lr": 0.0002, "elapsed_sec": 48563.34811902046, "step_time_sec": 8.239707084954716, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5842, "loss": 3.84189510345459, "lr": 0.0002, "elapsed_sec": 48571.586274147034, "step_time_sec": 8.23798918002285, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5843, "loss": 3.909205436706543, "lr": 0.0002, "elapsed_sec": 48579.826236248016, "step_time_sec": 8.239812849089503, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5844, "loss": 4.076709270477295, "lr": 0.0002, "elapsed_sec": 48588.06555581093, "step_time_sec": 8.239197363029234, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5845, "loss": 3.947563409805298, "lr": 0.0002, "elapsed_sec": 48596.30503106117, "step_time_sec": 8.239281660062261, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5846, "loss": 4.065851211547852, "lr": 0.0002, "elapsed_sec": 48604.54412126541, "step_time_sec": 8.238885844941251, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5847, "loss": 3.992034912109375, "lr": 0.0002, "elapsed_sec": 48612.78301048279, "step_time_sec": 8.238712699036114, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5848, "loss": 3.93412184715271, "lr": 0.0002, "elapsed_sec": 48621.02389383316, "step_time_sec": 8.240731582045555, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5849, "loss": 3.9888012409210205, "lr": 0.0002, "elapsed_sec": 48629.26354289055, "step_time_sec": 8.239496345049702, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5850, "loss": 3.8404242992401123, "lr": 0.0002, "elapsed_sec": 48637.500769138336, "step_time_sec": 8.237080389051698, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5851, "loss": 3.9212803840637207, "lr": 0.0002, "elapsed_sec": 48645.74044466019, "step_time_sec": 8.239503550925292, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5852, "loss": 3.9822964668273926, "lr": 0.0002, "elapsed_sec": 48653.979973077774, "step_time_sec": 8.239350957097486, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5853, "loss": 3.974266290664673, "lr": 0.0002, "elapsed_sec": 48662.220507860184, "step_time_sec": 8.240350892068818, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5854, "loss": 4.063751697540283, "lr": 0.0002, "elapsed_sec": 48670.46055603027, "step_time_sec": 8.239894136087969, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5855, "loss": 4.002129077911377, "lr": 0.0002, "elapsed_sec": 48678.69888544083, "step_time_sec": 8.238200346007943, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5856, "loss": 3.8210959434509277, "lr": 0.0002, "elapsed_sec": 48686.940126895905, "step_time_sec": 8.241083706030622, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5857, "loss": 4.00530481338501, "lr": 0.0002, "elapsed_sec": 48695.180402994156, "step_time_sec": 8.240047231083736, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5858, "loss": 4.026681900024414, "lr": 0.0002, "elapsed_sec": 48703.419516801834, "step_time_sec": 8.238976347958669, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5859, "loss": 4.0261616706848145, "lr": 0.0002, "elapsed_sec": 48711.65957784653, "step_time_sec": 8.239906547009014, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5860, "loss": 3.931396484375, "lr": 0.0002, "elapsed_sec": 48719.89827299118, "step_time_sec": 8.23854339797981, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5861, "loss": 3.9741556644439697, "lr": 0.0002, "elapsed_sec": 48728.1380341053, "step_time_sec": 8.23961819394026, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5862, "loss": 4.04969596862793, "lr": 0.0002, "elapsed_sec": 48736.37888836861, "step_time_sec": 8.240690067061223, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5863, "loss": 3.9729819297790527, "lr": 0.0002, "elapsed_sec": 48744.626873493195, "step_time_sec": 8.247848508064635, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5864, "loss": 4.0062737464904785, "lr": 0.0002, "elapsed_sec": 48752.88062572479, "step_time_sec": 8.253526396933012, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5865, "loss": 4.083975791931152, "lr": 0.0002, "elapsed_sec": 48761.120965480804, "step_time_sec": 8.240109439007938, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5866, "loss": 3.9599380493164062, "lr": 0.0002, "elapsed_sec": 48769.35969495773, "step_time_sec": 8.238610458094627, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5867, "loss": 4.015039920806885, "lr": 0.0002, "elapsed_sec": 48777.59782862663, "step_time_sec": 8.237947435001843, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5868, "loss": 4.055500030517578, "lr": 0.0002, "elapsed_sec": 48785.836676836014, "step_time_sec": 8.238645470933989, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5869, "loss": 4.074951648712158, "lr": 0.0002, "elapsed_sec": 48794.077003240585, "step_time_sec": 8.240200220025145, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5870, "loss": 4.031805992126465, "lr": 0.0002, "elapsed_sec": 48802.316294670105, "step_time_sec": 8.239119083038531, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5871, "loss": 4.050448417663574, "lr": 0.0002, "elapsed_sec": 48810.55610060692, "step_time_sec": 8.239638827042654, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5872, "loss": 3.93147873878479, "lr": 0.0002, "elapsed_sec": 48818.794729948044, "step_time_sec": 8.238506354973651, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5873, "loss": 3.8815948963165283, "lr": 0.0002, "elapsed_sec": 48827.0348303318, "step_time_sec": 8.23994242597837, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5874, "loss": 4.061783790588379, "lr": 0.0002, "elapsed_sec": 48835.274102687836, "step_time_sec": 8.239052336080931, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5875, "loss": 4.097681999206543, "lr": 0.0002, "elapsed_sec": 48843.51332163811, "step_time_sec": 8.239080414059572, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5876, "loss": 3.9308369159698486, "lr": 0.0002, "elapsed_sec": 48851.75279593468, "step_time_sec": 8.239324099966325, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5877, "loss": 3.8826582431793213, "lr": 0.0002, "elapsed_sec": 48859.99075913429, "step_time_sec": 8.237835325999185, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5878, "loss": 3.9675111770629883, "lr": 0.0002, "elapsed_sec": 48868.23060131073, "step_time_sec": 8.23963103198912, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5879, "loss": 4.036227703094482, "lr": 0.0002, "elapsed_sec": 48876.47009754181, "step_time_sec": 8.23939745593816, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5880, "loss": 3.901092529296875, "lr": 0.0002, "elapsed_sec": 48884.708178043365, "step_time_sec": 8.23784758290276, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5881, "loss": 4.073159217834473, "lr": 0.0002, "elapsed_sec": 48892.94860982895, "step_time_sec": 8.240260760998353, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5882, "loss": 3.979552984237671, "lr": 0.0002, "elapsed_sec": 48901.18670773506, "step_time_sec": 8.237954288022593, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5883, "loss": 4.043641090393066, "lr": 0.0002, "elapsed_sec": 48909.427195072174, "step_time_sec": 8.240327997948043, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5884, "loss": 4.116074562072754, "lr": 0.0002, "elapsed_sec": 48917.66708469391, "step_time_sec": 8.239755786955357, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5885, "loss": 4.115952014923096, "lr": 0.0002, "elapsed_sec": 48925.90479326248, "step_time_sec": 8.237519209971651, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5886, "loss": 3.90704607963562, "lr": 0.0002, "elapsed_sec": 48934.14389586449, "step_time_sec": 8.238968390971422, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5887, "loss": 4.041301250457764, "lr": 0.0002, "elapsed_sec": 48942.38338923454, "step_time_sec": 8.239305668976158, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5888, "loss": 3.802605390548706, "lr": 0.0002, "elapsed_sec": 48950.62215423584, "step_time_sec": 8.238633817993104, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5889, "loss": 4.117833137512207, "lr": 0.0002, "elapsed_sec": 48958.861016750336, "step_time_sec": 8.238678327063099, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5890, "loss": 4.147366046905518, "lr": 0.0002, "elapsed_sec": 48967.09905529022, "step_time_sec": 8.237833133083768, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5891, "loss": 4.045828342437744, "lr": 0.0002, "elapsed_sec": 48975.33951807022, "step_time_sec": 8.240293732029386, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5892, "loss": 4.023146629333496, "lr": 0.0002, "elapsed_sec": 48983.5796687603, "step_time_sec": 8.24000213900581, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5893, "loss": 3.9686217308044434, "lr": 0.0002, "elapsed_sec": 48991.819754362106, "step_time_sec": 8.239916725084186, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5894, "loss": 3.9339418411254883, "lr": 0.0002, "elapsed_sec": 49000.06079030037, "step_time_sec": 8.240860938094556, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5895, "loss": 3.8033974170684814, "lr": 0.0002, "elapsed_sec": 49008.30007266998, "step_time_sec": 8.239126958069392, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5896, "loss": 3.782716751098633, "lr": 0.0002, "elapsed_sec": 49016.53982543945, "step_time_sec": 8.23963524599094, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5897, "loss": 3.8845040798187256, "lr": 0.0002, "elapsed_sec": 49024.77912449837, "step_time_sec": 8.239066030015238, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5898, "loss": 4.024130821228027, "lr": 0.0002, "elapsed_sec": 49033.01739001274, "step_time_sec": 8.23809298290871, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5899, "loss": 3.981172561645508, "lr": 0.0002, "elapsed_sec": 49041.257717609406, "step_time_sec": 8.240169139928184, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5900, "loss": 3.9115171432495117, "lr": 0.0002, "elapsed_sec": 49049.4970266819, "step_time_sec": 8.239133668015711, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5901, "loss": 3.882962465286255, "lr": 0.0002, "elapsed_sec": 49057.7363820076, "step_time_sec": 8.239219595910981, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5902, "loss": 4.034497261047363, "lr": 0.0002, "elapsed_sec": 49065.9757771492, "step_time_sec": 8.239251298946328, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5903, "loss": 3.902211904525757, "lr": 0.0002, "elapsed_sec": 49074.21337199211, "step_time_sec": 8.237441097968258, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5904, "loss": 3.993246555328369, "lr": 0.0002, "elapsed_sec": 49082.452682971954, "step_time_sec": 8.239132982911542, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5905, "loss": 3.9492318630218506, "lr": 0.0002, "elapsed_sec": 49090.6917989254, "step_time_sec": 8.238984175026417, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5906, "loss": 4.0803375244140625, "lr": 0.0002, "elapsed_sec": 49098.93043112755, "step_time_sec": 8.23847369698342, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5907, "loss": 3.9447851181030273, "lr": 0.0002, "elapsed_sec": 49107.16909098625, "step_time_sec": 8.238476808997802, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5908, "loss": 3.9264719486236572, "lr": 0.0002, "elapsed_sec": 49115.40643930435, "step_time_sec": 8.237142289988697, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5909, "loss": 4.108121871948242, "lr": 0.0002, "elapsed_sec": 49123.64529109001, "step_time_sec": 8.23870658595115, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5910, "loss": 3.9113619327545166, "lr": 0.0002, "elapsed_sec": 49131.88369369507, "step_time_sec": 8.238213559961878, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5911, "loss": 3.9832496643066406, "lr": 0.0002, "elapsed_sec": 49140.12288451195, "step_time_sec": 8.238983690971509, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5912, "loss": 4.02502965927124, "lr": 0.0002, "elapsed_sec": 49148.36162018776, "step_time_sec": 8.238631236949004, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5913, "loss": 4.026751518249512, "lr": 0.0002, "elapsed_sec": 49156.60106468201, "step_time_sec": 8.239295899984427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5914, "loss": 3.947237253189087, "lr": 0.0002, "elapsed_sec": 49164.840458869934, "step_time_sec": 8.23918301903177, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5915, "loss": 3.9419631958007812, "lr": 0.0002, "elapsed_sec": 49173.07933759689, "step_time_sec": 8.238696073065512, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5916, "loss": 4.080405235290527, "lr": 0.0002, "elapsed_sec": 49181.31782269478, "step_time_sec": 8.23832175298594, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5917, "loss": 4.019593715667725, "lr": 0.0002, "elapsed_sec": 49189.55725979805, "step_time_sec": 8.239244448021054, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5918, "loss": 4.087812423706055, "lr": 0.0002, "elapsed_sec": 49197.79611444473, "step_time_sec": 8.2386970219668, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5919, "loss": 4.177035331726074, "lr": 0.0002, "elapsed_sec": 49206.03681373596, "step_time_sec": 8.240551979048178, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5920, "loss": 4.061131954193115, "lr": 0.0002, "elapsed_sec": 49214.276985645294, "step_time_sec": 8.240024178987369, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5921, "loss": 4.120114326477051, "lr": 0.0002, "elapsed_sec": 49222.515376091, "step_time_sec": 8.238216744037345, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5922, "loss": 4.060939311981201, "lr": 0.0002, "elapsed_sec": 49230.75585627556, "step_time_sec": 8.240303619997576, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5923, "loss": 4.195763111114502, "lr": 0.0002, "elapsed_sec": 49238.99398493767, "step_time_sec": 8.237968530971557, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5924, "loss": 3.9571285247802734, "lr": 0.0002, "elapsed_sec": 49247.23374462128, "step_time_sec": 8.239603777066804, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5925, "loss": 4.417896270751953, "lr": 0.0002, "elapsed_sec": 49255.47317862511, "step_time_sec": 8.239298939006403, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5926, "loss": 4.014237403869629, "lr": 0.0002, "elapsed_sec": 49263.711666345596, "step_time_sec": 8.238286421983503, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5927, "loss": 4.078775405883789, "lr": 0.0002, "elapsed_sec": 49271.953033685684, "step_time_sec": 8.241231629974209, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5928, "loss": 4.094433307647705, "lr": 0.0002, "elapsed_sec": 49280.19281435013, "step_time_sec": 8.239591849967837, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5929, "loss": 4.109241485595703, "lr": 0.0002, "elapsed_sec": 49288.43110537529, "step_time_sec": 8.238153188023716, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5930, "loss": 4.017411231994629, "lr": 0.0002, "elapsed_sec": 49296.67021584511, "step_time_sec": 8.239006020012312, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5931, "loss": 4.0406928062438965, "lr": 0.0002, "elapsed_sec": 49304.90859532356, "step_time_sec": 8.238174839993007, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5932, "loss": 4.135941505432129, "lr": 0.0002, "elapsed_sec": 49313.148352861404, "step_time_sec": 8.239587237942033, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5933, "loss": 4.121638774871826, "lr": 0.0002, "elapsed_sec": 49321.38920927048, "step_time_sec": 8.240736759034917, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5934, "loss": 4.066292762756348, "lr": 0.0002, "elapsed_sec": 49329.626705408096, "step_time_sec": 8.237360831000842, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5935, "loss": 3.977565050125122, "lr": 0.0002, "elapsed_sec": 49337.86748480797, "step_time_sec": 8.240586732048541, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5936, "loss": 4.0087456703186035, "lr": 0.0002, "elapsed_sec": 49346.10686135292, "step_time_sec": 8.239209830062464, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5937, "loss": 3.934199571609497, "lr": 0.0002, "elapsed_sec": 49354.34638285637, "step_time_sec": 8.239377422025427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5938, "loss": 3.953993797302246, "lr": 0.0002, "elapsed_sec": 49362.585973739624, "step_time_sec": 8.23938967299182, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5939, "loss": 4.029754638671875, "lr": 0.0002, "elapsed_sec": 49370.82258057594, "step_time_sec": 8.236440257984214, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5940, "loss": 4.050640106201172, "lr": 0.0002, "elapsed_sec": 49379.061504125595, "step_time_sec": 8.238757640006952, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5941, "loss": 4.00532865524292, "lr": 0.0002, "elapsed_sec": 49387.298979997635, "step_time_sec": 8.237320521031506, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5942, "loss": 4.131404399871826, "lr": 0.0002, "elapsed_sec": 49395.538090229034, "step_time_sec": 8.23898377304431, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5943, "loss": 4.023319244384766, "lr": 0.0002, "elapsed_sec": 49403.777114629745, "step_time_sec": 8.238841549027711, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5944, "loss": 4.038042068481445, "lr": 0.0002, "elapsed_sec": 49412.015194654465, "step_time_sec": 8.237911316915415, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5945, "loss": 3.948655128479004, "lr": 0.0002, "elapsed_sec": 49420.25440573692, "step_time_sec": 8.239042827975936, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5946, "loss": 4.099289894104004, "lr": 0.0002, "elapsed_sec": 49428.49379205704, "step_time_sec": 8.239198766998015, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5947, "loss": 3.8708231449127197, "lr": 0.0002, "elapsed_sec": 49436.73326706886, "step_time_sec": 8.239368855021894, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5948, "loss": 3.9496631622314453, "lr": 0.0002, "elapsed_sec": 49444.97361278534, "step_time_sec": 8.240116704953834, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5949, "loss": 4.0789794921875, "lr": 0.0002, "elapsed_sec": 49453.212636470795, "step_time_sec": 8.238877747906372, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5950, "loss": 4.037042617797852, "lr": 0.0002, "elapsed_sec": 49461.45366692543, "step_time_sec": 8.240847874083556, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5951, "loss": 3.945944309234619, "lr": 0.0002, "elapsed_sec": 49469.692215681076, "step_time_sec": 8.238406453980133, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5952, "loss": 3.9005236625671387, "lr": 0.0002, "elapsed_sec": 49477.930770397186, "step_time_sec": 8.23839204502292, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5953, "loss": 4.014499664306641, "lr": 0.0002, "elapsed_sec": 49486.17050409317, "step_time_sec": 8.239565409952775, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5954, "loss": 4.101596832275391, "lr": 0.0002, "elapsed_sec": 49494.40878200531, "step_time_sec": 8.2381438909797, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5955, "loss": 4.154485702514648, "lr": 0.0002, "elapsed_sec": 49502.64947319031, "step_time_sec": 8.2405251970049, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5956, "loss": 4.096892833709717, "lr": 0.0002, "elapsed_sec": 49510.88853597641, "step_time_sec": 8.23887648805976, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5957, "loss": 4.101762771606445, "lr": 0.0002, "elapsed_sec": 49519.12766170502, "step_time_sec": 8.2389534850372, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5958, "loss": 4.054455280303955, "lr": 0.0002, "elapsed_sec": 49527.366076231, "step_time_sec": 8.238262989092618, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5959, "loss": 3.862532615661621, "lr": 0.0002, "elapsed_sec": 49535.605008125305, "step_time_sec": 8.238754563964903, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5960, "loss": 3.958259344100952, "lr": 0.0002, "elapsed_sec": 49543.843672037125, "step_time_sec": 8.238531525013968, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5961, "loss": 3.8246421813964844, "lr": 0.0002, "elapsed_sec": 49552.08320260048, "step_time_sec": 8.239368784008548, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5962, "loss": 4.268807411193848, "lr": 0.0002, "elapsed_sec": 49560.32009267807, "step_time_sec": 8.236693670973182, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5963, "loss": 3.9540882110595703, "lr": 0.0002, "elapsed_sec": 49568.55853557587, "step_time_sec": 8.23830376600381, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5964, "loss": 3.9914321899414062, "lr": 0.0002, "elapsed_sec": 49576.797550439835, "step_time_sec": 8.238873903057538, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5965, "loss": 4.126914024353027, "lr": 0.0002, "elapsed_sec": 49585.037873506546, "step_time_sec": 8.24011046893429, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5966, "loss": 3.9900121688842773, "lr": 0.0002, "elapsed_sec": 49593.27666544914, "step_time_sec": 8.238638357026502, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5967, "loss": 3.9676673412323, "lr": 0.0002, "elapsed_sec": 49601.51579141617, "step_time_sec": 8.238984931027517, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5968, "loss": 4.033723831176758, "lr": 0.0002, "elapsed_sec": 49609.756410360336, "step_time_sec": 8.24042983201798, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5969, "loss": 3.981508255004883, "lr": 0.0002, "elapsed_sec": 49617.99574160576, "step_time_sec": 8.239197601913475, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5970, "loss": 3.9742870330810547, "lr": 0.0002, "elapsed_sec": 49626.23477721214, "step_time_sec": 8.238852916052565, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5971, "loss": 3.9183189868927, "lr": 0.0002, "elapsed_sec": 49634.47450733185, "step_time_sec": 8.239609951036982, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5972, "loss": 4.054561138153076, "lr": 0.0002, "elapsed_sec": 49642.71190428734, "step_time_sec": 8.237245280994102, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5973, "loss": 4.058853626251221, "lr": 0.0002, "elapsed_sec": 49650.95230102539, "step_time_sec": 8.240200503030792, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5974, "loss": 3.9222702980041504, "lr": 0.0002, "elapsed_sec": 49659.19081783295, "step_time_sec": 8.238305498030968, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5975, "loss": 3.9839508533477783, "lr": 0.0002, "elapsed_sec": 49667.429403066635, "step_time_sec": 8.238453953992575, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5976, "loss": 4.02211856842041, "lr": 0.0002, "elapsed_sec": 49675.667823791504, "step_time_sec": 8.238237426965497, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5977, "loss": 3.9019947052001953, "lr": 0.0002, "elapsed_sec": 49683.90465116501, "step_time_sec": 8.236663711024448, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5978, "loss": 4.032370090484619, "lr": 0.0002, "elapsed_sec": 49692.14322781563, "step_time_sec": 8.238443442969583, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5979, "loss": 4.018975734710693, "lr": 0.0002, "elapsed_sec": 49700.38126349449, "step_time_sec": 8.237881154986098, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5980, "loss": 4.004411697387695, "lr": 0.0002, "elapsed_sec": 49708.62110185623, "step_time_sec": 8.239643731969409, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5981, "loss": 3.8729560375213623, "lr": 0.0002, "elapsed_sec": 49716.86123466492, "step_time_sec": 8.240029646083713, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5982, "loss": 4.056183815002441, "lr": 0.0002, "elapsed_sec": 49725.099699258804, "step_time_sec": 8.238232373027131, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5983, "loss": 3.985917091369629, "lr": 0.0002, "elapsed_sec": 49733.338985443115, "step_time_sec": 8.239177625044249, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5984, "loss": 3.9955363273620605, "lr": 0.0002, "elapsed_sec": 49741.57879114151, "step_time_sec": 8.23958606889937, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5985, "loss": 4.096652030944824, "lr": 0.0002, "elapsed_sec": 49749.81722021103, "step_time_sec": 8.238304420025088, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5986, "loss": 3.9611642360687256, "lr": 0.0002, "elapsed_sec": 49758.05762577057, "step_time_sec": 8.24019340693485, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5987, "loss": 3.9448082447052, "lr": 0.0002, "elapsed_sec": 49766.296164274216, "step_time_sec": 8.238393713021651, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5988, "loss": 3.9196364879608154, "lr": 0.0002, "elapsed_sec": 49774.53666090965, "step_time_sec": 8.240370981977321, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5989, "loss": 4.0294694900512695, "lr": 0.0002, "elapsed_sec": 49782.77697634697, "step_time_sec": 8.240144914016128, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5990, "loss": 3.9100186824798584, "lr": 0.0002, "elapsed_sec": 49791.01647090912, "step_time_sec": 8.239325804053806, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5991, "loss": 3.870330333709717, "lr": 0.0002, "elapsed_sec": 49799.25538444519, "step_time_sec": 8.238705485011451, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5992, "loss": 3.9887337684631348, "lr": 0.0002, "elapsed_sec": 49807.494193315506, "step_time_sec": 8.238680681912228, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5993, "loss": 4.002125263214111, "lr": 0.0002, "elapsed_sec": 49815.734395980835, "step_time_sec": 8.24006082105916, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5994, "loss": 3.825486421585083, "lr": 0.0002, "elapsed_sec": 49823.9732530117, "step_time_sec": 8.238681136979721, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5995, "loss": 3.9616100788116455, "lr": 0.0002, "elapsed_sec": 49832.21164107323, "step_time_sec": 8.238182311062701, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5996, "loss": 4.051314830780029, "lr": 0.0002, "elapsed_sec": 49840.452461242676, "step_time_sec": 8.240690537029877, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5997, "loss": 4.0497636795043945, "lr": 0.0002, "elapsed_sec": 49848.6912381649, "step_time_sec": 8.238607765058987, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5998, "loss": 3.9698708057403564, "lr": 0.0002, "elapsed_sec": 49856.93049168587, "step_time_sec": 8.239092484931462, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5999, "loss": 3.958129644393921, "lr": 0.0002, "elapsed_sec": 49865.17013335228, "step_time_sec": 8.239428024971858, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6000, "loss": 3.95389461517334, "lr": 0.0002, "elapsed_sec": 49873.40903377533, "step_time_sec": 56.46131690405309, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": true, "validation_ran": true, "probe_ran": true, "probe_elapsed_sec": 0.9785656429594383, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6001, "loss": 4.06046199798584, "lr": 0.0002, "elapsed_sec": 49929.87630653381, "step_time_sec": 8.244642723933794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6002, "loss": 4.160254955291748, "lr": 0.0002, "elapsed_sec": 49938.11592411995, "step_time_sec": 8.239376084995456, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6003, "loss": 4.076103687286377, "lr": 0.0002, "elapsed_sec": 49946.357924222946, "step_time_sec": 8.241929784067906, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6004, "loss": 4.052543640136719, "lr": 0.0002, "elapsed_sec": 49954.59709525108, "step_time_sec": 8.238974025007337, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6005, "loss": 3.820432424545288, "lr": 0.0002, "elapsed_sec": 49962.83415102959, "step_time_sec": 8.236864533973858, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6006, "loss": 3.9852797985076904, "lr": 0.0002, "elapsed_sec": 49971.07397150993, "step_time_sec": 8.239659289014526, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6007, "loss": 3.979524612426758, "lr": 0.0002, "elapsed_sec": 49979.31368350983, "step_time_sec": 8.239538950962014, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6008, "loss": 3.898315668106079, "lr": 0.0002, "elapsed_sec": 49987.55262565613, "step_time_sec": 8.238812033901922, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6009, "loss": 3.9041085243225098, "lr": 0.0002, "elapsed_sec": 49995.79225373268, "step_time_sec": 8.239492319058627, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6010, "loss": 4.023970127105713, "lr": 0.0002, "elapsed_sec": 50004.03081345558, "step_time_sec": 8.238400108995847, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6011, "loss": 3.860846519470215, "lr": 0.0002, "elapsed_sec": 50012.271280288696, "step_time_sec": 8.240251998999156, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6012, "loss": 4.034785270690918, "lr": 0.0002, "elapsed_sec": 50020.51163101196, "step_time_sec": 8.240217447048053, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6013, "loss": 3.934361219406128, "lr": 0.0002, "elapsed_sec": 50028.74953198433, "step_time_sec": 8.237736409995705, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6014, "loss": 4.038136959075928, "lr": 0.0002, "elapsed_sec": 50036.98886704445, "step_time_sec": 8.239170960034244, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6015, "loss": 4.114266872406006, "lr": 0.0002, "elapsed_sec": 50045.228813648224, "step_time_sec": 8.239704534993507, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6016, "loss": 3.944587469100952, "lr": 0.0002, "elapsed_sec": 50053.46821808815, "step_time_sec": 8.239317034021951, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6017, "loss": 3.9228391647338867, "lr": 0.0002, "elapsed_sec": 50061.70858979225, "step_time_sec": 8.240171147044748, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6018, "loss": 4.123205184936523, "lr": 0.0002, "elapsed_sec": 50069.94730758667, "step_time_sec": 8.238587950007059, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6019, "loss": 4.109888553619385, "lr": 0.0002, "elapsed_sec": 50078.186296224594, "step_time_sec": 8.238778272992931, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6020, "loss": 4.00106954574585, "lr": 0.0002, "elapsed_sec": 50086.42553830147, "step_time_sec": 8.239180881995708, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6021, "loss": 4.009840488433838, "lr": 0.0002, "elapsed_sec": 50094.66489648819, "step_time_sec": 8.239145464031026, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6022, "loss": 4.067665100097656, "lr": 0.0002, "elapsed_sec": 50102.9042403698, "step_time_sec": 8.239158329903148, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6023, "loss": 4.026853084564209, "lr": 0.0002, "elapsed_sec": 50111.143808841705, "step_time_sec": 8.239381583058275, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6024, "loss": 3.994335174560547, "lr": 0.0002, "elapsed_sec": 50119.383878946304, "step_time_sec": 8.239915191894397, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6025, "loss": 3.961683750152588, "lr": 0.0002, "elapsed_sec": 50127.62401199341, "step_time_sec": 8.239985764957964, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6026, "loss": 4.0893988609313965, "lr": 0.0002, "elapsed_sec": 50135.86184310913, "step_time_sec": 8.23764805495739, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6027, "loss": 4.023416519165039, "lr": 0.0002, "elapsed_sec": 50144.10204911232, "step_time_sec": 8.240032764035277, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6028, "loss": 4.030804634094238, "lr": 0.0002, "elapsed_sec": 50152.34123158455, "step_time_sec": 8.239018433028832, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6029, "loss": 4.011767387390137, "lr": 0.0002, "elapsed_sec": 50160.58145689964, "step_time_sec": 8.240092354943044, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6030, "loss": 4.009069442749023, "lr": 0.0002, "elapsed_sec": 50168.82154417038, "step_time_sec": 8.239913080935366, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6031, "loss": 3.936631917953491, "lr": 0.0002, "elapsed_sec": 50177.061005592346, "step_time_sec": 8.23933217395097, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6032, "loss": 3.815511703491211, "lr": 0.0002, "elapsed_sec": 50185.301856040955, "step_time_sec": 8.240636660950258, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6033, "loss": 4.0199151039123535, "lr": 0.0002, "elapsed_sec": 50193.54157328606, "step_time_sec": 8.239567546057515, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6034, "loss": 4.054005146026611, "lr": 0.0002, "elapsed_sec": 50201.77952861786, "step_time_sec": 8.237781162955798, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6035, "loss": 4.048700332641602, "lr": 0.0002, "elapsed_sec": 50210.01973414421, "step_time_sec": 8.240085039986297, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6036, "loss": 3.864496946334839, "lr": 0.0002, "elapsed_sec": 50218.25917840004, "step_time_sec": 8.239241883973591, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6037, "loss": 3.747441053390503, "lr": 0.0002, "elapsed_sec": 50226.499577999115, "step_time_sec": 8.24031989404466, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6038, "loss": 3.96235728263855, "lr": 0.0002, "elapsed_sec": 50234.73909878731, "step_time_sec": 8.239337155013345, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6039, "loss": 3.9903507232666016, "lr": 0.0002, "elapsed_sec": 50242.97703719139, "step_time_sec": 8.237754608038813, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6040, "loss": 3.917466163635254, "lr": 0.0002, "elapsed_sec": 50251.21718668938, "step_time_sec": 8.239949595998041, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6041, "loss": 4.149703502655029, "lr": 0.0002, "elapsed_sec": 50259.45727133751, "step_time_sec": 8.239961111918092, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6042, "loss": 3.9557993412017822, "lr": 0.0002, "elapsed_sec": 50267.69617438316, "step_time_sec": 8.238763157976791, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6043, "loss": 3.954482316970825, "lr": 0.0002, "elapsed_sec": 50275.93620467186, "step_time_sec": 8.239851541002281, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6044, "loss": 4.019817352294922, "lr": 0.0002, "elapsed_sec": 50284.175111055374, "step_time_sec": 8.23878650192637, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6045, "loss": 4.057166576385498, "lr": 0.0002, "elapsed_sec": 50292.41486001015, "step_time_sec": 8.239540030015633, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6046, "loss": 4.043412685394287, "lr": 0.0002, "elapsed_sec": 50300.65560412407, "step_time_sec": 8.240621985984035, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6047, "loss": 3.918933629989624, "lr": 0.0002, "elapsed_sec": 50308.89364981651, "step_time_sec": 8.237853066064417, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6048, "loss": 4.088562965393066, "lr": 0.0002, "elapsed_sec": 50317.13445210457, "step_time_sec": 8.240673910011537, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6049, "loss": 4.005990982055664, "lr": 0.0002, "elapsed_sec": 50325.37287902832, "step_time_sec": 8.238219082006253, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6050, "loss": 4.018719673156738, "lr": 0.0002, "elapsed_sec": 50333.612278699875, "step_time_sec": 8.239232867956161, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6051, "loss": 3.848020315170288, "lr": 0.0002, "elapsed_sec": 50341.85198569298, "step_time_sec": 8.239530652062967, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6052, "loss": 4.016816139221191, "lr": 0.0002, "elapsed_sec": 50350.0905251503, "step_time_sec": 8.238415431929752, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6053, "loss": 3.981682777404785, "lr": 0.0002, "elapsed_sec": 50358.33104920387, "step_time_sec": 8.240319974953309, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6054, "loss": 3.935959815979004, "lr": 0.0002, "elapsed_sec": 50366.57070684433, "step_time_sec": 8.239566044998355, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6055, "loss": 4.024338722229004, "lr": 0.0002, "elapsed_sec": 50374.809000492096, "step_time_sec": 8.238085982040502, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6056, "loss": 4.053546905517578, "lr": 0.0002, "elapsed_sec": 50383.04940080643, "step_time_sec": 8.24023604101967, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6057, "loss": 4.043776512145996, "lr": 0.0002, "elapsed_sec": 50391.287380218506, "step_time_sec": 8.237765041994862, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6058, "loss": 4.122267246246338, "lr": 0.0002, "elapsed_sec": 50399.52697944641, "step_time_sec": 8.239461958990432, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6059, "loss": 4.031364917755127, "lr": 0.0002, "elapsed_sec": 50407.767548561096, "step_time_sec": 8.240416922955774, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6060, "loss": 4.148503303527832, "lr": 0.0002, "elapsed_sec": 50416.00580096245, "step_time_sec": 8.23807718604803, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6061, "loss": 3.963857650756836, "lr": 0.0002, "elapsed_sec": 50424.246564626694, "step_time_sec": 8.240608107997105, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6062, "loss": 4.033041954040527, "lr": 0.0002, "elapsed_sec": 50432.485270261765, "step_time_sec": 8.238512355950661, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6063, "loss": 3.920192241668701, "lr": 0.0002, "elapsed_sec": 50440.72465920448, "step_time_sec": 8.239236509078182, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6064, "loss": 4.038794994354248, "lr": 0.0002, "elapsed_sec": 50448.96293282509, "step_time_sec": 8.2381257629022, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6065, "loss": 3.9478325843811035, "lr": 0.0002, "elapsed_sec": 50457.200695991516, "step_time_sec": 8.237570866011083, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6066, "loss": 3.996044874191284, "lr": 0.0002, "elapsed_sec": 50465.441232442856, "step_time_sec": 8.240368025028147, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6067, "loss": 3.92384672164917, "lr": 0.0002, "elapsed_sec": 50473.68273162842, "step_time_sec": 8.24137807090301, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6068, "loss": 4.130105972290039, "lr": 0.0002, "elapsed_sec": 50481.92221784592, "step_time_sec": 8.239279624074697, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6069, "loss": 3.9795799255371094, "lr": 0.0002, "elapsed_sec": 50490.16195559502, "step_time_sec": 8.239594010054134, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6070, "loss": 3.987788438796997, "lr": 0.0002, "elapsed_sec": 50498.401809215546, "step_time_sec": 8.239688722998835, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6071, "loss": 3.893402099609375, "lr": 0.0002, "elapsed_sec": 50506.64212656021, "step_time_sec": 8.240199460065924, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6072, "loss": 4.008767127990723, "lr": 0.0002, "elapsed_sec": 50514.88056874275, "step_time_sec": 8.238258567987941, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6073, "loss": 4.054451942443848, "lr": 0.0002, "elapsed_sec": 50523.119320869446, "step_time_sec": 8.238562681013718, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6074, "loss": 3.875991106033325, "lr": 0.0002, "elapsed_sec": 50531.35767650604, "step_time_sec": 8.23820101202, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6075, "loss": 3.986220598220825, "lr": 0.0002, "elapsed_sec": 50539.59801840782, "step_time_sec": 8.240136984037235, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6076, "loss": 3.944179058074951, "lr": 0.0002, "elapsed_sec": 50547.83769488335, "step_time_sec": 8.239517844980583, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6077, "loss": 3.9217734336853027, "lr": 0.0002, "elapsed_sec": 50556.077931404114, "step_time_sec": 8.24011164798867, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6078, "loss": 3.921973466873169, "lr": 0.0002, "elapsed_sec": 50564.317475795746, "step_time_sec": 8.239388242014684, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6079, "loss": 3.955507278442383, "lr": 0.0002, "elapsed_sec": 50572.55744576454, "step_time_sec": 8.239805130986497, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6080, "loss": 3.988065719604492, "lr": 0.0002, "elapsed_sec": 50580.795919418335, "step_time_sec": 8.238295125076547, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6081, "loss": 4.028645992279053, "lr": 0.0002, "elapsed_sec": 50589.03448653221, "step_time_sec": 8.238388902042061, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6082, "loss": 4.007589340209961, "lr": 0.0002, "elapsed_sec": 50597.2740175724, "step_time_sec": 8.23935824399814, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6083, "loss": 3.9391584396362305, "lr": 0.0002, "elapsed_sec": 50605.51202726364, "step_time_sec": 8.237874533049762, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6084, "loss": 3.8043158054351807, "lr": 0.0002, "elapsed_sec": 50613.752192258835, "step_time_sec": 8.239958999911323, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6085, "loss": 3.9854233264923096, "lr": 0.0002, "elapsed_sec": 50621.99166417122, "step_time_sec": 8.239316595019773, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6086, "loss": 4.127425670623779, "lr": 0.0002, "elapsed_sec": 50630.22895312309, "step_time_sec": 8.23715429008007, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6087, "loss": 3.9241788387298584, "lr": 0.0002, "elapsed_sec": 50638.469918727875, "step_time_sec": 8.2407686089864, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6088, "loss": 3.9699556827545166, "lr": 0.0002, "elapsed_sec": 50646.708949804306, "step_time_sec": 8.23891380801797, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6089, "loss": 4.128028392791748, "lr": 0.0002, "elapsed_sec": 50654.94819641113, "step_time_sec": 8.239060731953941, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6090, "loss": 3.8436532020568848, "lr": 0.0002, "elapsed_sec": 50663.188843250275, "step_time_sec": 8.240473279030994, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6091, "loss": 3.9503817558288574, "lr": 0.0002, "elapsed_sec": 50671.42647457123, "step_time_sec": 8.237438047071919, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6092, "loss": 3.938999652862549, "lr": 0.0002, "elapsed_sec": 50679.66655111313, "step_time_sec": 8.239910261007026, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6093, "loss": 3.9089906215667725, "lr": 0.0002, "elapsed_sec": 50687.90297937393, "step_time_sec": 8.236233968986198, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6094, "loss": 3.959054946899414, "lr": 0.0002, "elapsed_sec": 50696.14174771309, "step_time_sec": 8.238614102941938, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6095, "loss": 4.034877300262451, "lr": 0.0002, "elapsed_sec": 50704.38057541847, "step_time_sec": 8.23868669604417, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6096, "loss": 3.826418161392212, "lr": 0.0002, "elapsed_sec": 50712.61927485466, "step_time_sec": 8.238524532993324, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6097, "loss": 3.9660098552703857, "lr": 0.0002, "elapsed_sec": 50720.85966515541, "step_time_sec": 8.240269356989302, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6098, "loss": 4.0474934577941895, "lr": 0.0002, "elapsed_sec": 50729.09995508194, "step_time_sec": 8.24009341292549, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6099, "loss": 4.122805595397949, "lr": 0.0002, "elapsed_sec": 50737.339003801346, "step_time_sec": 8.238905676058494, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6100, "loss": 3.775444984436035, "lr": 0.0002, "elapsed_sec": 50745.57815480232, "step_time_sec": 8.239002229995094, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6101, "loss": 3.9006035327911377, "lr": 0.0002, "elapsed_sec": 50753.81629061699, "step_time_sec": 8.237934298929758, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6102, "loss": 4.017724990844727, "lr": 0.0002, "elapsed_sec": 50762.05686044693, "step_time_sec": 8.240435193991289, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6103, "loss": 4.170105934143066, "lr": 0.0002, "elapsed_sec": 50770.295597076416, "step_time_sec": 8.238622521050274, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6104, "loss": 3.8258376121520996, "lr": 0.0002, "elapsed_sec": 50778.534548044205, "step_time_sec": 8.238800029968843, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6105, "loss": 3.8909265995025635, "lr": 0.0002, "elapsed_sec": 50786.774364709854, "step_time_sec": 8.239631184958853, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6106, "loss": 3.995993137359619, "lr": 0.0002, "elapsed_sec": 50795.013129234314, "step_time_sec": 8.238535761949606, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6107, "loss": 3.898972511291504, "lr": 0.0002, "elapsed_sec": 50803.253313302994, "step_time_sec": 8.240055983071215, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6108, "loss": 4.114335536956787, "lr": 0.0002, "elapsed_sec": 50811.49274802208, "step_time_sec": 8.239260967937298, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6109, "loss": 4.151114463806152, "lr": 0.0002, "elapsed_sec": 50819.72926545143, "step_time_sec": 8.236358991940506, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6110, "loss": 4.085884094238281, "lr": 0.0002, "elapsed_sec": 50827.96898031235, "step_time_sec": 8.239562888047658, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6111, "loss": 4.01243257522583, "lr": 0.0002, "elapsed_sec": 50836.20848894119, "step_time_sec": 8.239366222987883, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6112, "loss": 3.898677110671997, "lr": 0.0002, "elapsed_sec": 50844.44668507576, "step_time_sec": 8.238020527991466, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6113, "loss": 3.99857497215271, "lr": 0.0002, "elapsed_sec": 50852.68772959709, "step_time_sec": 8.240882128942758, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6114, "loss": 4.17493200302124, "lr": 0.0002, "elapsed_sec": 50860.92438530922, "step_time_sec": 8.236495961085893, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6115, "loss": 4.014724254608154, "lr": 0.0002, "elapsed_sec": 50869.16288137436, "step_time_sec": 8.238341860938817, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6116, "loss": 4.0503129959106445, "lr": 0.0002, "elapsed_sec": 50877.40262532234, "step_time_sec": 8.239591846941039, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6117, "loss": 3.9732418060302734, "lr": 0.0002, "elapsed_sec": 50885.64301109314, "step_time_sec": 8.240246615023352, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6118, "loss": 4.012448310852051, "lr": 0.0002, "elapsed_sec": 50893.88246178627, "step_time_sec": 8.239287220058031, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6119, "loss": 4.003515243530273, "lr": 0.0002, "elapsed_sec": 50902.12040615082, "step_time_sec": 8.237735624075867, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6120, "loss": 3.934934139251709, "lr": 0.0002, "elapsed_sec": 50910.36108255386, "step_time_sec": 8.240541180945002, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6121, "loss": 3.9893031120300293, "lr": 0.0002, "elapsed_sec": 50918.60008478165, "step_time_sec": 8.238831205992028, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6122, "loss": 4.0180792808532715, "lr": 0.0002, "elapsed_sec": 50926.83874464035, "step_time_sec": 8.238516320008785, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6123, "loss": 3.9599621295928955, "lr": 0.0002, "elapsed_sec": 50935.07768917084, "step_time_sec": 8.2387321209535, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6124, "loss": 3.879964590072632, "lr": 0.0002, "elapsed_sec": 50943.31620144844, "step_time_sec": 8.238374402048066, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6125, "loss": 4.000030517578125, "lr": 0.0002, "elapsed_sec": 50951.55741453171, "step_time_sec": 8.24102406797465, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6126, "loss": 3.997276544570923, "lr": 0.0002, "elapsed_sec": 50959.79806613922, "step_time_sec": 8.240473396959715, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6127, "loss": 3.9025261402130127, "lr": 0.0002, "elapsed_sec": 50968.03670978546, "step_time_sec": 8.238473561010323, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6128, "loss": 4.03404426574707, "lr": 0.0002, "elapsed_sec": 50976.27623295784, "step_time_sec": 8.239370187977329, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6129, "loss": 4.010164260864258, "lr": 0.0002, "elapsed_sec": 50984.5154504776, "step_time_sec": 8.239060074905865, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6130, "loss": 3.941983222961426, "lr": 0.0002, "elapsed_sec": 50992.75537919998, "step_time_sec": 8.239746778970584, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6131, "loss": 3.909740924835205, "lr": 0.0002, "elapsed_sec": 51000.99553775787, "step_time_sec": 8.24001634598244, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6132, "loss": 3.882082939147949, "lr": 0.0002, "elapsed_sec": 51009.232624292374, "step_time_sec": 8.236916067893617, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6133, "loss": 3.8827390670776367, "lr": 0.0002, "elapsed_sec": 51017.47322368622, "step_time_sec": 8.240435143001378, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6134, "loss": 4.010104179382324, "lr": 0.0002, "elapsed_sec": 51025.71297502518, "step_time_sec": 8.239612594945356, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6135, "loss": 3.9318342208862305, "lr": 0.0002, "elapsed_sec": 51033.951763391495, "step_time_sec": 8.238595478935167, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6136, "loss": 4.002337455749512, "lr": 0.0002, "elapsed_sec": 51042.192967653275, "step_time_sec": 8.240998225985095, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6137, "loss": 3.9282844066619873, "lr": 0.0002, "elapsed_sec": 51050.43198466301, "step_time_sec": 8.238862277008593, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6138, "loss": 3.7798233032226562, "lr": 0.0002, "elapsed_sec": 51058.673263549805, "step_time_sec": 8.24115650600288, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6139, "loss": 4.000411033630371, "lr": 0.0002, "elapsed_sec": 51066.913395643234, "step_time_sec": 8.239980223006569, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6140, "loss": 4.034682273864746, "lr": 0.0002, "elapsed_sec": 51075.15187764168, "step_time_sec": 8.2382535089273, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6141, "loss": 3.7949562072753906, "lr": 0.0002, "elapsed_sec": 51083.39304685593, "step_time_sec": 8.241065411013551, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6142, "loss": 3.972872495651245, "lr": 0.0002, "elapsed_sec": 51091.63392829895, "step_time_sec": 8.24066152703017, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6143, "loss": 3.914275884628296, "lr": 0.0002, "elapsed_sec": 51099.8725938797, "step_time_sec": 8.238518205936998, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6144, "loss": 4.070976257324219, "lr": 0.0002, "elapsed_sec": 51108.11226058006, "step_time_sec": 8.239476635004394, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6145, "loss": 3.891179084777832, "lr": 0.0002, "elapsed_sec": 51116.35181617737, "step_time_sec": 8.23944062110968, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6146, "loss": 4.01839017868042, "lr": 0.0002, "elapsed_sec": 51124.59281253815, "step_time_sec": 8.240818309946917, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6147, "loss": 4.02459192276001, "lr": 0.0002, "elapsed_sec": 51132.833403110504, "step_time_sec": 8.240486819995567, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6148, "loss": 3.9326562881469727, "lr": 0.0002, "elapsed_sec": 51141.07394003868, "step_time_sec": 8.240372360916808, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6149, "loss": 3.945894479751587, "lr": 0.0002, "elapsed_sec": 51149.32573533058, "step_time_sec": 8.251632223953493, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6150, "loss": 3.8873794078826904, "lr": 0.0002, "elapsed_sec": 51157.56598162651, "step_time_sec": 8.239971050992608, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6151, "loss": 3.87430739402771, "lr": 0.0002, "elapsed_sec": 51165.805077552795, "step_time_sec": 8.238953263033181, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6152, "loss": 3.9451723098754883, "lr": 0.0002, "elapsed_sec": 51174.042408943176, "step_time_sec": 8.23713282099925, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6153, "loss": 4.034451484680176, "lr": 0.0002, "elapsed_sec": 51182.2826294899, "step_time_sec": 8.240090526058339, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6154, "loss": 3.965689182281494, "lr": 0.0002, "elapsed_sec": 51190.520584106445, "step_time_sec": 8.237821877002716, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6155, "loss": 3.906275987625122, "lr": 0.0002, "elapsed_sec": 51198.75873732567, "step_time_sec": 8.238039429998025, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6156, "loss": 4.020291805267334, "lr": 0.0002, "elapsed_sec": 51206.99644064903, "step_time_sec": 8.237514114938676, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6157, "loss": 3.9981696605682373, "lr": 0.0002, "elapsed_sec": 51215.23706459999, "step_time_sec": 8.240403125993907, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6158, "loss": 3.955744743347168, "lr": 0.0002, "elapsed_sec": 51223.47611403465, "step_time_sec": 8.238911087973975, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6159, "loss": 3.9984822273254395, "lr": 0.0002, "elapsed_sec": 51231.71460771561, "step_time_sec": 8.238325768033974, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6160, "loss": 4.019150733947754, "lr": 0.0002, "elapsed_sec": 51239.955142736435, "step_time_sec": 8.240343640907668, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6161, "loss": 3.9956533908843994, "lr": 0.0002, "elapsed_sec": 51248.194440841675, "step_time_sec": 8.239168490981683, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6162, "loss": 3.868612766265869, "lr": 0.0002, "elapsed_sec": 51256.434962034225, "step_time_sec": 8.240390506922267, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6163, "loss": 3.986051559448242, "lr": 0.0002, "elapsed_sec": 51264.67436695099, "step_time_sec": 8.239187862025574, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6164, "loss": 3.766209602355957, "lr": 0.0002, "elapsed_sec": 51272.914132118225, "step_time_sec": 8.239597006002441, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6165, "loss": 4.013200283050537, "lr": 0.0002, "elapsed_sec": 51281.154546022415, "step_time_sec": 8.240276806987822, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6166, "loss": 3.908900499343872, "lr": 0.0002, "elapsed_sec": 51289.393726825714, "step_time_sec": 8.238994653103873, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6167, "loss": 3.9614174365997314, "lr": 0.0002, "elapsed_sec": 51297.63068962097, "step_time_sec": 8.236786627909169, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6168, "loss": 4.052263259887695, "lr": 0.0002, "elapsed_sec": 51305.87054014206, "step_time_sec": 8.239729357068427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6169, "loss": 3.9219510555267334, "lr": 0.0002, "elapsed_sec": 51314.11027574539, "step_time_sec": 8.239534329040907, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6170, "loss": 3.9380388259887695, "lr": 0.0002, "elapsed_sec": 51322.34986424446, "step_time_sec": 8.239452054025605, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6171, "loss": 3.9688801765441895, "lr": 0.0002, "elapsed_sec": 51330.590243816376, "step_time_sec": 8.240206512971781, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6172, "loss": 4.048956871032715, "lr": 0.0002, "elapsed_sec": 51338.8284907341, "step_time_sec": 8.238139141001739, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6173, "loss": 3.904109001159668, "lr": 0.0002, "elapsed_sec": 51347.06812787056, "step_time_sec": 8.23947109701112, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6174, "loss": 3.9023890495300293, "lr": 0.0002, "elapsed_sec": 51355.30639076233, "step_time_sec": 8.23805527901277, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6175, "loss": 3.925577163696289, "lr": 0.0002, "elapsed_sec": 51363.54559588432, "step_time_sec": 8.23907605803106, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6176, "loss": 4.0415472984313965, "lr": 0.0002, "elapsed_sec": 51371.78537940979, "step_time_sec": 8.239613070036285, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6177, "loss": 4.021612167358398, "lr": 0.0002, "elapsed_sec": 51380.0236120224, "step_time_sec": 8.238097170018591, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6178, "loss": 4.029481410980225, "lr": 0.0002, "elapsed_sec": 51388.264399290085, "step_time_sec": 8.240669492050074, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6179, "loss": 3.8437676429748535, "lr": 0.0002, "elapsed_sec": 51396.504056453705, "step_time_sec": 8.239482428994961, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6180, "loss": 4.035646915435791, "lr": 0.0002, "elapsed_sec": 51404.74112057686, "step_time_sec": 8.236851538997144, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6181, "loss": 3.7619943618774414, "lr": 0.0002, "elapsed_sec": 51412.982231378555, "step_time_sec": 8.24093916802667, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6182, "loss": 3.7287533283233643, "lr": 0.0002, "elapsed_sec": 51421.22099208832, "step_time_sec": 8.238624618039466, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6183, "loss": 4.001369476318359, "lr": 0.0002, "elapsed_sec": 51429.46090912819, "step_time_sec": 8.23981100297533, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6184, "loss": 3.8962366580963135, "lr": 0.0002, "elapsed_sec": 51437.70098185539, "step_time_sec": 8.239867733092979, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6185, "loss": 3.8980612754821777, "lr": 0.0002, "elapsed_sec": 51445.93911767006, "step_time_sec": 8.238024839083664, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6186, "loss": 3.869837999343872, "lr": 0.0002, "elapsed_sec": 51454.1773686409, "step_time_sec": 8.238042702083476, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6187, "loss": 3.858684539794922, "lr": 0.0002, "elapsed_sec": 51462.41589951515, "step_time_sec": 8.23839220998343, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6188, "loss": 3.870208263397217, "lr": 0.0002, "elapsed_sec": 51470.65533518791, "step_time_sec": 8.239254041924141, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6189, "loss": 3.822758674621582, "lr": 0.0002, "elapsed_sec": 51478.89512991905, "step_time_sec": 8.23967179004103, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6190, "loss": 3.9115161895751953, "lr": 0.0002, "elapsed_sec": 51487.13359975815, "step_time_sec": 8.238327141967602, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6191, "loss": 3.948577642440796, "lr": 0.0002, "elapsed_sec": 51495.3736884594, "step_time_sec": 8.239947335910983, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6192, "loss": 3.8597495555877686, "lr": 0.0002, "elapsed_sec": 51503.613203287125, "step_time_sec": 8.239292814047076, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6193, "loss": 3.9008748531341553, "lr": 0.0002, "elapsed_sec": 51511.8521091938, "step_time_sec": 8.238750661024824, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6194, "loss": 3.870375156402588, "lr": 0.0002, "elapsed_sec": 51520.09151053429, "step_time_sec": 8.239266515010968, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6195, "loss": 3.984143018722534, "lr": 0.0002, "elapsed_sec": 51528.330193281174, "step_time_sec": 8.23850879794918, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6196, "loss": 3.8575024604797363, "lr": 0.0002, "elapsed_sec": 51536.5696246624, "step_time_sec": 8.239339023944922, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6197, "loss": 4.046431541442871, "lr": 0.0002, "elapsed_sec": 51544.80906891823, "step_time_sec": 8.239241574076004, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6198, "loss": 3.9534308910369873, "lr": 0.0002, "elapsed_sec": 51553.048015356064, "step_time_sec": 8.23873356601689, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6199, "loss": 3.9312996864318848, "lr": 0.0002, "elapsed_sec": 51561.287222862244, "step_time_sec": 8.239086609100923, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6200, "loss": 3.8600518703460693, "lr": 0.0002, "elapsed_sec": 51569.52588176727, "step_time_sec": 8.238493761979043, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6201, "loss": 3.9489808082580566, "lr": 0.0002, "elapsed_sec": 51577.765988111496, "step_time_sec": 8.239947914029472, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6202, "loss": 3.9047274589538574, "lr": 0.0002, "elapsed_sec": 51586.0065946579, "step_time_sec": 8.240476020961069, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6203, "loss": 3.884876251220703, "lr": 0.0002, "elapsed_sec": 51594.24362325668, "step_time_sec": 8.236809748923406, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6204, "loss": 3.8752329349517822, "lr": 0.0002, "elapsed_sec": 51602.48331928253, "step_time_sec": 8.239536135923117, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6205, "loss": 3.8209705352783203, "lr": 0.0002, "elapsed_sec": 51610.72144269943, "step_time_sec": 8.23798126692418, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6206, "loss": 3.910114049911499, "lr": 0.0002, "elapsed_sec": 51618.960839271545, "step_time_sec": 8.2392564159818, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6207, "loss": 3.9346554279327393, "lr": 0.0002, "elapsed_sec": 51627.20142674446, "step_time_sec": 8.240414652042091, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6208, "loss": 3.82602596282959, "lr": 0.0002, "elapsed_sec": 51635.43955683708, "step_time_sec": 8.237915455014445, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6209, "loss": 3.852173089981079, "lr": 0.0002, "elapsed_sec": 51643.67893409729, "step_time_sec": 8.239261487033218, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6210, "loss": 3.8549232482910156, "lr": 0.0002, "elapsed_sec": 51651.918758392334, "step_time_sec": 8.23966219893191, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6211, "loss": 3.9504122734069824, "lr": 0.0002, "elapsed_sec": 51660.15891337395, "step_time_sec": 8.239954396034591, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6212, "loss": 4.129492282867432, "lr": 0.0002, "elapsed_sec": 51668.39903497696, "step_time_sec": 8.239981954917312, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6213, "loss": 3.9523584842681885, "lr": 0.0002, "elapsed_sec": 51676.6381688118, "step_time_sec": 8.239010138087906, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6214, "loss": 3.911808967590332, "lr": 0.0002, "elapsed_sec": 51684.8783595562, "step_time_sec": 8.239964948035777, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6215, "loss": 3.9782111644744873, "lr": 0.0002, "elapsed_sec": 51693.11886525154, "step_time_sec": 8.240324298967607, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6216, "loss": 4.144980430603027, "lr": 0.0002, "elapsed_sec": 51701.35685324669, "step_time_sec": 8.237865022965707, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6217, "loss": 4.11576509475708, "lr": 0.0002, "elapsed_sec": 51709.595420598984, "step_time_sec": 8.23839832295198, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6218, "loss": 3.8335764408111572, "lr": 0.0002, "elapsed_sec": 51717.83440923691, "step_time_sec": 8.238809333997779, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6219, "loss": 4.117641925811768, "lr": 0.0002, "elapsed_sec": 51726.07324099541, "step_time_sec": 8.238698067958467, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6220, "loss": 4.244569301605225, "lr": 0.0002, "elapsed_sec": 51734.31198596954, "step_time_sec": 8.238558895071037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6221, "loss": 3.78749418258667, "lr": 0.0002, "elapsed_sec": 51742.61864089966, "step_time_sec": 8.242188018979505, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6222, "loss": 3.778592109680176, "lr": 0.0002, "elapsed_sec": 51750.857033252716, "step_time_sec": 8.238235766999424, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6223, "loss": 4.000006675720215, "lr": 0.0002, "elapsed_sec": 51759.09585976601, "step_time_sec": 8.238672657054849, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6224, "loss": 3.9627270698547363, "lr": 0.0002, "elapsed_sec": 51767.335715293884, "step_time_sec": 8.239701960934326, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6225, "loss": 3.8988895416259766, "lr": 0.0002, "elapsed_sec": 51775.574139118195, "step_time_sec": 8.23821817105636, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6226, "loss": 3.672189474105835, "lr": 0.0002, "elapsed_sec": 51783.814158678055, "step_time_sec": 8.239858129993081, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6227, "loss": 3.712888240814209, "lr": 0.0002, "elapsed_sec": 51792.051693439484, "step_time_sec": 8.237384994979948, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6228, "loss": 3.9667468070983887, "lr": 0.0002, "elapsed_sec": 51800.291472673416, "step_time_sec": 8.239590178010985, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6229, "loss": 3.953507423400879, "lr": 0.0002, "elapsed_sec": 51808.53013944626, "step_time_sec": 8.238498297985643, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6230, "loss": 3.889178991317749, "lr": 0.0002, "elapsed_sec": 51816.76758265495, "step_time_sec": 8.237342073000036, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6231, "loss": 3.9128785133361816, "lr": 0.0002, "elapsed_sec": 51825.00616264343, "step_time_sec": 8.238368581980467, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6232, "loss": 4.040666580200195, "lr": 0.0002, "elapsed_sec": 51833.24361491203, "step_time_sec": 8.237282382906415, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6233, "loss": 3.7485551834106445, "lr": 0.0002, "elapsed_sec": 51841.48324561119, "step_time_sec": 8.23951867595315, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6234, "loss": 4.021271705627441, "lr": 0.0002, "elapsed_sec": 51849.72232246399, "step_time_sec": 8.238934095948935, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6235, "loss": 4.173309326171875, "lr": 0.0002, "elapsed_sec": 51857.961384058, "step_time_sec": 8.238904936006293, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6236, "loss": 3.913477897644043, "lr": 0.0002, "elapsed_sec": 51866.200023412704, "step_time_sec": 8.238458617008291, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6237, "loss": 3.917527437210083, "lr": 0.0002, "elapsed_sec": 51874.43776631355, "step_time_sec": 8.237589246942662, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6238, "loss": 4.001239776611328, "lr": 0.0002, "elapsed_sec": 51882.67875504494, "step_time_sec": 8.240793236065656, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6239, "loss": 3.777393341064453, "lr": 0.0002, "elapsed_sec": 51890.918113946915, "step_time_sec": 8.239211108069867, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6240, "loss": 3.8393330574035645, "lr": 0.0002, "elapsed_sec": 51899.15588378906, "step_time_sec": 8.237640229985118, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6241, "loss": 4.037397861480713, "lr": 0.0002, "elapsed_sec": 51907.394444942474, "step_time_sec": 8.238378452020697, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6242, "loss": 3.9109370708465576, "lr": 0.0002, "elapsed_sec": 51915.63201498985, "step_time_sec": 8.237458305084147, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6243, "loss": 4.004587173461914, "lr": 0.0002, "elapsed_sec": 51923.872089385986, "step_time_sec": 8.2398962599691, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6244, "loss": 4.081582546234131, "lr": 0.0002, "elapsed_sec": 51932.111169576645, "step_time_sec": 8.238901256932877, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6245, "loss": 4.054685592651367, "lr": 0.0002, "elapsed_sec": 51940.34950709343, "step_time_sec": 8.238170158932917, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6246, "loss": 4.0403828620910645, "lr": 0.0002, "elapsed_sec": 51948.58865618706, "step_time_sec": 8.239018381922506, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6247, "loss": 4.082411766052246, "lr": 0.0002, "elapsed_sec": 51956.82519721985, "step_time_sec": 8.2363897790201, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6248, "loss": 3.966099500656128, "lr": 0.0002, "elapsed_sec": 51965.064538002014, "step_time_sec": 8.239166854065843, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6249, "loss": 3.927065134048462, "lr": 0.0002, "elapsed_sec": 51973.30407953262, "step_time_sec": 8.239369145943783, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6250, "loss": 3.9831666946411133, "lr": 0.0002, "elapsed_sec": 51981.543999671936, "step_time_sec": 8.239765990991145, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6251, "loss": 3.9791693687438965, "lr": 0.0002, "elapsed_sec": 51989.7841129303, "step_time_sec": 8.239906604983844, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6252, "loss": 3.8673977851867676, "lr": 0.0002, "elapsed_sec": 51998.02358388901, "step_time_sec": 8.239341162960045, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6253, "loss": 3.8865041732788086, "lr": 0.0002, "elapsed_sec": 52006.263014793396, "step_time_sec": 8.239300378016196, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6254, "loss": 3.992187023162842, "lr": 0.0002, "elapsed_sec": 52014.5020775795, "step_time_sec": 8.238874593982473, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6255, "loss": 4.0109100341796875, "lr": 0.0002, "elapsed_sec": 52022.74010872841, "step_time_sec": 8.237872083089314, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6256, "loss": 3.979417324066162, "lr": 0.0002, "elapsed_sec": 52030.97882127762, "step_time_sec": 8.23852109303698, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6257, "loss": 4.195825099945068, "lr": 0.0002, "elapsed_sec": 52039.21852517128, "step_time_sec": 8.239584299968556, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6258, "loss": 4.257193565368652, "lr": 0.0002, "elapsed_sec": 52047.45866942406, "step_time_sec": 8.2399578030454, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6259, "loss": 3.928187131881714, "lr": 0.0002, "elapsed_sec": 52055.698684215546, "step_time_sec": 8.2398492959328, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6260, "loss": 3.955284833908081, "lr": 0.0002, "elapsed_sec": 52063.93681383133, "step_time_sec": 8.238018910982646, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6261, "loss": 4.072494029998779, "lr": 0.0002, "elapsed_sec": 52072.17669224739, "step_time_sec": 8.239677540026605, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6262, "loss": 4.0296478271484375, "lr": 0.0002, "elapsed_sec": 52080.416299819946, "step_time_sec": 8.239451122004539, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6263, "loss": 3.936002254486084, "lr": 0.0002, "elapsed_sec": 52088.65559864044, "step_time_sec": 8.239161216071807, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6264, "loss": 3.944843053817749, "lr": 0.0002, "elapsed_sec": 52096.894411087036, "step_time_sec": 8.238653574953787, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6265, "loss": 3.9524335861206055, "lr": 0.0002, "elapsed_sec": 52105.13201570511, "step_time_sec": 8.237388610024936, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6266, "loss": 4.034337520599365, "lr": 0.0002, "elapsed_sec": 52113.37080860138, "step_time_sec": 8.238654083106667, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6267, "loss": 3.891322135925293, "lr": 0.0002, "elapsed_sec": 52121.6103618145, "step_time_sec": 8.239427912980318, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6268, "loss": 4.013558387756348, "lr": 0.0002, "elapsed_sec": 52129.84859275818, "step_time_sec": 8.238041345961392, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6269, "loss": 3.9845006465911865, "lr": 0.0002, "elapsed_sec": 52138.08781027794, "step_time_sec": 8.239067553076893, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6270, "loss": 3.934425115585327, "lr": 0.0002, "elapsed_sec": 52146.32553601265, "step_time_sec": 8.237615171005018, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6271, "loss": 3.933718681335449, "lr": 0.0002, "elapsed_sec": 52154.56581759453, "step_time_sec": 8.240115112974308, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6272, "loss": 4.039102554321289, "lr": 0.0002, "elapsed_sec": 52162.80599999428, "step_time_sec": 8.240024948958308, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6273, "loss": 4.04306173324585, "lr": 0.0002, "elapsed_sec": 52171.043904066086, "step_time_sec": 8.23774092597887, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6274, "loss": 3.9281418323516846, "lr": 0.0002, "elapsed_sec": 52179.28360056877, "step_time_sec": 8.239519612980075, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6275, "loss": 4.036200523376465, "lr": 0.0002, "elapsed_sec": 52187.52222466469, "step_time_sec": 8.238475539023057, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6276, "loss": 3.973159074783325, "lr": 0.0002, "elapsed_sec": 52195.761502981186, "step_time_sec": 8.239128047018312, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6277, "loss": 4.024532794952393, "lr": 0.0002, "elapsed_sec": 52204.00143647194, "step_time_sec": 8.239744061022066, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6278, "loss": 3.846238613128662, "lr": 0.0002, "elapsed_sec": 52212.24091768265, "step_time_sec": 8.239309942000546, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6279, "loss": 3.910830497741699, "lr": 0.0002, "elapsed_sec": 52220.48188447952, "step_time_sec": 8.240857438067906, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6280, "loss": 4.193355083465576, "lr": 0.0002, "elapsed_sec": 52228.72259569168, "step_time_sec": 8.240542955929413, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6281, "loss": 3.999899387359619, "lr": 0.0002, "elapsed_sec": 52236.96207547188, "step_time_sec": 8.239368423004635, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6282, "loss": 3.892578601837158, "lr": 0.0002, "elapsed_sec": 52245.20175528526, "step_time_sec": 8.239464046084322, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6283, "loss": 4.040317058563232, "lr": 0.0002, "elapsed_sec": 52253.441076517105, "step_time_sec": 8.239181826938875, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6284, "loss": 3.96390700340271, "lr": 0.0002, "elapsed_sec": 52261.67976164818, "step_time_sec": 8.238559534074739, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6285, "loss": 4.092876434326172, "lr": 0.0002, "elapsed_sec": 52269.919358968735, "step_time_sec": 8.23936796898488, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6286, "loss": 4.079198360443115, "lr": 0.0002, "elapsed_sec": 52278.156567811966, "step_time_sec": 8.237037174054421, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6287, "loss": 4.113961696624756, "lr": 0.0002, "elapsed_sec": 52286.39690876007, "step_time_sec": 8.240242264000699, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6288, "loss": 3.9729199409484863, "lr": 0.0002, "elapsed_sec": 52294.63509893417, "step_time_sec": 8.237999335047789, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6289, "loss": 3.908883810043335, "lr": 0.0002, "elapsed_sec": 52302.87495446205, "step_time_sec": 8.239681822946295, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6290, "loss": 3.983381986618042, "lr": 0.0002, "elapsed_sec": 52311.11524558067, "step_time_sec": 8.24013821198605, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6291, "loss": 4.05990743637085, "lr": 0.0002, "elapsed_sec": 52319.35207939148, "step_time_sec": 8.236706262105145, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6292, "loss": 3.8849475383758545, "lr": 0.0002, "elapsed_sec": 52327.5919444561, "step_time_sec": 8.239678621990606, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6293, "loss": 3.8775641918182373, "lr": 0.0002, "elapsed_sec": 52335.83156490326, "step_time_sec": 8.23947828693781, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6294, "loss": 3.8519344329833984, "lr": 0.0002, "elapsed_sec": 52344.071638822556, "step_time_sec": 8.239955350058153, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6295, "loss": 4.107296943664551, "lr": 0.0002, "elapsed_sec": 52352.31200313568, "step_time_sec": 8.24014878901653, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6296, "loss": 3.9238011837005615, "lr": 0.0002, "elapsed_sec": 52360.55063080788, "step_time_sec": 8.238429127028212, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6297, "loss": 3.8822028636932373, "lr": 0.0002, "elapsed_sec": 52368.7904651165, "step_time_sec": 8.239707212080248, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6298, "loss": 3.962290048599243, "lr": 0.0002, "elapsed_sec": 52377.02867913246, "step_time_sec": 8.238054273999296, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6299, "loss": 3.9319992065429688, "lr": 0.0002, "elapsed_sec": 52385.26704931259, "step_time_sec": 8.238146471092477, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6300, "loss": 3.93154239654541, "lr": 0.0002, "elapsed_sec": 52393.508061885834, "step_time_sec": 8.240922824013978, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6301, "loss": 4.097264766693115, "lr": 0.0002, "elapsed_sec": 52401.74642634392, "step_time_sec": 8.23820406908635, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6302, "loss": 4.025487422943115, "lr": 0.0002, "elapsed_sec": 52409.98681116104, "step_time_sec": 8.240170488017611, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6303, "loss": 4.072266101837158, "lr": 0.0002, "elapsed_sec": 52418.227459430695, "step_time_sec": 8.24049339001067, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6304, "loss": 4.021612644195557, "lr": 0.0002, "elapsed_sec": 52426.46522831917, "step_time_sec": 8.237671626033261, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6305, "loss": 3.9134185314178467, "lr": 0.0002, "elapsed_sec": 52434.704179525375, "step_time_sec": 8.238763169036247, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6306, "loss": 3.9459683895111084, "lr": 0.0002, "elapsed_sec": 52442.94243431091, "step_time_sec": 8.238096551969647, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6307, "loss": 3.8667404651641846, "lr": 0.0002, "elapsed_sec": 52451.18213009834, "step_time_sec": 8.239455927046947, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6308, "loss": 4.044455528259277, "lr": 0.0002, "elapsed_sec": 52459.42187452316, "step_time_sec": 8.239636592916213, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6309, "loss": 4.086817264556885, "lr": 0.0002, "elapsed_sec": 52467.659317970276, "step_time_sec": 8.237254680949263, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6310, "loss": 4.049266338348389, "lr": 0.0002, "elapsed_sec": 52475.90066719055, "step_time_sec": 8.241196729009971, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6311, "loss": 4.079002857208252, "lr": 0.0002, "elapsed_sec": 52484.13962125778, "step_time_sec": 8.238767798990011, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6312, "loss": 4.026549339294434, "lr": 0.0002, "elapsed_sec": 52492.3793296814, "step_time_sec": 8.239546398050152, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6313, "loss": 4.098391056060791, "lr": 0.0002, "elapsed_sec": 52500.61861920357, "step_time_sec": 8.239136998075992, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6314, "loss": 4.085630893707275, "lr": 0.0002, "elapsed_sec": 52508.85776782036, "step_time_sec": 8.239003367954865, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6315, "loss": 4.006505012512207, "lr": 0.0002, "elapsed_sec": 52517.09940099716, "step_time_sec": 8.241489090025425, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6316, "loss": 3.896385669708252, "lr": 0.0002, "elapsed_sec": 52525.33838200569, "step_time_sec": 8.238756365026347, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6317, "loss": 3.914520502090454, "lr": 0.0002, "elapsed_sec": 52533.576333999634, "step_time_sec": 8.237820607959293, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6318, "loss": 3.8880369663238525, "lr": 0.0002, "elapsed_sec": 52541.81628727913, "step_time_sec": 8.239803811069578, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6319, "loss": 4.040862560272217, "lr": 0.0002, "elapsed_sec": 52550.05474114418, "step_time_sec": 8.238278753939085, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6320, "loss": 4.124720573425293, "lr": 0.0002, "elapsed_sec": 52558.29427790642, "step_time_sec": 8.239366119960323, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6321, "loss": 3.797121047973633, "lr": 0.0002, "elapsed_sec": 52566.53293323517, "step_time_sec": 8.238548953086138, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6322, "loss": 4.018161296844482, "lr": 0.0002, "elapsed_sec": 52574.77179932594, "step_time_sec": 8.23867326497566, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6323, "loss": 3.8725242614746094, "lr": 0.0002, "elapsed_sec": 52583.01162171364, "step_time_sec": 8.239641867927276, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6324, "loss": 4.031326770782471, "lr": 0.0002, "elapsed_sec": 52591.25075888634, "step_time_sec": 8.238977749017067, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6325, "loss": 4.000059604644775, "lr": 0.0002, "elapsed_sec": 52599.48956203461, "step_time_sec": 8.238630407955498, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6326, "loss": 3.878939151763916, "lr": 0.0002, "elapsed_sec": 52607.729362249374, "step_time_sec": 8.239634404075332, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6327, "loss": 4.016806125640869, "lr": 0.0002, "elapsed_sec": 52615.96684932709, "step_time_sec": 8.237338450970128, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6328, "loss": 3.937066078186035, "lr": 0.0002, "elapsed_sec": 52624.20643544197, "step_time_sec": 8.239387481939048, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6329, "loss": 3.697253942489624, "lr": 0.0002, "elapsed_sec": 52632.44639420509, "step_time_sec": 8.239801969961263, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6330, "loss": 3.978834867477417, "lr": 0.0002, "elapsed_sec": 52640.68313074112, "step_time_sec": 8.236598570947535, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6331, "loss": 3.9858667850494385, "lr": 0.0002, "elapsed_sec": 52648.92215847969, "step_time_sec": 8.238837211974896, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6332, "loss": 3.878140926361084, "lr": 0.0002, "elapsed_sec": 52657.158828258514, "step_time_sec": 8.236581623088568, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6333, "loss": 4.028689861297607, "lr": 0.0002, "elapsed_sec": 52665.39794445038, "step_time_sec": 8.238887807936408, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6334, "loss": 4.067182540893555, "lr": 0.0002, "elapsed_sec": 52673.63648056984, "step_time_sec": 8.238401259994134, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6335, "loss": 3.9347691535949707, "lr": 0.0002, "elapsed_sec": 52681.87626576424, "step_time_sec": 8.239631406962872, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6336, "loss": 3.9711802005767822, "lr": 0.0002, "elapsed_sec": 52690.115970373154, "step_time_sec": 8.239541216986254, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6337, "loss": 3.917801856994629, "lr": 0.0002, "elapsed_sec": 52698.3534617424, "step_time_sec": 8.237297810032032, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6338, "loss": 4.0574469566345215, "lr": 0.0002, "elapsed_sec": 52706.59379863739, "step_time_sec": 8.240217396989465, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6339, "loss": 3.9804086685180664, "lr": 0.0002, "elapsed_sec": 52714.83276343346, "step_time_sec": 8.238768895971589, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6340, "loss": 4.011727809906006, "lr": 0.0002, "elapsed_sec": 52723.07100701332, "step_time_sec": 8.23805421194993, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6341, "loss": 3.9170305728912354, "lr": 0.0002, "elapsed_sec": 52731.31104969978, "step_time_sec": 8.239871429977939, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6342, "loss": 3.9717540740966797, "lr": 0.0002, "elapsed_sec": 52739.54861330986, "step_time_sec": 8.237421295954846, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6343, "loss": 3.9827733039855957, "lr": 0.0002, "elapsed_sec": 52747.788523197174, "step_time_sec": 8.239749102038331, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6344, "loss": 4.147298336029053, "lr": 0.0002, "elapsed_sec": 52756.02977156639, "step_time_sec": 8.24110143398866, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6345, "loss": 4.1289963722229, "lr": 0.0002, "elapsed_sec": 52764.268476724625, "step_time_sec": 8.238494624034502, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6346, "loss": 3.9949729442596436, "lr": 0.0002, "elapsed_sec": 52772.506729602814, "step_time_sec": 8.238104657968506, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6347, "loss": 4.02663516998291, "lr": 0.0002, "elapsed_sec": 52780.74402666092, "step_time_sec": 8.23712656099815, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6348, "loss": 4.0110063552856445, "lr": 0.0002, "elapsed_sec": 52788.98325943947, "step_time_sec": 8.239092453033663, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6349, "loss": 3.937761068344116, "lr": 0.0002, "elapsed_sec": 52797.222168684006, "step_time_sec": 8.238753322977573, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6350, "loss": 3.9009320735931396, "lr": 0.0002, "elapsed_sec": 52805.46102786064, "step_time_sec": 8.238639864954166, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6351, "loss": 4.201737880706787, "lr": 0.0002, "elapsed_sec": 52813.70060420036, "step_time_sec": 8.239480728982016, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6352, "loss": 3.8922853469848633, "lr": 0.0002, "elapsed_sec": 52821.93931722641, "step_time_sec": 8.23854544700589, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6353, "loss": 4.086575031280518, "lr": 0.0002, "elapsed_sec": 52830.17927002907, "step_time_sec": 8.239767664927058, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6354, "loss": 4.031437397003174, "lr": 0.0002, "elapsed_sec": 52838.41879153252, "step_time_sec": 8.239339152001776, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6355, "loss": 3.9981095790863037, "lr": 0.0002, "elapsed_sec": 52846.657526016235, "step_time_sec": 8.238661102950573, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6356, "loss": 4.053003787994385, "lr": 0.0002, "elapsed_sec": 52854.898183345795, "step_time_sec": 8.240448081982322, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6357, "loss": 3.978165626525879, "lr": 0.0002, "elapsed_sec": 52863.13701581955, "step_time_sec": 8.23864215996582, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6358, "loss": 3.9415934085845947, "lr": 0.0002, "elapsed_sec": 52871.37561583519, "step_time_sec": 8.238445388968103, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6359, "loss": 3.90868878364563, "lr": 0.0002, "elapsed_sec": 52879.616273880005, "step_time_sec": 8.240554157062434, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6360, "loss": 3.997009038925171, "lr": 0.0002, "elapsed_sec": 52887.85529565811, "step_time_sec": 8.238873964059167, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6361, "loss": 3.967912197113037, "lr": 0.0002, "elapsed_sec": 52896.0958211422, "step_time_sec": 8.240306023042649, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6362, "loss": 4.086536884307861, "lr": 0.0002, "elapsed_sec": 52904.33419537544, "step_time_sec": 8.238215286051854, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6363, "loss": 3.859569787979126, "lr": 0.0002, "elapsed_sec": 52912.57281947136, "step_time_sec": 8.238469842006452, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6364, "loss": 3.9836769104003906, "lr": 0.0002, "elapsed_sec": 52920.81339049339, "step_time_sec": 8.240449566976167, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6365, "loss": 4.019556999206543, "lr": 0.0002, "elapsed_sec": 52929.05192065239, "step_time_sec": 8.238288388005458, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6366, "loss": 3.9208483695983887, "lr": 0.0002, "elapsed_sec": 52937.291625499725, "step_time_sec": 8.239606169052422, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6367, "loss": 4.057679653167725, "lr": 0.0002, "elapsed_sec": 52945.52986073494, "step_time_sec": 8.238016198971309, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6368, "loss": 3.994380235671997, "lr": 0.0002, "elapsed_sec": 52953.767740011215, "step_time_sec": 8.237749206018634, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6369, "loss": 4.139214992523193, "lr": 0.0002, "elapsed_sec": 52962.00800156593, "step_time_sec": 8.240123220952228, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6370, "loss": 4.00244140625, "lr": 0.0002, "elapsed_sec": 52970.24649333954, "step_time_sec": 8.238288864027709, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6371, "loss": 3.9234089851379395, "lr": 0.0002, "elapsed_sec": 52978.48552632332, "step_time_sec": 8.238866448053159, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6372, "loss": 4.073566913604736, "lr": 0.0002, "elapsed_sec": 52986.72526907921, "step_time_sec": 8.239630625001155, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6373, "loss": 3.997404098510742, "lr": 0.0002, "elapsed_sec": 52994.96325325966, "step_time_sec": 8.23778549907729, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6374, "loss": 4.019168376922607, "lr": 0.0002, "elapsed_sec": 53003.20355606079, "step_time_sec": 8.240137961925939, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6375, "loss": 3.9688844680786133, "lr": 0.0002, "elapsed_sec": 53011.44246053696, "step_time_sec": 8.238737758016214, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6376, "loss": 3.9906718730926514, "lr": 0.0002, "elapsed_sec": 53019.681732177734, "step_time_sec": 8.239125019987114, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6377, "loss": 3.8239247798919678, "lr": 0.0002, "elapsed_sec": 53027.921199798584, "step_time_sec": 8.239268039935268, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6378, "loss": 3.941298723220825, "lr": 0.0002, "elapsed_sec": 53036.15836071968, "step_time_sec": 8.237027456983924, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6379, "loss": 3.937743902206421, "lr": 0.0002, "elapsed_sec": 53044.39863848686, "step_time_sec": 8.240162947913632, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6380, "loss": 4.094953536987305, "lr": 0.0002, "elapsed_sec": 53052.63643479347, "step_time_sec": 8.237567279022187, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6381, "loss": 4.068798065185547, "lr": 0.0002, "elapsed_sec": 53060.8743994236, "step_time_sec": 8.237837417051196, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6382, "loss": 3.9606614112854004, "lr": 0.0002, "elapsed_sec": 53069.113519907, "step_time_sec": 8.238915735972114, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6383, "loss": 3.9414823055267334, "lr": 0.0002, "elapsed_sec": 53077.35258626938, "step_time_sec": 8.238955341046676, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6384, "loss": 4.029609680175781, "lr": 0.0002, "elapsed_sec": 53085.59392619133, "step_time_sec": 8.24111752898898, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6385, "loss": 3.859971523284912, "lr": 0.0002, "elapsed_sec": 53093.83404445648, "step_time_sec": 8.239974777912721, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6386, "loss": 3.9624977111816406, "lr": 0.0002, "elapsed_sec": 53102.072999715805, "step_time_sec": 8.238821109989658, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6387, "loss": 3.964022397994995, "lr": 0.0002, "elapsed_sec": 53110.31313085556, "step_time_sec": 8.239931076997891, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6388, "loss": 3.9717845916748047, "lr": 0.0002, "elapsed_sec": 53118.551161527634, "step_time_sec": 8.237916920916177, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6389, "loss": 4.018810272216797, "lr": 0.0002, "elapsed_sec": 53126.79122567177, "step_time_sec": 8.23988248698879, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6390, "loss": 4.025184631347656, "lr": 0.0002, "elapsed_sec": 53135.030168533325, "step_time_sec": 8.238754627062008, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6391, "loss": 3.859227180480957, "lr": 0.0002, "elapsed_sec": 53143.26859688759, "step_time_sec": 8.238271052017808, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6392, "loss": 4.136579990386963, "lr": 0.0002, "elapsed_sec": 53151.508016347885, "step_time_sec": 8.239262038958259, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6393, "loss": 4.103107452392578, "lr": 0.0002, "elapsed_sec": 53159.74604296684, "step_time_sec": 8.237861281028017, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6394, "loss": 4.007992744445801, "lr": 0.0002, "elapsed_sec": 53167.98485445976, "step_time_sec": 8.238646934973076, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6395, "loss": 3.9985740184783936, "lr": 0.0002, "elapsed_sec": 53176.226018190384, "step_time_sec": 8.241033739992417, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6396, "loss": 4.039003849029541, "lr": 0.0002, "elapsed_sec": 53184.46431016922, "step_time_sec": 8.23813292896375, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6397, "loss": 3.9883811473846436, "lr": 0.0002, "elapsed_sec": 53192.704474925995, "step_time_sec": 8.240019881981425, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6398, "loss": 3.9268858432769775, "lr": 0.0002, "elapsed_sec": 53200.94459986687, "step_time_sec": 8.239968999987468, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6399, "loss": 3.9906182289123535, "lr": 0.0002, "elapsed_sec": 53209.181843042374, "step_time_sec": 8.237055890029296, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6400, "loss": 4.017326831817627, "lr": 0.0002, "elapsed_sec": 53217.42137145996, "step_time_sec": 8.239408824010752, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6401, "loss": 3.9895129203796387, "lr": 0.0002, "elapsed_sec": 53225.65953230858, "step_time_sec": 8.237967096967623, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6402, "loss": 4.045604705810547, "lr": 0.0002, "elapsed_sec": 53233.899789094925, "step_time_sec": 8.24015754496213, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6403, "loss": 3.9778010845184326, "lr": 0.0002, "elapsed_sec": 53242.13934636116, "step_time_sec": 8.239335107966326, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6404, "loss": 4.0377044677734375, "lr": 0.0002, "elapsed_sec": 53250.37739300728, "step_time_sec": 8.237893434008583, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6405, "loss": 4.03519344329834, "lr": 0.0002, "elapsed_sec": 53258.61661195755, "step_time_sec": 8.239068091963418, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6406, "loss": 3.9213943481445312, "lr": 0.0002, "elapsed_sec": 53266.85523843765, "step_time_sec": 8.238499206956476, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6407, "loss": 4.0246758460998535, "lr": 0.0002, "elapsed_sec": 53275.09422636032, "step_time_sec": 8.23878512706142, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6408, "loss": 3.918342351913452, "lr": 0.0002, "elapsed_sec": 53283.33337807655, "step_time_sec": 8.238988725934178, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6409, "loss": 4.0077409744262695, "lr": 0.0002, "elapsed_sec": 53291.57249379158, "step_time_sec": 8.238993015955202, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6410, "loss": 3.9711015224456787, "lr": 0.0002, "elapsed_sec": 53299.81301045418, "step_time_sec": 8.240307786967605, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6411, "loss": 3.7828891277313232, "lr": 0.0002, "elapsed_sec": 53308.052641153336, "step_time_sec": 8.239469332038425, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6412, "loss": 4.049727916717529, "lr": 0.0002, "elapsed_sec": 53316.291395902634, "step_time_sec": 8.238615131936967, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6413, "loss": 3.8867034912109375, "lr": 0.0002, "elapsed_sec": 53324.530139923096, "step_time_sec": 8.238578189979307, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6414, "loss": 4.032956600189209, "lr": 0.0002, "elapsed_sec": 53332.7676255703, "step_time_sec": 8.237322399043478, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6415, "loss": 3.999136209487915, "lr": 0.0002, "elapsed_sec": 53341.005800008774, "step_time_sec": 8.23801673098933, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6416, "loss": 4.0927534103393555, "lr": 0.0002, "elapsed_sec": 53349.2452981472, "step_time_sec": 8.239328588009812, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6417, "loss": 3.9384095668792725, "lr": 0.0002, "elapsed_sec": 53357.4838950634, "step_time_sec": 8.238485398935154, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6418, "loss": 3.811631917953491, "lr": 0.0002, "elapsed_sec": 53365.72354531288, "step_time_sec": 8.23948847304564, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6419, "loss": 3.9281628131866455, "lr": 0.0002, "elapsed_sec": 53373.96184229851, "step_time_sec": 8.2381013330305, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6420, "loss": 3.9229602813720703, "lr": 0.0002, "elapsed_sec": 53382.20231771469, "step_time_sec": 8.240343221928924, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6421, "loss": 4.015458583831787, "lr": 0.0002, "elapsed_sec": 53390.44200396538, "step_time_sec": 8.239550928934477, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6422, "loss": 4.064155578613281, "lr": 0.0002, "elapsed_sec": 53398.681290864944, "step_time_sec": 8.239089949056506, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6423, "loss": 3.913038730621338, "lr": 0.0002, "elapsed_sec": 53406.91947340965, "step_time_sec": 8.238046001992188, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6424, "loss": 4.033903121948242, "lr": 0.0002, "elapsed_sec": 53415.157067775726, "step_time_sec": 8.237378555932082, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6425, "loss": 4.017990589141846, "lr": 0.0002, "elapsed_sec": 53423.3969874382, "step_time_sec": 8.239768864004873, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6426, "loss": 4.028083324432373, "lr": 0.0002, "elapsed_sec": 53431.63615369797, "step_time_sec": 8.238998651970178, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6427, "loss": 4.049785137176514, "lr": 0.0002, "elapsed_sec": 53439.875529527664, "step_time_sec": 8.23922848701477, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6428, "loss": 4.141582489013672, "lr": 0.0002, "elapsed_sec": 53448.11381983757, "step_time_sec": 8.238061084994115, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6429, "loss": 4.0733418464660645, "lr": 0.0002, "elapsed_sec": 53456.35161781311, "step_time_sec": 8.237650824012235, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6430, "loss": 3.898533344268799, "lr": 0.0002, "elapsed_sec": 53464.59106588364, "step_time_sec": 8.239286473020911, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6431, "loss": 3.7365360260009766, "lr": 0.0002, "elapsed_sec": 53472.829676389694, "step_time_sec": 8.238447292009369, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6432, "loss": 3.917518138885498, "lr": 0.0002, "elapsed_sec": 53481.06933259964, "step_time_sec": 8.239477450028062, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6433, "loss": 3.9910011291503906, "lr": 0.0002, "elapsed_sec": 53489.30883216858, "step_time_sec": 8.239349554060027, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6434, "loss": 3.9242360591888428, "lr": 0.0002, "elapsed_sec": 53497.548142910004, "step_time_sec": 8.239179600030184, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6435, "loss": 3.9263916015625, "lr": 0.0002, "elapsed_sec": 53505.78718805313, "step_time_sec": 8.23887315497268, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6436, "loss": 3.9073047637939453, "lr": 0.0002, "elapsed_sec": 53514.02677130699, "step_time_sec": 8.239411310991272, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6437, "loss": 4.083890438079834, "lr": 0.0002, "elapsed_sec": 53522.26331329346, "step_time_sec": 8.23640102497302, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6438, "loss": 4.054958820343018, "lr": 0.0002, "elapsed_sec": 53530.501987457275, "step_time_sec": 8.238522005965933, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6439, "loss": 3.9710559844970703, "lr": 0.0002, "elapsed_sec": 53538.740953445435, "step_time_sec": 8.238793728058226, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6440, "loss": 3.9014079570770264, "lr": 0.0002, "elapsed_sec": 53546.98218154907, "step_time_sec": 8.241087489994243, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6441, "loss": 3.9958155155181885, "lr": 0.0002, "elapsed_sec": 53555.224952697754, "step_time_sec": 8.24260342901107, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6442, "loss": 3.902186870574951, "lr": 0.0002, "elapsed_sec": 53563.46319580078, "step_time_sec": 8.238040250958875, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6443, "loss": 3.8567845821380615, "lr": 0.0002, "elapsed_sec": 53571.70252037048, "step_time_sec": 8.239246290992014, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6444, "loss": 3.8706493377685547, "lr": 0.0002, "elapsed_sec": 53579.940769433975, "step_time_sec": 8.237994836992584, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6445, "loss": 4.041216850280762, "lr": 0.0002, "elapsed_sec": 53588.17910504341, "step_time_sec": 8.238198960083537, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6446, "loss": 3.8657169342041016, "lr": 0.0002, "elapsed_sec": 53596.41901016235, "step_time_sec": 8.239710723049939, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6447, "loss": 3.670084238052368, "lr": 0.0002, "elapsed_sec": 53604.6574010849, "step_time_sec": 8.238240603008308, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6448, "loss": 3.828385829925537, "lr": 0.0002, "elapsed_sec": 53612.89703655243, "step_time_sec": 8.239483688026667, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6449, "loss": 3.8087260723114014, "lr": 0.0002, "elapsed_sec": 53621.13739275932, "step_time_sec": 8.24017584696412, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6450, "loss": 4.274536609649658, "lr": 0.0002, "elapsed_sec": 53629.37632083893, "step_time_sec": 8.238737424020655, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6451, "loss": 3.97835373878479, "lr": 0.0002, "elapsed_sec": 53637.616525411606, "step_time_sec": 8.240077469032258, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6452, "loss": 3.9082558155059814, "lr": 0.0002, "elapsed_sec": 53645.856226205826, "step_time_sec": 8.239514349028468, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6453, "loss": 4.051574230194092, "lr": 0.0002, "elapsed_sec": 53654.095291376114, "step_time_sec": 8.238893270026892, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6454, "loss": 3.916947603225708, "lr": 0.0002, "elapsed_sec": 53662.335087776184, "step_time_sec": 8.239675658987835, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6455, "loss": 4.112372875213623, "lr": 0.0002, "elapsed_sec": 53670.573289871216, "step_time_sec": 8.238002842990682, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6456, "loss": 4.018118381500244, "lr": 0.0002, "elapsed_sec": 53678.81394815445, "step_time_sec": 8.240475483005866, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6457, "loss": 3.955976724624634, "lr": 0.0002, "elapsed_sec": 53687.05308222771, "step_time_sec": 8.239004807081074, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6458, "loss": 3.913033962249756, "lr": 0.0002, "elapsed_sec": 53695.291988134384, "step_time_sec": 8.23869652301073, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6459, "loss": 3.984344482421875, "lr": 0.0002, "elapsed_sec": 53703.53130722046, "step_time_sec": 8.239167838008143, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6460, "loss": 3.9192605018615723, "lr": 0.0002, "elapsed_sec": 53711.76945900917, "step_time_sec": 8.237981923972256, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6461, "loss": 3.938890218734741, "lr": 0.0002, "elapsed_sec": 53720.011003255844, "step_time_sec": 8.24135231191758, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6462, "loss": 3.835263729095459, "lr": 0.0002, "elapsed_sec": 53728.25076150894, "step_time_sec": 8.239586689975113, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6463, "loss": 4.0271453857421875, "lr": 0.0002, "elapsed_sec": 53736.489396095276, "step_time_sec": 8.23854776204098, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6464, "loss": 3.834249496459961, "lr": 0.0002, "elapsed_sec": 53744.730170726776, "step_time_sec": 8.240556132048368, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6465, "loss": 3.9736225605010986, "lr": 0.0002, "elapsed_sec": 53752.96960520744, "step_time_sec": 8.239322689012624, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6466, "loss": 3.8961074352264404, "lr": 0.0002, "elapsed_sec": 53761.20887231827, "step_time_sec": 8.23905818699859, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6467, "loss": 4.051596641540527, "lr": 0.0002, "elapsed_sec": 53769.44743824005, "step_time_sec": 8.238423232920468, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6468, "loss": 4.025034427642822, "lr": 0.0002, "elapsed_sec": 53777.68610429764, "step_time_sec": 8.238528436049819, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6469, "loss": 4.110292911529541, "lr": 0.0002, "elapsed_sec": 53785.926224946976, "step_time_sec": 8.239958859980106, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6470, "loss": 4.022669315338135, "lr": 0.0002, "elapsed_sec": 53794.16584968567, "step_time_sec": 8.239455215982161, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6471, "loss": 3.946260690689087, "lr": 0.0002, "elapsed_sec": 53802.404052734375, "step_time_sec": 8.238040006021038, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6472, "loss": 4.05419921875, "lr": 0.0002, "elapsed_sec": 53810.64293551445, "step_time_sec": 8.238740738015622, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6473, "loss": 3.9374866485595703, "lr": 0.0002, "elapsed_sec": 53818.87975502014, "step_time_sec": 8.236680009984411, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6474, "loss": 4.02734375, "lr": 0.0002, "elapsed_sec": 53827.11905860901, "step_time_sec": 8.239185336977243, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6475, "loss": 4.076154708862305, "lr": 0.0002, "elapsed_sec": 53835.35911011696, "step_time_sec": 8.239868367090821, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6476, "loss": 3.961707830429077, "lr": 0.0002, "elapsed_sec": 53843.597068071365, "step_time_sec": 8.237763453973457, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6477, "loss": 3.9921796321868896, "lr": 0.0002, "elapsed_sec": 53851.837176799774, "step_time_sec": 8.239963270025328, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6478, "loss": 3.983067035675049, "lr": 0.0002, "elapsed_sec": 53860.07567572594, "step_time_sec": 8.238377950969152, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6479, "loss": 4.015637397766113, "lr": 0.0002, "elapsed_sec": 53868.31543827057, "step_time_sec": 8.239572998019867, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6480, "loss": 3.8053793907165527, "lr": 0.0002, "elapsed_sec": 53876.55414915085, "step_time_sec": 8.238599270931445, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6481, "loss": 3.943908929824829, "lr": 0.0002, "elapsed_sec": 53884.79280257225, "step_time_sec": 8.2384143009549, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6482, "loss": 3.8926515579223633, "lr": 0.0002, "elapsed_sec": 53893.032348155975, "step_time_sec": 8.239441704005003, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6483, "loss": 3.9629225730895996, "lr": 0.0002, "elapsed_sec": 53901.27320075035, "step_time_sec": 8.240621919976547, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6484, "loss": 4.014781951904297, "lr": 0.0002, "elapsed_sec": 53909.512904167175, "step_time_sec": 8.239566387957893, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6485, "loss": 3.9422295093536377, "lr": 0.0002, "elapsed_sec": 53917.75269842148, "step_time_sec": 8.239677536999807, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6486, "loss": 3.9891281127929688, "lr": 0.0002, "elapsed_sec": 53925.99208855629, "step_time_sec": 8.239215702051297, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6487, "loss": 4.000980377197266, "lr": 0.0002, "elapsed_sec": 53934.23220992088, "step_time_sec": 8.239921426051296, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6488, "loss": 3.740363836288452, "lr": 0.0002, "elapsed_sec": 53942.47128844261, "step_time_sec": 8.238913235953078, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6489, "loss": 3.8718509674072266, "lr": 0.0002, "elapsed_sec": 53962.213693618774, "step_time_sec": 19.74223051907029, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6490, "loss": 3.9221913814544678, "lr": 0.0002, "elapsed_sec": 53970.44507026672, "step_time_sec": 8.231180369970389, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6491, "loss": 3.9519145488739014, "lr": 0.0002, "elapsed_sec": 53978.68022489548, "step_time_sec": 8.235039399005473, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6492, "loss": 3.8771414756774902, "lr": 0.0002, "elapsed_sec": 53986.91870331764, "step_time_sec": 8.23834063403774, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6493, "loss": 3.843959093093872, "lr": 0.0002, "elapsed_sec": 53995.15961384773, "step_time_sec": 8.240702423034236, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6494, "loss": 3.980989694595337, "lr": 0.0002, "elapsed_sec": 54003.39773797989, "step_time_sec": 8.237998572993092, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6495, "loss": 4.017395496368408, "lr": 0.0002, "elapsed_sec": 54011.63724374771, "step_time_sec": 8.239308727090247, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6496, "loss": 3.8500113487243652, "lr": 0.0002, "elapsed_sec": 54019.87613558769, "step_time_sec": 8.238752758945338, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6497, "loss": 3.890122652053833, "lr": 0.0002, "elapsed_sec": 54028.11300492287, "step_time_sec": 8.236733826925047, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6498, "loss": 3.9844651222229004, "lr": 0.0002, "elapsed_sec": 54036.353004693985, "step_time_sec": 8.23980472004041, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6499, "loss": 4.089648723602295, "lr": 0.0002, "elapsed_sec": 54044.59302353859, "step_time_sec": 8.239856387954205, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6500, "loss": 3.9794609546661377, "lr": 0.0002, "elapsed_sec": 54052.83160996437, "step_time_sec": 24.267023687949404, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": true, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6501, "loss": 3.7978928089141846, "lr": 0.0002, "elapsed_sec": 54077.11147117615, "step_time_sec": 8.250798834953457, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6502, "loss": 3.837700128555298, "lr": 0.0002, "elapsed_sec": 54085.34780573845, "step_time_sec": 8.236084097996354, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6503, "loss": 3.9339959621429443, "lr": 0.0002, "elapsed_sec": 54093.58668780327, "step_time_sec": 8.23876063595526, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6504, "loss": 3.9382925033569336, "lr": 0.0002, "elapsed_sec": 54101.826103925705, "step_time_sec": 8.239205459947698, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6505, "loss": 3.927506685256958, "lr": 0.0002, "elapsed_sec": 54110.0642786026, "step_time_sec": 8.238027233979665, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6506, "loss": 3.865191698074341, "lr": 0.0002, "elapsed_sec": 54118.30373263359, "step_time_sec": 8.23933765606489, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6507, "loss": 3.787604808807373, "lr": 0.0002, "elapsed_sec": 54126.54325771332, "step_time_sec": 8.239376056008041, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6508, "loss": 4.090160369873047, "lr": 0.0002, "elapsed_sec": 54134.78363609314, "step_time_sec": 8.240164070972241, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6509, "loss": 3.9380533695220947, "lr": 0.0002, "elapsed_sec": 54143.02196645737, "step_time_sec": 8.238200878025964, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6510, "loss": 3.778937578201294, "lr": 0.0002, "elapsed_sec": 54151.26172494888, "step_time_sec": 8.239570830017328, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6511, "loss": 3.9313082695007324, "lr": 0.0002, "elapsed_sec": 54159.50153303146, "step_time_sec": 8.239611400058493, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6512, "loss": 3.946981430053711, "lr": 0.0002, "elapsed_sec": 54167.73970890045, "step_time_sec": 8.238002037047409, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6513, "loss": 3.833416223526001, "lr": 0.0002, "elapsed_sec": 54175.979370594025, "step_time_sec": 8.239504863042384, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6514, "loss": 3.918016195297241, "lr": 0.0002, "elapsed_sec": 54184.21980524063, "step_time_sec": 8.240309185930528, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6515, "loss": 3.9556992053985596, "lr": 0.0002, "elapsed_sec": 54192.45874238014, "step_time_sec": 8.238754076999612, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6516, "loss": 3.86407732963562, "lr": 0.0002, "elapsed_sec": 54200.69839715958, "step_time_sec": 8.239545232965611, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6517, "loss": 3.848188638687134, "lr": 0.0002, "elapsed_sec": 54208.935695171356, "step_time_sec": 8.237075700075366, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6518, "loss": 3.978983163833618, "lr": 0.0002, "elapsed_sec": 54217.175228595734, "step_time_sec": 8.23940889898222, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6519, "loss": 4.077030658721924, "lr": 0.0002, "elapsed_sec": 54225.414257764816, "step_time_sec": 8.238836553995498, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6520, "loss": 3.904581069946289, "lr": 0.0002, "elapsed_sec": 54233.65355706215, "step_time_sec": 8.23913070897106, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6521, "loss": 4.048017978668213, "lr": 0.0002, "elapsed_sec": 54241.89408874512, "step_time_sec": 8.240334070986137, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6522, "loss": 3.8387184143066406, "lr": 0.0002, "elapsed_sec": 54250.13286614418, "step_time_sec": 8.238643698045053, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6523, "loss": 3.8492720127105713, "lr": 0.0002, "elapsed_sec": 54258.37253379822, "step_time_sec": 8.239549912977964, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6524, "loss": 3.967698574066162, "lr": 0.0002, "elapsed_sec": 54266.612258672714, "step_time_sec": 8.239573196973652, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6525, "loss": 4.073078155517578, "lr": 0.0002, "elapsed_sec": 54274.85079240799, "step_time_sec": 8.238308480009437, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6526, "loss": 4.117469787597656, "lr": 0.0002, "elapsed_sec": 54283.08917117119, "step_time_sec": 8.238272188929841, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6527, "loss": 4.089370250701904, "lr": 0.0002, "elapsed_sec": 54291.3288359642, "step_time_sec": 8.239475107053295, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6528, "loss": 3.863710403442383, "lr": 0.0002, "elapsed_sec": 54299.568281412125, "step_time_sec": 8.239272249047644, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6529, "loss": 3.9767725467681885, "lr": 0.0002, "elapsed_sec": 54307.80820465088, "step_time_sec": 8.239738570060581, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6530, "loss": 4.079622268676758, "lr": 0.0002, "elapsed_sec": 54316.04609656334, "step_time_sec": 8.237747219973244, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6531, "loss": 3.861502170562744, "lr": 0.0002, "elapsed_sec": 54324.28579926491, "step_time_sec": 8.239569719997235, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6532, "loss": 3.8877389430999756, "lr": 0.0002, "elapsed_sec": 54332.52520918846, "step_time_sec": 8.239239171030931, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6533, "loss": 3.8362438678741455, "lr": 0.0002, "elapsed_sec": 54340.763847351074, "step_time_sec": 8.238424950977787, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6534, "loss": 3.8889973163604736, "lr": 0.0002, "elapsed_sec": 54349.002744197845, "step_time_sec": 8.238757996936329, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6535, "loss": 3.9374473094940186, "lr": 0.0002, "elapsed_sec": 54357.241349220276, "step_time_sec": 8.238464619964361, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6536, "loss": 3.9579226970672607, "lr": 0.0002, "elapsed_sec": 54365.48194241524, "step_time_sec": 8.240402025054209, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6537, "loss": 3.875885248184204, "lr": 0.0002, "elapsed_sec": 54373.721420526505, "step_time_sec": 8.23933961498551, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6538, "loss": 3.8740944862365723, "lr": 0.0002, "elapsed_sec": 54381.96117782593, "step_time_sec": 8.239551636041142, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6539, "loss": 4.1128950119018555, "lr": 0.0002, "elapsed_sec": 54390.201780080795, "step_time_sec": 8.240473759942688, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6540, "loss": 3.975843667984009, "lr": 0.0002, "elapsed_sec": 54398.440551519394, "step_time_sec": 8.23863405000884, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6541, "loss": 3.9902212619781494, "lr": 0.0002, "elapsed_sec": 54406.68127131462, "step_time_sec": 8.240551283000968, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6542, "loss": 3.854616165161133, "lr": 0.0002, "elapsed_sec": 54414.92043352127, "step_time_sec": 8.23898011399433, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6543, "loss": 3.7875750064849854, "lr": 0.0002, "elapsed_sec": 54423.15957713127, "step_time_sec": 8.238949035992846, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6544, "loss": 3.8802876472473145, "lr": 0.0002, "elapsed_sec": 54431.40095758438, "step_time_sec": 8.241205780999735, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6545, "loss": 3.9340388774871826, "lr": 0.0002, "elapsed_sec": 54439.64140915871, "step_time_sec": 8.240272292052396, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6546, "loss": 3.9086413383483887, "lr": 0.0002, "elapsed_sec": 54447.87934088707, "step_time_sec": 8.237727814936079, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6547, "loss": 4.068613529205322, "lr": 0.0002, "elapsed_sec": 54456.11799263954, "step_time_sec": 8.238516445970163, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6548, "loss": 3.9635984897613525, "lr": 0.0002, "elapsed_sec": 54464.3569996357, "step_time_sec": 8.238854172057472, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6549, "loss": 3.884291648864746, "lr": 0.0002, "elapsed_sec": 54472.59831929207, "step_time_sec": 8.241185996914282, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6550, "loss": 4.05339241027832, "lr": 0.0002, "elapsed_sec": 54480.838805913925, "step_time_sec": 8.240291594062, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6551, "loss": 3.859222412109375, "lr": 0.0002, "elapsed_sec": 54489.0771009922, "step_time_sec": 8.238135709078051, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6552, "loss": 4.0503034591674805, "lr": 0.0002, "elapsed_sec": 54497.31811833382, "step_time_sec": 8.24087256193161, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6553, "loss": 3.9407498836517334, "lr": 0.0002, "elapsed_sec": 54505.557839632034, "step_time_sec": 8.23960473202169, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6554, "loss": 3.9117777347564697, "lr": 0.0002, "elapsed_sec": 54513.79741358757, "step_time_sec": 8.239326117094606, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6555, "loss": 3.9086337089538574, "lr": 0.0002, "elapsed_sec": 54522.03703212738, "step_time_sec": 8.239474994014017, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6556, "loss": 3.832933187484741, "lr": 0.0002, "elapsed_sec": 54530.276773929596, "step_time_sec": 8.239595578983426, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6557, "loss": 3.9622228145599365, "lr": 0.0002, "elapsed_sec": 54538.5171854496, "step_time_sec": 8.24028707598336, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6558, "loss": 4.01061487197876, "lr": 0.0002, "elapsed_sec": 54546.757061719894, "step_time_sec": 8.239723594975658, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6559, "loss": 3.819549798965454, "lr": 0.0002, "elapsed_sec": 54554.996106386185, "step_time_sec": 8.238863955950364, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6560, "loss": 3.868824005126953, "lr": 0.0002, "elapsed_sec": 54563.23623466492, "step_time_sec": 8.239973439951427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6561, "loss": 3.8709588050842285, "lr": 0.0002, "elapsed_sec": 54571.47501993179, "step_time_sec": 8.23863961501047, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6562, "loss": 3.7746405601501465, "lr": 0.0002, "elapsed_sec": 54579.714829444885, "step_time_sec": 8.239597384002991, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6563, "loss": 3.8435442447662354, "lr": 0.0002, "elapsed_sec": 54587.953453063965, "step_time_sec": 8.238454408943653, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6564, "loss": 3.9504125118255615, "lr": 0.0002, "elapsed_sec": 54596.19080543518, "step_time_sec": 8.23718256293796, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6565, "loss": 3.9594128131866455, "lr": 0.0002, "elapsed_sec": 54604.42988181114, "step_time_sec": 8.238972703926265, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6566, "loss": 3.939363479614258, "lr": 0.0002, "elapsed_sec": 54612.667244672775, "step_time_sec": 8.237171343993396, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6567, "loss": 3.971686363220215, "lr": 0.0002, "elapsed_sec": 54620.907242536545, "step_time_sec": 8.239804880111478, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6568, "loss": 4.048854827880859, "lr": 0.0002, "elapsed_sec": 54629.145854234695, "step_time_sec": 8.238455526996404, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6569, "loss": 3.940915584564209, "lr": 0.0002, "elapsed_sec": 54637.38444328308, "step_time_sec": 8.238444017013535, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6570, "loss": 3.869466543197632, "lr": 0.0002, "elapsed_sec": 54645.62411689758, "step_time_sec": 8.23952980095055, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6571, "loss": 3.823763847351074, "lr": 0.0002, "elapsed_sec": 54653.86376810074, "step_time_sec": 8.23946176096797, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6572, "loss": 4.002265453338623, "lr": 0.0002, "elapsed_sec": 54662.103066921234, "step_time_sec": 8.239129166933708, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6573, "loss": 3.93039870262146, "lr": 0.0002, "elapsed_sec": 54670.34312796593, "step_time_sec": 8.23990419099573, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6574, "loss": 3.9023923873901367, "lr": 0.0002, "elapsed_sec": 54678.58308887482, "step_time_sec": 8.239848766941577, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6575, "loss": 3.8738303184509277, "lr": 0.0002, "elapsed_sec": 54686.823835134506, "step_time_sec": 8.240567079046741, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6576, "loss": 3.864245653152466, "lr": 0.0002, "elapsed_sec": 54695.06351852417, "step_time_sec": 8.239467981038615, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6577, "loss": 3.8908464908599854, "lr": 0.0002, "elapsed_sec": 54703.30290412903, "step_time_sec": 8.23923243698664, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6578, "loss": 3.9399354457855225, "lr": 0.0002, "elapsed_sec": 54711.54300761223, "step_time_sec": 8.239967857953161, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6579, "loss": 3.8897719383239746, "lr": 0.0002, "elapsed_sec": 54719.78187131882, "step_time_sec": 8.2386871969793, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6580, "loss": 3.7720999717712402, "lr": 0.0002, "elapsed_sec": 54728.02093935013, "step_time_sec": 8.238958379020914, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6581, "loss": 3.798823356628418, "lr": 0.0002, "elapsed_sec": 54736.26085090637, "step_time_sec": 8.239765061065555, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6582, "loss": 3.8271706104278564, "lr": 0.0002, "elapsed_sec": 54744.50019907951, "step_time_sec": 8.239164595026523, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6583, "loss": 3.951929807662964, "lr": 0.0002, "elapsed_sec": 54752.74189400673, "step_time_sec": 8.241534324013628, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6584, "loss": 4.085700988769531, "lr": 0.0002, "elapsed_sec": 54760.981164455414, "step_time_sec": 8.239065844099969, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6585, "loss": 3.880056142807007, "lr": 0.0002, "elapsed_sec": 54769.22036123276, "step_time_sec": 8.23902356508188, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6586, "loss": 4.045399188995361, "lr": 0.0002, "elapsed_sec": 54777.461614608765, "step_time_sec": 8.241108700982295, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6587, "loss": 3.927363634109497, "lr": 0.0002, "elapsed_sec": 54785.70121407509, "step_time_sec": 8.23948495707009, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6588, "loss": 4.083092212677002, "lr": 0.0002, "elapsed_sec": 54793.93975043297, "step_time_sec": 8.23835667502135, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6589, "loss": 3.8661930561065674, "lr": 0.0002, "elapsed_sec": 54802.17883658409, "step_time_sec": 8.238878365023993, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6590, "loss": 3.9746358394622803, "lr": 0.0002, "elapsed_sec": 54810.417328596115, "step_time_sec": 8.238348647020757, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6591, "loss": 3.9825148582458496, "lr": 0.0002, "elapsed_sec": 54818.65832781792, "step_time_sec": 8.240860752994195, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6592, "loss": 3.778369188308716, "lr": 0.0002, "elapsed_sec": 54826.89835166931, "step_time_sec": 8.239883755915798, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6593, "loss": 4.01329231262207, "lr": 0.0002, "elapsed_sec": 54835.1364068985, "step_time_sec": 8.237834158004262, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6594, "loss": 3.9211032390594482, "lr": 0.0002, "elapsed_sec": 54843.37569141388, "step_time_sec": 8.239121324033476, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6595, "loss": 4.065788745880127, "lr": 0.0002, "elapsed_sec": 54851.61521100998, "step_time_sec": 8.239365766989067, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6596, "loss": 3.901047468185425, "lr": 0.0002, "elapsed_sec": 54859.8545665741, "step_time_sec": 8.239171304972842, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6597, "loss": 4.0573811531066895, "lr": 0.0002, "elapsed_sec": 54868.09441947937, "step_time_sec": 8.23968181910459, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6598, "loss": 3.939014196395874, "lr": 0.0002, "elapsed_sec": 54876.33343744278, "step_time_sec": 8.238861380028538, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6599, "loss": 3.9165046215057373, "lr": 0.0002, "elapsed_sec": 54884.57447504997, "step_time_sec": 8.240877370000817, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6600, "loss": 3.8959567546844482, "lr": 0.0002, "elapsed_sec": 54892.81233406067, "step_time_sec": 8.237695319927298, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6601, "loss": 3.893806219100952, "lr": 0.0002, "elapsed_sec": 54901.05160331726, "step_time_sec": 8.239106049994007, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6602, "loss": 4.137752532958984, "lr": 0.0002, "elapsed_sec": 54909.29157042503, "step_time_sec": 8.239881596062332, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6603, "loss": 3.8622825145721436, "lr": 0.0002, "elapsed_sec": 54917.53050708771, "step_time_sec": 8.23864491097629, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6604, "loss": 3.9099481105804443, "lr": 0.0002, "elapsed_sec": 54925.77000498772, "step_time_sec": 8.239342659944668, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6605, "loss": 3.889763832092285, "lr": 0.0002, "elapsed_sec": 54934.00826072693, "step_time_sec": 8.238058555056341, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6606, "loss": 3.8710849285125732, "lr": 0.0002, "elapsed_sec": 54942.24657320976, "step_time_sec": 8.238153886049986, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6607, "loss": 3.763559341430664, "lr": 0.0002, "elapsed_sec": 54950.48664045334, "step_time_sec": 8.239904552930966, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6608, "loss": 4.047165393829346, "lr": 0.0002, "elapsed_sec": 54958.72392964363, "step_time_sec": 8.237151547917165, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6609, "loss": 3.906049966812134, "lr": 0.0002, "elapsed_sec": 54966.96337723732, "step_time_sec": 8.239295764011331, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6610, "loss": 3.927391529083252, "lr": 0.0002, "elapsed_sec": 54975.20217490196, "step_time_sec": 8.238590324064717, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6611, "loss": 4.072024822235107, "lr": 0.0002, "elapsed_sec": 54983.44052362442, "step_time_sec": 8.238174076075666, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6612, "loss": 4.064703464508057, "lr": 0.0002, "elapsed_sec": 54991.68061423302, "step_time_sec": 8.239937612088397, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6613, "loss": 3.9143333435058594, "lr": 0.0002, "elapsed_sec": 54999.920305252075, "step_time_sec": 8.239534295978956, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6614, "loss": 3.957768678665161, "lr": 0.0002, "elapsed_sec": 55008.15906000137, "step_time_sec": 8.238599207019433, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6615, "loss": 4.146961688995361, "lr": 0.0002, "elapsed_sec": 55016.39851784706, "step_time_sec": 8.239301412948407, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6616, "loss": 3.8527486324310303, "lr": 0.0002, "elapsed_sec": 55024.6364736557, "step_time_sec": 8.237780272960663, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6617, "loss": 3.9940199851989746, "lr": 0.0002, "elapsed_sec": 55032.876440286636, "step_time_sec": 8.239800832001492, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6618, "loss": 3.9568986892700195, "lr": 0.0002, "elapsed_sec": 55041.11737656593, "step_time_sec": 8.240775879006833, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6619, "loss": 4.010634899139404, "lr": 0.0002, "elapsed_sec": 55049.35732984543, "step_time_sec": 8.239755384041928, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6620, "loss": 4.002954006195068, "lr": 0.0002, "elapsed_sec": 55057.59722805023, "step_time_sec": 8.23974090802949, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6621, "loss": 3.8365378379821777, "lr": 0.0002, "elapsed_sec": 55065.835814237595, "step_time_sec": 8.238455633050762, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6622, "loss": 3.7971620559692383, "lr": 0.0002, "elapsed_sec": 55074.07611465454, "step_time_sec": 8.240123988944106, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6623, "loss": 4.208259582519531, "lr": 0.0002, "elapsed_sec": 55082.31515598297, "step_time_sec": 8.238841203041375, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6624, "loss": 3.912167549133301, "lr": 0.0002, "elapsed_sec": 55090.553813934326, "step_time_sec": 8.238497590995394, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6625, "loss": 3.976762533187866, "lr": 0.0002, "elapsed_sec": 55098.79346251488, "step_time_sec": 8.239535331027582, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6626, "loss": 3.991295576095581, "lr": 0.0002, "elapsed_sec": 55107.03160190582, "step_time_sec": 8.238023902056739, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6627, "loss": 3.9359960556030273, "lr": 0.0002, "elapsed_sec": 55115.27198624611, "step_time_sec": 8.240122598945163, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6628, "loss": 3.8803560733795166, "lr": 0.0002, "elapsed_sec": 55123.5125772953, "step_time_sec": 8.240416462998837, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6629, "loss": 4.1165876388549805, "lr": 0.0002, "elapsed_sec": 55131.75057601929, "step_time_sec": 8.23784474504646, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6630, "loss": 3.889921188354492, "lr": 0.0002, "elapsed_sec": 55139.98915052414, "step_time_sec": 8.238394037936814, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6631, "loss": 4.063501358032227, "lr": 0.0002, "elapsed_sec": 55148.22809648514, "step_time_sec": 8.238816094002686, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6632, "loss": 3.985719919204712, "lr": 0.0002, "elapsed_sec": 55156.467164993286, "step_time_sec": 8.238945206976496, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6633, "loss": 3.8931703567504883, "lr": 0.0002, "elapsed_sec": 55164.706815481186, "step_time_sec": 8.23943301208783, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6634, "loss": 3.971184253692627, "lr": 0.0002, "elapsed_sec": 55172.94476890564, "step_time_sec": 8.23779480101075, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6635, "loss": 3.8315467834472656, "lr": 0.0002, "elapsed_sec": 55181.18295145035, "step_time_sec": 8.238014831091277, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6636, "loss": 3.9330763816833496, "lr": 0.0002, "elapsed_sec": 55189.4218044281, "step_time_sec": 8.23867342597805, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6637, "loss": 3.826455593109131, "lr": 0.0002, "elapsed_sec": 55197.66067767143, "step_time_sec": 8.238770261988975, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6638, "loss": 3.733225107192993, "lr": 0.0002, "elapsed_sec": 55205.89917016029, "step_time_sec": 8.238265128922649, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6639, "loss": 3.8909246921539307, "lr": 0.0002, "elapsed_sec": 55214.139169216156, "step_time_sec": 8.239897135994397, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6640, "loss": 3.881666660308838, "lr": 0.0002, "elapsed_sec": 55222.37937045097, "step_time_sec": 8.24000592099037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6641, "loss": 4.0049285888671875, "lr": 0.0002, "elapsed_sec": 55230.61878871918, "step_time_sec": 8.239273356972262, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6642, "loss": 3.9752657413482666, "lr": 0.0002, "elapsed_sec": 55238.857909202576, "step_time_sec": 8.23900744994171, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6643, "loss": 3.9462063312530518, "lr": 0.0002, "elapsed_sec": 55247.09742283821, "step_time_sec": 8.239356354926713, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6644, "loss": 3.8836402893066406, "lr": 0.0002, "elapsed_sec": 55255.3363571167, "step_time_sec": 8.238700060988776, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6645, "loss": 3.874324321746826, "lr": 0.0002, "elapsed_sec": 55263.57619547844, "step_time_sec": 8.239687149995007, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6646, "loss": 3.9866647720336914, "lr": 0.0002, "elapsed_sec": 55271.81526637077, "step_time_sec": 8.2389186139917, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6647, "loss": 4.043267250061035, "lr": 0.0002, "elapsed_sec": 55280.05376601219, "step_time_sec": 8.23834616097156, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6648, "loss": 3.8357131481170654, "lr": 0.0002, "elapsed_sec": 55288.294534921646, "step_time_sec": 8.24061597196851, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6649, "loss": 3.970144271850586, "lr": 0.0002, "elapsed_sec": 55296.53381586075, "step_time_sec": 8.239166036015376, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6650, "loss": 3.8688595294952393, "lr": 0.0002, "elapsed_sec": 55304.77367424965, "step_time_sec": 8.239638376049697, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6651, "loss": 4.035975933074951, "lr": 0.0002, "elapsed_sec": 55313.01161360741, "step_time_sec": 8.237768749939278, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6652, "loss": 3.891925573348999, "lr": 0.0002, "elapsed_sec": 55321.2477118969, "step_time_sec": 8.235947229084559, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6653, "loss": 3.912071943283081, "lr": 0.0002, "elapsed_sec": 55329.48645186424, "step_time_sec": 8.238578893011436, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6654, "loss": 3.8877389430999756, "lr": 0.0002, "elapsed_sec": 55337.725189208984, "step_time_sec": 8.23855409794487, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6655, "loss": 3.926835775375366, "lr": 0.0002, "elapsed_sec": 55345.96470570564, "step_time_sec": 8.239338234998286, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6656, "loss": 4.00197172164917, "lr": 0.0002, "elapsed_sec": 55354.204228162766, "step_time_sec": 8.239404473919421, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6657, "loss": 3.777670383453369, "lr": 0.0002, "elapsed_sec": 55362.44233608246, "step_time_sec": 8.237929919036105, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6658, "loss": 3.977037191390991, "lr": 0.0002, "elapsed_sec": 55370.6813147068, "step_time_sec": 8.238808172056451, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6659, "loss": 3.912471294403076, "lr": 0.0002, "elapsed_sec": 55378.919998168945, "step_time_sec": 8.23858047707472, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6660, "loss": 3.9285879135131836, "lr": 0.0002, "elapsed_sec": 55387.16007065773, "step_time_sec": 8.239902682020329, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6661, "loss": 3.9520676136016846, "lr": 0.0002, "elapsed_sec": 55395.40071892738, "step_time_sec": 8.240479858941399, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6662, "loss": 3.880357265472412, "lr": 0.0002, "elapsed_sec": 55403.639141082764, "step_time_sec": 8.238203807035461, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6663, "loss": 4.0207600593566895, "lr": 0.0002, "elapsed_sec": 55411.87970209122, "step_time_sec": 8.240415081963874, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6664, "loss": 3.9793343544006348, "lr": 0.0002, "elapsed_sec": 55420.11915946007, "step_time_sec": 8.239330600015819, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6665, "loss": 4.079327583312988, "lr": 0.0002, "elapsed_sec": 55428.35740232468, "step_time_sec": 8.238058095099404, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6666, "loss": 4.018498420715332, "lr": 0.0002, "elapsed_sec": 55436.595933675766, "step_time_sec": 8.238428737968206, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6667, "loss": 4.200689315795898, "lr": 0.0002, "elapsed_sec": 55444.83440423012, "step_time_sec": 8.238246012013406, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6668, "loss": 4.0111870765686035, "lr": 0.0002, "elapsed_sec": 55453.074290037155, "step_time_sec": 8.239723053062335, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6669, "loss": 3.9728660583496094, "lr": 0.0002, "elapsed_sec": 55461.314428567886, "step_time_sec": 8.240002881968394, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6670, "loss": 4.048582553863525, "lr": 0.0002, "elapsed_sec": 55469.552203178406, "step_time_sec": 8.23759706306737, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6671, "loss": 3.9950292110443115, "lr": 0.0002, "elapsed_sec": 55477.7907640934, "step_time_sec": 8.238390943035483, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6672, "loss": 4.0873517990112305, "lr": 0.0002, "elapsed_sec": 55486.02803564072, "step_time_sec": 8.237097934936173, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6673, "loss": 4.171553611755371, "lr": 0.0002, "elapsed_sec": 55494.267614126205, "step_time_sec": 8.239436277071945, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6674, "loss": 4.006600856781006, "lr": 0.0002, "elapsed_sec": 55502.50806903839, "step_time_sec": 8.240274291019887, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6675, "loss": 3.953357219696045, "lr": 0.0002, "elapsed_sec": 55510.74659204483, "step_time_sec": 8.238355900975876, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6676, "loss": 3.8015999794006348, "lr": 0.0002, "elapsed_sec": 55518.98655986786, "step_time_sec": 8.239864350063726, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6677, "loss": 3.8228461742401123, "lr": 0.0002, "elapsed_sec": 55527.22481870651, "step_time_sec": 8.238070238032378, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6678, "loss": 3.8822410106658936, "lr": 0.0002, "elapsed_sec": 55535.464571237564, "step_time_sec": 8.23960543598514, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6679, "loss": 3.9822230339050293, "lr": 0.0002, "elapsed_sec": 55543.70403981209, "step_time_sec": 8.239305802970193, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6680, "loss": 3.944591999053955, "lr": 0.0002, "elapsed_sec": 55551.941126823425, "step_time_sec": 8.23690645606257, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6681, "loss": 3.983664035797119, "lr": 0.0002, "elapsed_sec": 55560.18083357811, "step_time_sec": 8.239527350990102, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6682, "loss": 4.0357747077941895, "lr": 0.0002, "elapsed_sec": 55568.41937375069, "step_time_sec": 8.238412020960823, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6683, "loss": 3.8619461059570312, "lr": 0.0002, "elapsed_sec": 55576.659856796265, "step_time_sec": 8.240375189925544, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6684, "loss": 3.8740506172180176, "lr": 0.0002, "elapsed_sec": 55584.89965701103, "step_time_sec": 8.2395733629819, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6685, "loss": 4.170729637145996, "lr": 0.0002, "elapsed_sec": 55593.13767385483, "step_time_sec": 8.237839858047664, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6686, "loss": 3.7871010303497314, "lr": 0.0002, "elapsed_sec": 55601.37607073784, "step_time_sec": 8.238270901958458, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6687, "loss": 3.9037411212921143, "lr": 0.0002, "elapsed_sec": 55609.61586689949, "step_time_sec": 8.239638075930998, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6688, "loss": 4.019525527954102, "lr": 0.0002, "elapsed_sec": 55617.85455942154, "step_time_sec": 8.23855955991894, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6689, "loss": 4.015413761138916, "lr": 0.0002, "elapsed_sec": 55626.0940926075, "step_time_sec": 8.239344025030732, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6690, "loss": 4.032644271850586, "lr": 0.0002, "elapsed_sec": 55634.33274316788, "step_time_sec": 8.238513287040405, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6691, "loss": 3.9470326900482178, "lr": 0.0002, "elapsed_sec": 55642.572874069214, "step_time_sec": 8.239959580008872, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6692, "loss": 4.063872814178467, "lr": 0.0002, "elapsed_sec": 55650.811950445175, "step_time_sec": 8.238917271024548, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6693, "loss": 3.9170897006988525, "lr": 0.0002, "elapsed_sec": 55659.05072617531, "step_time_sec": 8.238584437058307, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6694, "loss": 4.084837913513184, "lr": 0.0002, "elapsed_sec": 55667.29084134102, "step_time_sec": 8.239958724938333, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6695, "loss": 4.03239631652832, "lr": 0.0002, "elapsed_sec": 55675.52978157997, "step_time_sec": 8.238799899932928, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6696, "loss": 4.027815341949463, "lr": 0.0002, "elapsed_sec": 55683.770384550095, "step_time_sec": 8.240425423020497, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6697, "loss": 3.9475221633911133, "lr": 0.0002, "elapsed_sec": 55692.0105714798, "step_time_sec": 8.24005170201417, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6698, "loss": 3.9188523292541504, "lr": 0.0002, "elapsed_sec": 55700.24877977371, "step_time_sec": 8.238037204020657, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6699, "loss": 4.185210227966309, "lr": 0.0002, "elapsed_sec": 55708.48851656914, "step_time_sec": 8.239594486076385, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6700, "loss": 3.992060899734497, "lr": 0.0002, "elapsed_sec": 55716.728951215744, "step_time_sec": 8.240320035954937, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6701, "loss": 4.0528154373168945, "lr": 0.0002, "elapsed_sec": 55724.96808242798, "step_time_sec": 8.238898935029283, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6702, "loss": 4.019493103027344, "lr": 0.0002, "elapsed_sec": 55733.206879377365, "step_time_sec": 8.238636018009856, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6703, "loss": 3.9890120029449463, "lr": 0.0002, "elapsed_sec": 55741.44433879852, "step_time_sec": 8.237323807901703, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6704, "loss": 3.9527440071105957, "lr": 0.0002, "elapsed_sec": 55749.68398189545, "step_time_sec": 8.239477458992042, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6705, "loss": 3.942643642425537, "lr": 0.0002, "elapsed_sec": 55757.92392063141, "step_time_sec": 8.239801150979474, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6706, "loss": 4.023979187011719, "lr": 0.0002, "elapsed_sec": 55766.161467552185, "step_time_sec": 8.237353946082294, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6707, "loss": 3.9708914756774902, "lr": 0.0002, "elapsed_sec": 55774.40103435516, "step_time_sec": 8.239422293030657, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6708, "loss": 3.867598533630371, "lr": 0.0002, "elapsed_sec": 55782.63821768761, "step_time_sec": 8.237020900007337, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6709, "loss": 3.9271702766418457, "lr": 0.0002, "elapsed_sec": 55790.87702655792, "step_time_sec": 8.238652331987396, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6710, "loss": 3.8618805408477783, "lr": 0.0002, "elapsed_sec": 55799.11693954468, "step_time_sec": 8.239727784995921, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6711, "loss": 4.007408618927002, "lr": 0.0002, "elapsed_sec": 55807.355855703354, "step_time_sec": 8.238747593015432, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6712, "loss": 3.7499094009399414, "lr": 0.0002, "elapsed_sec": 55815.595291376114, "step_time_sec": 8.23934214306064, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6713, "loss": 3.919638156890869, "lr": 0.0002, "elapsed_sec": 55823.83276486397, "step_time_sec": 8.237304496928118, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6714, "loss": 3.991117238998413, "lr": 0.0002, "elapsed_sec": 55832.0722694397, "step_time_sec": 8.239273241953924, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6715, "loss": 3.9559006690979004, "lr": 0.0002, "elapsed_sec": 55840.311439991, "step_time_sec": 8.23900393594522, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6716, "loss": 3.925703525543213, "lr": 0.0002, "elapsed_sec": 55848.55000448227, "step_time_sec": 8.238450602977537, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6717, "loss": 4.012415409088135, "lr": 0.0002, "elapsed_sec": 55856.78853535652, "step_time_sec": 8.238386393990368, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6718, "loss": 4.007360935211182, "lr": 0.0002, "elapsed_sec": 55865.02622103691, "step_time_sec": 8.237468376988545, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6719, "loss": 3.9465885162353516, "lr": 0.0002, "elapsed_sec": 55873.26676154137, "step_time_sec": 8.240385020035319, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6720, "loss": 3.9813544750213623, "lr": 0.0002, "elapsed_sec": 55881.507966041565, "step_time_sec": 8.241053306031972, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6721, "loss": 3.9898557662963867, "lr": 0.0002, "elapsed_sec": 55889.745465278625, "step_time_sec": 8.237373214098625, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6722, "loss": 3.9974632263183594, "lr": 0.0002, "elapsed_sec": 55897.98527121544, "step_time_sec": 8.239566990057938, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6723, "loss": 3.950545310974121, "lr": 0.0002, "elapsed_sec": 55906.224179029465, "step_time_sec": 8.238764209090732, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6724, "loss": 4.0245466232299805, "lr": 0.0002, "elapsed_sec": 55914.4636015892, "step_time_sec": 8.239305692957714, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6725, "loss": 3.970896005630493, "lr": 0.0002, "elapsed_sec": 55922.704050540924, "step_time_sec": 8.240262666950002, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6726, "loss": 3.9845261573791504, "lr": 0.0002, "elapsed_sec": 55930.94278597832, "step_time_sec": 8.238612817018293, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6727, "loss": 3.959556818008423, "lr": 0.0002, "elapsed_sec": 55939.18514609337, "step_time_sec": 8.242148532997817, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6728, "loss": 3.969811201095581, "lr": 0.0002, "elapsed_sec": 55947.42676639557, "step_time_sec": 8.241485040984116, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6729, "loss": 3.965841054916382, "lr": 0.0002, "elapsed_sec": 55955.68483734131, "step_time_sec": 8.25792992906645, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6730, "loss": 3.8990705013275146, "lr": 0.0002, "elapsed_sec": 55963.92291855812, "step_time_sec": 8.237865369999781, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6731, "loss": 3.9475784301757812, "lr": 0.0002, "elapsed_sec": 55972.163120269775, "step_time_sec": 8.240010709036142, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6732, "loss": 4.0894317626953125, "lr": 0.0002, "elapsed_sec": 55980.401668310165, "step_time_sec": 8.238440632936545, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6733, "loss": 3.891030788421631, "lr": 0.0002, "elapsed_sec": 55988.640169382095, "step_time_sec": 8.23832003294956, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6734, "loss": 3.884631872177124, "lr": 0.0002, "elapsed_sec": 55996.880425691605, "step_time_sec": 8.240113325067796, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6735, "loss": 4.079953193664551, "lr": 0.0002, "elapsed_sec": 56005.118170022964, "step_time_sec": 8.237541264970787, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6736, "loss": 3.890700101852417, "lr": 0.0002, "elapsed_sec": 56013.358325481415, "step_time_sec": 8.240008111926727, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6737, "loss": 3.949012041091919, "lr": 0.0002, "elapsed_sec": 56021.59824132919, "step_time_sec": 8.239752334076911, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6738, "loss": 3.851555109024048, "lr": 0.0002, "elapsed_sec": 56029.83755350113, "step_time_sec": 8.239164761966094, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6739, "loss": 3.9673545360565186, "lr": 0.0002, "elapsed_sec": 56038.07816195488, "step_time_sec": 8.240437645930797, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6740, "loss": 3.8637146949768066, "lr": 0.0002, "elapsed_sec": 56046.317084789276, "step_time_sec": 8.238761696964502, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6741, "loss": 4.042762279510498, "lr": 0.0002, "elapsed_sec": 56054.55746006966, "step_time_sec": 8.240252700983547, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6742, "loss": 3.944520950317383, "lr": 0.0002, "elapsed_sec": 56062.7981095314, "step_time_sec": 8.240442232927307, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6743, "loss": 3.9835116863250732, "lr": 0.0002, "elapsed_sec": 56071.035825014114, "step_time_sec": 8.237573502003215, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6744, "loss": 3.934366464614868, "lr": 0.0002, "elapsed_sec": 56079.27453875542, "step_time_sec": 8.238593630958349, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6745, "loss": 3.80177903175354, "lr": 0.0002, "elapsed_sec": 56087.513223171234, "step_time_sec": 8.23846636293456, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6746, "loss": 3.9998373985290527, "lr": 0.0002, "elapsed_sec": 56095.75191259384, "step_time_sec": 8.238528353977017, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6747, "loss": 3.9654529094696045, "lr": 0.0002, "elapsed_sec": 56103.99050664902, "step_time_sec": 8.238438223022968, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6748, "loss": 3.9995243549346924, "lr": 0.0002, "elapsed_sec": 56112.22932219505, "step_time_sec": 8.238640449941158, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6749, "loss": 3.9193077087402344, "lr": 0.0002, "elapsed_sec": 56120.46979260445, "step_time_sec": 8.240355661022477, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6750, "loss": 3.893812656402588, "lr": 0.0002, "elapsed_sec": 56128.709404706955, "step_time_sec": 8.239399048965424, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6751, "loss": 4.041822910308838, "lr": 0.0002, "elapsed_sec": 56136.9484937191, "step_time_sec": 8.238994714920409, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6752, "loss": 4.067927360534668, "lr": 0.0002, "elapsed_sec": 56145.18717908859, "step_time_sec": 8.238455118960701, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6753, "loss": 3.9007465839385986, "lr": 0.0002, "elapsed_sec": 56153.42563080788, "step_time_sec": 8.23829933302477, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6754, "loss": 3.9308595657348633, "lr": 0.0002, "elapsed_sec": 56161.66654372215, "step_time_sec": 8.24076550710015, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6755, "loss": 3.9945907592773438, "lr": 0.0002, "elapsed_sec": 56169.90506505966, "step_time_sec": 8.238373095053248, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6756, "loss": 4.0883708000183105, "lr": 0.0002, "elapsed_sec": 56178.14387154579, "step_time_sec": 8.23860247596167, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6757, "loss": 3.951169729232788, "lr": 0.0002, "elapsed_sec": 56186.38409781456, "step_time_sec": 8.240064159035683, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6758, "loss": 3.920699119567871, "lr": 0.0002, "elapsed_sec": 56194.62311220169, "step_time_sec": 8.238901307922788, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6759, "loss": 3.9878122806549072, "lr": 0.0002, "elapsed_sec": 56202.86346197128, "step_time_sec": 8.240167025011033, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6760, "loss": 3.92866849899292, "lr": 0.0002, "elapsed_sec": 56211.102956056595, "step_time_sec": 8.239307207986712, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6761, "loss": 3.955334186553955, "lr": 0.0002, "elapsed_sec": 56219.34162521362, "step_time_sec": 8.2385293119587, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6762, "loss": 3.9029929637908936, "lr": 0.0002, "elapsed_sec": 56227.5796084404, "step_time_sec": 8.237794385990128, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6763, "loss": 3.893589735031128, "lr": 0.0002, "elapsed_sec": 56235.817739248276, "step_time_sec": 8.238013415015303, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6764, "loss": 3.9720425605773926, "lr": 0.0002, "elapsed_sec": 56244.05703687668, "step_time_sec": 8.239124638959765, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6765, "loss": 3.905052661895752, "lr": 0.0002, "elapsed_sec": 56252.29710364342, "step_time_sec": 8.23988394800108, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6766, "loss": 3.771193265914917, "lr": 0.0002, "elapsed_sec": 56260.53605413437, "step_time_sec": 8.23878484906163, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6767, "loss": 3.9718267917633057, "lr": 0.0002, "elapsed_sec": 56268.77595973015, "step_time_sec": 8.239750735927373, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6768, "loss": 3.8832814693450928, "lr": 0.0002, "elapsed_sec": 56277.015100479126, "step_time_sec": 8.238980351947248, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6769, "loss": 3.9897029399871826, "lr": 0.0002, "elapsed_sec": 56285.25422215462, "step_time_sec": 8.238918364979327, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6770, "loss": 4.0935211181640625, "lr": 0.0002, "elapsed_sec": 56293.49582695961, "step_time_sec": 8.241480352939107, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6771, "loss": 3.9439094066619873, "lr": 0.0002, "elapsed_sec": 56301.73360013962, "step_time_sec": 8.23757287801709, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6772, "loss": 3.9709792137145996, "lr": 0.0002, "elapsed_sec": 56309.97507953644, "step_time_sec": 8.24136454006657, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6773, "loss": 3.794989585876465, "lr": 0.0002, "elapsed_sec": 56318.21548843384, "step_time_sec": 8.24024248495698, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6774, "loss": 4.034719467163086, "lr": 0.0002, "elapsed_sec": 56326.45273923874, "step_time_sec": 8.237058129976504, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6775, "loss": 4.010144233703613, "lr": 0.0002, "elapsed_sec": 56334.693244457245, "step_time_sec": 8.2403781110188, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6776, "loss": 3.945713758468628, "lr": 0.0002, "elapsed_sec": 56342.93143391609, "step_time_sec": 8.238022834993899, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6777, "loss": 3.9730939865112305, "lr": 0.0002, "elapsed_sec": 56351.171090602875, "step_time_sec": 8.239498057053424, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6778, "loss": 3.935067892074585, "lr": 0.0002, "elapsed_sec": 56359.41155338287, "step_time_sec": 8.240301962010562, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6779, "loss": 4.10628604888916, "lr": 0.0002, "elapsed_sec": 56367.650418519974, "step_time_sec": 8.238621867028996, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6780, "loss": 4.015390872955322, "lr": 0.0002, "elapsed_sec": 56375.88972043991, "step_time_sec": 8.239152908907272, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6781, "loss": 3.9159133434295654, "lr": 0.0002, "elapsed_sec": 56384.12908864021, "step_time_sec": 8.239221101976, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6782, "loss": 3.9133598804473877, "lr": 0.0002, "elapsed_sec": 56392.36827373505, "step_time_sec": 8.239016092964448, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6783, "loss": 3.809271812438965, "lr": 0.0002, "elapsed_sec": 56400.6074552536, "step_time_sec": 8.239053464960307, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6784, "loss": 3.8544843196868896, "lr": 0.0002, "elapsed_sec": 56408.84645485878, "step_time_sec": 8.238795222947374, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6785, "loss": 3.90889835357666, "lr": 0.0002, "elapsed_sec": 56417.08763074875, "step_time_sec": 8.241069872980006, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6786, "loss": 4.095098972320557, "lr": 0.0002, "elapsed_sec": 56425.32664847374, "step_time_sec": 8.238815653021447, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6787, "loss": 4.018412113189697, "lr": 0.0002, "elapsed_sec": 56433.564754486084, "step_time_sec": 8.237971294904128, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6788, "loss": 3.903874635696411, "lr": 0.0002, "elapsed_sec": 56441.80589175224, "step_time_sec": 8.240983473020606, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6789, "loss": 3.8759546279907227, "lr": 0.0002, "elapsed_sec": 56450.0455725193, "step_time_sec": 8.2395234759897, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6790, "loss": 4.094009876251221, "lr": 0.0002, "elapsed_sec": 56458.28520679474, "step_time_sec": 8.239439523895271, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6791, "loss": 3.95501708984375, "lr": 0.0002, "elapsed_sec": 56466.52537679672, "step_time_sec": 8.240051762084477, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6792, "loss": 3.875995397567749, "lr": 0.0002, "elapsed_sec": 56474.76459860802, "step_time_sec": 8.239040708052926, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6793, "loss": 3.9707772731781006, "lr": 0.0002, "elapsed_sec": 56483.00500655174, "step_time_sec": 8.240221660933457, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6794, "loss": 4.027758598327637, "lr": 0.0002, "elapsed_sec": 56491.24517464638, "step_time_sec": 8.23999685596209, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6795, "loss": 3.916443109512329, "lr": 0.0002, "elapsed_sec": 56499.48303794861, "step_time_sec": 8.237697947071865, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6796, "loss": 3.8691611289978027, "lr": 0.0002, "elapsed_sec": 56507.7235519886, "step_time_sec": 8.240345634985715, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6797, "loss": 3.822002649307251, "lr": 0.0002, "elapsed_sec": 56515.961987257004, "step_time_sec": 8.238284251070581, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6798, "loss": 4.048954963684082, "lr": 0.0002, "elapsed_sec": 56524.20177292824, "step_time_sec": 8.239644619054161, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6799, "loss": 3.8818204402923584, "lr": 0.0002, "elapsed_sec": 56532.43992781639, "step_time_sec": 8.237969795940444, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6800, "loss": 3.935504674911499, "lr": 0.0002, "elapsed_sec": 56540.67901134491, "step_time_sec": 8.238930820953101, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6801, "loss": 3.922694683074951, "lr": 0.0002, "elapsed_sec": 56548.9194893837, "step_time_sec": 8.240292115020566, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6802, "loss": 3.8548552989959717, "lr": 0.0002, "elapsed_sec": 56557.15853476524, "step_time_sec": 8.238946003955789, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6803, "loss": 3.8505609035491943, "lr": 0.0002, "elapsed_sec": 56565.39687514305, "step_time_sec": 8.238103908021003, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6804, "loss": 3.9853274822235107, "lr": 0.0002, "elapsed_sec": 56573.63667964935, "step_time_sec": 8.23963987303432, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6805, "loss": 3.990708827972412, "lr": 0.0002, "elapsed_sec": 56581.87350511551, "step_time_sec": 8.236721471999772, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6806, "loss": 3.8787577152252197, "lr": 0.0002, "elapsed_sec": 56590.11494231224, "step_time_sec": 8.2412778850412, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6807, "loss": 4.102675914764404, "lr": 0.0002, "elapsed_sec": 56598.35461425781, "step_time_sec": 8.239483320969157, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6808, "loss": 3.8517584800720215, "lr": 0.0002, "elapsed_sec": 56606.59345579147, "step_time_sec": 8.238710402045399, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6809, "loss": 4.046047687530518, "lr": 0.0002, "elapsed_sec": 56614.83329963684, "step_time_sec": 8.23960536904633, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6810, "loss": 3.84338116645813, "lr": 0.0002, "elapsed_sec": 56623.07207322121, "step_time_sec": 8.238615487003699, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6811, "loss": 3.929734945297241, "lr": 0.0002, "elapsed_sec": 56631.311333179474, "step_time_sec": 8.23909545794595, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6812, "loss": 3.8426361083984375, "lr": 0.0002, "elapsed_sec": 56639.551409482956, "step_time_sec": 8.239901127992198, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6813, "loss": 3.8896753787994385, "lr": 0.0002, "elapsed_sec": 56647.78955769539, "step_time_sec": 8.237978061079048, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6814, "loss": 4.052138328552246, "lr": 0.0002, "elapsed_sec": 56656.030650138855, "step_time_sec": 8.240989223006181, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6815, "loss": 3.9559285640716553, "lr": 0.0002, "elapsed_sec": 56664.26999807358, "step_time_sec": 8.239135002018884, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6816, "loss": 4.138584136962891, "lr": 0.0002, "elapsed_sec": 56672.50922894478, "step_time_sec": 8.239105056971312, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6817, "loss": 4.014448642730713, "lr": 0.0002, "elapsed_sec": 56680.75024700165, "step_time_sec": 8.240832956973463, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6818, "loss": 3.7932889461517334, "lr": 0.0002, "elapsed_sec": 56688.988250494, "step_time_sec": 8.237885536975227, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6819, "loss": 3.9979569911956787, "lr": 0.0002, "elapsed_sec": 56697.22872233391, "step_time_sec": 8.240276386030018, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6820, "loss": 4.103504657745361, "lr": 0.0002, "elapsed_sec": 56705.46765422821, "step_time_sec": 8.238775820005685, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6821, "loss": 3.9023244380950928, "lr": 0.0002, "elapsed_sec": 56713.70684957504, "step_time_sec": 8.239046992035583, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6822, "loss": 3.8349740505218506, "lr": 0.0002, "elapsed_sec": 56721.94688248634, "step_time_sec": 8.23981256398838, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6823, "loss": 3.9083383083343506, "lr": 0.0002, "elapsed_sec": 56730.186074733734, "step_time_sec": 8.23906795901712, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6824, "loss": 3.8983101844787598, "lr": 0.0002, "elapsed_sec": 56738.425030231476, "step_time_sec": 8.23878950998187, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6825, "loss": 3.9206979274749756, "lr": 0.0002, "elapsed_sec": 56746.663113832474, "step_time_sec": 8.237944729044102, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6826, "loss": 3.9600157737731934, "lr": 0.0002, "elapsed_sec": 56754.90157985687, "step_time_sec": 8.238292046939023, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6827, "loss": 3.8884143829345703, "lr": 0.0002, "elapsed_sec": 56763.14130783081, "step_time_sec": 8.2395346990088, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6828, "loss": 4.01203727722168, "lr": 0.0002, "elapsed_sec": 56771.38040447235, "step_time_sec": 8.238931917934678, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6829, "loss": 3.834864616394043, "lr": 0.0002, "elapsed_sec": 56779.62012028694, "step_time_sec": 8.23956835793797, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6830, "loss": 4.04175329208374, "lr": 0.0002, "elapsed_sec": 56787.85979628563, "step_time_sec": 8.23951115400996, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6831, "loss": 3.861520290374756, "lr": 0.0002, "elapsed_sec": 56796.096064805984, "step_time_sec": 8.236122485017404, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6832, "loss": 4.014881610870361, "lr": 0.0002, "elapsed_sec": 56804.33524751663, "step_time_sec": 8.239036077051423, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6833, "loss": 3.9952096939086914, "lr": 0.0002, "elapsed_sec": 56812.57373905182, "step_time_sec": 8.238358599017374, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6834, "loss": 4.035658359527588, "lr": 0.0002, "elapsed_sec": 56820.814016103745, "step_time_sec": 8.240110595943406, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6835, "loss": 3.864020586013794, "lr": 0.0002, "elapsed_sec": 56829.05397963524, "step_time_sec": 8.239761703996919, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6836, "loss": 3.9471356868743896, "lr": 0.0002, "elapsed_sec": 56837.29209113121, "step_time_sec": 8.237963840947486, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6837, "loss": 3.937443256378174, "lr": 0.0002, "elapsed_sec": 56845.53112912178, "step_time_sec": 8.238860429963097, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6838, "loss": 3.9402518272399902, "lr": 0.0002, "elapsed_sec": 56853.76918244362, "step_time_sec": 8.237882410059683, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6839, "loss": 3.887843608856201, "lr": 0.0002, "elapsed_sec": 56862.007563591, "step_time_sec": 8.238220291095786, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6840, "loss": 3.8801660537719727, "lr": 0.0002, "elapsed_sec": 56870.246743917465, "step_time_sec": 8.239055143087171, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6841, "loss": 3.9841349124908447, "lr": 0.0002, "elapsed_sec": 56878.484545469284, "step_time_sec": 8.237654404016212, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6842, "loss": 3.7853598594665527, "lr": 0.0002, "elapsed_sec": 56886.7241165638, "step_time_sec": 8.239418660989031, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6843, "loss": 3.9081602096557617, "lr": 0.0002, "elapsed_sec": 56894.963156700134, "step_time_sec": 8.238804778084159, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6844, "loss": 3.8797054290771484, "lr": 0.0002, "elapsed_sec": 56903.20333480835, "step_time_sec": 8.240027079940774, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6845, "loss": 3.807162046432495, "lr": 0.0002, "elapsed_sec": 56911.44284415245, "step_time_sec": 8.239358195103705, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6846, "loss": 3.814424991607666, "lr": 0.0002, "elapsed_sec": 56919.75998592377, "step_time_sec": 8.251397878979333, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6847, "loss": 3.8734042644500732, "lr": 0.0002, "elapsed_sec": 56927.9992518425, "step_time_sec": 8.239093436975963, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6848, "loss": 3.987304925918579, "lr": 0.0002, "elapsed_sec": 56936.23676657677, "step_time_sec": 8.23735656507779, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6849, "loss": 4.007880210876465, "lr": 0.0002, "elapsed_sec": 56944.47569894791, "step_time_sec": 8.238807122921571, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6850, "loss": 3.9433038234710693, "lr": 0.0002, "elapsed_sec": 56952.71447658539, "step_time_sec": 8.238582917954773, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6851, "loss": 3.9696097373962402, "lr": 0.0002, "elapsed_sec": 56960.953714609146, "step_time_sec": 8.239042514003813, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6852, "loss": 3.8390719890594482, "lr": 0.0002, "elapsed_sec": 56969.19295692444, "step_time_sec": 8.239088490023278, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6853, "loss": 3.910698890686035, "lr": 0.0002, "elapsed_sec": 56977.43131017685, "step_time_sec": 8.238205787958577, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6854, "loss": 4.014161586761475, "lr": 0.0002, "elapsed_sec": 56985.67117547989, "step_time_sec": 8.239690032904036, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6855, "loss": 4.018764495849609, "lr": 0.0002, "elapsed_sec": 56993.91097211838, "step_time_sec": 8.239620565087534, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6856, "loss": 3.8802828788757324, "lr": 0.0002, "elapsed_sec": 57002.15035676956, "step_time_sec": 8.239238891052082, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6857, "loss": 3.6155269145965576, "lr": 0.0002, "elapsed_sec": 57010.38965010643, "step_time_sec": 8.239187608938664, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6858, "loss": 3.970142126083374, "lr": 0.0002, "elapsed_sec": 57018.62861585617, "step_time_sec": 8.238812204101123, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6859, "loss": 3.9995944499969482, "lr": 0.0002, "elapsed_sec": 57026.86958646774, "step_time_sec": 8.240819655009545, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6860, "loss": 3.8535232543945312, "lr": 0.0002, "elapsed_sec": 57035.109607219696, "step_time_sec": 8.239806008990854, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6861, "loss": 3.7913994789123535, "lr": 0.0002, "elapsed_sec": 57043.34857106209, "step_time_sec": 8.23881375300698, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6862, "loss": 3.9262924194335938, "lr": 0.0002, "elapsed_sec": 57051.588633298874, "step_time_sec": 8.239901105989702, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6863, "loss": 3.8837149143218994, "lr": 0.0002, "elapsed_sec": 57059.826600551605, "step_time_sec": 8.237775614950806, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6864, "loss": 4.012428283691406, "lr": 0.0002, "elapsed_sec": 57068.06701683998, "step_time_sec": 8.240293807000853, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6865, "loss": 4.012096405029297, "lr": 0.0002, "elapsed_sec": 57076.3065431118, "step_time_sec": 8.239308924996294, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6866, "loss": 3.879359722137451, "lr": 0.0002, "elapsed_sec": 57084.54500579834, "step_time_sec": 8.23829687293619, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6867, "loss": 3.891937255859375, "lr": 0.0002, "elapsed_sec": 57092.78482723236, "step_time_sec": 8.239647828042507, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6868, "loss": 3.816730260848999, "lr": 0.0002, "elapsed_sec": 57101.02367949486, "step_time_sec": 8.238697319990024, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6869, "loss": 3.7463693618774414, "lr": 0.0002, "elapsed_sec": 57109.26221370697, "step_time_sec": 8.238350606057793, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6870, "loss": 3.799851179122925, "lr": 0.0002, "elapsed_sec": 57117.50258588791, "step_time_sec": 8.240232763928361, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6871, "loss": 3.8576853275299072, "lr": 0.0002, "elapsed_sec": 57125.73937225342, "step_time_sec": 8.236588317900896, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6872, "loss": 4.045327663421631, "lr": 0.0002, "elapsed_sec": 57133.98106408119, "step_time_sec": 8.241588167962618, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6873, "loss": 4.066464900970459, "lr": 0.0002, "elapsed_sec": 57142.22366809845, "step_time_sec": 8.248486071941443, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6874, "loss": 3.909850835800171, "lr": 0.0002, "elapsed_sec": 57150.482556819916, "step_time_sec": 8.252625779015943, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6875, "loss": 3.9889845848083496, "lr": 0.0002, "elapsed_sec": 57158.72006106377, "step_time_sec": 8.237318426021375, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6876, "loss": 3.979848861694336, "lr": 0.0002, "elapsed_sec": 57166.95835971832, "step_time_sec": 8.238154002930969, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6877, "loss": 3.8614957332611084, "lr": 0.0002, "elapsed_sec": 57175.19641828537, "step_time_sec": 8.237865571049042, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6878, "loss": 3.885133743286133, "lr": 0.0002, "elapsed_sec": 57183.43652868271, "step_time_sec": 8.23990208900068, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6879, "loss": 3.869816780090332, "lr": 0.0002, "elapsed_sec": 57191.675097465515, "step_time_sec": 8.238434080965817, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6880, "loss": 3.866891622543335, "lr": 0.0002, "elapsed_sec": 57199.91360282898, "step_time_sec": 8.238314846064895, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6881, "loss": 3.808710813522339, "lr": 0.0002, "elapsed_sec": 57208.15274786949, "step_time_sec": 8.238965129945427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6882, "loss": 3.7501537799835205, "lr": 0.0002, "elapsed_sec": 57216.38983845711, "step_time_sec": 8.236927081015892, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6883, "loss": 4.076931953430176, "lr": 0.0002, "elapsed_sec": 57224.628829956055, "step_time_sec": 8.238868362037465, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6884, "loss": 4.071195602416992, "lr": 0.0002, "elapsed_sec": 57232.86709976196, "step_time_sec": 8.238098868983798, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6885, "loss": 3.93562912940979, "lr": 0.0002, "elapsed_sec": 57241.10455703735, "step_time_sec": 8.237319371080957, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6886, "loss": 3.9140260219573975, "lr": 0.0002, "elapsed_sec": 57249.341886520386, "step_time_sec": 8.237120221019723, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6887, "loss": 3.9429738521575928, "lr": 0.0002, "elapsed_sec": 57257.57803058624, "step_time_sec": 8.236007325933315, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6888, "loss": 3.9790146350860596, "lr": 0.0002, "elapsed_sec": 57265.81717944145, "step_time_sec": 8.238963332027197, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6889, "loss": 3.9484457969665527, "lr": 0.0002, "elapsed_sec": 57274.05632805824, "step_time_sec": 8.23902401002124, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6890, "loss": 3.9477999210357666, "lr": 0.0002, "elapsed_sec": 57282.295390844345, "step_time_sec": 8.238881160970777, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6891, "loss": 3.889065980911255, "lr": 0.0002, "elapsed_sec": 57290.53549361229, "step_time_sec": 8.23991515301168, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6892, "loss": 3.9766905307769775, "lr": 0.0002, "elapsed_sec": 57298.77434301376, "step_time_sec": 8.238648190977983, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6893, "loss": 3.9211018085479736, "lr": 0.0002, "elapsed_sec": 57307.014241695404, "step_time_sec": 8.239757133997045, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6894, "loss": 3.89943265914917, "lr": 0.0002, "elapsed_sec": 57315.254106760025, "step_time_sec": 8.23966936103534, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6895, "loss": 3.864006280899048, "lr": 0.0002, "elapsed_sec": 57323.49265265465, "step_time_sec": 8.23837260901928, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6896, "loss": 3.896686315536499, "lr": 0.0002, "elapsed_sec": 57331.731261491776, "step_time_sec": 8.238427720032632, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6897, "loss": 3.9245963096618652, "lr": 0.0002, "elapsed_sec": 57339.97094106674, "step_time_sec": 8.239527183002792, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6898, "loss": 3.8235599994659424, "lr": 0.0002, "elapsed_sec": 57348.21004199982, "step_time_sec": 8.238970900070854, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6899, "loss": 3.9694838523864746, "lr": 0.0002, "elapsed_sec": 57356.44795060158, "step_time_sec": 8.237740569980815, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6900, "loss": 3.724323034286499, "lr": 0.0002, "elapsed_sec": 57364.68597340584, "step_time_sec": 8.23786001210101, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6901, "loss": 3.921781539916992, "lr": 0.0002, "elapsed_sec": 57372.92680287361, "step_time_sec": 8.240645833080634, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6902, "loss": 3.94497013092041, "lr": 0.0002, "elapsed_sec": 57381.16713690758, "step_time_sec": 8.240143998060375, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6903, "loss": 3.8964366912841797, "lr": 0.0002, "elapsed_sec": 57389.40711593628, "step_time_sec": 8.239832098945044, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6904, "loss": 3.7624034881591797, "lr": 0.0002, "elapsed_sec": 57397.647247076035, "step_time_sec": 8.239973253104836, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6905, "loss": 3.7756850719451904, "lr": 0.0002, "elapsed_sec": 57405.8862888813, "step_time_sec": 8.23891068994999, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6906, "loss": 3.7058703899383545, "lr": 0.0002, "elapsed_sec": 57414.12806510925, "step_time_sec": 8.241582025075331, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6907, "loss": 3.9590909481048584, "lr": 0.0002, "elapsed_sec": 57422.367638111115, "step_time_sec": 8.239438547054306, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6908, "loss": 3.8024141788482666, "lr": 0.0002, "elapsed_sec": 57430.60700750351, "step_time_sec": 8.23921008396428, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6909, "loss": 3.9931225776672363, "lr": 0.0002, "elapsed_sec": 57438.846901893616, "step_time_sec": 8.239738479955122, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6910, "loss": 3.9236245155334473, "lr": 0.0002, "elapsed_sec": 57447.08662223816, "step_time_sec": 8.239568040939048, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6911, "loss": 3.9811575412750244, "lr": 0.0002, "elapsed_sec": 57455.32529902458, "step_time_sec": 8.238459631917067, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6912, "loss": 3.812302350997925, "lr": 0.0002, "elapsed_sec": 57463.56458044052, "step_time_sec": 8.239175060996786, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6913, "loss": 3.9757046699523926, "lr": 0.0002, "elapsed_sec": 57471.8026368618, "step_time_sec": 8.237844125018455, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6914, "loss": 3.8986141681671143, "lr": 0.0002, "elapsed_sec": 57480.0412709713, "step_time_sec": 8.238478942075744, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6915, "loss": 3.833346366882324, "lr": 0.0002, "elapsed_sec": 57488.28023004532, "step_time_sec": 8.238810093025677, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6916, "loss": 4.038203239440918, "lr": 0.0002, "elapsed_sec": 57496.51933312416, "step_time_sec": 8.238984956988133, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6917, "loss": 3.8505828380584717, "lr": 0.0002, "elapsed_sec": 57504.75808477402, "step_time_sec": 8.238537058932707, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6918, "loss": 3.708422899246216, "lr": 0.0002, "elapsed_sec": 57512.99546647072, "step_time_sec": 8.237208361038938, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6919, "loss": 4.0798492431640625, "lr": 0.0002, "elapsed_sec": 57521.236567020416, "step_time_sec": 8.240931201027706, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6920, "loss": 3.9599578380584717, "lr": 0.0002, "elapsed_sec": 57529.47641444206, "step_time_sec": 8.23970696690958, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6921, "loss": 3.686624765396118, "lr": 0.0002, "elapsed_sec": 57537.71478295326, "step_time_sec": 8.238195121055469, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6922, "loss": 3.8904640674591064, "lr": 0.0002, "elapsed_sec": 57545.95392417908, "step_time_sec": 8.238992710015737, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6923, "loss": 3.9353647232055664, "lr": 0.0002, "elapsed_sec": 57554.19190478325, "step_time_sec": 8.23781491490081, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6924, "loss": 3.8693578243255615, "lr": 0.0002, "elapsed_sec": 57562.432233810425, "step_time_sec": 8.240159112028778, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6925, "loss": 4.250406742095947, "lr": 0.0002, "elapsed_sec": 57570.67080140114, "step_time_sec": 8.238448270014487, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6926, "loss": 3.8532674312591553, "lr": 0.0002, "elapsed_sec": 57578.90977668762, "step_time_sec": 8.238796927034855, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6927, "loss": 3.9416322708129883, "lr": 0.0002, "elapsed_sec": 57587.14923930168, "step_time_sec": 8.239327042014338, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6928, "loss": 3.850220203399658, "lr": 0.0002, "elapsed_sec": 57595.38863635063, "step_time_sec": 8.239176169969141, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6929, "loss": 3.7979867458343506, "lr": 0.0002, "elapsed_sec": 57603.62828063965, "step_time_sec": 8.239500677096657, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6930, "loss": 3.9429917335510254, "lr": 0.0002, "elapsed_sec": 57611.866458415985, "step_time_sec": 8.237995645962656, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6931, "loss": 3.8972301483154297, "lr": 0.0002, "elapsed_sec": 57620.1058011055, "step_time_sec": 8.239192089997232, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6932, "loss": 3.7657034397125244, "lr": 0.0002, "elapsed_sec": 57628.344887018204, "step_time_sec": 8.23893797409255, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6933, "loss": 3.9922735691070557, "lr": 0.0002, "elapsed_sec": 57636.58317232132, "step_time_sec": 8.238178087049164, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6934, "loss": 3.8855488300323486, "lr": 0.0002, "elapsed_sec": 57644.82182765007, "step_time_sec": 8.238464572932571, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6935, "loss": 4.014326572418213, "lr": 0.0002, "elapsed_sec": 57653.061614990234, "step_time_sec": 8.239648443064652, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6936, "loss": 3.922701120376587, "lr": 0.0002, "elapsed_sec": 57661.299939632416, "step_time_sec": 8.238122535985895, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6937, "loss": 3.803334951400757, "lr": 0.0002, "elapsed_sec": 57669.54025912285, "step_time_sec": 8.2401575720869, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6938, "loss": 3.912752628326416, "lr": 0.0002, "elapsed_sec": 57677.780782699585, "step_time_sec": 8.240352030028589, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6939, "loss": 3.9787991046905518, "lr": 0.0002, "elapsed_sec": 57686.019587516785, "step_time_sec": 8.238675717962906, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6940, "loss": 4.036231994628906, "lr": 0.0002, "elapsed_sec": 57694.26034450531, "step_time_sec": 8.2405685960548, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6941, "loss": 3.7441458702087402, "lr": 0.0002, "elapsed_sec": 57702.49802660942, "step_time_sec": 8.237490359926596, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6942, "loss": 3.79610276222229, "lr": 0.0002, "elapsed_sec": 57710.73734688759, "step_time_sec": 8.23922153993044, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6943, "loss": 3.820888042449951, "lr": 0.0002, "elapsed_sec": 57718.97652745247, "step_time_sec": 8.238956167944707, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6944, "loss": 4.009273529052734, "lr": 0.0002, "elapsed_sec": 57727.21363210678, "step_time_sec": 8.236985648982227, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6945, "loss": 3.8881216049194336, "lr": 0.0002, "elapsed_sec": 57735.45327758789, "step_time_sec": 8.239470736007206, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6946, "loss": 3.9550445079803467, "lr": 0.0002, "elapsed_sec": 57743.6903731823, "step_time_sec": 8.236939257010818, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6947, "loss": 3.8989667892456055, "lr": 0.0002, "elapsed_sec": 57751.92974066734, "step_time_sec": 8.239120991085656, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6948, "loss": 3.833167791366577, "lr": 0.0002, "elapsed_sec": 57760.169258117676, "step_time_sec": 8.239358577993698, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6949, "loss": 3.8995072841644287, "lr": 0.0002, "elapsed_sec": 57768.40726542473, "step_time_sec": 8.237834207015112, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6950, "loss": 4.0185441970825195, "lr": 0.0002, "elapsed_sec": 57776.64611387253, "step_time_sec": 8.238728041062132, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6951, "loss": 3.9581220149993896, "lr": 0.0002, "elapsed_sec": 57784.88431596756, "step_time_sec": 8.23797082703095, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6952, "loss": 3.6669986248016357, "lr": 0.0002, "elapsed_sec": 57793.12459087372, "step_time_sec": 8.240155511070043, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6953, "loss": 4.056606292724609, "lr": 0.0002, "elapsed_sec": 57801.36557006836, "step_time_sec": 8.240789624978788, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6954, "loss": 3.8747103214263916, "lr": 0.0002, "elapsed_sec": 57809.60423231125, "step_time_sec": 8.238499950035475, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6955, "loss": 4.181436061859131, "lr": 0.0002, "elapsed_sec": 57817.845093011856, "step_time_sec": 8.240711213904433, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6956, "loss": 3.8199656009674072, "lr": 0.0002, "elapsed_sec": 57826.08360147476, "step_time_sec": 8.238364927936345, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6957, "loss": 4.031128883361816, "lr": 0.0002, "elapsed_sec": 57834.322649240494, "step_time_sec": 8.238858412019908, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6958, "loss": 4.085410118103027, "lr": 0.0002, "elapsed_sec": 57842.56247282028, "step_time_sec": 8.239660931983963, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6959, "loss": 3.9409029483795166, "lr": 0.0002, "elapsed_sec": 57850.80082321167, "step_time_sec": 8.238183224922977, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6960, "loss": 3.8967654705047607, "lr": 0.0002, "elapsed_sec": 57859.041107177734, "step_time_sec": 8.240130011923611, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6961, "loss": 3.9280426502227783, "lr": 0.0002, "elapsed_sec": 57867.28004908562, "step_time_sec": 8.238758540945128, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6962, "loss": 4.038933753967285, "lr": 0.0002, "elapsed_sec": 57875.51988363266, "step_time_sec": 8.239674028009176, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6963, "loss": 3.9737164974212646, "lr": 0.0002, "elapsed_sec": 57883.75922489166, "step_time_sec": 8.23923260497395, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6964, "loss": 3.9324471950531006, "lr": 0.0002, "elapsed_sec": 57891.997420072556, "step_time_sec": 8.237978687975556, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6965, "loss": 3.9885506629943848, "lr": 0.0002, "elapsed_sec": 57900.23709321022, "step_time_sec": 8.239490682026371, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6966, "loss": 4.030643939971924, "lr": 0.0002, "elapsed_sec": 57908.47722816467, "step_time_sec": 8.240001181024127, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6967, "loss": 3.8555548191070557, "lr": 0.0002, "elapsed_sec": 57916.716027736664, "step_time_sec": 8.238678885041736, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6968, "loss": 4.0219879150390625, "lr": 0.0002, "elapsed_sec": 57924.95648980141, "step_time_sec": 8.24024982796982, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6969, "loss": 3.925548553466797, "lr": 0.0002, "elapsed_sec": 57933.195613861084, "step_time_sec": 8.238952717976645, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6970, "loss": 3.974717378616333, "lr": 0.0002, "elapsed_sec": 57941.43436169624, "step_time_sec": 8.238586362916976, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6971, "loss": 3.9277613162994385, "lr": 0.0002, "elapsed_sec": 57949.67343854904, "step_time_sec": 8.23892911709845, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6972, "loss": 3.8517203330993652, "lr": 0.0002, "elapsed_sec": 57957.91063404083, "step_time_sec": 8.237041061045602, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6973, "loss": 3.9814000129699707, "lr": 0.0002, "elapsed_sec": 57966.1500043869, "step_time_sec": 8.23918826202862, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6974, "loss": 3.856527090072632, "lr": 0.0002, "elapsed_sec": 57974.38907670975, "step_time_sec": 8.238889912958257, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6975, "loss": 4.062811851501465, "lr": 0.0002, "elapsed_sec": 57982.62835597992, "step_time_sec": 8.239129147026688, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6976, "loss": 3.8391611576080322, "lr": 0.0002, "elapsed_sec": 57990.86782288551, "step_time_sec": 8.239324346999638, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6977, "loss": 3.9512951374053955, "lr": 0.0002, "elapsed_sec": 57999.10647082329, "step_time_sec": 8.23845488904044, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6978, "loss": 3.823701858520508, "lr": 0.0002, "elapsed_sec": 58007.34645700455, "step_time_sec": 8.23981785599608, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6979, "loss": 4.027743339538574, "lr": 0.0002, "elapsed_sec": 58015.584992170334, "step_time_sec": 8.238378316047601, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6980, "loss": 3.823763132095337, "lr": 0.0002, "elapsed_sec": 58023.82385993004, "step_time_sec": 8.238680662936531, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6981, "loss": 3.790135383605957, "lr": 0.0002, "elapsed_sec": 58032.062103033066, "step_time_sec": 8.238085602060892, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6982, "loss": 3.8334641456604004, "lr": 0.0002, "elapsed_sec": 58040.30098462105, "step_time_sec": 8.238695241045207, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6983, "loss": 3.6735806465148926, "lr": 0.0002, "elapsed_sec": 58048.54060935974, "step_time_sec": 8.239513971027918, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6984, "loss": 3.8791663646698, "lr": 0.0002, "elapsed_sec": 58056.78026652336, "step_time_sec": 8.239517678972334, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6985, "loss": 3.9501030445098877, "lr": 0.0002, "elapsed_sec": 58065.02022051811, "step_time_sec": 8.239725245977752, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6986, "loss": 3.918354034423828, "lr": 0.0002, "elapsed_sec": 58073.25998353958, "step_time_sec": 8.239612095989287, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6987, "loss": 3.8627519607543945, "lr": 0.0002, "elapsed_sec": 58081.497849702835, "step_time_sec": 8.237731240922585, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6988, "loss": 3.832257032394409, "lr": 0.0002, "elapsed_sec": 58089.73781991005, "step_time_sec": 8.239831297076307, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6989, "loss": 3.909646511077881, "lr": 0.0002, "elapsed_sec": 58097.97812914848, "step_time_sec": 8.240093633998185, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6990, "loss": 3.9600861072540283, "lr": 0.0002, "elapsed_sec": 58106.2170855999, "step_time_sec": 8.238836817909032, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6991, "loss": 3.8311331272125244, "lr": 0.0002, "elapsed_sec": 58114.457260131836, "step_time_sec": 8.240012925933115, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6992, "loss": 4.02207612991333, "lr": 0.0002, "elapsed_sec": 58122.69691133499, "step_time_sec": 8.239438010030426, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6993, "loss": 3.8972513675689697, "lr": 0.0002, "elapsed_sec": 58130.936242580414, "step_time_sec": 8.239161240984686, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6994, "loss": 4.023190021514893, "lr": 0.0002, "elapsed_sec": 58139.17570447922, "step_time_sec": 8.239330857992172, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6995, "loss": 3.7011289596557617, "lr": 0.0002, "elapsed_sec": 58147.41354084015, "step_time_sec": 8.23766115005128, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6996, "loss": 3.8905019760131836, "lr": 0.0002, "elapsed_sec": 58155.65384101868, "step_time_sec": 8.240181319997646, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6997, "loss": 4.237015247344971, "lr": 0.0002, "elapsed_sec": 58163.89397382736, "step_time_sec": 8.239957263926044, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6998, "loss": 3.8947505950927734, "lr": 0.0002, "elapsed_sec": 58172.13315033913, "step_time_sec": 8.238975999993272, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6999, "loss": 3.949688196182251, "lr": 0.0002, "elapsed_sec": 58180.37306714058, "step_time_sec": 8.239752939902246, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7000, "loss": 3.800400733947754, "lr": 0.0002, "elapsed_sec": 58188.61170172691, "step_time_sec": 51.464248298085295, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": true, "validation_ran": true, "probe_ran": true, "probe_elapsed_sec": 0.9931593659566715, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7001, "loss": 3.9858007431030273, "lr": 0.0002, "elapsed_sec": 58240.08399629593, "step_time_sec": 8.246434325003065, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7002, "loss": 3.9183297157287598, "lr": 0.00019999996991350928, "elapsed_sec": 58248.3156414032, "step_time_sec": 8.23147397406865, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7003, "loss": 4.02330207824707, "lr": 0.00019999987965405567, "elapsed_sec": 58256.54871916771, "step_time_sec": 8.232971205958165, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7004, "loss": 3.973808765411377, "lr": 0.00019999972922169488, "elapsed_sec": 58264.78070092201, "step_time_sec": 8.23180077096913, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7005, "loss": 4.041426658630371, "lr": 0.0001999995186165197, "elapsed_sec": 58273.013077020645, "step_time_sec": 8.232159227016382, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7006, "loss": 3.9395899772644043, "lr": 0.0001999992478386602, "elapsed_sec": 58281.24521970749, "step_time_sec": 8.232009210041724, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7007, "loss": 3.848543167114258, "lr": 0.00019999891688828346, "elapsed_sec": 58289.47784805298, "step_time_sec": 8.232487606932409, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7008, "loss": 3.9062623977661133, "lr": 0.0001999985257655937, "elapsed_sec": 58297.709980010986, "step_time_sec": 8.231950081069954, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7009, "loss": 3.9422075748443604, "lr": 0.00019999807447083228, "elapsed_sec": 58305.94340276718, "step_time_sec": 8.233272564946674, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7010, "loss": 3.994739532470703, "lr": 0.00019999756300427782, "elapsed_sec": 58314.176651477814, "step_time_sec": 8.233031803974882, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7011, "loss": 3.7951300144195557, "lr": 0.00019999699136624588, "elapsed_sec": 58322.409430503845, "step_time_sec": 8.23261006094981, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7012, "loss": 3.895678997039795, "lr": 0.00019999635955708932, "elapsed_sec": 58330.643684625626, "step_time_sec": 8.234154690988362, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7013, "loss": 3.915358781814575, "lr": 0.00019999566757719801, "elapsed_sec": 58338.87756276131, "step_time_sec": 8.233711843029596, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7014, "loss": 3.9107820987701416, "lr": 0.00019999491542699904, "elapsed_sec": 58347.13297343254, "step_time_sec": 8.255239189020358, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7015, "loss": 3.839203357696533, "lr": 0.00019999410310695664, "elapsed_sec": 58355.3666677475, "step_time_sec": 8.23352738795802, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7016, "loss": 3.8659942150115967, "lr": 0.00019999323061757208, "elapsed_sec": 58363.598766088486, "step_time_sec": 8.231895940029062, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7017, "loss": 3.981718063354492, "lr": 0.00019999229795938383, "elapsed_sec": 58371.831679821014, "step_time_sec": 8.232757866964675, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7018, "loss": 3.9170122146606445, "lr": 0.0001999913051329675, "elapsed_sec": 58380.063638687134, "step_time_sec": 8.231804135022685, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7019, "loss": 3.992027759552002, "lr": 0.00019999025213893583, "elapsed_sec": 58388.297040224075, "step_time_sec": 8.233268472016789, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7020, "loss": 3.8913891315460205, "lr": 0.00019998913897793872, "elapsed_sec": 58396.529000520706, "step_time_sec": 8.23180842003785, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7021, "loss": 4.046229362487793, "lr": 0.0001999879656506631, "elapsed_sec": 58404.7631444931, "step_time_sec": 8.233936705975793, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7022, "loss": 3.900265693664551, "lr": 0.00019998673215783315, "elapsed_sec": 58413.00177717209, "step_time_sec": 8.238461189088412, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7023, "loss": 3.929584264755249, "lr": 0.0001999854385002101, "elapsed_sec": 58421.24205350876, "step_time_sec": 8.240108605008572, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7024, "loss": 3.968212366104126, "lr": 0.00019998408467859232, "elapsed_sec": 58429.48272681236, "step_time_sec": 8.240534210926853, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7025, "loss": 4.012511730194092, "lr": 0.0001999826706938154, "elapsed_sec": 58437.72150802612, "step_time_sec": 8.238606380997226, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7026, "loss": 3.807037115097046, "lr": 0.00019998119654675195, "elapsed_sec": 58445.95980620384, "step_time_sec": 8.238128107972443, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7027, "loss": 3.8193233013153076, "lr": 0.00019997966223831175, "elapsed_sec": 58454.19861578941, "step_time_sec": 8.238646642072126, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7028, "loss": 4.062386989593506, "lr": 0.00019997806776944174, "elapsed_sec": 58462.4397046566, "step_time_sec": 8.240966917015612, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7029, "loss": 4.0127058029174805, "lr": 0.00019997641314112592, "elapsed_sec": 58470.679759025574, "step_time_sec": 8.239877699990757, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7030, "loss": 3.8546671867370605, "lr": 0.0001999746983543855, "elapsed_sec": 58478.91769218445, "step_time_sec": 8.237748344079591, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7031, "loss": 3.8656041622161865, "lr": 0.00019997292341027877, "elapsed_sec": 58487.15685772896, "step_time_sec": 8.23905431793537, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7032, "loss": 3.9723312854766846, "lr": 0.0001999710883099011, "elapsed_sec": 58495.39581441879, "step_time_sec": 8.238799309940077, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7033, "loss": 3.944380760192871, "lr": 0.0001999691930543851, "elapsed_sec": 58503.63426113129, "step_time_sec": 8.238245569053106, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7034, "loss": 3.9380836486816406, "lr": 0.0001999672376449004, "elapsed_sec": 58511.87332224846, "step_time_sec": 8.238903642981313, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7035, "loss": 3.8519129753112793, "lr": 0.00019996522208265384, "elapsed_sec": 58520.111354351044, "step_time_sec": 8.23790640803054, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7036, "loss": 4.000858783721924, "lr": 0.00019996314636888934, "elapsed_sec": 58528.34889984131, "step_time_sec": 8.237357453093864, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7037, "loss": 3.917731761932373, "lr": 0.0001999610105048879, "elapsed_sec": 58536.58659386635, "step_time_sec": 8.237587367999367, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7038, "loss": 3.977034330368042, "lr": 0.0001999588144919677, "elapsed_sec": 58544.825994968414, "step_time_sec": 8.239178383955732, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7039, "loss": 3.962132692337036, "lr": 0.00019995655833148407, "elapsed_sec": 58553.06533694267, "step_time_sec": 8.239173824083991, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7040, "loss": 3.773845911026001, "lr": 0.0001999542420248294, "elapsed_sec": 58561.30417084694, "step_time_sec": 8.238661637995392, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7041, "loss": 3.947843313217163, "lr": 0.00019995186557343318, "elapsed_sec": 58569.54264783859, "step_time_sec": 8.238352659041993, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7042, "loss": 3.8045239448547363, "lr": 0.0001999494289787621, "elapsed_sec": 58577.78157043457, "step_time_sec": 8.238741644076072, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7043, "loss": 3.893159866333008, "lr": 0.00019994693224231993, "elapsed_sec": 58586.02255535126, "step_time_sec": 8.24084257602226, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7044, "loss": 3.8795289993286133, "lr": 0.0001999443753656475, "elapsed_sec": 58594.26182746887, "step_time_sec": 8.239091194001958, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7045, "loss": 3.7613003253936768, "lr": 0.0001999417583503229, "elapsed_sec": 58602.49995303154, "step_time_sec": 8.237950181006454, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7046, "loss": 3.912895679473877, "lr": 0.00019993908119796114, "elapsed_sec": 58610.73884558678, "step_time_sec": 8.238749764976092, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7047, "loss": 3.9183876514434814, "lr": 0.00019993634391021452, "elapsed_sec": 58618.977598428726, "step_time_sec": 8.238590892986394, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7048, "loss": 3.984307289123535, "lr": 0.00019993354648877235, "elapsed_sec": 58627.21514868736, "step_time_sec": 8.237449976964854, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7049, "loss": 4.04412317276001, "lr": 0.00019993068893536112, "elapsed_sec": 58635.45608639717, "step_time_sec": 8.24077563197352, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7050, "loss": 4.211026668548584, "lr": 0.00019992777125174435, "elapsed_sec": 58643.69453001022, "step_time_sec": 8.238283755956218, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7051, "loss": 3.9420669078826904, "lr": 0.0001999247934397227, "elapsed_sec": 58651.93407654762, "step_time_sec": 8.239351995056495, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7052, "loss": 3.9112892150878906, "lr": 0.00019992175550113402, "elapsed_sec": 58660.172867298126, "step_time_sec": 8.238635826972313, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7053, "loss": 3.867034912109375, "lr": 0.00019991865743785315, "elapsed_sec": 58668.41218137741, "step_time_sec": 8.23914998199325, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7054, "loss": 4.060445308685303, "lr": 0.00019991549925179212, "elapsed_sec": 58676.6526017189, "step_time_sec": 8.240296815987676, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7055, "loss": 3.9334640502929688, "lr": 0.0001999122809449, "elapsed_sec": 58684.890328884125, "step_time_sec": 8.237486295984127, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7056, "loss": 3.785857677459717, "lr": 0.00019990900251916298, "elapsed_sec": 58693.1285803318, "step_time_sec": 8.238080299925059, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7057, "loss": 4.001068115234375, "lr": 0.00019990566397660442, "elapsed_sec": 58701.369059085846, "step_time_sec": 8.24032446090132, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7058, "loss": 3.9448368549346924, "lr": 0.00019990226531928473, "elapsed_sec": 58709.60824966431, "step_time_sec": 8.239029265008867, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7059, "loss": 4.031815528869629, "lr": 0.00019989880654930136, "elapsed_sec": 58717.848584890366, "step_time_sec": 8.240190858952701, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7060, "loss": 3.929867744445801, "lr": 0.000199895287668789, "elapsed_sec": 58726.088050842285, "step_time_sec": 8.239327696035616, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7061, "loss": 4.048248291015625, "lr": 0.00019989170867991927, "elapsed_sec": 58734.32680439949, "step_time_sec": 8.238552124937996, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7062, "loss": 3.941105604171753, "lr": 0.00019988806958490108, "elapsed_sec": 58742.56554245949, "step_time_sec": 8.238581636920571, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7063, "loss": 3.880868911743164, "lr": 0.00019988437038598024, "elapsed_sec": 58750.80383682251, "step_time_sec": 8.238157370011322, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7064, "loss": 3.9185709953308105, "lr": 0.0001998806110854398, "elapsed_sec": 58759.043870687485, "step_time_sec": 8.239928564988077, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7065, "loss": 3.8621578216552734, "lr": 0.00019987679168559982, "elapsed_sec": 58767.28332352638, "step_time_sec": 8.239209207007661, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7066, "loss": 3.92179536819458, "lr": 0.00019987291218881746, "elapsed_sec": 58775.523122787476, "step_time_sec": 8.239666978945024, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7067, "loss": 3.8909459114074707, "lr": 0.000199868972597487, "elapsed_sec": 58783.762781620026, "step_time_sec": 8.239503358956426, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7068, "loss": 3.8511528968811035, "lr": 0.00019986497291403981, "elapsed_sec": 58792.00162959099, "step_time_sec": 8.238703754963353, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7069, "loss": 3.9412312507629395, "lr": 0.00019986091314094432, "elapsed_sec": 58800.242676734924, "step_time_sec": 8.24086921592243, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7070, "loss": 4.170441150665283, "lr": 0.00019985679328070604, "elapsed_sec": 58808.48216128349, "step_time_sec": 8.239312547026202, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7071, "loss": 3.91555118560791, "lr": 0.0001998526133358676, "elapsed_sec": 58816.72167849541, "step_time_sec": 8.239414692972787, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7072, "loss": 3.7467615604400635, "lr": 0.00019984837330900865, "elapsed_sec": 58824.96174573898, "step_time_sec": 8.239850340993144, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7073, "loss": 3.8193507194519043, "lr": 0.00019984407320274602, "elapsed_sec": 58833.199597120285, "step_time_sec": 8.237679944024421, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7074, "loss": 3.8752939701080322, "lr": 0.00019983971301973353, "elapsed_sec": 58841.439348459244, "step_time_sec": 8.239597564097494, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7075, "loss": 3.997201919555664, "lr": 0.00019983529276266205, "elapsed_sec": 58849.679023981094, "step_time_sec": 8.239504546974786, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7076, "loss": 4.0475921630859375, "lr": 0.0001998308124342597, "elapsed_sec": 58857.917567014694, "step_time_sec": 8.238395027001388, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7077, "loss": 3.896418571472168, "lr": 0.00019982627203729146, "elapsed_sec": 58866.15626454353, "step_time_sec": 8.238537751021795, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7078, "loss": 3.933353900909424, "lr": 0.0001998216715745595, "elapsed_sec": 58874.394793987274, "step_time_sec": 8.238346687983721, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7079, "loss": 3.8446850776672363, "lr": 0.00019981701104890303, "elapsed_sec": 58882.633970975876, "step_time_sec": 8.239018713007681, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7080, "loss": 3.859673023223877, "lr": 0.00019981229046319834, "elapsed_sec": 58890.87320661545, "step_time_sec": 8.239099452039227, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7081, "loss": 3.872600555419922, "lr": 0.0001998075098203588, "elapsed_sec": 58899.11050462723, "step_time_sec": 8.23710646003019, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7082, "loss": 3.9172801971435547, "lr": 0.0001998026691233348, "elapsed_sec": 58907.35072302818, "step_time_sec": 8.240054273046553, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7083, "loss": 3.9236807823181152, "lr": 0.00019979776837511382, "elapsed_sec": 58915.589584350586, "step_time_sec": 8.23868917208165, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7084, "loss": 3.8819851875305176, "lr": 0.00019979280757872044, "elapsed_sec": 58923.82774925232, "step_time_sec": 8.237990163033828, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7085, "loss": 4.041921138763428, "lr": 0.00019978778673721622, "elapsed_sec": 58932.066487550735, "step_time_sec": 8.238620244082995, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7086, "loss": 3.894882917404175, "lr": 0.00019978270585369978, "elapsed_sec": 58940.305369377136, "step_time_sec": 8.238699261914007, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7087, "loss": 3.9128599166870117, "lr": 0.00019977756493130692, "elapsed_sec": 58948.54556441307, "step_time_sec": 8.239995728014037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7088, "loss": 3.888892650604248, "lr": 0.00019977236397321035, "elapsed_sec": 58956.78436207771, "step_time_sec": 8.238700015004724, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7089, "loss": 3.8675830364227295, "lr": 0.00019976710298261988, "elapsed_sec": 58965.02383637428, "step_time_sec": 8.239260001922958, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7090, "loss": 3.977267026901245, "lr": 0.0001997617819627824, "elapsed_sec": 58973.2634935379, "step_time_sec": 8.239486765000038, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7091, "loss": 4.157339572906494, "lr": 0.0001997564009169818, "elapsed_sec": 58981.501740932465, "step_time_sec": 8.238093350082636, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7092, "loss": 3.929396867752075, "lr": 0.00019975095984853911, "elapsed_sec": 58989.74221992493, "step_time_sec": 8.240319389034994, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7093, "loss": 3.9324307441711426, "lr": 0.00019974545876081225, "elapsed_sec": 58997.98234844208, "step_time_sec": 8.23996067000553, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7094, "loss": 4.00989294052124, "lr": 0.0001997398976571963, "elapsed_sec": 59006.22050404549, "step_time_sec": 8.238008129992522, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7095, "loss": 3.9798779487609863, "lr": 0.00019973427654112333, "elapsed_sec": 59014.46060228348, "step_time_sec": 8.239970099064521, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7096, "loss": 3.8862857818603516, "lr": 0.0001997285954160625, "elapsed_sec": 59022.69937539101, "step_time_sec": 8.238572854083031, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7097, "loss": 3.842067003250122, "lr": 0.0001997228542855199, "elapsed_sec": 59030.93911099434, "step_time_sec": 8.239606763003394, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7098, "loss": 3.8274643421173096, "lr": 0.00019971705315303877, "elapsed_sec": 59039.178150177, "step_time_sec": 8.23886160098482, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7099, "loss": 3.914236545562744, "lr": 0.00019971119202219934, "elapsed_sec": 59047.41639113426, "step_time_sec": 8.238083326024935, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7100, "loss": 3.930976152420044, "lr": 0.0001997052708966188, "elapsed_sec": 59055.65543103218, "step_time_sec": 8.238905129022896, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7101, "loss": 4.005894660949707, "lr": 0.00019969928977995149, "elapsed_sec": 59063.89263176918, "step_time_sec": 8.237020718981512, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7102, "loss": 3.8683738708496094, "lr": 0.00019969324867588864, "elapsed_sec": 59072.13108301163, "step_time_sec": 8.238286362960935, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7103, "loss": 3.8764548301696777, "lr": 0.00019968714758815864, "elapsed_sec": 59080.37043213844, "step_time_sec": 8.239172724075615, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7104, "loss": 4.106833457946777, "lr": 0.00019968098652052677, "elapsed_sec": 59088.60779595375, "step_time_sec": 8.237164190970361, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7105, "loss": 3.901313066482544, "lr": 0.00019967476547679545, "elapsed_sec": 59096.84744858742, "step_time_sec": 8.239514726912603, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7106, "loss": 3.8141636848449707, "lr": 0.00019966848446080398, "elapsed_sec": 59105.085907936096, "step_time_sec": 8.238244869979098, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7107, "loss": 3.8825747966766357, "lr": 0.00019966214347642882, "elapsed_sec": 59113.325021743774, "step_time_sec": 8.238979870104231, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7108, "loss": 4.250698089599609, "lr": 0.0001996557425275833, "elapsed_sec": 59121.564475774765, "step_time_sec": 8.23927885107696, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7109, "loss": 3.94180965423584, "lr": 0.00019964928161821785, "elapsed_sec": 59129.80313396454, "step_time_sec": 8.238493082928471, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7110, "loss": 3.844170331954956, "lr": 0.0001996427607523199, "elapsed_sec": 59138.04352068901, "step_time_sec": 8.240211228025146, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7111, "loss": 3.9029741287231445, "lr": 0.0001996361799339138, "elapsed_sec": 59146.28104329109, "step_time_sec": 8.237365519045852, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7112, "loss": 3.7763302326202393, "lr": 0.000199629539167061, "elapsed_sec": 59154.519914865494, "step_time_sec": 8.238727297983132, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7113, "loss": 3.8821933269500732, "lr": 0.0001996228384558599, "elapsed_sec": 59162.759227752686, "step_time_sec": 8.239135895040818, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7114, "loss": 3.875852346420288, "lr": 0.00019961607780444592, "elapsed_sec": 59170.99718165398, "step_time_sec": 8.237779012066312, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7115, "loss": 3.963602066040039, "lr": 0.00019960925721699142, "elapsed_sec": 59179.23781085014, "step_time_sec": 8.240483180969022, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7116, "loss": 4.030581951141357, "lr": 0.00019960237669770583, "elapsed_sec": 59187.477153778076, "step_time_sec": 8.23917434096802, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7117, "loss": 3.6593403816223145, "lr": 0.00019959543625083545, "elapsed_sec": 59195.7162258625, "step_time_sec": 8.238952061044984, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7118, "loss": 3.9686825275421143, "lr": 0.00019958843588066373, "elapsed_sec": 59203.95505237579, "step_time_sec": 8.23866845795419, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7119, "loss": 3.9111764430999756, "lr": 0.000199581375591511, "elapsed_sec": 59212.1923725605, "step_time_sec": 8.237178342067637, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7120, "loss": 3.7561566829681396, "lr": 0.00019957425538773448, "elapsed_sec": 59220.430346012115, "step_time_sec": 8.237801914918236, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7121, "loss": 3.9495651721954346, "lr": 0.00019956707527372854, "elapsed_sec": 59228.66760659218, "step_time_sec": 8.237087204004638, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7122, "loss": 3.7848761081695557, "lr": 0.00019955983525392446, "elapsed_sec": 59236.90665793419, "step_time_sec": 8.2389494930394, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7123, "loss": 3.8958377838134766, "lr": 0.00019955253533279044, "elapsed_sec": 59245.14582514763, "step_time_sec": 8.238944715005346, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7124, "loss": 4.088960647583008, "lr": 0.00019954517551483175, "elapsed_sec": 59253.38390493393, "step_time_sec": 8.237940026912838, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7125, "loss": 3.974393367767334, "lr": 0.0001995377558045905, "elapsed_sec": 59261.62476229668, "step_time_sec": 8.240702874958515, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7126, "loss": 3.8498754501342773, "lr": 0.00019953027620664586, "elapsed_sec": 59269.86268687248, "step_time_sec": 8.237777382950298, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7127, "loss": 4.139691352844238, "lr": 0.00019952273672561394, "elapsed_sec": 59278.10141992569, "step_time_sec": 8.238536695949733, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7128, "loss": 3.879945755004883, "lr": 0.0001995151373661478, "elapsed_sec": 59286.34021735191, "step_time_sec": 8.23865175107494, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7129, "loss": 3.8814446926116943, "lr": 0.0001995074781329374, "elapsed_sec": 59294.57888221741, "step_time_sec": 8.238505506073125, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7130, "loss": 3.8880443572998047, "lr": 0.00019949975903070977, "elapsed_sec": 59302.81827354431, "step_time_sec": 8.239206294063479, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7131, "loss": 3.940706491470337, "lr": 0.00019949198006422882, "elapsed_sec": 59311.056975364685, "step_time_sec": 8.238585861050524, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7132, "loss": 3.8161144256591797, "lr": 0.00019948414123829533, "elapsed_sec": 59319.29583859444, "step_time_sec": 8.238628842984326, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7133, "loss": 3.7605338096618652, "lr": 0.00019947624255774716, "elapsed_sec": 59327.53443336487, "step_time_sec": 8.238483385997824, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7134, "loss": 3.909794569015503, "lr": 0.00019946828402745903, "elapsed_sec": 59335.7717590332, "step_time_sec": 8.237121165031567, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7135, "loss": 3.9052693843841553, "lr": 0.00019946026565234267, "elapsed_sec": 59344.01079082489, "step_time_sec": 8.238841640995815, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7136, "loss": 3.910626173019409, "lr": 0.00019945218743734662, "elapsed_sec": 59352.24891805649, "step_time_sec": 8.237997601972893, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7137, "loss": 3.959479570388794, "lr": 0.00019944404938745643, "elapsed_sec": 59360.48791766167, "step_time_sec": 8.238827875000425, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7138, "loss": 3.972780227661133, "lr": 0.00019943585150769463, "elapsed_sec": 59368.728422641754, "step_time_sec": 8.240335143054835, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7139, "loss": 3.9304888248443604, "lr": 0.00019942759380312054, "elapsed_sec": 59376.96681571007, "step_time_sec": 8.23825154604856, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7140, "loss": 3.897388219833374, "lr": 0.00019941927627883054, "elapsed_sec": 59385.207033872604, "step_time_sec": 8.23999916901812, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7141, "loss": 3.9715309143066406, "lr": 0.0001994108989399578, "elapsed_sec": 59393.44659900665, "step_time_sec": 8.23941703804303, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7142, "loss": 3.9186222553253174, "lr": 0.0001994024617916725, "elapsed_sec": 59401.68597650528, "step_time_sec": 8.239211060921662, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7143, "loss": 3.8234221935272217, "lr": 0.00019939396483918171, "elapsed_sec": 59409.924834012985, "step_time_sec": 8.238695589010604, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7144, "loss": 3.8074657917022705, "lr": 0.00019938540808772938, "elapsed_sec": 59418.16306948662, "step_time_sec": 8.238063577911817, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7145, "loss": 4.01168155670166, "lr": 0.0001993767915425964, "elapsed_sec": 59426.40284919739, "step_time_sec": 8.239623397006653, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7146, "loss": 4.006702423095703, "lr": 0.0001993681152091005, "elapsed_sec": 59434.640896081924, "step_time_sec": 8.237910658936016, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7147, "loss": 3.8790740966796875, "lr": 0.00019935937909259645, "elapsed_sec": 59442.87881875038, "step_time_sec": 8.237742522032931, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7148, "loss": 3.8348329067230225, "lr": 0.00019935058319847575, "elapsed_sec": 59451.11858868599, "step_time_sec": 8.239584635943174, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7149, "loss": 3.7756030559539795, "lr": 0.00019934172753216685, "elapsed_sec": 59459.35743546486, "step_time_sec": 8.23870678700041, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7150, "loss": 4.01802396774292, "lr": 0.00019933281209913515, "elapsed_sec": 59467.59877848625, "step_time_sec": 8.241203691926785, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7151, "loss": 3.9360387325286865, "lr": 0.00019932383690488285, "elapsed_sec": 59475.83742809296, "step_time_sec": 8.238510013907216, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7152, "loss": 3.9489526748657227, "lr": 0.0001993148019549491, "elapsed_sec": 59484.07464003563, "step_time_sec": 8.237043169909157, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7153, "loss": 4.128907203674316, "lr": 0.0001993057072549099, "elapsed_sec": 59492.31444120407, "step_time_sec": 8.23961284197867, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7154, "loss": 4.045854091644287, "lr": 0.00019929655281037808, "elapsed_sec": 59500.55361199379, "step_time_sec": 8.239019777975045, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7155, "loss": 4.025400638580322, "lr": 0.00019928733862700348, "elapsed_sec": 59508.79270839691, "step_time_sec": 8.238977695000358, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7156, "loss": 3.9354145526885986, "lr": 0.00019927806471047257, "elapsed_sec": 59517.03297114372, "step_time_sec": 8.240094303968363, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7157, "loss": 3.760288953781128, "lr": 0.00019926873106650896, "elapsed_sec": 59525.270330667496, "step_time_sec": 8.237125602085143, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7158, "loss": 3.861440420150757, "lr": 0.00019925933770087293, "elapsed_sec": 59533.51089930534, "step_time_sec": 8.240500051062554, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7159, "loss": 3.8619275093078613, "lr": 0.00019924988461936168, "elapsed_sec": 59541.75475335121, "step_time_sec": 8.243675006902777, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7160, "loss": 3.8744704723358154, "lr": 0.0001992403718278093, "elapsed_sec": 59549.996641397476, "step_time_sec": 8.241756523959339, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7161, "loss": 4.081801891326904, "lr": 0.00019923079933208663, "elapsed_sec": 59558.241507053375, "step_time_sec": 8.244517253013328, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7162, "loss": 3.9008681774139404, "lr": 0.00019922116713810145, "elapsed_sec": 59566.48226952553, "step_time_sec": 8.240579869016074, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7163, "loss": 3.7990450859069824, "lr": 0.00019921147525179837, "elapsed_sec": 59574.72106742859, "step_time_sec": 8.238649635924958, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7164, "loss": 4.080687999725342, "lr": 0.00019920172367915879, "elapsed_sec": 59582.95999336243, "step_time_sec": 8.238718420965597, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7165, "loss": 3.85418963432312, "lr": 0.000199191912426201, "elapsed_sec": 59591.199424505234, "step_time_sec": 8.239303544978611, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7166, "loss": 3.9777586460113525, "lr": 0.00019918204149898012, "elapsed_sec": 59599.439257860184, "step_time_sec": 8.239608863950707, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7167, "loss": 3.783754825592041, "lr": 0.00019917211090358805, "elapsed_sec": 59607.67836976051, "step_time_sec": 8.238958367961459, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7168, "loss": 3.735064744949341, "lr": 0.00019916212064615356, "elapsed_sec": 59615.91692471504, "step_time_sec": 8.238374101929367, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7169, "loss": 3.8976454734802246, "lr": 0.0001991520707328422, "elapsed_sec": 59624.157170295715, "step_time_sec": 8.240064619923942, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7170, "loss": 3.9372799396514893, "lr": 0.0001991419611698564, "elapsed_sec": 59632.39716696739, "step_time_sec": 8.239878575084731, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7171, "loss": 3.851034641265869, "lr": 0.0001991317919634353, "elapsed_sec": 59640.636630535126, "step_time_sec": 8.23929119296372, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7172, "loss": 3.909344434738159, "lr": 0.00019912156311985502, "elapsed_sec": 59648.876983881, "step_time_sec": 8.240213723038323, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7173, "loss": 3.9078001976013184, "lr": 0.0001991112746454283, "elapsed_sec": 59657.11545586586, "step_time_sec": 8.238311894005165, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7174, "loss": 3.7012133598327637, "lr": 0.0001991009265465048, "elapsed_sec": 59665.35502624512, "step_time_sec": 8.23936651297845, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7175, "loss": 4.054723262786865, "lr": 0.0001990905188294709, "elapsed_sec": 59673.59388804436, "step_time_sec": 8.238673551008105, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7176, "loss": 3.79856276512146, "lr": 0.00019908005150074982, "elapsed_sec": 59681.83275485039, "step_time_sec": 8.23871271405369, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7177, "loss": 3.7930350303649902, "lr": 0.0001990695245668016, "elapsed_sec": 59690.07262659073, "step_time_sec": 8.239706971100532, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7178, "loss": 3.9529807567596436, "lr": 0.000199058938034123, "elapsed_sec": 59698.31186389923, "step_time_sec": 8.239107056055218, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7179, "loss": 3.9742956161499023, "lr": 0.00019904829190924762, "elapsed_sec": 59706.55049204826, "step_time_sec": 8.238470928976312, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7180, "loss": 3.9326162338256836, "lr": 0.0001990375861987458, "elapsed_sec": 59714.79048037529, "step_time_sec": 8.239799889968708, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7181, "loss": 3.970034122467041, "lr": 0.00019902682090922465, "elapsed_sec": 59723.0276362896, "step_time_sec": 8.23695420299191, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7182, "loss": 3.910111904144287, "lr": 0.00019901599604732806, "elapsed_sec": 59731.268361091614, "step_time_sec": 8.240541354985908, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7183, "loss": 3.855403184890747, "lr": 0.00019900511161973668, "elapsed_sec": 59739.506471157074, "step_time_sec": 8.2379507960286, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7184, "loss": 3.9479598999023438, "lr": 0.00019899416763316795, "elapsed_sec": 59747.74425768852, "step_time_sec": 8.237640404026024, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7185, "loss": 3.933682918548584, "lr": 0.00019898316409437607, "elapsed_sec": 59755.982033491135, "step_time_sec": 8.237590425065719, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7186, "loss": 3.9054596424102783, "lr": 0.00019897210101015196, "elapsed_sec": 59764.22061705589, "step_time_sec": 8.238418876077048, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7187, "loss": 3.879938840866089, "lr": 0.00019896097838732326, "elapsed_sec": 59772.46083641052, "step_time_sec": 8.24008959904313, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7188, "loss": 3.741886854171753, "lr": 0.00019894979623275442, "elapsed_sec": 59780.70045495033, "step_time_sec": 8.239448211970739, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7189, "loss": 3.903092861175537, "lr": 0.00019893855455334661, "elapsed_sec": 59788.94078326225, "step_time_sec": 8.240168382995762, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7190, "loss": 3.85062313079834, "lr": 0.0001989272533560377, "elapsed_sec": 59797.179736852646, "step_time_sec": 8.238790174014866, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7191, "loss": 3.7672975063323975, "lr": 0.00019891589264780238, "elapsed_sec": 59805.41913962364, "step_time_sec": 8.23918854596559, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7192, "loss": 3.8985791206359863, "lr": 0.00019890447243565195, "elapsed_sec": 59813.659041404724, "step_time_sec": 8.239733802038245, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7193, "loss": 3.7781753540039062, "lr": 0.00019889299272663456, "elapsed_sec": 59821.89853310585, "step_time_sec": 8.23933288094122, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7194, "loss": 3.816899061203003, "lr": 0.00019888145352783495, "elapsed_sec": 59830.13613700867, "step_time_sec": 8.237444260972552, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7195, "loss": 3.8033447265625, "lr": 0.0001988698548463747, "elapsed_sec": 59838.37597775459, "step_time_sec": 8.239658496924676, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7196, "loss": 3.9348785877227783, "lr": 0.00019885819668941196, "elapsed_sec": 59846.61597466469, "step_time_sec": 8.239885966992006, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7197, "loss": 3.941911220550537, "lr": 0.0001988464790641417, "elapsed_sec": 59854.85594582558, "step_time_sec": 8.239789591985755, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7198, "loss": 3.939427137374878, "lr": 0.0001988347019777956, "elapsed_sec": 59863.09592247009, "step_time_sec": 8.239774859976023, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7199, "loss": 3.849695920944214, "lr": 0.00019882286543764193, "elapsed_sec": 59871.33384871483, "step_time_sec": 8.237740156939253, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7200, "loss": 4.0115509033203125, "lr": 0.00019881096945098577, "elapsed_sec": 59879.57383441925, "step_time_sec": 8.239860410918482, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7201, "loss": 3.8023996353149414, "lr": 0.0001987990140251688, "elapsed_sec": 59887.81230926514, "step_time_sec": 8.238313317997381, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7202, "loss": 4.014066219329834, "lr": 0.00019878699916756938, "elapsed_sec": 59896.05182862282, "step_time_sec": 8.239347485941835, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7203, "loss": 3.9074301719665527, "lr": 0.00019877492488560263, "elapsed_sec": 59904.29188513756, "step_time_sec": 8.239864520961419, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7204, "loss": 3.945211887359619, "lr": 0.0001987627911867203, "elapsed_sec": 59912.53014588356, "step_time_sec": 8.23813060997054, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7205, "loss": 3.7717556953430176, "lr": 0.00019875059807841083, "elapsed_sec": 59920.770473480225, "step_time_sec": 8.240145985968411, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7206, "loss": 3.976238250732422, "lr": 0.00019873834556819923, "elapsed_sec": 59929.010194540024, "step_time_sec": 8.239580559078604, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7207, "loss": 3.9013543128967285, "lr": 0.00019872603366364734, "elapsed_sec": 59937.24855303764, "step_time_sec": 8.238207418937236, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7208, "loss": 3.856022357940674, "lr": 0.00019871366237235347, "elapsed_sec": 59945.48767518997, "step_time_sec": 8.238908187951893, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7209, "loss": 3.9889872074127197, "lr": 0.0001987012317019527, "elapsed_sec": 59953.72521853447, "step_time_sec": 8.2374142940389, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7210, "loss": 4.060306072235107, "lr": 0.00019868874166011676, "elapsed_sec": 59961.965280771255, "step_time_sec": 8.23984527005814, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7211, "loss": 3.972097158432007, "lr": 0.00019867619225455397, "elapsed_sec": 59970.20462822914, "step_time_sec": 8.239201643038541, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7212, "loss": 3.8064627647399902, "lr": 0.0001986635834930093, "elapsed_sec": 59978.442685842514, "step_time_sec": 8.237929422990419, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7213, "loss": 3.9544286727905273, "lr": 0.00019865091538326436, "elapsed_sec": 59986.682638168335, "step_time_sec": 8.23982459004037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7214, "loss": 3.9566338062286377, "lr": 0.0001986381879331374, "elapsed_sec": 59994.920622348785, "step_time_sec": 8.237742864061147, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7215, "loss": 3.925288438796997, "lr": 0.00019862540115048324, "elapsed_sec": 60003.15860414505, "step_time_sec": 8.23782645596657, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7216, "loss": 4.047096252441406, "lr": 0.00019861255504319337, "elapsed_sec": 60011.39708662033, "step_time_sec": 8.23834026302211, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7217, "loss": 3.8022420406341553, "lr": 0.0001985996496191959, "elapsed_sec": 60019.63672184944, "step_time_sec": 8.239463102072477, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7218, "loss": 3.684274196624756, "lr": 0.00019858668488645552, "elapsed_sec": 60027.8770904541, "step_time_sec": 8.240205642068759, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7219, "loss": 3.8797943592071533, "lr": 0.00019857366085297354, "elapsed_sec": 60036.115971803665, "step_time_sec": 8.23867329210043, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7220, "loss": 3.9482781887054443, "lr": 0.0001985605775267878, "elapsed_sec": 60044.35572767258, "step_time_sec": 8.23957734706346, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7221, "loss": 3.9733288288116455, "lr": 0.00019854743491597285, "elapsed_sec": 60052.594526052475, "step_time_sec": 8.238652772037312, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7222, "loss": 3.7737746238708496, "lr": 0.00019853423302863973, "elapsed_sec": 60060.83292222023, "step_time_sec": 8.238216584082693, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7223, "loss": 3.9050183296203613, "lr": 0.00019852097187293612, "elapsed_sec": 60069.07406640053, "step_time_sec": 8.240989449899644, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7224, "loss": 3.843703031539917, "lr": 0.00019850765145704627, "elapsed_sec": 60077.31331825256, "step_time_sec": 8.239057963015512, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7225, "loss": 3.9426755905151367, "lr": 0.00019849427178919097, "elapsed_sec": 60085.55282878876, "step_time_sec": 8.2393599570496, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7226, "loss": 3.9626386165618896, "lr": 0.00019848083287762758, "elapsed_sec": 60093.79237699509, "step_time_sec": 8.239361925050616, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7227, "loss": 3.741938352584839, "lr": 0.0001984673347306501, "elapsed_sec": 60102.030126810074, "step_time_sec": 8.23759308399167, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7228, "loss": 3.722644805908203, "lr": 0.00019845377735658898, "elapsed_sec": 60110.270550489426, "step_time_sec": 8.240257628029212, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7229, "loss": 3.8841662406921387, "lr": 0.0001984401607638113, "elapsed_sec": 60118.509689092636, "step_time_sec": 8.23901029699482, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7230, "loss": 3.923830986022949, "lr": 0.00019842648496072064, "elapsed_sec": 60126.747874975204, "step_time_sec": 8.238048461964354, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7231, "loss": 3.7801620960235596, "lr": 0.00019841274995575716, "elapsed_sec": 60134.98790359497, "step_time_sec": 8.239800269016996, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7232, "loss": 3.8577868938446045, "lr": 0.00019839895575739753, "elapsed_sec": 60143.227776288986, "step_time_sec": 8.239712905022316, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7233, "loss": 3.9232263565063477, "lr": 0.00019838510237415498, "elapsed_sec": 60151.46726202965, "step_time_sec": 8.239341126056388, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7234, "loss": 3.7891602516174316, "lr": 0.0001983711898145792, "elapsed_sec": 60159.706981658936, "step_time_sec": 8.239551134058274, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7235, "loss": 3.9798920154571533, "lr": 0.0001983572180872565, "elapsed_sec": 60167.94491529465, "step_time_sec": 8.23777392692864, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7236, "loss": 3.75579833984375, "lr": 0.00019834318720080964, "elapsed_sec": 60176.18475461006, "step_time_sec": 8.23964452103246, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7237, "loss": 4.010015964508057, "lr": 0.00019832909716389785, "elapsed_sec": 60184.42468261719, "step_time_sec": 8.239776736008935, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7238, "loss": 3.870189666748047, "lr": 0.00019831494798521702, "elapsed_sec": 60192.66341662407, "step_time_sec": 8.238554767915048, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7239, "loss": 3.9814679622650146, "lr": 0.00019830073967349937, "elapsed_sec": 60200.904826164246, "step_time_sec": 8.24122178601101, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7240, "loss": 3.9396872520446777, "lr": 0.00019828647223751368, "elapsed_sec": 60209.14326071739, "step_time_sec": 8.238262505969033, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7241, "loss": 3.8335142135620117, "lr": 0.00019827214568606528, "elapsed_sec": 60217.38274073601, "step_time_sec": 8.239324206020683, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7242, "loss": 3.8212077617645264, "lr": 0.00019825776002799586, "elapsed_sec": 60225.62071275711, "step_time_sec": 8.237860005930997, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7243, "loss": 3.843325138092041, "lr": 0.0001982433152721837, "elapsed_sec": 60233.8586127758, "step_time_sec": 8.23765495303087, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7244, "loss": 3.8134586811065674, "lr": 0.00019822881142754348, "elapsed_sec": 60242.097543001175, "step_time_sec": 8.238781980937347, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7245, "loss": 3.995920181274414, "lr": 0.0001982142485030264, "elapsed_sec": 60250.33648490906, "step_time_sec": 8.23875768110156, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7246, "loss": 3.9128003120422363, "lr": 0.0001981996265076201, "elapsed_sec": 60258.57618331909, "step_time_sec": 8.239573750062846, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7247, "loss": 3.9717350006103516, "lr": 0.00019818494545034864, "elapsed_sec": 60266.81457233429, "step_time_sec": 8.238239760976285, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7248, "loss": 3.9128427505493164, "lr": 0.0001981702053402726, "elapsed_sec": 60275.05303668976, "step_time_sec": 8.238219626015052, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7249, "loss": 3.7899186611175537, "lr": 0.00019815540618648895, "elapsed_sec": 60283.29357838631, "step_time_sec": 8.240416763932444, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7250, "loss": 3.855646848678589, "lr": 0.0001981405479981311, "elapsed_sec": 60291.53182840347, "step_time_sec": 8.238094873959199, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7251, "loss": 3.7697532176971436, "lr": 0.00019812563078436897, "elapsed_sec": 60299.77168297768, "step_time_sec": 8.239678644924425, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7252, "loss": 3.9198789596557617, "lr": 0.00019811065455440876, "elapsed_sec": 60308.01148414612, "step_time_sec": 8.239691014052369, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7253, "loss": 3.8590919971466064, "lr": 0.00019809561931749328, "elapsed_sec": 60316.251051187515, "step_time_sec": 8.239354292978533, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7254, "loss": 3.8203935623168945, "lr": 0.00019808052508290156, "elapsed_sec": 60324.49311590195, "step_time_sec": 8.241921035922132, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7255, "loss": 3.926313638687134, "lr": 0.00019806537185994925, "elapsed_sec": 60332.732404470444, "step_time_sec": 8.23912298004143, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7256, "loss": 3.9278831481933594, "lr": 0.00019805015965798818, "elapsed_sec": 60340.971153497696, "step_time_sec": 8.238558908109553, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7257, "loss": 3.793017625808716, "lr": 0.0001980348884864068, "elapsed_sec": 60349.21119213104, "step_time_sec": 8.239877389045432, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7258, "loss": 3.9048640727996826, "lr": 0.00019801955835462975, "elapsed_sec": 60357.4485912323, "step_time_sec": 8.237244852003641, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7259, "loss": 3.8548974990844727, "lr": 0.0001980041692721182, "elapsed_sec": 60365.68772959709, "step_time_sec": 8.238978169974871, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7260, "loss": 3.8698198795318604, "lr": 0.0001979887212483697, "elapsed_sec": 60373.927755355835, "step_time_sec": 8.239857517997734, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7261, "loss": 3.8876144886016846, "lr": 0.00019797321429291807, "elapsed_sec": 60382.16540122032, "step_time_sec": 8.237448294064961, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7262, "loss": 3.9708659648895264, "lr": 0.0001979576484153336, "elapsed_sec": 60390.4054210186, "step_time_sec": 8.239895100006834, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7263, "loss": 3.952486753463745, "lr": 0.00019794202362522287, "elapsed_sec": 60398.6443169117, "step_time_sec": 8.238696392974816, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7264, "loss": 3.9064433574676514, "lr": 0.00019792633993222892, "elapsed_sec": 60406.8831717968, "step_time_sec": 8.238742141984403, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7265, "loss": 3.913958787918091, "lr": 0.00019791059734603102, "elapsed_sec": 60415.123004198074, "step_time_sec": 8.239614143967628, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7266, "loss": 3.7721898555755615, "lr": 0.00019789479587634484, "elapsed_sec": 60423.3602604866, "step_time_sec": 8.237119007972069, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7267, "loss": 3.9070851802825928, "lr": 0.00019787893553292244, "elapsed_sec": 60431.599418878555, "step_time_sec": 8.239039979991503, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7268, "loss": 3.8933706283569336, "lr": 0.00019786301632555216, "elapsed_sec": 60439.838963747025, "step_time_sec": 8.239340128027834, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7269, "loss": 3.9464874267578125, "lr": 0.00019784703826405866, "elapsed_sec": 60448.07895350456, "step_time_sec": 8.23982558795251, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7270, "loss": 3.9327104091644287, "lr": 0.00019783100135830297, "elapsed_sec": 60456.31898474693, "step_time_sec": 8.239819320035167, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7271, "loss": 3.9594321250915527, "lr": 0.00019781490561818235, "elapsed_sec": 60464.55844759941, "step_time_sec": 8.239329615025781, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7272, "loss": 4.047475337982178, "lr": 0.0001977987510536305, "elapsed_sec": 60472.79753494263, "step_time_sec": 8.238872008980252, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7273, "loss": 3.777529716491699, "lr": 0.0001977825376746173, "elapsed_sec": 60481.03678178787, "step_time_sec": 8.239060249994509, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7274, "loss": 3.9934284687042236, "lr": 0.00019776626549114902, "elapsed_sec": 60489.2759873867, "step_time_sec": 8.239074343000539, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7275, "loss": 3.853342294692993, "lr": 0.00019774993451326814, "elapsed_sec": 60497.51382899284, "step_time_sec": 8.237683737068437, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7276, "loss": 3.901582956314087, "lr": 0.0001977335447510535, "elapsed_sec": 60505.752489328384, "step_time_sec": 8.238511770963669, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7277, "loss": 4.000204563140869, "lr": 0.00019771709621462018, "elapsed_sec": 60513.9927854538, "step_time_sec": 8.240107900928706, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7278, "loss": 3.9092884063720703, "lr": 0.0001977005889141195, "elapsed_sec": 60522.231244802475, "step_time_sec": 8.238296907045878, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7279, "loss": 4.086248874664307, "lr": 0.0001976840228597391, "elapsed_sec": 60530.47015094757, "step_time_sec": 8.23878919298295, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7280, "loss": 3.7995030879974365, "lr": 0.0001976673980617029, "elapsed_sec": 60538.709684848785, "step_time_sec": 8.239340206957422, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7281, "loss": 4.145299911499023, "lr": 0.00019765071453027103, "elapsed_sec": 60546.94617486, "step_time_sec": 8.236353332991712, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7282, "loss": 3.9467670917510986, "lr": 0.00019763397227573986, "elapsed_sec": 60555.18733882904, "step_time_sec": 8.241000571055338, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7283, "loss": 4.033139228820801, "lr": 0.00019761717130844202, "elapsed_sec": 60563.4273416996, "step_time_sec": 8.239836400025524, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7284, "loss": 3.6930809020996094, "lr": 0.0001976003116387464, "elapsed_sec": 60571.66465854645, "step_time_sec": 8.23715712595731, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7285, "loss": 3.8681039810180664, "lr": 0.00019758339327705806, "elapsed_sec": 60579.904888153076, "step_time_sec": 8.240039800992236, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7286, "loss": 3.6989309787750244, "lr": 0.00019756641623381834, "elapsed_sec": 60588.14461064339, "step_time_sec": 8.239589334931225, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7287, "loss": 3.9847450256347656, "lr": 0.00019754938051950476, "elapsed_sec": 60596.38441514969, "step_time_sec": 8.239587374031544, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7288, "loss": 3.7526326179504395, "lr": 0.00019753228614463107, "elapsed_sec": 60604.62357354164, "step_time_sec": 8.239029140910134, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7289, "loss": 3.863887310028076, "lr": 0.00019751513311974718, "elapsed_sec": 60612.86124444008, "step_time_sec": 8.2375178799266, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7290, "loss": 3.936598300933838, "lr": 0.00019749792145543925, "elapsed_sec": 60621.101702451706, "step_time_sec": 8.240307282074355, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7291, "loss": 3.9814164638519287, "lr": 0.0001974806511623297, "elapsed_sec": 60629.34047579765, "step_time_sec": 8.23856883996632, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7292, "loss": 3.8396806716918945, "lr": 0.00019746332225107686, "elapsed_sec": 60637.57876968384, "step_time_sec": 8.238141590030864, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7293, "loss": 3.9091548919677734, "lr": 0.00019744593473237557, "elapsed_sec": 60645.818212509155, "step_time_sec": 8.239283949020319, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7294, "loss": 3.8369946479797363, "lr": 0.0001974284886169566, "elapsed_sec": 60654.05644083023, "step_time_sec": 8.238030678010546, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7295, "loss": 3.994140148162842, "lr": 0.00019741098391558708, "elapsed_sec": 60662.29661631584, "step_time_sec": 8.240015801973641, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7296, "loss": 3.9627156257629395, "lr": 0.0001973934206390701, "elapsed_sec": 60670.53584742546, "step_time_sec": 8.239071748917922, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7297, "loss": 3.7638466358184814, "lr": 0.00019737579879824498, "elapsed_sec": 60678.776895046234, "step_time_sec": 8.240924476995133, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7298, "loss": 3.838434934616089, "lr": 0.00019735811840398726, "elapsed_sec": 60687.0165207386, "step_time_sec": 8.239478292991407, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7299, "loss": 3.8822007179260254, "lr": 0.00019734037946720847, "elapsed_sec": 60695.25637936592, "step_time_sec": 8.239611136028543, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7300, "loss": 3.859532356262207, "lr": 0.00019732258199885644, "elapsed_sec": 60714.926121234894, "step_time_sec": 19.669596012099646, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7301, "loss": 3.8124749660491943, "lr": 0.00019730472600991498, "elapsed_sec": 60723.15822339058, "step_time_sec": 8.23196036403533, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7302, "loss": 3.9028878211975098, "lr": 0.0001972868115114041, "elapsed_sec": 60731.39177083969, "step_time_sec": 8.233389385044575, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7303, "loss": 3.757430076599121, "lr": 0.00019726883851437987, "elapsed_sec": 60739.62961316109, "step_time_sec": 8.23770534596406, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7304, "loss": 3.8894765377044678, "lr": 0.00019725080702993447, "elapsed_sec": 60747.86873936653, "step_time_sec": 8.238906477112323, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7305, "loss": 3.941800117492676, "lr": 0.0001972327170691962, "elapsed_sec": 60756.11011362076, "step_time_sec": 8.24114113394171, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7306, "loss": 3.9174158573150635, "lr": 0.00019721456864332944, "elapsed_sec": 60764.34991645813, "step_time_sec": 8.2396487389924, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7307, "loss": 3.8182497024536133, "lr": 0.00019719636176353467, "elapsed_sec": 60772.58964729309, "step_time_sec": 8.239575557992794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7308, "loss": 3.965500831604004, "lr": 0.00019717809644104843, "elapsed_sec": 60780.829419612885, "step_time_sec": 8.239643941051327, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7309, "loss": 3.9339511394500732, "lr": 0.00019715977268714327, "elapsed_sec": 60789.066578149796, "step_time_sec": 8.236943506984971, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7310, "loss": 3.9285078048706055, "lr": 0.0001971413905131279, "elapsed_sec": 60797.30569911003, "step_time_sec": 8.238954996922985, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7311, "loss": 4.012170791625977, "lr": 0.00019712294993034702, "elapsed_sec": 60805.545062065125, "step_time_sec": 8.23922827502247, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7312, "loss": 3.963243007659912, "lr": 0.00019710445095018141, "elapsed_sec": 60813.78418302536, "step_time_sec": 8.238942321040668, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7313, "loss": 3.884058713912964, "lr": 0.0001970858935840479, "elapsed_sec": 60822.022609472275, "step_time_sec": 8.238317398936488, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7314, "loss": 3.734272003173828, "lr": 0.0001970672778433993, "elapsed_sec": 60830.26149368286, "step_time_sec": 8.238656473928131, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7315, "loss": 3.860262155532837, "lr": 0.00019704860373972446, "elapsed_sec": 60838.501105070114, "step_time_sec": 8.239487976999953, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7316, "loss": 3.8662030696868896, "lr": 0.0001970298712845483, "elapsed_sec": 60846.73988842964, "step_time_sec": 8.23862983006984, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7317, "loss": 3.8838868141174316, "lr": 0.00019701108048943172, "elapsed_sec": 60854.97913432121, "step_time_sec": 8.23903433792293, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7318, "loss": 3.906519889831543, "lr": 0.0001969922313659716, "elapsed_sec": 60863.21759271622, "step_time_sec": 8.238297895994037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7319, "loss": 3.919675588607788, "lr": 0.00019697332392580087, "elapsed_sec": 60871.45692181587, "step_time_sec": 8.23919880494941, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7320, "loss": 3.920018196105957, "lr": 0.0001969543581805884, "elapsed_sec": 60879.69700694084, "step_time_sec": 8.239975516917184, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7321, "loss": 3.951638698577881, "lr": 0.0001969353341420391, "elapsed_sec": 60887.93701553345, "step_time_sec": 8.239795097964816, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7322, "loss": 3.9191081523895264, "lr": 0.0001969162518218938, "elapsed_sec": 60896.17471718788, "step_time_sec": 8.237502763047814, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7323, "loss": 3.882855176925659, "lr": 0.0001968971112319293, "elapsed_sec": 60904.415212631226, "step_time_sec": 8.240322716999799, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7324, "loss": 3.8840699195861816, "lr": 0.0001968779123839584, "elapsed_sec": 60912.65345430374, "step_time_sec": 8.238077581045218, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7325, "loss": 3.9390904903411865, "lr": 0.00019685865528982984, "elapsed_sec": 60920.89314508438, "step_time_sec": 8.239533475018106, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7326, "loss": 3.8317816257476807, "lr": 0.0001968393399614283, "elapsed_sec": 60929.13327240944, "step_time_sec": 8.239962356048636, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7327, "loss": 3.748913526535034, "lr": 0.00019681996641067443, "elapsed_sec": 60937.37188029289, "step_time_sec": 8.238444162067026, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7328, "loss": 3.9430699348449707, "lr": 0.00019680053464952478, "elapsed_sec": 60945.611771821976, "step_time_sec": 8.23975629301276, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7329, "loss": 3.7671520709991455, "lr": 0.00019678104468997176, "elapsed_sec": 60953.85206365585, "step_time_sec": 8.240117077948526, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7330, "loss": 3.934274196624756, "lr": 0.0001967614965440439, "elapsed_sec": 60962.09024119377, "step_time_sec": 8.238062525982969, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7331, "loss": 3.9863247871398926, "lr": 0.0001967418902238054, "elapsed_sec": 60970.33069586754, "step_time_sec": 8.240244974964298, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7332, "loss": 3.992899179458618, "lr": 0.00019672222574135655, "elapsed_sec": 60978.570302248, "step_time_sec": 8.239515562076122, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7333, "loss": 3.837690830230713, "lr": 0.00019670250310883336, "elapsed_sec": 60986.80998969078, "step_time_sec": 8.239506704965606, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7334, "loss": 4.056400775909424, "lr": 0.00019668272233840794, "elapsed_sec": 60995.0497405529, "step_time_sec": 8.239528905949555, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7335, "loss": 3.9298641681671143, "lr": 0.00019666288344228808, "elapsed_sec": 61003.28628516197, "step_time_sec": 8.236382167902775, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7336, "loss": 3.703770399093628, "lr": 0.00019664298643271756, "elapsed_sec": 61011.5261323452, "step_time_sec": 8.239654178963974, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7337, "loss": 3.8692381381988525, "lr": 0.00019662303132197604, "elapsed_sec": 61019.76457595825, "step_time_sec": 8.238273745053448, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7338, "loss": 3.9294114112854004, "lr": 0.00019660301812237892, "elapsed_sec": 61028.00410985947, "step_time_sec": 8.239413885981776, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7339, "loss": 3.8731112480163574, "lr": 0.00019658294684627754, "elapsed_sec": 61036.24393939972, "step_time_sec": 8.239652927964926, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7340, "loss": 3.8922622203826904, "lr": 0.00019656281750605908, "elapsed_sec": 61044.48186469078, "step_time_sec": 8.237758266972378, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7341, "loss": 3.9358015060424805, "lr": 0.00019654263011414653, "elapsed_sec": 61052.721333265305, "step_time_sec": 8.239309857948683, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7342, "loss": 3.8829426765441895, "lr": 0.00019652238468299873, "elapsed_sec": 61060.95950961113, "step_time_sec": 8.238037626026198, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7343, "loss": 3.868126153945923, "lr": 0.0001965020812251103, "elapsed_sec": 61069.197966337204, "step_time_sec": 8.238284401013516, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7344, "loss": 3.704892158508301, "lr": 0.00019648171975301173, "elapsed_sec": 61077.43718481064, "step_time_sec": 8.239047897048295, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7345, "loss": 4.0527448654174805, "lr": 0.00019646130027926927, "elapsed_sec": 61085.675528764725, "step_time_sec": 8.238200973020867, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7346, "loss": 3.900930881500244, "lr": 0.00019644082281648493, "elapsed_sec": 61093.91430592537, "step_time_sec": 8.238656994071789, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7347, "loss": 3.7921485900878906, "lr": 0.00019642028737729662, "elapsed_sec": 61102.15418553352, "step_time_sec": 8.239694328978658, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7348, "loss": 3.9187347888946533, "lr": 0.00019639969397437795, "elapsed_sec": 61110.39396595955, "step_time_sec": 8.23955904901959, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7349, "loss": 3.815164089202881, "lr": 0.0001963790426204383, "elapsed_sec": 61118.633922576904, "step_time_sec": 8.239864708040841, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7350, "loss": 3.957667589187622, "lr": 0.00019635833332822287, "elapsed_sec": 61126.87279868126, "step_time_sec": 8.238702232949436, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7351, "loss": 3.9925689697265625, "lr": 0.00019633756611051254, "elapsed_sec": 61135.11396861076, "step_time_sec": 8.240973924985155, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7352, "loss": 3.942871570587158, "lr": 0.000196316740980124, "elapsed_sec": 61143.353549957275, "step_time_sec": 8.239415168063715, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7353, "loss": 3.991389036178589, "lr": 0.0001962958579499097, "elapsed_sec": 61151.59117650986, "step_time_sec": 8.237456624978222, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7354, "loss": 3.8917582035064697, "lr": 0.0001962749170327577, "elapsed_sec": 61159.82933449745, "step_time_sec": 8.237979181925766, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7355, "loss": 3.922870397567749, "lr": 0.00019625391824159194, "elapsed_sec": 61168.06734132767, "step_time_sec": 8.237858009990305, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7356, "loss": 3.669074296951294, "lr": 0.00019623286158937197, "elapsed_sec": 61176.30821108818, "step_time_sec": 8.240690303966403, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7357, "loss": 3.9782440662384033, "lr": 0.00019621174708909313, "elapsed_sec": 61184.54794359207, "step_time_sec": 8.239579950924963, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7358, "loss": 3.887333631515503, "lr": 0.0001961905747537864, "elapsed_sec": 61192.78635716438, "step_time_sec": 8.238241941086017, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7359, "loss": 3.963322401046753, "lr": 0.00019616934459651845, "elapsed_sec": 61201.025292634964, "step_time_sec": 8.238763996050693, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7360, "loss": 3.8847591876983643, "lr": 0.00019614805663039164, "elapsed_sec": 61209.26278591156, "step_time_sec": 8.237367094028741, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7361, "loss": 3.973238229751587, "lr": 0.00019612671086854412, "elapsed_sec": 61217.50191092491, "step_time_sec": 8.238981861970387, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7362, "loss": 3.899449348449707, "lr": 0.00019610530732414955, "elapsed_sec": 61225.74146437645, "step_time_sec": 8.23932713503018, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7363, "loss": 3.98972749710083, "lr": 0.0001960838460104173, "elapsed_sec": 61233.97918462753, "step_time_sec": 8.237548481905833, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7364, "loss": 3.7726030349731445, "lr": 0.00019606232694059243, "elapsed_sec": 61242.217480659485, "step_time_sec": 8.238171156030148, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7365, "loss": 3.9144093990325928, "lr": 0.0001960407501279556, "elapsed_sec": 61250.45643424988, "step_time_sec": 8.238762460066937, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7366, "loss": 3.914170503616333, "lr": 0.0001960191155858232, "elapsed_sec": 61258.69554567337, "step_time_sec": 8.238982754992321, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7367, "loss": 4.0265302658081055, "lr": 0.0001959974233275471, "elapsed_sec": 61266.93428182602, "step_time_sec": 8.238574311952107, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7368, "loss": 3.8618171215057373, "lr": 0.0001959756733665149, "elapsed_sec": 61275.17364215851, "step_time_sec": 8.239140028017573, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7369, "loss": 3.9969310760498047, "lr": 0.00019595386571614977, "elapsed_sec": 61283.41456365585, "step_time_sec": 8.240755071048625, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7370, "loss": 3.8589038848876953, "lr": 0.00019593200038991052, "elapsed_sec": 61291.653490543365, "step_time_sec": 8.23876402399037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7371, "loss": 3.95186710357666, "lr": 0.0001959100774012915, "elapsed_sec": 61299.89272618294, "step_time_sec": 8.239062042091973, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7372, "loss": 3.972378969192505, "lr": 0.00019588809676382267, "elapsed_sec": 61308.13164424896, "step_time_sec": 8.238750157062896, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7373, "loss": 4.011697769165039, "lr": 0.00019586605849106957, "elapsed_sec": 61316.3702545166, "step_time_sec": 8.238419622997753, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7374, "loss": 3.9286177158355713, "lr": 0.00019584396259663336, "elapsed_sec": 61324.609946489334, "step_time_sec": 8.239537487970665, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7375, "loss": 3.9928388595581055, "lr": 0.0001958218090941507, "elapsed_sec": 61332.84790587425, "step_time_sec": 8.237774636945687, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7376, "loss": 3.9107882976531982, "lr": 0.00019579959799729377, "elapsed_sec": 61341.08758473396, "step_time_sec": 8.239506825921126, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7377, "loss": 3.950211524963379, "lr": 0.00019577732931977038, "elapsed_sec": 61349.327104091644, "step_time_sec": 8.239356806036085, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7378, "loss": 3.936357259750366, "lr": 0.00019575500307532386, "elapsed_sec": 61357.563876628876, "step_time_sec": 8.236656988970935, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7379, "loss": 3.8745925426483154, "lr": 0.000195732619277733, "elapsed_sec": 61365.803013563156, "step_time_sec": 8.238933423999697, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7380, "loss": 3.9370148181915283, "lr": 0.00019571017794081219, "elapsed_sec": 61374.04112863541, "step_time_sec": 8.237967684981413, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7381, "loss": 3.9336905479431152, "lr": 0.0001956876790784113, "elapsed_sec": 61382.280540943146, "step_time_sec": 8.239235813962296, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7382, "loss": 4.0277557373046875, "lr": 0.0001956651227044157, "elapsed_sec": 61390.52111983299, "step_time_sec": 8.240408021956682, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7383, "loss": 4.030332088470459, "lr": 0.00019564250883274622, "elapsed_sec": 61398.76030063629, "step_time_sec": 8.239049068070017, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7384, "loss": 3.936760187149048, "lr": 0.00019561983747735923, "elapsed_sec": 61406.99882435799, "step_time_sec": 8.238346816971898, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7385, "loss": 3.9393386840820312, "lr": 0.0001955971086522466, "elapsed_sec": 61415.23890399933, "step_time_sec": 8.23986008297652, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7386, "loss": 4.00948429107666, "lr": 0.00019557432237143554, "elapsed_sec": 61423.47783899307, "step_time_sec": 8.238803218002431, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7387, "loss": 3.8820419311523438, "lr": 0.00019555147864898885, "elapsed_sec": 61431.71845412254, "step_time_sec": 8.240409519057721, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7388, "loss": 3.9413490295410156, "lr": 0.00019552857749900477, "elapsed_sec": 61439.95790028572, "step_time_sec": 8.239266120013781, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7389, "loss": 3.796140432357788, "lr": 0.00019550561893561688, "elapsed_sec": 61448.19730615616, "step_time_sec": 8.239272767910734, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7390, "loss": 3.7784206867218018, "lr": 0.0001954826029729943, "elapsed_sec": 61456.43689060211, "step_time_sec": 8.239398880978115, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7391, "loss": 3.9167356491088867, "lr": 0.00019545952962534154, "elapsed_sec": 61464.67578101158, "step_time_sec": 8.238730069017038, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7392, "loss": 3.794403076171875, "lr": 0.00019543639890689846, "elapsed_sec": 61472.916387081146, "step_time_sec": 8.240430042031221, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7393, "loss": 3.9226341247558594, "lr": 0.00019541321083194044, "elapsed_sec": 61481.15447664261, "step_time_sec": 8.237953345989808, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7394, "loss": 3.8577346801757812, "lr": 0.0001953899654147782, "elapsed_sec": 61489.39367508888, "step_time_sec": 8.238996927975677, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7395, "loss": 3.8688323497772217, "lr": 0.00019536666266975784, "elapsed_sec": 61497.63227748871, "step_time_sec": 8.238428392913193, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7396, "loss": 3.764160633087158, "lr": 0.00019534330261126087, "elapsed_sec": 61505.87055134773, "step_time_sec": 8.238088348996826, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7397, "loss": 3.8700599670410156, "lr": 0.00019531988525370411, "elapsed_sec": 61514.11193227768, "step_time_sec": 8.241241103969514, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7398, "loss": 3.965003252029419, "lr": 0.00019529641061153983, "elapsed_sec": 61522.35212492943, "step_time_sec": 8.240027346997522, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7399, "loss": 3.8530359268188477, "lr": 0.0001952728786992556, "elapsed_sec": 61530.59142255783, "step_time_sec": 8.23911444703117, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7400, "loss": 3.8476977348327637, "lr": 0.00019524928953137437, "elapsed_sec": 61538.82971215248, "step_time_sec": 8.238123744959012, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7401, "loss": 3.7620391845703125, "lr": 0.00019522564312245437, "elapsed_sec": 61547.06701898575, "step_time_sec": 8.237168536987156, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7402, "loss": 3.8335368633270264, "lr": 0.00019520193948708922, "elapsed_sec": 61555.30597758293, "step_time_sec": 8.238721312955022, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7403, "loss": 3.976346492767334, "lr": 0.00019517817863990775, "elapsed_sec": 61563.54591345787, "step_time_sec": 8.239806372090243, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7404, "loss": 3.8924245834350586, "lr": 0.0001951543605955743, "elapsed_sec": 61571.78339767456, "step_time_sec": 8.237268936005421, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7405, "loss": 3.9841837882995605, "lr": 0.00019513048536878828, "elapsed_sec": 61580.024082660675, "step_time_sec": 8.240497615071945, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7406, "loss": 3.9900248050689697, "lr": 0.00019510655297428454, "elapsed_sec": 61588.26283836365, "step_time_sec": 8.238594515947625, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7407, "loss": 3.8967983722686768, "lr": 0.00019508256342683314, "elapsed_sec": 61596.50180721283, "step_time_sec": 8.238846888998523, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7408, "loss": 3.967808485031128, "lr": 0.00019505851674123952, "elapsed_sec": 61604.74115180969, "step_time_sec": 8.239154625916854, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7409, "loss": 3.9016809463500977, "lr": 0.00019503441293234418, "elapsed_sec": 61612.978632450104, "step_time_sec": 8.237323724897578, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7410, "loss": 3.9937210083007812, "lr": 0.0001950102520150231, "elapsed_sec": 61621.217843294144, "step_time_sec": 8.239033840945922, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7411, "loss": 3.9380147457122803, "lr": 0.00019498603400418734, "elapsed_sec": 61629.45721888542, "step_time_sec": 8.239219111972488, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7412, "loss": 4.037240028381348, "lr": 0.0001949617589147833, "elapsed_sec": 61637.69777679443, "step_time_sec": 8.240396447014064, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7413, "loss": 3.8898377418518066, "lr": 0.00019493742676179252, "elapsed_sec": 61645.937848329544, "step_time_sec": 8.239883289090358, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7414, "loss": 3.9544410705566406, "lr": 0.00019491303756023184, "elapsed_sec": 61654.17628502846, "step_time_sec": 8.238365707919002, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7415, "loss": 3.9058399200439453, "lr": 0.00019488859132515328, "elapsed_sec": 61662.41602253914, "step_time_sec": 8.239518593996763, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7416, "loss": 3.9350454807281494, "lr": 0.000194864088071644, "elapsed_sec": 61670.655718803406, "step_time_sec": 8.239528078935109, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7417, "loss": 4.0052080154418945, "lr": 0.00019483952781482647, "elapsed_sec": 61678.89498758316, "step_time_sec": 8.239090134971775, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7418, "loss": 3.865211248397827, "lr": 0.0001948149105698582, "elapsed_sec": 61687.135125398636, "step_time_sec": 8.2399869550718, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7419, "loss": 3.8091604709625244, "lr": 0.00019479023635193195, "elapsed_sec": 61695.37447333336, "step_time_sec": 8.239231227897108, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7420, "loss": 3.815140724182129, "lr": 0.00019476550517627563, "elapsed_sec": 61703.613634586334, "step_time_sec": 8.238885707920417, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7421, "loss": 3.904154062271118, "lr": 0.00019474071705815233, "elapsed_sec": 61711.85124325752, "step_time_sec": 8.237431021989323, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7422, "loss": 3.9698874950408936, "lr": 0.00019471587201286026, "elapsed_sec": 61720.08875250816, "step_time_sec": 8.23740498709958, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7423, "loss": 3.7768497467041016, "lr": 0.0001946909700557327, "elapsed_sec": 61728.32902312279, "step_time_sec": 8.24002078990452, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7424, "loss": 3.838301658630371, "lr": 0.0001946660112021382, "elapsed_sec": 61736.566692352295, "step_time_sec": 8.23756116302684, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7425, "loss": 3.9634037017822266, "lr": 0.00019464099546748022, "elapsed_sec": 61744.80685329437, "step_time_sec": 8.239962943014689, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7426, "loss": 3.929621934890747, "lr": 0.00019461592286719754, "elapsed_sec": 61753.046006679535, "step_time_sec": 8.238963715033606, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7427, "loss": 4.063206195831299, "lr": 0.00019459079341676392, "elapsed_sec": 61761.28444337845, "step_time_sec": 8.238256001961417, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7428, "loss": 3.6977217197418213, "lr": 0.0001945656071316882, "elapsed_sec": 61769.523809194565, "step_time_sec": 8.239207804901525, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7429, "loss": 3.9070560932159424, "lr": 0.00019454036402751429, "elapsed_sec": 61777.7623064518, "step_time_sec": 8.23835079697892, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7430, "loss": 4.033854007720947, "lr": 0.00019451506411982122, "elapsed_sec": 61786.00115299225, "step_time_sec": 8.238631110987626, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7431, "loss": 4.119658946990967, "lr": 0.0001944897074242231, "elapsed_sec": 61794.2406270504, "step_time_sec": 8.239368235925213, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7432, "loss": 3.965151786804199, "lr": 0.00019446429395636894, "elapsed_sec": 61802.47779273987, "step_time_sec": 8.23694847500883, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7433, "loss": 3.9704785346984863, "lr": 0.00019443882373194296, "elapsed_sec": 61810.71524691582, "step_time_sec": 8.237330638919957, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7434, "loss": 3.860110282897949, "lr": 0.0001944132967666643, "elapsed_sec": 61818.954419612885, "step_time_sec": 8.239043212961406, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7435, "loss": 3.9247312545776367, "lr": 0.00019438771307628717, "elapsed_sec": 61827.19412302971, "step_time_sec": 8.239495946094394, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7436, "loss": 3.84090256690979, "lr": 0.00019436207267660075, "elapsed_sec": 61835.4332344532, "step_time_sec": 8.238915762049146, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7437, "loss": 3.7465288639068604, "lr": 0.00019433637558342924, "elapsed_sec": 61843.67146396637, "step_time_sec": 8.238092888030224, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7438, "loss": 3.851323366165161, "lr": 0.00019431062181263185, "elapsed_sec": 61851.91119122505, "step_time_sec": 8.239535039989278, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7439, "loss": 3.8469278812408447, "lr": 0.00019428481138010273, "elapsed_sec": 61860.14941287041, "step_time_sec": 8.238038880983368, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7440, "loss": 3.8557543754577637, "lr": 0.000194258944301771, "elapsed_sec": 61868.3887655735, "step_time_sec": 8.239202230004594, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7441, "loss": 4.022245407104492, "lr": 0.00019423302059360075, "elapsed_sec": 61876.62844824791, "step_time_sec": 8.239575284067541, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7442, "loss": 3.9394149780273438, "lr": 0.00019420704027159105, "elapsed_sec": 61884.86686229706, "step_time_sec": 8.238178619998507, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7443, "loss": 3.7579901218414307, "lr": 0.00019418100335177592, "elapsed_sec": 61893.105343580246, "step_time_sec": 8.23827795300167, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7444, "loss": 3.8259809017181396, "lr": 0.0001941549098502242, "elapsed_sec": 61901.34319925308, "step_time_sec": 8.237677137949504, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7445, "loss": 3.92838978767395, "lr": 0.00019412875978303974, "elapsed_sec": 61909.58165001869, "step_time_sec": 8.238328654086217, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7446, "loss": 3.9045796394348145, "lr": 0.00019410255316636133, "elapsed_sec": 61917.8216650486, "step_time_sec": 8.239825653959997, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7447, "loss": 3.8780288696289062, "lr": 0.00019407629001636257, "elapsed_sec": 61926.05839085579, "step_time_sec": 8.236529979971237, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7448, "loss": 3.7380728721618652, "lr": 0.000194049970349252, "elapsed_sec": 61934.29883861542, "step_time_sec": 8.240322593948804, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7449, "loss": 3.8363921642303467, "lr": 0.00019402359418127307, "elapsed_sec": 61942.53752350807, "step_time_sec": 8.238519343896769, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7450, "loss": 3.9489753246307373, "lr": 0.00019399716152870397, "elapsed_sec": 61950.77825427055, "step_time_sec": 8.240547864930704, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7451, "loss": 3.7154018878936768, "lr": 0.00019397067240785795, "elapsed_sec": 61959.01873707771, "step_time_sec": 8.240256129996851, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7452, "loss": 3.765594959259033, "lr": 0.00019394412683508292, "elapsed_sec": 61967.257719278336, "step_time_sec": 8.23887530399952, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7453, "loss": 3.829072952270508, "lr": 0.0001939175248267618, "elapsed_sec": 61975.49701142311, "step_time_sec": 8.23906573606655, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7454, "loss": 3.8710482120513916, "lr": 0.00019389086639931219, "elapsed_sec": 61983.73689508438, "step_time_sec": 8.239731810055673, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7455, "loss": 4.00183629989624, "lr": 0.00019386415156918662, "elapsed_sec": 61991.97544026375, "step_time_sec": 8.238396720029414, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7456, "loss": 4.236107349395752, "lr": 0.0001938373803528723, "elapsed_sec": 62000.21585035324, "step_time_sec": 8.24027017201297, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7457, "loss": 3.830724000930786, "lr": 0.00019381055276689139, "elapsed_sec": 62008.455493927, "step_time_sec": 8.239418370067142, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7458, "loss": 3.9146883487701416, "lr": 0.00019378366882780076, "elapsed_sec": 62016.69467830658, "step_time_sec": 8.239084862056188, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7459, "loss": 3.8417696952819824, "lr": 0.00019375672855219208, "elapsed_sec": 62024.93341612816, "step_time_sec": 8.2385466969572, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7460, "loss": 3.9326250553131104, "lr": 0.00019372973195669172, "elapsed_sec": 62033.1707713604, "step_time_sec": 8.237145926919766, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7461, "loss": 3.9330856800079346, "lr": 0.00019370267905796093, "elapsed_sec": 62041.4050681591, "step_time_sec": 8.234114723978564, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7462, "loss": 3.965386390686035, "lr": 0.00019367556987269561, "elapsed_sec": 62049.64268040657, "step_time_sec": 8.23743983602617, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7463, "loss": 3.8016178607940674, "lr": 0.00019364840441762644, "elapsed_sec": 62057.8827214241, "step_time_sec": 8.239899497013539, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7464, "loss": 3.7990806102752686, "lr": 0.00019362118270951884, "elapsed_sec": 62066.12196111679, "step_time_sec": 8.23905424005352, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7465, "loss": 3.82407283782959, "lr": 0.0001935939047651729, "elapsed_sec": 62074.358578681946, "step_time_sec": 8.236498673912138, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7466, "loss": 3.8841662406921387, "lr": 0.00019356657060142345, "elapsed_sec": 62082.67212796211, "step_time_sec": 8.248161321971565, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7467, "loss": 3.836625099182129, "lr": 0.00019353918023513997, "elapsed_sec": 62090.90446615219, "step_time_sec": 8.232177799101919, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7468, "loss": 3.796442985534668, "lr": 0.00019351173368322677, "elapsed_sec": 62099.13595890999, "step_time_sec": 8.231324799009599, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7469, "loss": 3.6902503967285156, "lr": 0.00019348423096262262, "elapsed_sec": 62107.37442612648, "step_time_sec": 8.238261281978339, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7470, "loss": 3.7594571113586426, "lr": 0.00019345667209030116, "elapsed_sec": 62115.61231446266, "step_time_sec": 8.237792066996917, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7471, "loss": 3.9010000228881836, "lr": 0.0001934290570832705, "elapsed_sec": 62123.85106897354, "step_time_sec": 8.238596700015478, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7472, "loss": 4.043856620788574, "lr": 0.00019340138595857353, "elapsed_sec": 62132.090465545654, "step_time_sec": 8.23918643698562, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7473, "loss": 4.1107401847839355, "lr": 0.00019337365873328772, "elapsed_sec": 62140.328424692154, "step_time_sec": 8.237756254035048, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7474, "loss": 3.8471031188964844, "lr": 0.00019334587542452517, "elapsed_sec": 62148.567462444305, "step_time_sec": 8.238897028961219, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7475, "loss": 3.8563191890716553, "lr": 0.0001933180360494326, "elapsed_sec": 62156.80505871773, "step_time_sec": 8.237428510095924, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7476, "loss": 3.9466030597686768, "lr": 0.00019329014062519136, "elapsed_sec": 62165.04418206215, "step_time_sec": 8.238904775003903, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7477, "loss": 3.993959903717041, "lr": 0.0001932621891690173, "elapsed_sec": 62173.28211927414, "step_time_sec": 8.237749784020707, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7478, "loss": 3.776327133178711, "lr": 0.00019323418169816088, "elapsed_sec": 62181.52021408081, "step_time_sec": 8.237881485954858, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7479, "loss": 3.707653284072876, "lr": 0.00019320611822990725, "elapsed_sec": 62189.75953125954, "step_time_sec": 8.239204328972846, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7480, "loss": 3.8659656047821045, "lr": 0.00019317799878157594, "elapsed_sec": 62197.99811172485, "step_time_sec": 8.238365032011643, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7481, "loss": 3.8804397583007812, "lr": 0.0001931498233705212, "elapsed_sec": 62206.23778915405, "step_time_sec": 8.239479981013574, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7482, "loss": 3.906066656112671, "lr": 0.00019312159201413163, "elapsed_sec": 62214.4775865078, "step_time_sec": 8.239664177992381, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7483, "loss": 3.8740732669830322, "lr": 0.00019309330472983056, "elapsed_sec": 62222.71609830856, "step_time_sec": 8.23830218007788, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7484, "loss": 3.9664998054504395, "lr": 0.00019306496153507566, "elapsed_sec": 62230.956843853, "step_time_sec": 8.240591033012606, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7485, "loss": 3.9324302673339844, "lr": 0.00019303656244735922, "elapsed_sec": 62239.196590662, "step_time_sec": 8.23958562396001, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7486, "loss": 3.9247312545776367, "lr": 0.000193008107484208, "elapsed_sec": 62247.436737537384, "step_time_sec": 8.239995167008601, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7487, "loss": 3.9212377071380615, "lr": 0.00019297959666318316, "elapsed_sec": 62255.6737742424, "step_time_sec": 8.236883130040951, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7488, "loss": 3.9123008251190186, "lr": 0.00019295103000188047, "elapsed_sec": 62263.91078329086, "step_time_sec": 8.236816314049065, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7489, "loss": 3.9372832775115967, "lr": 0.0001929224075179301, "elapsed_sec": 62272.148376226425, "step_time_sec": 8.23739372601267, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7490, "loss": 3.658118486404419, "lr": 0.00019289372922899663, "elapsed_sec": 62280.38689947128, "step_time_sec": 8.238354237983003, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7491, "loss": 3.860814094543457, "lr": 0.00019286499515277914, "elapsed_sec": 62288.62669014931, "step_time_sec": 8.239686423912644, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7492, "loss": 3.939877986907959, "lr": 0.00019283620530701108, "elapsed_sec": 62296.866859674454, "step_time_sec": 8.240015401039273, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7493, "loss": 3.8778436183929443, "lr": 0.00019280735970946043, "elapsed_sec": 62305.10582756996, "step_time_sec": 8.238779219915159, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7494, "loss": 3.8288817405700684, "lr": 0.00019277845837792945, "elapsed_sec": 62313.346698760986, "step_time_sec": 8.240714638028294, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7495, "loss": 3.74941086769104, "lr": 0.00019274950133025485, "elapsed_sec": 62321.58476233482, "step_time_sec": 8.237878090003505, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7496, "loss": 3.8654940128326416, "lr": 0.00019272048858430773, "elapsed_sec": 62329.82304739952, "step_time_sec": 8.23815458000172, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7497, "loss": 3.878284215927124, "lr": 0.0001926914201579936, "elapsed_sec": 62338.061228990555, "step_time_sec": 8.238007117062807, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7498, "loss": 3.8938522338867188, "lr": 0.00019266229606925222, "elapsed_sec": 62346.298874378204, "step_time_sec": 8.237451019929722, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7499, "loss": 3.898564577102661, "lr": 0.00019263311633605784, "elapsed_sec": 62354.53846406937, "step_time_sec": 8.239475541049615, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7500, "loss": 3.965228796005249, "lr": 0.00019260388097641897, "elapsed_sec": 62362.7766623497, "step_time_sec": 35.82628510403447, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": true, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7501, "loss": 3.8115923404693604, "lr": 0.0001925745900083785, "elapsed_sec": 62398.6120057106, "step_time_sec": 8.246544204070233, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7502, "loss": 3.7784199714660645, "lr": 0.00019254524345001353, "elapsed_sec": 62406.84237623215, "step_time_sec": 8.230273027089424, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7503, "loss": 3.806361198425293, "lr": 0.0001925158413194356, "elapsed_sec": 62415.075791597366, "step_time_sec": 8.233225036063232, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7504, "loss": 3.854295253753662, "lr": 0.0001924863836347905, "elapsed_sec": 62423.30780506134, "step_time_sec": 8.231822905945592, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7505, "loss": 3.9562172889709473, "lr": 0.00019245687041425828, "elapsed_sec": 62431.54114627838, "step_time_sec": 8.233173205051571, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7506, "loss": 3.897641181945801, "lr": 0.0001924273016760533, "elapsed_sec": 62439.773970365524, "step_time_sec": 8.232676709070802, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7507, "loss": 3.8785018920898438, "lr": 0.00019239767743842414, "elapsed_sec": 62448.00557088852, "step_time_sec": 8.231439238996245, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7508, "loss": 3.7657463550567627, "lr": 0.00019236799771965368, "elapsed_sec": 62456.23819255829, "step_time_sec": 8.232444397057407, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7509, "loss": 3.844693183898926, "lr": 0.000192338262538059, "elapsed_sec": 62464.474952697754, "step_time_sec": 8.236587076913565, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7510, "loss": 3.895798444747925, "lr": 0.00019230847191199145, "elapsed_sec": 62472.71467542648, "step_time_sec": 8.239540062029846, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7511, "loss": 3.725402355194092, "lr": 0.00019227862585983658, "elapsed_sec": 62480.95547914505, "step_time_sec": 8.24062185396906, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7512, "loss": 3.9780726432800293, "lr": 0.0001922487244000141, "elapsed_sec": 62489.1940202713, "step_time_sec": 8.23840302194003, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7513, "loss": 3.8313660621643066, "lr": 0.00019221876755097804, "elapsed_sec": 62497.4334859848, "step_time_sec": 8.239290546975099, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7514, "loss": 3.9433095455169678, "lr": 0.00019218875533121642, "elapsed_sec": 62505.67270898819, "step_time_sec": 8.23903579695616, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7515, "loss": 3.9575917720794678, "lr": 0.00019215868775925165, "elapsed_sec": 62513.91101074219, "step_time_sec": 8.23812713404186, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7516, "loss": 3.9929754734039307, "lr": 0.0001921285648536401, "elapsed_sec": 62522.14960002899, "step_time_sec": 8.238426418043673, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7517, "loss": 3.8256592750549316, "lr": 0.0001920983866329725, "elapsed_sec": 62530.38618373871, "step_time_sec": 8.236414251034148, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7518, "loss": 3.7887842655181885, "lr": 0.00019206815311587351, "elapsed_sec": 62538.6261575222, "step_time_sec": 8.239845941076055, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7519, "loss": 3.9439868927001953, "lr": 0.00019203786432100205, "elapsed_sec": 62546.865040540695, "step_time_sec": 8.238709985045716, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7520, "loss": 4.001900672912598, "lr": 0.00019200752026705113, "elapsed_sec": 62555.1032102108, "step_time_sec": 8.237950368085876, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7521, "loss": 3.823472499847412, "lr": 0.0001919771209727478, "elapsed_sec": 62563.34349703789, "step_time_sec": 8.240150186000392, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7522, "loss": 3.8013241291046143, "lr": 0.0001919466664568533, "elapsed_sec": 62571.58212327957, "step_time_sec": 8.238421693095006, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7523, "loss": 3.7753946781158447, "lr": 0.00019191615673816288, "elapsed_sec": 62579.82252264023, "step_time_sec": 8.240264424937777, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7524, "loss": 3.8003408908843994, "lr": 0.00019188559183550585, "elapsed_sec": 62588.06114959717, "step_time_sec": 8.238454342936166, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7525, "loss": 3.8492867946624756, "lr": 0.00019185497176774568, "elapsed_sec": 62596.30190730095, "step_time_sec": 8.240603611106053, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7526, "loss": 3.7600085735321045, "lr": 0.00019182429655377974, "elapsed_sec": 62604.542867422104, "step_time_sec": 8.2407926969463, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7527, "loss": 4.029182434082031, "lr": 0.00019179356621253958, "elapsed_sec": 62612.782094955444, "step_time_sec": 8.239078290062025, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7528, "loss": 4.050063133239746, "lr": 0.00019176278076299063, "elapsed_sec": 62621.021980285645, "step_time_sec": 8.23967132391408, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7529, "loss": 3.872410774230957, "lr": 0.00019173194022413244, "elapsed_sec": 62629.26133489609, "step_time_sec": 8.239219006965868, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7530, "loss": 3.9849965572357178, "lr": 0.00019170104461499853, "elapsed_sec": 62637.49990987778, "step_time_sec": 8.238398736109957, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7531, "loss": 3.7907145023345947, "lr": 0.00019167009395465638, "elapsed_sec": 62645.74061584473, "step_time_sec": 8.240548797999509, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7532, "loss": 4.003468036651611, "lr": 0.00019163908826220745, "elapsed_sec": 62653.97903871536, "step_time_sec": 8.238272445974872, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7533, "loss": 3.723675489425659, "lr": 0.00019160802755678716, "elapsed_sec": 62662.21741271019, "step_time_sec": 8.238208213937469, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7534, "loss": 3.8277573585510254, "lr": 0.00019157691185756498, "elapsed_sec": 62670.455517053604, "step_time_sec": 8.237907179049216, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7535, "loss": 3.8510565757751465, "lr": 0.00019154574118374415, "elapsed_sec": 62678.69300723076, "step_time_sec": 8.237322882981971, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7536, "loss": 3.7630202770233154, "lr": 0.00019151451555456196, "elapsed_sec": 62686.93171310425, "step_time_sec": 8.238586657913402, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7537, "loss": 3.7246038913726807, "lr": 0.00019148323498928958, "elapsed_sec": 62695.16995406151, "step_time_sec": 8.238022711011581, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7538, "loss": 3.863685131072998, "lr": 0.00019145189950723208, "elapsed_sec": 62703.407291173935, "step_time_sec": 8.237171329092234, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7539, "loss": 3.8736462593078613, "lr": 0.00019142050912772843, "elapsed_sec": 62711.64666390419, "step_time_sec": 8.239213071996346, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7540, "loss": 3.8526437282562256, "lr": 0.0001913890638701515, "elapsed_sec": 62719.8838019371, "step_time_sec": 8.23697028902825, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7541, "loss": 3.8082191944122314, "lr": 0.00019135756375390794, "elapsed_sec": 62728.122646808624, "step_time_sec": 8.238699145964347, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7542, "loss": 3.8749420642852783, "lr": 0.0001913260087984384, "elapsed_sec": 62736.35892057419, "step_time_sec": 8.236131864017807, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7543, "loss": 4.082320213317871, "lr": 0.00019129439902321725, "elapsed_sec": 62744.598853349686, "step_time_sec": 8.239730191999115, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7544, "loss": 3.9619216918945312, "lr": 0.0001912627344477528, "elapsed_sec": 62752.83832883835, "step_time_sec": 8.239355406025425, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7545, "loss": 3.743030309677124, "lr": 0.00019123101509158706, "elapsed_sec": 62761.07700204849, "step_time_sec": 8.238494627992623, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7546, "loss": 3.598586320877075, "lr": 0.0001911992409742959, "elapsed_sec": 62769.31635427475, "step_time_sec": 8.23915334593039, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7547, "loss": 3.834761142730713, "lr": 0.00019116741211548907, "elapsed_sec": 62777.55560207367, "step_time_sec": 8.239114516065456, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7548, "loss": 3.903672456741333, "lr": 0.00019113552853480996, "elapsed_sec": 62785.79701924324, "step_time_sec": 8.241232500062324, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7549, "loss": 3.924748659133911, "lr": 0.0001911035902519358, "elapsed_sec": 62794.034709692, "step_time_sec": 8.237535458989441, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7550, "loss": 3.860429286956787, "lr": 0.00019107159728657766, "elapsed_sec": 62802.273617744446, "step_time_sec": 8.23874544899445, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7551, "loss": 3.956538677215576, "lr": 0.00019103954965848016, "elapsed_sec": 62810.512325525284, "step_time_sec": 8.238565196981654, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7552, "loss": 3.8387625217437744, "lr": 0.00019100744738742188, "elapsed_sec": 62818.750705718994, "step_time_sec": 8.23823039594572, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7553, "loss": 3.7967207431793213, "lr": 0.000190975290493215, "elapsed_sec": 62826.98803973198, "step_time_sec": 8.237205290002748, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7554, "loss": 3.8206114768981934, "lr": 0.0001909430789957054, "elapsed_sec": 62835.22776675224, "step_time_sec": 8.239551631966606, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7555, "loss": 3.738189220428467, "lr": 0.00019091081291477271, "elapsed_sec": 62843.46451830864, "step_time_sec": 8.236566098057665, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7556, "loss": 3.8232762813568115, "lr": 0.00019087849227033017, "elapsed_sec": 62851.70355796814, "step_time_sec": 8.238896289956756, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7557, "loss": 3.837310791015625, "lr": 0.00019084611708232484, "elapsed_sec": 62859.94129872322, "step_time_sec": 8.23757862392813, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7558, "loss": 3.769810199737549, "lr": 0.00019081368737073728, "elapsed_sec": 62868.18026304245, "step_time_sec": 8.238770478987135, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7559, "loss": 3.8988075256347656, "lr": 0.00019078120315558182, "elapsed_sec": 62876.41903471947, "step_time_sec": 8.238600759068504, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7560, "loss": 3.630375385284424, "lr": 0.00019074866445690632, "elapsed_sec": 62884.657027721405, "step_time_sec": 8.237829103018157, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7561, "loss": 3.7421576976776123, "lr": 0.00019071607129479237, "elapsed_sec": 62892.89736247063, "step_time_sec": 8.24016277701594, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7562, "loss": 3.9084346294403076, "lr": 0.00019068342368935514, "elapsed_sec": 62901.13800215721, "step_time_sec": 8.24046575895045, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7563, "loss": 3.921710968017578, "lr": 0.00019065072166074334, "elapsed_sec": 62909.37761974335, "step_time_sec": 8.23947714397218, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7564, "loss": 3.6991686820983887, "lr": 0.00019061796522913934, "elapsed_sec": 62917.61644124985, "step_time_sec": 8.23869168094825, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7565, "loss": 3.9254863262176514, "lr": 0.0001905851544147591, "elapsed_sec": 62925.853519678116, "step_time_sec": 8.236874397029169, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7566, "loss": 3.8985402584075928, "lr": 0.000190552289237852, "elapsed_sec": 62934.09262442589, "step_time_sec": 8.238934073015116, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7567, "loss": 3.90183162689209, "lr": 0.00019051936971870115, "elapsed_sec": 62942.3322622776, "step_time_sec": 8.23946240695659, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7568, "loss": 3.8568265438079834, "lr": 0.00019048639587762313, "elapsed_sec": 62950.5712761879, "step_time_sec": 8.238867126055993, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7569, "loss": 3.8417418003082275, "lr": 0.00019045336773496795, "elapsed_sec": 62958.809586286545, "step_time_sec": 8.238177569000982, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7570, "loss": 3.951331615447998, "lr": 0.00019042028531111933, "elapsed_sec": 62967.047154426575, "step_time_sec": 8.237389573943801, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7571, "loss": 3.836060047149658, "lr": 0.00019038714862649428, "elapsed_sec": 62975.28558135033, "step_time_sec": 8.23825278901495, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7572, "loss": 3.977962017059326, "lr": 0.0001903539577015435, "elapsed_sec": 62983.5239136219, "step_time_sec": 8.238135737017728, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7573, "loss": 3.8980231285095215, "lr": 0.00019032071255675094, "elapsed_sec": 62991.76269888878, "step_time_sec": 8.238641318981536, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7574, "loss": 3.931554079055786, "lr": 0.0001902874132126342, "elapsed_sec": 63000.001551389694, "step_time_sec": 8.238707362907007, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7575, "loss": 3.912302017211914, "lr": 0.00019025405968974428, "elapsed_sec": 63008.240280389786, "step_time_sec": 8.238585268030874, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7576, "loss": 3.8698737621307373, "lr": 0.00019022065200866557, "elapsed_sec": 63016.481367349625, "step_time_sec": 8.240898768999614, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7577, "loss": 3.8299407958984375, "lr": 0.00019018719019001592, "elapsed_sec": 63024.720959186554, "step_time_sec": 8.2394188339822, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7578, "loss": 3.8074607849121094, "lr": 0.00019015367425444658, "elapsed_sec": 63032.960310935974, "step_time_sec": 8.239178440067917, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7579, "loss": 3.8936493396759033, "lr": 0.0001901201042226422, "elapsed_sec": 63041.19967365265, "step_time_sec": 8.239195000031032, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7580, "loss": 3.8838937282562256, "lr": 0.00019008648011532088, "elapsed_sec": 63049.43844604492, "step_time_sec": 8.238628357998095, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7581, "loss": 3.8108904361724854, "lr": 0.00019005280195323395, "elapsed_sec": 63057.67806529999, "step_time_sec": 8.239469015039504, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7582, "loss": 4.04097318649292, "lr": 0.00019001906975716623, "elapsed_sec": 63065.91705775261, "step_time_sec": 8.238794909906574, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7583, "loss": 3.9090282917022705, "lr": 0.00018998528354793577, "elapsed_sec": 63074.155757665634, "step_time_sec": 8.238533569965512, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7584, "loss": 3.962059259414673, "lr": 0.00018995144334639413, "elapsed_sec": 63082.39554095268, "step_time_sec": 8.239610175951384, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7585, "loss": 3.9152324199676514, "lr": 0.00018991754917342603, "elapsed_sec": 63090.63432049751, "step_time_sec": 8.238655716995709, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7586, "loss": 3.740884304046631, "lr": 0.0001898836010499496, "elapsed_sec": 63098.87386274338, "step_time_sec": 8.239381272927858, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7587, "loss": 3.989720344543457, "lr": 0.00018984959899691613, "elapsed_sec": 63107.11369252205, "step_time_sec": 8.239711915026419, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7588, "loss": 3.8730874061584473, "lr": 0.0001898155430353104, "elapsed_sec": 63115.35060739517, "step_time_sec": 8.236730135977268, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7589, "loss": 3.7556793689727783, "lr": 0.00018978143318615026, "elapsed_sec": 63123.58975458145, "step_time_sec": 8.238936624024063, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7590, "loss": 3.9133048057556152, "lr": 0.00018974726947048695, "elapsed_sec": 63131.82882761955, "step_time_sec": 8.238960655988194, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7591, "loss": 3.9022116661071777, "lr": 0.0001897130519094049, "elapsed_sec": 63140.06912636757, "step_time_sec": 8.24012786091771, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7592, "loss": 3.9904239177703857, "lr": 0.00018967878052402175, "elapsed_sec": 63148.30896401405, "step_time_sec": 8.23967096710112, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7593, "loss": 3.926175594329834, "lr": 0.00018964445533548843, "elapsed_sec": 63156.54657912254, "step_time_sec": 8.23748395498842, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7594, "loss": 3.7817773818969727, "lr": 0.00018961007636498898, "elapsed_sec": 63164.786053180695, "step_time_sec": 8.239240953000262, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7595, "loss": 3.7475719451904297, "lr": 0.00018957564363374075, "elapsed_sec": 63173.02365851402, "step_time_sec": 8.237445133039728, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7596, "loss": 3.826120376586914, "lr": 0.00018954115716299415, "elapsed_sec": 63181.26299595833, "step_time_sec": 8.239157794974744, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7597, "loss": 3.924988031387329, "lr": 0.0001895066169740328, "elapsed_sec": 63189.50084018707, "step_time_sec": 8.237699544988573, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7598, "loss": 3.848879098892212, "lr": 0.0001894720230881735, "elapsed_sec": 63197.73907876015, "step_time_sec": 8.238084729062393, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7599, "loss": 3.928081750869751, "lr": 0.00018943737552676621, "elapsed_sec": 63205.97875881195, "step_time_sec": 8.239508841070347, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7600, "loss": 3.966672658920288, "lr": 0.00018940267431119393, "elapsed_sec": 63214.21588778496, "step_time_sec": 8.23693698598072, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7601, "loss": 3.8548803329467773, "lr": 0.00018936791946287283, "elapsed_sec": 63222.457292318344, "step_time_sec": 8.241251140949316, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7602, "loss": 3.962650775909424, "lr": 0.00018933311100325216, "elapsed_sec": 63230.69667887688, "step_time_sec": 8.239225536002778, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7603, "loss": 3.952545642852783, "lr": 0.00018929824895381429, "elapsed_sec": 63238.93513584137, "step_time_sec": 8.23828923504334, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7604, "loss": 3.8358755111694336, "lr": 0.00018926333333607462, "elapsed_sec": 63247.17432665825, "step_time_sec": 8.239060592954047, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7605, "loss": 3.9669992923736572, "lr": 0.00018922836417158164, "elapsed_sec": 63255.414630651474, "step_time_sec": 8.24013452802319, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7606, "loss": 3.9146080017089844, "lr": 0.00018919334148191687, "elapsed_sec": 63263.65410256386, "step_time_sec": 8.239251303020865, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7607, "loss": 3.897716999053955, "lr": 0.0001891582652886949, "elapsed_sec": 63271.893981695175, "step_time_sec": 8.239754972979426, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7608, "loss": 3.7783310413360596, "lr": 0.0001891231356135633, "elapsed_sec": 63280.132271289825, "step_time_sec": 8.238146687974222, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7609, "loss": 3.970529794692993, "lr": 0.0001890879524782026, "elapsed_sec": 63288.373324394226, "step_time_sec": 8.240846932982095, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7610, "loss": 3.893612861633301, "lr": 0.00018905271590432643, "elapsed_sec": 63296.61311650276, "step_time_sec": 8.239650840056129, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7611, "loss": 3.8415443897247314, "lr": 0.00018901742591368137, "elapsed_sec": 63304.85165977478, "step_time_sec": 8.238339618896134, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7612, "loss": 3.974698781967163, "lr": 0.0001889820825280469, "elapsed_sec": 63313.0926322937, "step_time_sec": 8.24083910905756, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7613, "loss": 3.9444494247436523, "lr": 0.00018894668576923554, "elapsed_sec": 63321.33271384239, "step_time_sec": 8.239887451985851, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7614, "loss": 3.886890172958374, "lr": 0.00018891123565909268, "elapsed_sec": 63329.5718562603, "step_time_sec": 8.239006518968381, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7615, "loss": 3.9314053058624268, "lr": 0.0001888757322194967, "elapsed_sec": 63337.81146025658, "step_time_sec": 8.23942985094618, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7616, "loss": 3.985753297805786, "lr": 0.00018884017547235883, "elapsed_sec": 63346.04813456535, "step_time_sec": 8.236518076970242, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7617, "loss": 3.9333457946777344, "lr": 0.00018880456543962325, "elapsed_sec": 63354.2881295681, "step_time_sec": 8.239838397013955, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7618, "loss": 3.927985191345215, "lr": 0.000188768902143267, "elapsed_sec": 63362.52813386917, "step_time_sec": 8.239808924030513, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7619, "loss": 3.9492225646972656, "lr": 0.0001887331856053, "elapsed_sec": 63370.7662396431, "step_time_sec": 8.237968407920562, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7620, "loss": 4.029506206512451, "lr": 0.00018869741584776498, "elapsed_sec": 63379.006299734116, "step_time_sec": 8.239900766988285, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7621, "loss": 3.886504888534546, "lr": 0.00018866159289273763, "elapsed_sec": 63387.2439095974, "step_time_sec": 8.237481177900918, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7622, "loss": 3.8339149951934814, "lr": 0.00018862571676232635, "elapsed_sec": 63395.4832675457, "step_time_sec": 8.239180548931472, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7623, "loss": 3.888695001602173, "lr": 0.00018858978747867246, "elapsed_sec": 63403.722798109055, "step_time_sec": 8.239366371999495, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7624, "loss": 3.886491537094116, "lr": 0.00018855380506394996, "elapsed_sec": 63411.96119594574, "step_time_sec": 8.238220146973617, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7625, "loss": 3.843332052230835, "lr": 0.00018851776954036578, "elapsed_sec": 63420.20143246651, "step_time_sec": 8.24007954809349, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7626, "loss": 3.7481753826141357, "lr": 0.0001884816809301595, "elapsed_sec": 63428.44073367119, "step_time_sec": 8.239119023899548, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7627, "loss": 3.887763500213623, "lr": 0.0001884455392556036, "elapsed_sec": 63436.67947912216, "step_time_sec": 8.23862595204264, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7628, "loss": 3.8224620819091797, "lr": 0.00018840934453900318, "elapsed_sec": 63444.91993713379, "step_time_sec": 8.24025126802735, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7629, "loss": 3.8324058055877686, "lr": 0.00018837309680269613, "elapsed_sec": 63453.15866422653, "step_time_sec": 8.238561430945992, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7630, "loss": 3.8095884323120117, "lr": 0.00018833679606905304, "elapsed_sec": 63461.397906541824, "step_time_sec": 8.239071343909018, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7631, "loss": 3.8475968837738037, "lr": 0.0001883004423604773, "elapsed_sec": 63469.63716721535, "step_time_sec": 8.23909156897571, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7632, "loss": 3.97385573387146, "lr": 0.00018826403569940483, "elapsed_sec": 63477.876202344894, "step_time_sec": 8.238890034961514, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7633, "loss": 3.88061261177063, "lr": 0.00018822757610830438, "elapsed_sec": 63486.11569881439, "step_time_sec": 8.239315940998495, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7634, "loss": 3.9187159538269043, "lr": 0.00018819106360967727, "elapsed_sec": 63494.353996276855, "step_time_sec": 8.238145941053517, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7635, "loss": 3.7942240238189697, "lr": 0.00018815449822605755, "elapsed_sec": 63502.592398405075, "step_time_sec": 8.238216512952931, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7636, "loss": 3.8638055324554443, "lr": 0.00018811787998001186, "elapsed_sec": 63510.83164405823, "step_time_sec": 8.23917329206597, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7637, "loss": 3.8448097705841064, "lr": 0.00018808120889413947, "elapsed_sec": 63519.070264577866, "step_time_sec": 8.238403424969874, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7638, "loss": 3.819643497467041, "lr": 0.0001880444849910722, "elapsed_sec": 63527.30986571312, "step_time_sec": 8.239395680022426, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7639, "loss": 3.80765962600708, "lr": 0.00018800770829347462, "elapsed_sec": 63535.546927928925, "step_time_sec": 8.23695810395293, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7640, "loss": 3.90531587600708, "lr": 0.00018797087882404373, "elapsed_sec": 63543.78764081001, "step_time_sec": 8.240506384056062, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7641, "loss": 4.135818004608154, "lr": 0.0001879339966055092, "elapsed_sec": 63552.02797079086, "step_time_sec": 8.240182457957417, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7642, "loss": 3.840778350830078, "lr": 0.0001878970616606332, "elapsed_sec": 63560.26692223549, "step_time_sec": 8.238761884975247, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7643, "loss": 3.9577815532684326, "lr": 0.00018786007401221043, "elapsed_sec": 63568.50651407242, "step_time_sec": 8.239451773930341, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7644, "loss": 3.728720188140869, "lr": 0.00018782303368306818, "elapsed_sec": 63576.744500637054, "step_time_sec": 8.237860764958896, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7645, "loss": 3.792410373687744, "lr": 0.00018778594069606623, "elapsed_sec": 63584.983713150024, "step_time_sec": 8.239006522926502, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7646, "loss": 3.9119443893432617, "lr": 0.00018774879507409676, "elapsed_sec": 63593.22377324104, "step_time_sec": 8.239872385980561, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7647, "loss": 3.8151304721832275, "lr": 0.00018771159684008462, "elapsed_sec": 63601.46181321144, "step_time_sec": 8.237860964029096, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7648, "loss": 3.739166259765625, "lr": 0.00018767434601698693, "elapsed_sec": 63609.701374292374, "step_time_sec": 8.239392149960622, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7649, "loss": 3.895867109298706, "lr": 0.00018763704262779346, "elapsed_sec": 63617.9419298172, "step_time_sec": 8.240414172061719, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7650, "loss": 3.901589870452881, "lr": 0.00018759968669552622, "elapsed_sec": 63626.18169283867, "step_time_sec": 8.23962661193218, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7651, "loss": 3.7983744144439697, "lr": 0.00018756227824323987, "elapsed_sec": 63634.420573949814, "step_time_sec": 8.238684323965572, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7652, "loss": 3.688936471939087, "lr": 0.00018752481729402125, "elapsed_sec": 63642.65850067139, "step_time_sec": 8.237765196943656, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7653, "loss": 3.9239518642425537, "lr": 0.00018748730387098977, "elapsed_sec": 63650.8990046978, "step_time_sec": 8.24034652498085, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7654, "loss": 3.743298292160034, "lr": 0.0001874497379972972, "elapsed_sec": 63659.13812112808, "step_time_sec": 8.238976516993716, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7655, "loss": 3.7642228603363037, "lr": 0.00018741211969612758, "elapsed_sec": 63667.37724328041, "step_time_sec": 8.23890240304172, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7656, "loss": 3.9679698944091797, "lr": 0.00018737444899069743, "elapsed_sec": 63675.616490364075, "step_time_sec": 8.239141857018694, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7657, "loss": 3.90944504737854, "lr": 0.00018733672590425548, "elapsed_sec": 63683.85485482216, "step_time_sec": 8.23819911503233, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7658, "loss": 3.812129497528076, "lr": 0.000187298950460083, "elapsed_sec": 63692.09519457817, "step_time_sec": 8.240141715039499, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7659, "loss": 3.8706140518188477, "lr": 0.0001872611226814933, "elapsed_sec": 63700.334745168686, "step_time_sec": 8.23939181806054, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7660, "loss": 3.9208273887634277, "lr": 0.0001872232425918322, "elapsed_sec": 63708.57174873352, "step_time_sec": 8.23685777396895, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7661, "loss": 3.8097269535064697, "lr": 0.00018718531021447773, "elapsed_sec": 63716.810764312744, "step_time_sec": 8.238871181034483, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7662, "loss": 3.858442544937134, "lr": 0.00018714732557284015, "elapsed_sec": 63725.04763388634, "step_time_sec": 8.236648836056702, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7663, "loss": 3.8649704456329346, "lr": 0.000187109288690362, "elapsed_sec": 63733.286444664, "step_time_sec": 8.238638321054168, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7664, "loss": 3.875459909439087, "lr": 0.00018707119959051818, "elapsed_sec": 63741.52531862259, "step_time_sec": 8.238695755950175, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7665, "loss": 4.001216411590576, "lr": 0.0001870330582968156, "elapsed_sec": 63749.7639605999, "step_time_sec": 8.238516349927522, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7666, "loss": 3.953852653503418, "lr": 0.00018699486483279357, "elapsed_sec": 63758.00397109985, "step_time_sec": 8.239813952008262, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7667, "loss": 3.8250982761383057, "lr": 0.00018695661922202346, "elapsed_sec": 63766.24273395538, "step_time_sec": 8.238613954978064, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7668, "loss": 3.799288272857666, "lr": 0.00018691832148810895, "elapsed_sec": 63774.48233413696, "step_time_sec": 8.239456921000965, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7669, "loss": 3.751052141189575, "lr": 0.00018687997165468576, "elapsed_sec": 63782.723049879074, "step_time_sec": 8.240558349061757, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7670, "loss": 3.801146984100342, "lr": 0.00018684156974542186, "elapsed_sec": 63790.96076178551, "step_time_sec": 8.237566341995262, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7671, "loss": 3.9806747436523438, "lr": 0.00018680311578401734, "elapsed_sec": 63799.20114946365, "step_time_sec": 8.240206612972543, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7672, "loss": 3.699627161026001, "lr": 0.00018676460979420438, "elapsed_sec": 63807.43970346451, "step_time_sec": 8.23838076100219, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7673, "loss": 3.828214168548584, "lr": 0.00018672605179974729, "elapsed_sec": 63815.67932033539, "step_time_sec": 8.239500775001943, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7674, "loss": 3.9440982341766357, "lr": 0.00018668744182444245, "elapsed_sec": 63823.91854071617, "step_time_sec": 8.239046334987506, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7675, "loss": 3.7152199745178223, "lr": 0.0001866487798921184, "elapsed_sec": 63832.15771460533, "step_time_sec": 8.238978989073075, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7676, "loss": 3.9296858310699463, "lr": 0.00018661006602663566, "elapsed_sec": 63840.39708662033, "step_time_sec": 8.239249618025497, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7677, "loss": 3.913140296936035, "lr": 0.00018657130025188682, "elapsed_sec": 63848.63645505905, "step_time_sec": 8.239205028046854, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7678, "loss": 3.8678138256073, "lr": 0.00018653248259179654, "elapsed_sec": 63856.87654733658, "step_time_sec": 8.239966495078988, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7679, "loss": 3.839991807937622, "lr": 0.00018649361307032148, "elapsed_sec": 63865.115785598755, "step_time_sec": 8.23903383500874, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7680, "loss": 3.7778704166412354, "lr": 0.00018645469171145029, "elapsed_sec": 63873.35273194313, "step_time_sec": 8.236789294052869, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7681, "loss": 4.002618789672852, "lr": 0.00018641571853920367, "elapsed_sec": 63881.59304070473, "step_time_sec": 8.240142527967691, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7682, "loss": 3.790642738342285, "lr": 0.0001863766935776342, "elapsed_sec": 63889.832760334015, "step_time_sec": 8.239571231999435, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7683, "loss": 4.047682762145996, "lr": 0.00018633761685082648, "elapsed_sec": 63898.0715739727, "step_time_sec": 8.238642985001206, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7684, "loss": 3.7605957984924316, "lr": 0.0001862984883828971, "elapsed_sec": 63906.31194400787, "step_time_sec": 8.240198710933328, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7685, "loss": 3.6479313373565674, "lr": 0.00018625930819799453, "elapsed_sec": 63914.55237555504, "step_time_sec": 8.240283088060096, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7686, "loss": 3.8985095024108887, "lr": 0.0001862200763202991, "elapsed_sec": 63922.792209386826, "step_time_sec": 8.239652782096528, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7687, "loss": 3.851168155670166, "lr": 0.00018618079277402322, "elapsed_sec": 63931.032458782196, "step_time_sec": 8.240085321012884, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7688, "loss": 3.8849937915802, "lr": 0.000186141457583411, "elapsed_sec": 63939.27069234848, "step_time_sec": 8.238030630047433, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7689, "loss": 3.849583148956299, "lr": 0.00018610207077273847, "elapsed_sec": 63947.509667396545, "step_time_sec": 8.238820401020348, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7690, "loss": 3.9432685375213623, "lr": 0.00018606263236631363, "elapsed_sec": 63955.749922037125, "step_time_sec": 8.24012513901107, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7691, "loss": 3.760927438735962, "lr": 0.00018602314238847618, "elapsed_sec": 63963.989065647125, "step_time_sec": 8.238934500026517, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7692, "loss": 3.8608829975128174, "lr": 0.00018598360086359773, "elapsed_sec": 63972.22830200195, "step_time_sec": 8.239044974907301, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7693, "loss": 3.8817927837371826, "lr": 0.00018594400781608167, "elapsed_sec": 63980.466292619705, "step_time_sec": 8.237883387948386, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7694, "loss": 3.6714277267456055, "lr": 0.0001859043632703632, "elapsed_sec": 63988.7065179348, "step_time_sec": 8.240060860058293, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7695, "loss": 3.956608772277832, "lr": 0.0001858646672509093, "elapsed_sec": 63996.94699001312, "step_time_sec": 8.240320694982074, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7696, "loss": 3.7835211753845215, "lr": 0.00018582491978221872, "elapsed_sec": 64005.18437123299, "step_time_sec": 8.237160805030726, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7697, "loss": 3.7602102756500244, "lr": 0.00018578512088882198, "elapsed_sec": 64013.42053079605, "step_time_sec": 8.235999932978302, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7698, "loss": 3.889376163482666, "lr": 0.00018574527059528125, "elapsed_sec": 64021.65916991234, "step_time_sec": 8.238489110022783, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7699, "loss": 3.926515579223633, "lr": 0.00018570536892619055, "elapsed_sec": 64029.89986515045, "step_time_sec": 8.240526276058517, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7700, "loss": 3.942775011062622, "lr": 0.00018566541590617553, "elapsed_sec": 64038.14020442963, "step_time_sec": 8.240198473911732, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7701, "loss": 3.7383852005004883, "lr": 0.00018562541155989356, "elapsed_sec": 64046.37987112999, "step_time_sec": 8.239499581977725, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7702, "loss": 3.8726584911346436, "lr": 0.00018558535591203364, "elapsed_sec": 64054.61828637123, "step_time_sec": 8.238279462908395, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7703, "loss": 3.8980209827423096, "lr": 0.00018554524898731648, "elapsed_sec": 64062.857770204544, "step_time_sec": 8.239327672985382, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7704, "loss": 3.9224324226379395, "lr": 0.0001855050908104944, "elapsed_sec": 64071.09798645973, "step_time_sec": 8.2400524020195, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7705, "loss": 3.956200122833252, "lr": 0.00018546488140635145, "elapsed_sec": 64079.338129520416, "step_time_sec": 8.239942594897002, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7706, "loss": 3.8792521953582764, "lr": 0.00018542462079970313, "elapsed_sec": 64087.577023506165, "step_time_sec": 8.238777091028169, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7707, "loss": 3.8899455070495605, "lr": 0.00018538430901539668, "elapsed_sec": 64095.817804813385, "step_time_sec": 8.240632731933147, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7708, "loss": 3.8709702491760254, "lr": 0.00018534394607831083, "elapsed_sec": 64104.056530714035, "step_time_sec": 8.238575347000733, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7709, "loss": 3.785229206085205, "lr": 0.00018530353201335597, "elapsed_sec": 64112.29535126686, "step_time_sec": 8.238617633935064, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7710, "loss": 3.530397891998291, "lr": 0.00018526306684547394, "elapsed_sec": 64120.53383374214, "step_time_sec": 8.238361065974459, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7711, "loss": 3.599289894104004, "lr": 0.0001852225505996382, "elapsed_sec": 64128.77134037018, "step_time_sec": 8.237332180025987, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7712, "loss": 3.9554829597473145, "lr": 0.00018518198330085375, "elapsed_sec": 64137.009428977966, "step_time_sec": 8.23794264497701, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7713, "loss": 3.8489551544189453, "lr": 0.00018514136497415698, "elapsed_sec": 64145.248302936554, "step_time_sec": 8.2386527529452, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7714, "loss": 3.806079626083374, "lr": 0.00018510069564461589, "elapsed_sec": 64153.48646354675, "step_time_sec": 8.238040501018986, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7715, "loss": 3.8429653644561768, "lr": 0.0001850599753373299, "elapsed_sec": 64161.72679114342, "step_time_sec": 8.240137584041804, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7716, "loss": 3.87306547164917, "lr": 0.00018501920407742992, "elapsed_sec": 64169.96415281296, "step_time_sec": 8.237220245064236, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7717, "loss": 3.7439918518066406, "lr": 0.00018497838189007827, "elapsed_sec": 64178.20519256592, "step_time_sec": 8.240896333940327, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7718, "loss": 3.874000072479248, "lr": 0.00018493750880046872, "elapsed_sec": 64186.444871902466, "step_time_sec": 8.239486454986036, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7719, "loss": 3.903355598449707, "lr": 0.0001848965848338265, "elapsed_sec": 64194.68165254593, "step_time_sec": 8.236640551011078, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7720, "loss": 3.8213722705841064, "lr": 0.00018485561001540816, "elapsed_sec": 64202.920758247375, "step_time_sec": 8.238930830033496, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7721, "loss": 3.861029624938965, "lr": 0.00018481458437050166, "elapsed_sec": 64211.15893697739, "step_time_sec": 8.238034292007796, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7722, "loss": 3.7927207946777344, "lr": 0.00018477350792442637, "elapsed_sec": 64219.397981882095, "step_time_sec": 8.23885592899751, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7723, "loss": 3.8237709999084473, "lr": 0.00018473238070253295, "elapsed_sec": 64227.63807463646, "step_time_sec": 8.23991055903025, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7724, "loss": 3.9179489612579346, "lr": 0.00018469120273020343, "elapsed_sec": 64235.87692785263, "step_time_sec": 8.238691140897572, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7725, "loss": 3.8870556354522705, "lr": 0.00018464997403285117, "elapsed_sec": 64244.11619138718, "step_time_sec": 8.239099520025775, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7726, "loss": 3.787580728530884, "lr": 0.0001846086946359208, "elapsed_sec": 64252.354714632034, "step_time_sec": 8.238380966009572, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7727, "loss": 3.786285400390625, "lr": 0.00018456736456488832, "elapsed_sec": 64260.59350466728, "step_time_sec": 8.238659271970391, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7728, "loss": 3.918842077255249, "lr": 0.00018452598384526087, "elapsed_sec": 64268.83319425583, "step_time_sec": 8.23947557900101, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7729, "loss": 3.94724702835083, "lr": 0.00018448455250257695, "elapsed_sec": 64277.071823596954, "step_time_sec": 8.238515663077123, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7730, "loss": 3.822064161300659, "lr": 0.00018444307056240635, "elapsed_sec": 64285.31206536293, "step_time_sec": 8.24003313400317, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7731, "loss": 3.952911376953125, "lr": 0.00018440153805034988, "elapsed_sec": 64293.55154585838, "step_time_sec": 8.239305302035064, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7732, "loss": 3.8133883476257324, "lr": 0.0001843599549920398, "elapsed_sec": 64301.79132246971, "step_time_sec": 8.239594992948696, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7733, "loss": 3.8384385108947754, "lr": 0.00018431832141313942, "elapsed_sec": 64310.03136229515, "step_time_sec": 8.239900317974389, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7734, "loss": 3.795400857925415, "lr": 0.00018427663733934328, "elapsed_sec": 64318.26914143562, "step_time_sec": 8.237636286998168, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7735, "loss": 3.8912665843963623, "lr": 0.00018423490279637707, "elapsed_sec": 64326.50882267952, "step_time_sec": 8.239499869057909, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7736, "loss": 3.9460690021514893, "lr": 0.0001841931178099976, "elapsed_sec": 64334.75133681297, "step_time_sec": 8.24238339997828, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7737, "loss": 3.88720440864563, "lr": 0.00018415128240599287, "elapsed_sec": 64342.990228414536, "step_time_sec": 8.23876430594828, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7738, "loss": 3.8265416622161865, "lr": 0.00018410939661018202, "elapsed_sec": 64351.232825517654, "step_time_sec": 8.242358562070876, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7739, "loss": 3.990535259246826, "lr": 0.00018406746044841513, "elapsed_sec": 64359.47157096863, "step_time_sec": 8.238556983065791, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7740, "loss": 3.842360496520996, "lr": 0.00018402547394657352, "elapsed_sec": 64367.7116792202, "step_time_sec": 8.239889266085811, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7741, "loss": 3.8970935344696045, "lr": 0.00018398343713056954, "elapsed_sec": 64375.950820446014, "step_time_sec": 8.238987602991983, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7742, "loss": 3.8520491123199463, "lr": 0.00018394135002634658, "elapsed_sec": 64384.1892478466, "step_time_sec": 8.238275478943251, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7743, "loss": 3.873136281967163, "lr": 0.00018389921265987905, "elapsed_sec": 64392.42965698242, "step_time_sec": 8.24024370405823, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7744, "loss": 3.7611589431762695, "lr": 0.00018385702505717243, "elapsed_sec": 64400.66990041733, "step_time_sec": 8.240100310998969, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7745, "loss": 3.7733654975891113, "lr": 0.0001838147872442631, "elapsed_sec": 64408.909813165665, "step_time_sec": 8.239744817023166, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7746, "loss": 3.950350761413574, "lr": 0.0001837724992472186, "elapsed_sec": 64417.15007162094, "step_time_sec": 8.240115000982769, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7747, "loss": 3.881033182144165, "lr": 0.0001837301610921372, "elapsed_sec": 64425.38791871071, "step_time_sec": 8.237632691045292, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7748, "loss": 3.9314591884613037, "lr": 0.00018368777280514838, "elapsed_sec": 64433.62948513031, "step_time_sec": 8.241405516047962, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7749, "loss": 3.8810853958129883, "lr": 0.00018364533441241242, "elapsed_sec": 64441.86847925186, "step_time_sec": 8.238844112027436, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7750, "loss": 4.013406276702881, "lr": 0.00018360284594012052, "elapsed_sec": 64450.1059782505, "step_time_sec": 8.2373447510181, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7751, "loss": 3.80700421333313, "lr": 0.00018356030741449483, "elapsed_sec": 64458.34459710121, "step_time_sec": 8.238450317061506, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7752, "loss": 4.014397621154785, "lr": 0.00018351771886178836, "elapsed_sec": 64466.583891391754, "step_time_sec": 8.239166737068444, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7753, "loss": 3.8899612426757812, "lr": 0.00018347508030828503, "elapsed_sec": 64474.82249879837, "step_time_sec": 8.238404850941151, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7754, "loss": 3.8807225227355957, "lr": 0.00018343239178029958, "elapsed_sec": 64483.062338113785, "step_time_sec": 8.239646141068079, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7755, "loss": 3.828062057495117, "lr": 0.00018338965330417763, "elapsed_sec": 64491.300114393234, "step_time_sec": 8.23764894506894, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7756, "loss": 3.8387930393218994, "lr": 0.00018334686490629562, "elapsed_sec": 64499.541246175766, "step_time_sec": 8.240986148943193, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7757, "loss": 3.847644329071045, "lr": 0.00018330402661306073, "elapsed_sec": 64507.78076457977, "step_time_sec": 8.239307256997563, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7758, "loss": 3.8623106479644775, "lr": 0.00018326113845091102, "elapsed_sec": 64516.02084374428, "step_time_sec": 8.239933026954532, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7759, "loss": 3.909534215927124, "lr": 0.00018321820044631533, "elapsed_sec": 64524.2603931427, "step_time_sec": 8.23939565406181, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7760, "loss": 3.9342689514160156, "lr": 0.00018317521262577313, "elapsed_sec": 64532.4979197979, "step_time_sec": 8.237345741013996, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7761, "loss": 3.949199676513672, "lr": 0.00018313217501581485, "elapsed_sec": 64540.73762345314, "step_time_sec": 8.239577600033954, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7762, "loss": 4.136767387390137, "lr": 0.00018308908764300144, "elapsed_sec": 64548.97836494446, "step_time_sec": 8.240555574069731, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7763, "loss": 3.996854305267334, "lr": 0.00018304595053392472, "elapsed_sec": 64557.21590042114, "step_time_sec": 8.237395612057298, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7764, "loss": 3.7912075519561768, "lr": 0.00018300276371520708, "elapsed_sec": 64565.4559841156, "step_time_sec": 8.239893137943, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7765, "loss": 3.960057258605957, "lr": 0.00018295952721350165, "elapsed_sec": 64573.69508934021, "step_time_sec": 8.239002893096767, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7766, "loss": 4.0456414222717285, "lr": 0.00018291624105549223, "elapsed_sec": 64581.934824705124, "step_time_sec": 8.239505014033057, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7767, "loss": 3.8382863998413086, "lr": 0.00018287290526789327, "elapsed_sec": 64590.17420864105, "step_time_sec": 8.23926360195037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7768, "loss": 3.8449723720550537, "lr": 0.0001828295198774498, "elapsed_sec": 64598.4137506485, "step_time_sec": 8.239347520051524, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7769, "loss": 3.7932634353637695, "lr": 0.00018278608491093752, "elapsed_sec": 64606.65370416641, "step_time_sec": 8.239852549973875, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7770, "loss": 3.972240924835205, "lr": 0.00018274260039516268, "elapsed_sec": 64614.89375495911, "step_time_sec": 8.23981154395733, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7771, "loss": 3.6245810985565186, "lr": 0.00018269906635696217, "elapsed_sec": 64623.13312983513, "step_time_sec": 8.239216752932407, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7772, "loss": 3.8710644245147705, "lr": 0.00018265548282320334, "elapsed_sec": 64631.372216939926, "step_time_sec": 8.238965208060108, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7773, "loss": 3.645946502685547, "lr": 0.00018261184982078422, "elapsed_sec": 64639.612419843674, "step_time_sec": 8.240017750067636, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7774, "loss": 3.8399558067321777, "lr": 0.00018256816737663326, "elapsed_sec": 64647.852685689926, "step_time_sec": 8.240081679075956, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7775, "loss": 4.110079765319824, "lr": 0.00018252443551770945, "elapsed_sec": 64656.09168291092, "step_time_sec": 8.238821948994882, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7776, "loss": 3.9993388652801514, "lr": 0.0001824806542710023, "elapsed_sec": 64664.33060216904, "step_time_sec": 8.238789616036229, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7777, "loss": 4.050775051116943, "lr": 0.00018243682366353184, "elapsed_sec": 64672.570100069046, "step_time_sec": 8.239313545986079, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7778, "loss": 3.7528433799743652, "lr": 0.00018239294372234848, "elapsed_sec": 64680.8093187809, "step_time_sec": 8.239085301058367, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7779, "loss": 3.6911213397979736, "lr": 0.00018234901447453312, "elapsed_sec": 64689.0507068634, "step_time_sec": 8.241196264978498, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7780, "loss": 3.900099515914917, "lr": 0.00018230503594719703, "elapsed_sec": 64697.29112958908, "step_time_sec": 8.240316738025285, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7781, "loss": 3.8738083839416504, "lr": 0.000182261008167482, "elapsed_sec": 64705.52973675728, "step_time_sec": 8.238411294994876, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7782, "loss": 3.804414987564087, "lr": 0.00018221693116256014, "elapsed_sec": 64713.769443035126, "step_time_sec": 8.23951936303638, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7783, "loss": 3.8411083221435547, "lr": 0.000182172804959634, "elapsed_sec": 64722.007234334946, "step_time_sec": 8.237662003026344, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7784, "loss": 3.8911428451538086, "lr": 0.00018212862958593638, "elapsed_sec": 64730.24566960335, "step_time_sec": 8.23831467400305, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7785, "loss": 3.871626853942871, "lr": 0.00018208440506873057, "elapsed_sec": 64738.484472990036, "step_time_sec": 8.238588899024762, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7786, "loss": 3.8419477939605713, "lr": 0.00018204013143531002, "elapsed_sec": 64746.72121810913, "step_time_sec": 8.236648999969475, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7787, "loss": 3.9475619792938232, "lr": 0.00018199580871299873, "elapsed_sec": 64754.96118092537, "step_time_sec": 8.2397181839915, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7788, "loss": 3.872295379638672, "lr": 0.0001819514369291508, "elapsed_sec": 64763.20074510574, "step_time_sec": 8.23941537400242, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7789, "loss": 3.802177667617798, "lr": 0.0001819070161111506, "elapsed_sec": 64771.44064474106, "step_time_sec": 8.239767896011472, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7790, "loss": 3.955873489379883, "lr": 0.00018186254628641292, "elapsed_sec": 64779.68024754524, "step_time_sec": 8.23938444396481, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7791, "loss": 3.8192434310913086, "lr": 0.00018181802748238265, "elapsed_sec": 64787.91803908348, "step_time_sec": 8.237636490957811, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7792, "loss": 3.753603219985962, "lr": 0.00018177345972653498, "elapsed_sec": 64796.15769505501, "step_time_sec": 8.239508638042025, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7793, "loss": 3.7116477489471436, "lr": 0.00018172884304637534, "elapsed_sec": 64804.39625239372, "step_time_sec": 8.238392719998956, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7794, "loss": 3.9002389907836914, "lr": 0.0001816841774694392, "elapsed_sec": 64812.63458490372, "step_time_sec": 8.238147440017201, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7795, "loss": 3.8049941062927246, "lr": 0.00018163946302329236, "elapsed_sec": 64820.87575006485, "step_time_sec": 8.24099521106109, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7796, "loss": 3.802579164505005, "lr": 0.00018159469973553076, "elapsed_sec": 64829.11586380005, "step_time_sec": 8.239936189027503, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7797, "loss": 3.837769031524658, "lr": 0.00018154988763378042, "elapsed_sec": 64837.35520553589, "step_time_sec": 8.239230822073296, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7798, "loss": 3.695228338241577, "lr": 0.00018150502674569755, "elapsed_sec": 64845.59573197365, "step_time_sec": 8.240353710949421, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7799, "loss": 3.980443239212036, "lr": 0.00018146011709896845, "elapsed_sec": 64853.83442020416, "step_time_sec": 8.23849465395324, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7800, "loss": 3.964552164077759, "lr": 0.00018141515872130946, "elapsed_sec": 64862.075550079346, "step_time_sec": 8.240937524940819, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7801, "loss": 3.92673921585083, "lr": 0.00018137015164046706, "elapsed_sec": 64870.316009283066, "step_time_sec": 8.240300116012804, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7802, "loss": 3.8177473545074463, "lr": 0.00018132509588421778, "elapsed_sec": 64878.55518221855, "step_time_sec": 8.239031216013245, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7803, "loss": 3.832573652267456, "lr": 0.00018127999148036817, "elapsed_sec": 64886.79499936104, "step_time_sec": 8.239657283993438, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7804, "loss": 3.852079391479492, "lr": 0.0001812348384567548, "elapsed_sec": 64895.03503084183, "step_time_sec": 8.239812008920126, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7805, "loss": 3.860182523727417, "lr": 0.00018118963684124424, "elapsed_sec": 64903.27326154709, "step_time_sec": 8.23806456103921, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7806, "loss": 3.89016056060791, "lr": 0.0001811443866617331, "elapsed_sec": 64911.512320518494, "step_time_sec": 8.238905076985247, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7807, "loss": 3.8745410442352295, "lr": 0.0001810990879461479, "elapsed_sec": 64919.75037193298, "step_time_sec": 8.237852122983895, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7808, "loss": 3.8549726009368896, "lr": 0.00018105374072244513, "elapsed_sec": 64927.99064898491, "step_time_sec": 8.240132701932453, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7809, "loss": 3.890024185180664, "lr": 0.00018100834501861125, "elapsed_sec": 64936.23117399216, "step_time_sec": 8.240349457017146, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7810, "loss": 4.014494895935059, "lr": 0.00018096290086266263, "elapsed_sec": 64944.47069787979, "step_time_sec": 8.23938037303742, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7811, "loss": 3.8686776161193848, "lr": 0.0001809174082826455, "elapsed_sec": 64952.71094417572, "step_time_sec": 8.240090253995731, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7812, "loss": 3.999635934829712, "lr": 0.000180871867306636, "elapsed_sec": 64960.949528217316, "step_time_sec": 8.238394755055197, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7813, "loss": 3.849032402038574, "lr": 0.00018082627796274018, "elapsed_sec": 64969.18843340874, "step_time_sec": 8.238707815064117, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7814, "loss": 3.8924171924591064, "lr": 0.0001807806402790939, "elapsed_sec": 64977.42704963684, "step_time_sec": 8.238460654974915, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7815, "loss": 3.938573122024536, "lr": 0.0001807349542838628, "elapsed_sec": 64985.66454267502, "step_time_sec": 8.237349442904815, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7816, "loss": 3.8856053352355957, "lr": 0.00018068922000524245, "elapsed_sec": 64993.904399871826, "step_time_sec": 8.239713615039364, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7817, "loss": 4.061390399932861, "lr": 0.00018064343747145815, "elapsed_sec": 65002.13731265068, "step_time_sec": 8.232688612071797, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7818, "loss": 3.8275046348571777, "lr": 0.00018059760671076502, "elapsed_sec": 65010.37136769295, "step_time_sec": 8.233920336002484, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7819, "loss": 3.942307233810425, "lr": 0.00018055172775144788, "elapsed_sec": 65018.61003947258, "step_time_sec": 8.238493305980228, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7820, "loss": 3.7726519107818604, "lr": 0.0001805058006218213, "elapsed_sec": 65026.84998488426, "step_time_sec": 8.239824996097013, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7821, "loss": 3.862459897994995, "lr": 0.00018045982535022968, "elapsed_sec": 65035.08963727951, "step_time_sec": 8.239421428996138, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7822, "loss": 3.9595396518707275, "lr": 0.000180413801965047, "elapsed_sec": 65043.32780337334, "step_time_sec": 8.238007239997387, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7823, "loss": 3.9212534427642822, "lr": 0.00018036773049467708, "elapsed_sec": 65051.567885398865, "step_time_sec": 8.239944075932726, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7824, "loss": 3.8788578510284424, "lr": 0.00018032161096755324, "elapsed_sec": 65059.80841636658, "step_time_sec": 8.240319833043031, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7825, "loss": 3.969602584838867, "lr": 0.0001802754434121386, "elapsed_sec": 65068.048523426056, "step_time_sec": 8.239958098041825, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7826, "loss": 3.848072052001953, "lr": 0.00018022922785692588, "elapsed_sec": 65076.28842711449, "step_time_sec": 8.239715725998394, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7827, "loss": 3.7258129119873047, "lr": 0.0001801829643304374, "elapsed_sec": 65084.521624326706, "step_time_sec": 8.233055879012682, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7828, "loss": 3.8010220527648926, "lr": 0.0001801366528612251, "elapsed_sec": 65092.76033473015, "step_time_sec": 8.238565651001409, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7829, "loss": 3.768705368041992, "lr": 0.00018009029347787052, "elapsed_sec": 65100.99747180939, "step_time_sec": 8.23700403701514, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7830, "loss": 3.9640309810638428, "lr": 0.00018004388620898476, "elapsed_sec": 65109.235263347626, "step_time_sec": 8.237545924028382, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7831, "loss": 4.124197959899902, "lr": 0.00017999743108320843, "elapsed_sec": 65117.47539114952, "step_time_sec": 8.239974939031526, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7832, "loss": 4.0095391273498535, "lr": 0.00017995092812921182, "elapsed_sec": 65125.713842868805, "step_time_sec": 8.238287291955203, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7833, "loss": 4.101194381713867, "lr": 0.0001799043773756945, "elapsed_sec": 65133.95354962349, "step_time_sec": 8.239520482951775, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7834, "loss": 3.822277545928955, "lr": 0.00017985777885138582, "elapsed_sec": 65142.19210290909, "step_time_sec": 8.238390597049147, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7835, "loss": 3.849451780319214, "lr": 0.00017981113258504436, "elapsed_sec": 65150.4310863018, "step_time_sec": 8.23885538103059, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7836, "loss": 3.6731796264648438, "lr": 0.0001797644386054583, "elapsed_sec": 65158.670018196106, "step_time_sec": 8.238746994989924, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7837, "loss": 3.7602388858795166, "lr": 0.00017971769694144528, "elapsed_sec": 65166.908794641495, "step_time_sec": 8.238647571997717, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7838, "loss": 3.9103329181671143, "lr": 0.0001796709076218523, "elapsed_sec": 65175.14998316765, "step_time_sec": 8.240961408941075, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7839, "loss": 3.7431588172912598, "lr": 0.00017962407067555577, "elapsed_sec": 65183.38965511322, "step_time_sec": 8.239546664990485, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7840, "loss": 3.64092755317688, "lr": 0.00017957718613146155, "elapsed_sec": 65191.62965631485, "step_time_sec": 8.23987625609152, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7841, "loss": 3.8197989463806152, "lr": 0.0001795302540185049, "elapsed_sec": 65199.86781215668, "step_time_sec": 8.237920655985363, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7842, "loss": 3.7517669200897217, "lr": 0.00017948327436565032, "elapsed_sec": 65208.10621833801, "step_time_sec": 8.238270581001416, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7843, "loss": 3.8997647762298584, "lr": 0.0001794362472018917, "elapsed_sec": 65216.34524822235, "step_time_sec": 8.23886061203666, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7844, "loss": 3.8865106105804443, "lr": 0.0001793891725562523, "elapsed_sec": 65224.583597421646, "step_time_sec": 8.238241404062137, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7845, "loss": 3.8123321533203125, "lr": 0.00017934205045778468, "elapsed_sec": 65232.82280874252, "step_time_sec": 8.239012499921955, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7846, "loss": 3.7457597255706787, "lr": 0.00017929488093557062, "elapsed_sec": 65241.06337761879, "step_time_sec": 8.240407697041519, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7847, "loss": 3.899806261062622, "lr": 0.00017924766401872122, "elapsed_sec": 65249.300154447556, "step_time_sec": 8.236630133003928, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7848, "loss": 3.902069568634033, "lr": 0.00017920039973637675, "elapsed_sec": 65257.54015016556, "step_time_sec": 8.239843055023812, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7849, "loss": 4.021007061004639, "lr": 0.0001791530881177069, "elapsed_sec": 65265.78075385094, "step_time_sec": 8.240461440989748, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7850, "loss": 3.8705878257751465, "lr": 0.00017910572919191035, "elapsed_sec": 65274.018800497055, "step_time_sec": 8.23784906999208, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7851, "loss": 3.753777503967285, "lr": 0.00017905832298821515, "elapsed_sec": 65282.25858950615, "step_time_sec": 8.239656916935928, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7852, "loss": 3.94191837310791, "lr": 0.00017901086953587837, "elapsed_sec": 65290.496196746826, "step_time_sec": 8.23745948902797, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7853, "loss": 3.8107283115386963, "lr": 0.0001789633688641864, "elapsed_sec": 65298.73646378517, "step_time_sec": 8.240179212065414, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7854, "loss": 4.154809951782227, "lr": 0.00017891582100245464, "elapsed_sec": 65306.97617149353, "step_time_sec": 8.239507003920153, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7855, "loss": 3.888310194015503, "lr": 0.00017886822598002774, "elapsed_sec": 65315.21455192566, "step_time_sec": 8.23817403393332, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7856, "loss": 4.137253284454346, "lr": 0.00017882058382627934, "elapsed_sec": 65323.45479655266, "step_time_sec": 8.240126307937317, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7857, "loss": 3.709979295730591, "lr": 0.0001787728945706122, "elapsed_sec": 65331.69318842888, "step_time_sec": 8.238195377052762, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7858, "loss": 3.8162903785705566, "lr": 0.00017872515824245817, "elapsed_sec": 65339.93162727356, "step_time_sec": 8.238263727980666, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7859, "loss": 3.69720196723938, "lr": 0.0001786773748712782, "elapsed_sec": 65348.170503139496, "step_time_sec": 8.238725262926891, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7860, "loss": 3.9009439945220947, "lr": 0.00017862954448656215, "elapsed_sec": 65356.40847373009, "step_time_sec": 8.237772793974727, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7861, "loss": 3.877044200897217, "lr": 0.00017858166711782897, "elapsed_sec": 65364.64752006531, "step_time_sec": 8.238885261001997, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7862, "loss": 3.7782530784606934, "lr": 0.00017853374279462662, "elapsed_sec": 65372.886182546616, "step_time_sec": 8.238468562019989, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7863, "loss": 3.932659864425659, "lr": 0.000178485771546532, "elapsed_sec": 65381.12581992149, "step_time_sec": 8.239475445938297, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7864, "loss": 3.7797882556915283, "lr": 0.00017843775340315105, "elapsed_sec": 65389.36497807503, "step_time_sec": 8.239020724897273, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7865, "loss": 3.865159511566162, "lr": 0.0001783896883941185, "elapsed_sec": 65397.603507995605, "step_time_sec": 8.238383595016785, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7866, "loss": 4.013564109802246, "lr": 0.00017834157654909811, "elapsed_sec": 65405.84273123741, "step_time_sec": 8.239074869081378, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7867, "loss": 3.7777366638183594, "lr": 0.0001782934178977826, "elapsed_sec": 65414.0834646225, "step_time_sec": 8.240524447988719, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7868, "loss": 3.9354002475738525, "lr": 0.0001782452124698934, "elapsed_sec": 65422.32237792015, "step_time_sec": 8.238751851953566, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7869, "loss": 4.000233173370361, "lr": 0.00017819696029518103, "elapsed_sec": 65430.555842638016, "step_time_sec": 8.23331301100552, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7870, "loss": 3.897179126739502, "lr": 0.0001781486614034246, "elapsed_sec": 65438.7940158844, "step_time_sec": 8.237999887089245, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7871, "loss": 3.798095703125, "lr": 0.00017810031582443233, "elapsed_sec": 65447.034042835236, "step_time_sec": 8.239912902005017, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7872, "loss": 3.7922379970550537, "lr": 0.00017805192358804104, "elapsed_sec": 65455.27221131325, "step_time_sec": 8.237955429009162, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7873, "loss": 3.9494755268096924, "lr": 0.00017800348472411646, "elapsed_sec": 65463.51166462898, "step_time_sec": 8.239302813075483, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7874, "loss": 3.9406466484069824, "lr": 0.00017795499926255307, "elapsed_sec": 65471.751561403275, "step_time_sec": 8.239759820979089, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7875, "loss": 3.995558023452759, "lr": 0.00017790646723327404, "elapsed_sec": 65479.98995375633, "step_time_sec": 8.238201536936685, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7876, "loss": 3.7756030559539795, "lr": 0.00017785788866623138, "elapsed_sec": 65488.230712890625, "step_time_sec": 8.24061892204918, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7877, "loss": 3.971370220184326, "lr": 0.00017780926359140577, "elapsed_sec": 65496.469277858734, "step_time_sec": 8.238422152935527, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7878, "loss": 3.7428457736968994, "lr": 0.0001777605920388066, "elapsed_sec": 65504.709263801575, "step_time_sec": 8.23979231400881, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7879, "loss": 3.7337987422943115, "lr": 0.000177711874038472, "elapsed_sec": 65512.94988274574, "step_time_sec": 8.240459574037232, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7880, "loss": 3.8019614219665527, "lr": 0.00017766310962046862, "elapsed_sec": 65521.18998217583, "step_time_sec": 8.23994339199271, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7881, "loss": 3.9000778198242188, "lr": 0.0001776142988148919, "elapsed_sec": 65529.4294462204, "step_time_sec": 8.239344947040081, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7882, "loss": 3.8431715965270996, "lr": 0.00017756544165186585, "elapsed_sec": 65537.67116570473, "step_time_sec": 8.241486403043382, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7883, "loss": 3.797532081604004, "lr": 0.0001775165381615431, "elapsed_sec": 65545.90923714638, "step_time_sec": 8.237953294068575, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7884, "loss": 3.9207193851470947, "lr": 0.00017746758837410486, "elapsed_sec": 65554.15134501457, "step_time_sec": 8.24196874606423, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7885, "loss": 3.865002155303955, "lr": 0.00017741859231976094, "elapsed_sec": 65562.38490509987, "step_time_sec": 8.233371378970332, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7886, "loss": 3.932575225830078, "lr": 0.00017736955002874965, "elapsed_sec": 65570.61867809296, "step_time_sec": 8.233604624983855, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7887, "loss": 3.881237030029297, "lr": 0.00017732046153133793, "elapsed_sec": 65578.85255169868, "step_time_sec": 8.23369298398029, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7888, "loss": 3.942897319793701, "lr": 0.00017727132685782111, "elapsed_sec": 65587.08642697334, "step_time_sec": 8.233732850989327, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7889, "loss": 3.719203233718872, "lr": 0.00017722214603852312, "elapsed_sec": 65595.31981873512, "step_time_sec": 8.23316458298359, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7890, "loss": 3.8159306049346924, "lr": 0.00017717291910379634, "elapsed_sec": 65603.55320906639, "step_time_sec": 8.233232058002613, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7891, "loss": 3.9601552486419678, "lr": 0.0001771236460840216, "elapsed_sec": 65611.79327368736, "step_time_sec": 8.239926186040975, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7892, "loss": 3.76326322555542, "lr": 0.00017707432700960816, "elapsed_sec": 65620.03251123428, "step_time_sec": 8.239070840994827, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7893, "loss": 4.059528350830078, "lr": 0.00017702496191099375, "elapsed_sec": 65628.27279353142, "step_time_sec": 8.240120129077695, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7894, "loss": 3.8819162845611572, "lr": 0.00017697555081864444, "elapsed_sec": 65636.51258301735, "step_time_sec": 8.239647403010167, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7895, "loss": 3.875457763671875, "lr": 0.00017692609376305474, "elapsed_sec": 65644.75182032585, "step_time_sec": 8.239068221999332, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7896, "loss": 3.823995590209961, "lr": 0.0001768765907747475, "elapsed_sec": 65652.98966479301, "step_time_sec": 8.237637848011218, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7897, "loss": 3.691021203994751, "lr": 0.00017682704188427397, "elapsed_sec": 65661.22801566124, "step_time_sec": 8.238204455934465, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7898, "loss": 3.897937536239624, "lr": 0.00017677744712221358, "elapsed_sec": 65669.4617741108, "step_time_sec": 8.233565922942944, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7899, "loss": 3.9455325603485107, "lr": 0.00017672780651917426, "elapsed_sec": 65677.69936728477, "step_time_sec": 8.237433727947064, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7900, "loss": 3.8928966522216797, "lr": 0.0001766781201057921, "elapsed_sec": 65685.93877625465, "step_time_sec": 8.2393174439203, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7901, "loss": 3.88558030128479, "lr": 0.00017662838791273155, "elapsed_sec": 65694.17659902573, "step_time_sec": 8.237647469039075, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7902, "loss": 4.111001491546631, "lr": 0.00017657860997068522, "elapsed_sec": 65702.41493916512, "step_time_sec": 8.238153109094128, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7903, "loss": 3.8637845516204834, "lr": 0.00017652878631037406, "elapsed_sec": 65710.65370702744, "step_time_sec": 8.23859575507231, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7904, "loss": 3.890920877456665, "lr": 0.0001764789169625471, "elapsed_sec": 65718.89292669296, "step_time_sec": 8.23909512895625, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7905, "loss": 4.01633358001709, "lr": 0.00017642900195798174, "elapsed_sec": 65727.13230633736, "step_time_sec": 8.239207133068703, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7906, "loss": 3.904926061630249, "lr": 0.0001763790413274834, "elapsed_sec": 65735.36479878426, "step_time_sec": 8.232291197986342, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7907, "loss": 4.009045600891113, "lr": 0.00017632903510188575, "elapsed_sec": 65743.59831762314, "step_time_sec": 8.233359089004807, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7908, "loss": 3.7531652450561523, "lr": 0.00017627898331205055, "elapsed_sec": 65751.82992959023, "step_time_sec": 8.23143344395794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7909, "loss": 3.914695978164673, "lr": 0.0001762288859888677, "elapsed_sec": 65760.06953549385, "step_time_sec": 8.239427636028267, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7910, "loss": 3.948024034500122, "lr": 0.00017617874316325522, "elapsed_sec": 65768.30887722969, "step_time_sec": 8.239213544991799, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7911, "loss": 3.8855772018432617, "lr": 0.0001761285548661592, "elapsed_sec": 65776.54859733582, "step_time_sec": 8.239583494956605, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7912, "loss": 3.945404291152954, "lr": 0.0001760783211285538, "elapsed_sec": 65784.78776049614, "step_time_sec": 8.23894253198523, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7913, "loss": 3.9118235111236572, "lr": 0.00017602804198144117, "elapsed_sec": 65793.02524447441, "step_time_sec": 8.237295662984252, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7914, "loss": 3.876732349395752, "lr": 0.00017597771745585155, "elapsed_sec": 65801.26603674889, "step_time_sec": 8.240621081902646, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7915, "loss": 4.010912895202637, "lr": 0.00017592734758284313, "elapsed_sec": 65809.50529670715, "step_time_sec": 8.2390849259682, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7916, "loss": 3.8468000888824463, "lr": 0.0001758769323935022, "elapsed_sec": 65817.74348926544, "step_time_sec": 8.238006004947238, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7917, "loss": 3.8767828941345215, "lr": 0.00017582647191894285, "elapsed_sec": 65825.97742390633, "step_time_sec": 8.233781709102914, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7918, "loss": 4.050091743469238, "lr": 0.00017577596619030726, "elapsed_sec": 65834.21086621284, "step_time_sec": 8.233290563919581, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7919, "loss": 3.912580966949463, "lr": 0.00017572541523876547, "elapsed_sec": 65842.445592165, "step_time_sec": 8.23456151096616, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7920, "loss": 3.855491876602173, "lr": 0.00017567481909551544, "elapsed_sec": 65850.68376159668, "step_time_sec": 8.238004660001025, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7921, "loss": 3.917487382888794, "lr": 0.00017562417779178302, "elapsed_sec": 65858.92321944237, "step_time_sec": 8.23927179700695, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7922, "loss": 3.930168390274048, "lr": 0.00017557349135882195, "elapsed_sec": 65867.16177797318, "step_time_sec": 8.238447032985277, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7923, "loss": 3.845285415649414, "lr": 0.0001755227598279138, "elapsed_sec": 65875.3954615593, "step_time_sec": 8.233455360052176, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7924, "loss": 3.9592273235321045, "lr": 0.000175471983230368, "elapsed_sec": 65883.62845754623, "step_time_sec": 8.232869378989562, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7925, "loss": 3.920599937438965, "lr": 0.0001754211615975217, "elapsed_sec": 65891.86345934868, "step_time_sec": 8.234778758022003, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7926, "loss": 3.906799554824829, "lr": 0.00017537029496074, "elapsed_sec": 65900.09786987305, "step_time_sec": 8.234229734982364, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7927, "loss": 3.9259092807769775, "lr": 0.00017531938335141567, "elapsed_sec": 65908.3318631649, "step_time_sec": 8.233869908959605, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7928, "loss": 3.752406120300293, "lr": 0.00017526842680096923, "elapsed_sec": 65916.56524467468, "step_time_sec": 8.233253198908642, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7929, "loss": 3.927983283996582, "lr": 0.000175217425340849, "elapsed_sec": 65924.80528092384, "step_time_sec": 8.239832959021442, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7930, "loss": 4.015666484832764, "lr": 0.0001751663790025309, "elapsed_sec": 65933.04500174522, "step_time_sec": 8.239516900037415, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7931, "loss": 4.167203426361084, "lr": 0.00017511528781751875, "elapsed_sec": 65941.28384375572, "step_time_sec": 8.238699859008193, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7932, "loss": 3.9410409927368164, "lr": 0.0001750641518173438, "elapsed_sec": 65949.52413392067, "step_time_sec": 8.240124532021582, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7933, "loss": 3.840787410736084, "lr": 0.0001750129710335652, "elapsed_sec": 65957.76422357559, "step_time_sec": 8.239888776908629, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7934, "loss": 3.8876748085021973, "lr": 0.00017496174549776954, "elapsed_sec": 65966.00324296951, "step_time_sec": 8.238855966017582, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7935, "loss": 3.992156744003296, "lr": 0.00017491047524157113, "elapsed_sec": 65974.24371767044, "step_time_sec": 8.240383383003063, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7936, "loss": 3.9379115104675293, "lr": 0.00017485916029661184, "elapsed_sec": 65982.48060894012, "step_time_sec": 8.236734976060688, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7937, "loss": 3.767059803009033, "lr": 0.00017480780069456119, "elapsed_sec": 65990.72106099129, "step_time_sec": 8.240259637008421, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7938, "loss": 3.9390921592712402, "lr": 0.00017475639646711617, "elapsed_sec": 65998.95947742462, "step_time_sec": 8.238237697980367, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7939, "loss": 3.870267629623413, "lr": 0.00017470494764600136, "elapsed_sec": 66007.19815206528, "step_time_sec": 8.23854457505513, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7940, "loss": 3.8280603885650635, "lr": 0.00017465345426296883, "elapsed_sec": 66015.43819332123, "step_time_sec": 8.239860650035553, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7941, "loss": 4.044775009155273, "lr": 0.00017460191634979818, "elapsed_sec": 66023.67691040039, "step_time_sec": 8.238558318000287, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7942, "loss": 3.8426551818847656, "lr": 0.00017455033393829654, "elapsed_sec": 66031.9166893959, "step_time_sec": 8.239571555983275, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7943, "loss": 4.028317451477051, "lr": 0.0001744987070602984, "elapsed_sec": 66040.15725898743, "step_time_sec": 8.240402462077327, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7944, "loss": 3.996246337890625, "lr": 0.00017444703574766572, "elapsed_sec": 66048.38931298256, "step_time_sec": 8.231896101031452, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7945, "loss": 4.002457141876221, "lr": 0.00017439532003228798, "elapsed_sec": 66056.62227129936, "step_time_sec": 8.232843574951403, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7946, "loss": 4.055463790893555, "lr": 0.00017434355994608194, "elapsed_sec": 66064.85562109947, "step_time_sec": 8.233170124003664, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7947, "loss": 3.9700608253479004, "lr": 0.00017429175552099175, "elapsed_sec": 66073.09438538551, "step_time_sec": 8.238555693998933, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7948, "loss": 3.9059014320373535, "lr": 0.00017423990678898903, "elapsed_sec": 66081.32689023018, "step_time_sec": 8.232359869056381, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7949, "loss": 3.956294298171997, "lr": 0.00017418801378207266, "elapsed_sec": 66089.55964684486, "step_time_sec": 8.232649205019698, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7950, "loss": 4.025996685028076, "lr": 0.00017413607653226886, "elapsed_sec": 66097.79258871078, "step_time_sec": 8.232711391989142, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7951, "loss": 3.8319640159606934, "lr": 0.00017408409507163117, "elapsed_sec": 66106.02615690231, "step_time_sec": 8.233384734019637, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7952, "loss": 4.0269341468811035, "lr": 0.0001740320694322404, "elapsed_sec": 66114.25765061378, "step_time_sec": 8.23136047902517, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7953, "loss": 3.948648691177368, "lr": 0.00017397999964620462, "elapsed_sec": 66122.49685287476, "step_time_sec": 8.239056021906435, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7954, "loss": 3.884169340133667, "lr": 0.0001739278857456592, "elapsed_sec": 66130.73449373245, "step_time_sec": 8.237461182987317, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7955, "loss": 3.9797921180725098, "lr": 0.00017387572776276662, "elapsed_sec": 66138.97320771217, "step_time_sec": 8.238573543960229, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7956, "loss": 3.9159276485443115, "lr": 0.0001738235257297167, "elapsed_sec": 66147.2122824192, "step_time_sec": 8.23891585599631, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7957, "loss": 4.038614749908447, "lr": 0.00017377127967872636, "elapsed_sec": 66155.45056128502, "step_time_sec": 8.23809059499763, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7958, "loss": 3.836695909500122, "lr": 0.0001737189896420397, "elapsed_sec": 66163.6906285286, "step_time_sec": 8.23990138899535, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7959, "loss": 3.988147258758545, "lr": 0.000173666655651928, "elapsed_sec": 66171.92995500565, "step_time_sec": 8.2391361000482, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7960, "loss": 4.104804515838623, "lr": 0.00017361427774068957, "elapsed_sec": 66180.16883563995, "step_time_sec": 8.238701481954195, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7961, "loss": 3.920534372329712, "lr": 0.00017356185594065, "elapsed_sec": 66188.40799331665, "step_time_sec": 8.239048536051996, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7962, "loss": 3.97385835647583, "lr": 0.0001735093902841618, "elapsed_sec": 66196.6468706131, "step_time_sec": 8.238743967958726, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7963, "loss": 4.0251264572143555, "lr": 0.00017345688080360464, "elapsed_sec": 66204.88580393791, "step_time_sec": 8.2386913029477, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7964, "loss": 4.008279323577881, "lr": 0.0001734043275313852, "elapsed_sec": 66213.12429475784, "step_time_sec": 8.238336352980696, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7965, "loss": 3.643368721008301, "lr": 0.00017335173049993718, "elapsed_sec": 66221.36355495453, "step_time_sec": 8.239140250952914, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7966, "loss": 4.076653003692627, "lr": 0.00017329908974172132, "elapsed_sec": 66229.60258483887, "step_time_sec": 8.238826189073734, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7967, "loss": 3.9162352085113525, "lr": 0.0001732464052892253, "elapsed_sec": 66237.84133505821, "step_time_sec": 8.238602273981087, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7968, "loss": 3.8708696365356445, "lr": 0.0001731936771749638, "elapsed_sec": 66246.074706316, "step_time_sec": 8.233177327085286, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7969, "loss": 3.8447465896606445, "lr": 0.00017314090543147845, "elapsed_sec": 66254.31273913383, "step_time_sec": 8.237898787949234, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7970, "loss": 3.903233528137207, "lr": 0.0001730880900913378, "elapsed_sec": 66262.5526278019, "step_time_sec": 8.239754152018577, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7971, "loss": 4.005638599395752, "lr": 0.0001730352311871373, "elapsed_sec": 66270.7899928093, "step_time_sec": 8.237188895000145, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7972, "loss": 4.082365989685059, "lr": 0.00017298232875149927, "elapsed_sec": 66279.02997636795, "step_time_sec": 8.239795789006166, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7973, "loss": 3.908205509185791, "lr": 0.00017292938281707293, "elapsed_sec": 66287.27019619942, "step_time_sec": 8.240114821004681, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7974, "loss": 3.93871808052063, "lr": 0.0001728763934165343, "elapsed_sec": 66295.50822281837, "step_time_sec": 8.237840574001893, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7975, "loss": 3.8573243618011475, "lr": 0.00017282336058258629, "elapsed_sec": 66303.74801135063, "step_time_sec": 8.239599928027019, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7976, "loss": 3.8701915740966797, "lr": 0.00017277028434795858, "elapsed_sec": 66311.98806333542, "step_time_sec": 8.239868890028447, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7977, "loss": 3.9340107440948486, "lr": 0.0001727171647454076, "elapsed_sec": 66320.22773861885, "step_time_sec": 8.239518888993189, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7978, "loss": 3.9354066848754883, "lr": 0.0001726640018077166, "elapsed_sec": 66328.4672665596, "step_time_sec": 8.239353239070624, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7979, "loss": 3.988467216491699, "lr": 0.00017261079556769556, "elapsed_sec": 66336.70540213585, "step_time_sec": 8.238040441065095, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7980, "loss": 3.884220838546753, "lr": 0.00017255754605818117, "elapsed_sec": 66344.94582271576, "step_time_sec": 8.240196216967888, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7981, "loss": 3.8898918628692627, "lr": 0.00017250425331203682, "elapsed_sec": 66353.18587779999, "step_time_sec": 8.239872752921656, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7982, "loss": 3.99863600730896, "lr": 0.00017245091736215262, "elapsed_sec": 66361.42295742035, "step_time_sec": 8.236920663970523, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7983, "loss": 3.88153076171875, "lr": 0.0001723975382414453, "elapsed_sec": 66369.6583635807, "step_time_sec": 8.235242184950039, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7984, "loss": 3.9389312267303467, "lr": 0.00017234411598285828, "elapsed_sec": 66377.89598345757, "step_time_sec": 8.23750083707273, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7985, "loss": 3.963521718978882, "lr": 0.00017229065061936155, "elapsed_sec": 66386.13399910927, "step_time_sec": 8.237799735972658, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7986, "loss": 3.918477773666382, "lr": 0.00017223714218395173, "elapsed_sec": 66394.37257933617, "step_time_sec": 8.238441321067512, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7987, "loss": 3.864250421524048, "lr": 0.00017218359070965205, "elapsed_sec": 66402.61160349846, "step_time_sec": 8.238902309094556, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7988, "loss": 4.0089569091796875, "lr": 0.00017212999622951225, "elapsed_sec": 66410.85081362724, "step_time_sec": 8.239040155080147, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7989, "loss": 3.842142105102539, "lr": 0.0001720763587766086, "elapsed_sec": 66419.08951950073, "step_time_sec": 8.238536054966971, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7990, "loss": 3.9514777660369873, "lr": 0.00017202267838404398, "elapsed_sec": 66427.32914566994, "step_time_sec": 8.239502883981913, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7991, "loss": 3.9502904415130615, "lr": 0.00017196895508494772, "elapsed_sec": 66435.56181049347, "step_time_sec": 8.232499876990914, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7992, "loss": 3.9418954849243164, "lr": 0.00017191518891247558, "elapsed_sec": 66443.79541230202, "step_time_sec": 8.233444849029183, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7993, "loss": 3.9522526264190674, "lr": 0.00017186137989980987, "elapsed_sec": 66452.02798390388, "step_time_sec": 8.232350141042843, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7994, "loss": 3.90483021736145, "lr": 0.00017180752808015922, "elapsed_sec": 66460.26754522324, "step_time_sec": 8.239401203929447, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7995, "loss": 4.0219550132751465, "lr": 0.00017175363348675887, "elapsed_sec": 66468.50626444817, "step_time_sec": 8.238568447995931, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7996, "loss": 3.9396519660949707, "lr": 0.00017169969615287025, "elapsed_sec": 66476.74523115158, "step_time_sec": 8.238864037091844, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7997, "loss": 3.960247039794922, "lr": 0.00017164571611178128, "elapsed_sec": 66484.98425936699, "step_time_sec": 8.238799343002029, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7998, "loss": 3.8819594383239746, "lr": 0.00017159169339680624, "elapsed_sec": 66493.22174406052, "step_time_sec": 8.23732946801465, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7999, "loss": 3.906001567840576, "lr": 0.0001715376280412857, "elapsed_sec": 66501.46143341064, "step_time_sec": 8.239506461075507, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8000, "loss": 3.9110074043273926, "lr": 0.00017148352007858659, "elapsed_sec": 66509.70039057732, "step_time_sec": 56.53131353505887, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": true, "validation_ran": true, "probe_ran": true, "probe_elapsed_sec": 0.993807731079869, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8001, "loss": 3.8855881690979004, "lr": 0.00017142936954210207, "elapsed_sec": 66566.23958396912, "step_time_sec": 8.246609268942848, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8002, "loss": 4.041001319885254, "lr": 0.00017137517646525168, "elapsed_sec": 66574.47324323654, "step_time_sec": 8.233457847964019, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8003, "loss": 4.029276371002197, "lr": 0.0001713209408814811, "elapsed_sec": 66582.70551681519, "step_time_sec": 8.232136244070716, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8004, "loss": 3.9419610500335693, "lr": 0.00017126666282426232, "elapsed_sec": 66590.94186782837, "step_time_sec": 8.236187722999603, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8005, "loss": 4.031802654266357, "lr": 0.00017121234232709355, "elapsed_sec": 66599.17355561256, "step_time_sec": 8.23152812209446, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8006, "loss": 3.982752799987793, "lr": 0.0001711579794234991, "elapsed_sec": 66607.40650033951, "step_time_sec": 8.232777026947588, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8007, "loss": 4.052074909210205, "lr": 0.0001711035741470296, "elapsed_sec": 66615.64036297798, "step_time_sec": 8.233687817002647, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8008, "loss": 4.130804538726807, "lr": 0.00017104912653126162, "elapsed_sec": 66623.87389802933, "step_time_sec": 8.233369127032347, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8009, "loss": 4.0373616218566895, "lr": 0.0001709946366097981, "elapsed_sec": 66632.10703897476, "step_time_sec": 8.232942527043633, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8010, "loss": 3.94659686088562, "lr": 0.0001709401044162679, "elapsed_sec": 66640.33998775482, "step_time_sec": 8.232775061042048, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8011, "loss": 4.0069403648376465, "lr": 0.00017088552998432608, "elapsed_sec": 66648.57806658745, "step_time_sec": 8.237923404900357, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8012, "loss": 4.062084674835205, "lr": 0.00017083091334765374, "elapsed_sec": 66656.81682944298, "step_time_sec": 8.238623942947015, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8013, "loss": 4.037883281707764, "lr": 0.000170776254539958, "elapsed_sec": 66665.05561041832, "step_time_sec": 8.238585126004182, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8014, "loss": 3.9491188526153564, "lr": 0.00017072155359497198, "elapsed_sec": 66673.29234361649, "step_time_sec": 8.236545993946493, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8015, "loss": 3.903566598892212, "lr": 0.00017066681054645491, "elapsed_sec": 66681.53334975243, "step_time_sec": 8.240817685960792, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8016, "loss": 3.8971590995788574, "lr": 0.00017061202542819194, "elapsed_sec": 66689.7723441124, "step_time_sec": 8.238880889024585, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8017, "loss": 3.9710421562194824, "lr": 0.00017055719827399415, "elapsed_sec": 66698.01160120964, "step_time_sec": 8.239085632027127, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8018, "loss": 3.8696320056915283, "lr": 0.00017050232911769864, "elapsed_sec": 66706.25132226944, "step_time_sec": 8.239501832984388, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8019, "loss": 3.993074655532837, "lr": 0.00017044741799316835, "elapsed_sec": 66714.48981976509, "step_time_sec": 8.23834325291682, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8020, "loss": 4.046352863311768, "lr": 0.0001703924649342922, "elapsed_sec": 66722.72860884666, "step_time_sec": 8.238629373954609, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8021, "loss": 3.997844696044922, "lr": 0.00017033746997498493, "elapsed_sec": 66730.96858096123, "step_time_sec": 8.239873360027559, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8022, "loss": 3.91964054107666, "lr": 0.00017028243314918718, "elapsed_sec": 66739.20879077911, "step_time_sec": 8.24002178106457, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8023, "loss": 3.8972535133361816, "lr": 0.00017022735449086535, "elapsed_sec": 66747.45157647133, "step_time_sec": 8.242627673083916, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8024, "loss": 3.828913450241089, "lr": 0.00017017223403401183, "elapsed_sec": 66755.6918785572, "step_time_sec": 8.240089764934964, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8025, "loss": 4.055810928344727, "lr": 0.00017011707181264457, "elapsed_sec": 66763.93102145195, "step_time_sec": 8.239001150010154, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8026, "loss": 3.900284767150879, "lr": 0.00017006186786080755, "elapsed_sec": 66772.16545557976, "step_time_sec": 8.234255359973758, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8027, "loss": 3.9621853828430176, "lr": 0.00017000662221257026, "elapsed_sec": 66780.40345692635, "step_time_sec": 8.237804948003031, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8028, "loss": 4.068194389343262, "lr": 0.0001699513349020281, "elapsed_sec": 66788.6430068016, "step_time_sec": 8.239391410024837, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8029, "loss": 4.1078782081604, "lr": 0.00016989600596330212, "elapsed_sec": 66796.88151168823, "step_time_sec": 8.238394801039249, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8030, "loss": 3.8878276348114014, "lr": 0.000169840635430539, "elapsed_sec": 66805.12191104889, "step_time_sec": 8.240190890035592, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8031, "loss": 3.974545955657959, "lr": 0.00016978522333791122, "elapsed_sec": 66813.36145877838, "step_time_sec": 8.239418759010732, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8032, "loss": 4.027039051055908, "lr": 0.00016972976971961682, "elapsed_sec": 66821.60034370422, "step_time_sec": 8.238675517961383, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8033, "loss": 3.9361987113952637, "lr": 0.00016967427460987946, "elapsed_sec": 66829.84065294266, "step_time_sec": 8.240129500976764, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8034, "loss": 3.983686685562134, "lr": 0.00016961873804294847, "elapsed_sec": 66838.07376670837, "step_time_sec": 8.23299368005246, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8035, "loss": 3.9799232482910156, "lr": 0.00016956316005309867, "elapsed_sec": 66846.30784702301, "step_time_sec": 8.233878963044845, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8036, "loss": 4.031805038452148, "lr": 0.0001695075406746306, "elapsed_sec": 66854.54118800163, "step_time_sec": 8.233192220912315, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8037, "loss": 3.904049873352051, "lr": 0.00016945187994187017, "elapsed_sec": 66862.7811331749, "step_time_sec": 8.239781248033978, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8038, "loss": 3.9850616455078125, "lr": 0.00016939617788916896, "elapsed_sec": 66871.01892185211, "step_time_sec": 8.23767059110105, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8039, "loss": 3.945091485977173, "lr": 0.0001693404345509039, "elapsed_sec": 66879.25714969635, "step_time_sec": 8.238071831059642, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8040, "loss": 4.154594898223877, "lr": 0.00016928464996147753, "elapsed_sec": 66887.49725651741, "step_time_sec": 8.239886702038348, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8041, "loss": 4.142704486846924, "lr": 0.00016922882415531778, "elapsed_sec": 66895.7353477478, "step_time_sec": 8.237952096038498, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8042, "loss": 4.100566864013672, "lr": 0.00016917295716687808, "elapsed_sec": 66903.97539138794, "step_time_sec": 8.239900021930225, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8043, "loss": 3.895878314971924, "lr": 0.00016911704903063715, "elapsed_sec": 66912.21564936638, "step_time_sec": 8.240086368052289, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8044, "loss": 4.013742446899414, "lr": 0.0001690610997810993, "elapsed_sec": 66920.45471596718, "step_time_sec": 8.238937870948575, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8045, "loss": 4.0698041915893555, "lr": 0.00016900510945279395, "elapsed_sec": 66928.6941792965, "step_time_sec": 8.239261719980277, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8046, "loss": 3.9106407165527344, "lr": 0.00016894907808027616, "elapsed_sec": 66936.93402934074, "step_time_sec": 8.239743212936446, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8047, "loss": 3.9586076736450195, "lr": 0.00016889300569812612, "elapsed_sec": 66945.17345285416, "step_time_sec": 8.239222441101447, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8048, "loss": 4.073708534240723, "lr": 0.00016883689234094943, "elapsed_sec": 66953.40719389915, "step_time_sec": 8.233597901999019, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8049, "loss": 4.334839344024658, "lr": 0.0001687807380433769, "elapsed_sec": 66961.64513778687, "step_time_sec": 8.237743948004209, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8050, "loss": 4.211881160736084, "lr": 0.00016872454284006468, "elapsed_sec": 66969.8851044178, "step_time_sec": 8.239794805995189, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8051, "loss": 4.035243988037109, "lr": 0.0001686683067656941, "elapsed_sec": 66978.12341141701, "step_time_sec": 8.238172670011409, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8052, "loss": 3.967625617980957, "lr": 0.00016861202985497177, "elapsed_sec": 66986.36252522469, "step_time_sec": 8.238936953945085, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8053, "loss": 3.9868650436401367, "lr": 0.00016855571214262948, "elapsed_sec": 66994.60133910179, "step_time_sec": 8.238653624081053, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8054, "loss": 4.012800216674805, "lr": 0.0001684993536634242, "elapsed_sec": 67002.84027171135, "step_time_sec": 8.238788632093929, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8055, "loss": 4.0418381690979, "lr": 0.00016844295445213802, "elapsed_sec": 67011.0741136074, "step_time_sec": 8.23365911806468, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8056, "loss": 4.001042366027832, "lr": 0.0001683865145435783, "elapsed_sec": 67019.3111615181, "step_time_sec": 8.236891962005757, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8057, "loss": 3.8773374557495117, "lr": 0.00016833003397257737, "elapsed_sec": 67027.54991579056, "step_time_sec": 8.23863071098458, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8058, "loss": 3.815748691558838, "lr": 0.00016827351277399268, "elapsed_sec": 67035.78932881355, "step_time_sec": 8.23926024697721, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8059, "loss": 3.8922510147094727, "lr": 0.00016821695098270684, "elapsed_sec": 67044.02336478233, "step_time_sec": 8.233865955960937, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8060, "loss": 3.9573001861572266, "lr": 0.0001681603486336274, "elapsed_sec": 67052.25592446327, "step_time_sec": 8.232364525902085, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8061, "loss": 3.969155788421631, "lr": 0.00016810370576168706, "elapsed_sec": 67060.48914122581, "step_time_sec": 8.233051848015748, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8062, "loss": 4.10532808303833, "lr": 0.00016804702240184343, "elapsed_sec": 67068.72878217697, "step_time_sec": 8.23950240504928, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8063, "loss": 3.9971725940704346, "lr": 0.00016799029858907913, "elapsed_sec": 67076.96699428558, "step_time_sec": 8.238041257951409, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8064, "loss": 3.962886095046997, "lr": 0.0001679335343584018, "elapsed_sec": 67085.20717215538, "step_time_sec": 8.239975751028396, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8065, "loss": 4.015621185302734, "lr": 0.00016787672974484394, "elapsed_sec": 67093.44577646255, "step_time_sec": 8.238488130038604, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8066, "loss": 3.955733299255371, "lr": 0.0001678198847834631, "elapsed_sec": 67101.68519330025, "step_time_sec": 8.23920451302547, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8067, "loss": 4.064785003662109, "lr": 0.00016776299950934151, "elapsed_sec": 67109.92402935028, "step_time_sec": 8.2386580419261, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8068, "loss": 4.0604987144470215, "lr": 0.0001677060739575865, "elapsed_sec": 67118.16270470619, "step_time_sec": 8.23853892995976, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8069, "loss": 4.049531936645508, "lr": 0.0001676491081633302, "elapsed_sec": 67126.40167212486, "step_time_sec": 8.2387880430324, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8070, "loss": 4.043437480926514, "lr": 0.00016759210216172953, "elapsed_sec": 67134.63994383812, "step_time_sec": 8.238125376054086, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8071, "loss": 4.075193405151367, "lr": 0.0001675350559879662, "elapsed_sec": 67142.87942171097, "step_time_sec": 8.239299839013256, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8072, "loss": 3.8698835372924805, "lr": 0.0001674779696772468, "elapsed_sec": 67151.11856365204, "step_time_sec": 8.239021079032682, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8073, "loss": 3.911945104598999, "lr": 0.00016742084326480262, "elapsed_sec": 67159.3561038971, "step_time_sec": 8.237388815032318, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8074, "loss": 4.216342926025391, "lr": 0.00016736367678588974, "elapsed_sec": 67167.596098423, "step_time_sec": 8.239861254929565, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8075, "loss": 3.910693407058716, "lr": 0.00016730647027578897, "elapsed_sec": 67175.8352394104, "step_time_sec": 8.23893401992973, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8076, "loss": 3.9660568237304688, "lr": 0.00016724922376980584, "elapsed_sec": 67184.07460308075, "step_time_sec": 8.23919878504239, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8077, "loss": 3.933084726333618, "lr": 0.00016719193730327045, "elapsed_sec": 67192.30777215958, "step_time_sec": 8.233036253019236, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8078, "loss": 4.049871444702148, "lr": 0.00016713461091153766, "elapsed_sec": 67200.54044318199, "step_time_sec": 8.232514658011496, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8079, "loss": 4.081300735473633, "lr": 0.00016707724462998704, "elapsed_sec": 67208.77457690239, "step_time_sec": 8.234004579950124, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8080, "loss": 4.13040828704834, "lr": 0.0001670198384940226, "elapsed_sec": 67217.01447916031, "step_time_sec": 8.23974839807488, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8081, "loss": 4.097232818603516, "lr": 0.0001669623925390731, "elapsed_sec": 67225.25416278839, "step_time_sec": 8.239440305973403, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8082, "loss": 3.9894847869873047, "lr": 0.0001669049068005918, "elapsed_sec": 67233.49202561378, "step_time_sec": 8.237728421925567, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8083, "loss": 4.025264739990234, "lr": 0.00016684738131405653, "elapsed_sec": 67241.73212981224, "step_time_sec": 8.239924611989409, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8084, "loss": 3.876084804534912, "lr": 0.00016678981611496964, "elapsed_sec": 67249.97259497643, "step_time_sec": 8.240343251964077, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8085, "loss": 4.13810920715332, "lr": 0.00016673221123885798, "elapsed_sec": 67258.2117152214, "step_time_sec": 8.238899238058366, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8086, "loss": 4.051367282867432, "lr": 0.00016667456672127295, "elapsed_sec": 67266.46055293083, "step_time_sec": 8.240730496007018, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8087, "loss": 4.008640766143799, "lr": 0.00016661688259779034, "elapsed_sec": 67274.70170521736, "step_time_sec": 8.240928187966347, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8088, "loss": 4.1228928565979, "lr": 0.0001665591589040104, "elapsed_sec": 67282.93996930122, "step_time_sec": 8.23810869990848, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8089, "loss": 4.081700801849365, "lr": 0.0001665013956755579, "elapsed_sec": 67291.17914175987, "step_time_sec": 8.239055097918026, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8090, "loss": 4.002479553222656, "lr": 0.0001664435929480818, "elapsed_sec": 67299.41813015938, "step_time_sec": 8.238813460920937, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8091, "loss": 3.9923787117004395, "lr": 0.00016638575075725564, "elapsed_sec": 67307.65780472755, "step_time_sec": 8.239551187958568, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8092, "loss": 4.102980613708496, "lr": 0.00016632786913877722, "elapsed_sec": 67315.89692854881, "step_time_sec": 8.238935936009511, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8093, "loss": 4.128912448883057, "lr": 0.0001662699481283687, "elapsed_sec": 67324.13488483429, "step_time_sec": 8.237778641050681, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8094, "loss": 3.9787750244140625, "lr": 0.0001662119877617765, "elapsed_sec": 67332.37556362152, "step_time_sec": 8.240531874005683, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8095, "loss": 4.128488063812256, "lr": 0.0001661539880747714, "elapsed_sec": 67340.6163725853, "step_time_sec": 8.240660440991633, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8096, "loss": 4.071262836456299, "lr": 0.00016609594910314842, "elapsed_sec": 67348.85491514206, "step_time_sec": 8.238383230986074, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8097, "loss": 3.768571138381958, "lr": 0.00016603787088272684, "elapsed_sec": 67357.0880355835, "step_time_sec": 8.23297345195897, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8098, "loss": 4.206910133361816, "lr": 0.00016597975344935007, "elapsed_sec": 67365.32131719589, "step_time_sec": 8.233077725977637, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8099, "loss": 4.0262651443481445, "lr": 0.00016592159683888588, "elapsed_sec": 67373.55476164818, "step_time_sec": 8.233308984083124, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8100, "loss": 4.111379623413086, "lr": 0.00016586340108722608, "elapsed_sec": 67381.79218816757, "step_time_sec": 8.237232161918655, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8101, "loss": 4.067412376403809, "lr": 0.0001658051662302867, "elapsed_sec": 67390.03092718124, "step_time_sec": 8.238616668968461, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8102, "loss": 3.983187437057495, "lr": 0.0001657468923040079, "elapsed_sec": 67398.26979541779, "step_time_sec": 8.238655989989638, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8103, "loss": 3.9992547035217285, "lr": 0.00016568857934435395, "elapsed_sec": 67406.50832533836, "step_time_sec": 8.238373263040558, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8104, "loss": 4.024644374847412, "lr": 0.0001656302273873132, "elapsed_sec": 67414.74683594704, "step_time_sec": 8.238370305043645, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8105, "loss": 4.274021625518799, "lr": 0.00016557183646889807, "elapsed_sec": 67422.98565530777, "step_time_sec": 8.23862022999674, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8106, "loss": 4.072221279144287, "lr": 0.000165513406625145, "elapsed_sec": 67431.22463965416, "step_time_sec": 8.238834909978323, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8107, "loss": 4.030247211456299, "lr": 0.00016545493789211455, "elapsed_sec": 67439.4634809494, "step_time_sec": 8.238686292083003, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8108, "loss": 4.1733198165893555, "lr": 0.00016539643030589115, "elapsed_sec": 67447.70362830162, "step_time_sec": 8.239993260009214, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8109, "loss": 4.065905570983887, "lr": 0.0001653378839025833, "elapsed_sec": 67455.94265246391, "step_time_sec": 8.238826319924556, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8110, "loss": 4.0966877937316895, "lr": 0.00016527929871832343, "elapsed_sec": 67464.18002080917, "step_time_sec": 8.237206417019479, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8111, "loss": 4.0502400398254395, "lr": 0.0001652206747892679, "elapsed_sec": 67483.91248822212, "step_time_sec": 19.732351226964965, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8112, "loss": 4.054891586303711, "lr": 0.00016516201215159695, "elapsed_sec": 67492.14530205727, "step_time_sec": 8.232598886010237, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8113, "loss": 4.014624118804932, "lr": 0.0001651033108415148, "elapsed_sec": 67500.37733507156, "step_time_sec": 8.231896238983609, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8114, "loss": 4.075477123260498, "lr": 0.00016504457089524944, "elapsed_sec": 67508.61425995827, "step_time_sec": 8.236797024030238, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8115, "loss": 4.060040473937988, "lr": 0.00016498579234905272, "elapsed_sec": 67516.84746432304, "step_time_sec": 8.233026519068517, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8116, "loss": 3.9968273639678955, "lr": 0.0001649269752392004, "elapsed_sec": 67525.08079576492, "step_time_sec": 8.233117841999047, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8117, "loss": 4.138562202453613, "lr": 0.00016486811960199192, "elapsed_sec": 67533.31236553192, "step_time_sec": 8.231419130926952, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8118, "loss": 4.061086177825928, "lr": 0.00016480922547375056, "elapsed_sec": 67541.54683852196, "step_time_sec": 8.234319463022985, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8119, "loss": 3.8870737552642822, "lr": 0.0001647502928908234, "elapsed_sec": 67549.78060150146, "step_time_sec": 8.233600831008516, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8120, "loss": 4.083533763885498, "lr": 0.00016469132188958116, "elapsed_sec": 67558.01415252686, "step_time_sec": 8.233375353040174, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8121, "loss": 4.336026668548584, "lr": 0.00016463231250641832, "elapsed_sec": 67566.2477991581, "step_time_sec": 8.233495685039088, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8122, "loss": 4.157345294952393, "lr": 0.000164573264777753, "elapsed_sec": 67574.48118543625, "step_time_sec": 8.233208123012446, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8123, "loss": 5.175028324127197, "lr": 0.0001645141787400271, "elapsed_sec": 67582.71325707436, "step_time_sec": 8.23194780794438, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8124, "loss": 4.426313400268555, "lr": 0.00016445505442970603, "elapsed_sec": 67590.9457757473, "step_time_sec": 8.23235871992074, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8125, "loss": 4.109997749328613, "lr": 0.0001643958918832789, "elapsed_sec": 67599.17865633965, "step_time_sec": 8.232689714990556, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8126, "loss": 4.212658405303955, "lr": 0.0001643366911372584, "elapsed_sec": 67607.41195368767, "step_time_sec": 8.233168978011236, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8127, "loss": 4.2358174324035645, "lr": 0.00016427745222818076, "elapsed_sec": 67615.64539813995, "step_time_sec": 8.23326799296774, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8128, "loss": 4.132638931274414, "lr": 0.00016421817519260586, "elapsed_sec": 67623.87921071053, "step_time_sec": 8.233646569075063, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8129, "loss": 4.125500679016113, "lr": 0.00016415886006711694, "elapsed_sec": 67632.11888194084, "step_time_sec": 8.239484282094054, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8130, "loss": 4.281131744384766, "lr": 0.00016409950688832094, "elapsed_sec": 67640.35943102837, "step_time_sec": 8.240396564011462, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8131, "loss": 4.075543403625488, "lr": 0.00016404011569284816, "elapsed_sec": 67648.59760856628, "step_time_sec": 8.238061834010296, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8132, "loss": 4.0961456298828125, "lr": 0.0001639806865173524, "elapsed_sec": 67656.83877158165, "step_time_sec": 8.241017747088335, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8133, "loss": 3.9886162281036377, "lr": 0.0001639212193985109, "elapsed_sec": 67665.07864379883, "step_time_sec": 8.239640702027828, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8134, "loss": 4.144123077392578, "lr": 0.00016386171437302429, "elapsed_sec": 67673.31767988205, "step_time_sec": 8.23889156198129, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8135, "loss": 4.072083473205566, "lr": 0.00016380217147761667, "elapsed_sec": 67681.557315588, "step_time_sec": 8.239441595040262, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8136, "loss": 4.0790696144104, "lr": 0.00016374259074903542, "elapsed_sec": 67689.79447817802, "step_time_sec": 8.237015212071128, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8137, "loss": 4.154244422912598, "lr": 0.00016368297222405136, "elapsed_sec": 67698.0343439579, "step_time_sec": 8.239709464949556, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8138, "loss": 3.990003824234009, "lr": 0.0001636233159394585, "elapsed_sec": 67706.27302503586, "step_time_sec": 8.238490783027373, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8139, "loss": 4.13132381439209, "lr": 0.0001635636219320743, "elapsed_sec": 67714.51051688194, "step_time_sec": 8.237333515076898, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8140, "loss": 4.168988227844238, "lr": 0.00016350389023873945, "elapsed_sec": 67722.75061559677, "step_time_sec": 8.23996866599191, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8141, "loss": 4.161397457122803, "lr": 0.0001634441208963179, "elapsed_sec": 67730.98967003822, "step_time_sec": 8.238862431026064, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8142, "loss": 4.115817546844482, "lr": 0.00016338431394169673, "elapsed_sec": 67739.23011875153, "step_time_sec": 8.240292728994973, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8143, "loss": 4.05761194229126, "lr": 0.00016332446941178644, "elapsed_sec": 67747.46994066238, "step_time_sec": 8.239694788935594, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8144, "loss": 3.8802452087402344, "lr": 0.00016326458734352054, "elapsed_sec": 67755.70729088783, "step_time_sec": 8.237215331057087, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8145, "loss": 4.2282562255859375, "lr": 0.0001632046677738558, "elapsed_sec": 67763.9413614273, "step_time_sec": 8.233898976934142, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8146, "loss": 4.226682186126709, "lr": 0.00016314471073977213, "elapsed_sec": 67772.17448163033, "step_time_sec": 8.232873979024589, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8147, "loss": 4.102356433868408, "lr": 0.00016308471627827245, "elapsed_sec": 67780.40832662582, "step_time_sec": 8.23368569102604, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8148, "loss": 4.269361972808838, "lr": 0.00016302468442638298, "elapsed_sec": 67788.64082622528, "step_time_sec": 8.232374408049509, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8149, "loss": 4.293551921844482, "lr": 0.00016296461522115284, "elapsed_sec": 67796.87414956093, "step_time_sec": 8.23320320493076, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8150, "loss": 4.071702003479004, "lr": 0.0001629045086996543, "elapsed_sec": 67805.1063053608, "step_time_sec": 8.231926776934415, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8151, "loss": 4.476241588592529, "lr": 0.0001628443648989826, "elapsed_sec": 67813.33970403671, "step_time_sec": 8.233287741080858, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8152, "loss": 4.152519226074219, "lr": 0.00016278418385625606, "elapsed_sec": 67821.57312393188, "step_time_sec": 8.233201588969678, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8153, "loss": 4.1324028968811035, "lr": 0.0001627239656086159, "elapsed_sec": 67829.8071346283, "step_time_sec": 8.23383606504649, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8154, "loss": 4.177358627319336, "lr": 0.00016266371019322631, "elapsed_sec": 67838.04594111443, "step_time_sec": 8.238674812018871, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8155, "loss": 4.2325592041015625, "lr": 0.00016260341764727453, "elapsed_sec": 67846.28522658348, "step_time_sec": 8.23908885801211, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8156, "loss": 3.9272260665893555, "lr": 0.00016254308800797056, "elapsed_sec": 67854.5247194767, "step_time_sec": 8.239313906989992, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8157, "loss": 4.164758205413818, "lr": 0.0001624827213125474, "elapsed_sec": 67862.76402115822, "step_time_sec": 8.239184887032025, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8158, "loss": 4.063919544219971, "lr": 0.00016242231759826088, "elapsed_sec": 67871.00500011444, "step_time_sec": 8.240770664997399, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8159, "loss": 3.8967440128326416, "lr": 0.00016236187690238968, "elapsed_sec": 67879.24404144287, "step_time_sec": 8.238944620010443, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8160, "loss": 3.9230856895446777, "lr": 0.00016230139926223534, "elapsed_sec": 67887.48376011848, "step_time_sec": 8.239515452063642, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8161, "loss": 4.1337056159973145, "lr": 0.00016224088471512212, "elapsed_sec": 67895.72400569916, "step_time_sec": 8.240112168947235, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8162, "loss": 3.8567399978637695, "lr": 0.00016218033329839715, "elapsed_sec": 67903.96246170998, "step_time_sec": 8.238297987030819, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8163, "loss": 4.087108135223389, "lr": 0.00016211974504943023, "elapsed_sec": 67912.20276427269, "step_time_sec": 8.240087832091376, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8164, "loss": 4.053261756896973, "lr": 0.00016205912000561391, "elapsed_sec": 67920.44020366669, "step_time_sec": 8.237270359997638, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8165, "loss": 4.227436065673828, "lr": 0.00016199845820436356, "elapsed_sec": 67928.67974734306, "step_time_sec": 8.239453047979623, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8166, "loss": 4.205412864685059, "lr": 0.00016193775968311705, "elapsed_sec": 67936.91964244843, "step_time_sec": 8.239722294034436, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8167, "loss": 4.0697550773620605, "lr": 0.00016187702447933506, "elapsed_sec": 67945.15827202797, "step_time_sec": 8.238431444042362, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8168, "loss": 3.8663601875305176, "lr": 0.00016181625263050079, "elapsed_sec": 67953.39812850952, "step_time_sec": 8.239706737920642, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8169, "loss": 4.08713960647583, "lr": 0.0001617554441741202, "elapsed_sec": 67961.6393084526, "step_time_sec": 8.240967377903871, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8170, "loss": 4.121424198150635, "lr": 0.00016169459914772174, "elapsed_sec": 67969.87913060188, "step_time_sec": 8.239656341029331, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8171, "loss": 4.0382914543151855, "lr": 0.00016163371758885642, "elapsed_sec": 67978.11921334267, "step_time_sec": 8.23993470496498, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8172, "loss": 3.920246124267578, "lr": 0.00016157279953509784, "elapsed_sec": 67986.35699439049, "step_time_sec": 8.237568655982614, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8173, "loss": 4.158060073852539, "lr": 0.00016151184502404217, "elapsed_sec": 67994.5971262455, "step_time_sec": 8.239991359063424, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8174, "loss": 4.060757637023926, "lr": 0.00016145085409330793, "elapsed_sec": 68002.837403059, "step_time_sec": 8.240115693071857, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8175, "loss": 4.108434200286865, "lr": 0.00016138982678053624, "elapsed_sec": 68011.07623744011, "step_time_sec": 8.238698197063059, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8176, "loss": 4.026072978973389, "lr": 0.0001613287631233907, "elapsed_sec": 68019.3156542778, "step_time_sec": 8.239258158020675, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8177, "loss": 3.968248128890991, "lr": 0.00016126766315955723, "elapsed_sec": 68027.55458903313, "step_time_sec": 8.238712201011367, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8178, "loss": 4.072576522827148, "lr": 0.00016120652692674418, "elapsed_sec": 68035.79477000237, "step_time_sec": 8.2400443010265, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8179, "loss": 4.314075469970703, "lr": 0.00016114535446268237, "elapsed_sec": 68044.03492498398, "step_time_sec": 8.24000586103648, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8180, "loss": 4.236372470855713, "lr": 0.00016108414580512492, "elapsed_sec": 68052.27245020866, "step_time_sec": 8.23732039600145, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8181, "loss": 4.076773643493652, "lr": 0.00016102290099184724, "elapsed_sec": 68060.51357388496, "step_time_sec": 8.240994491963647, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8182, "loss": 4.047784805297852, "lr": 0.0001609616200606472, "elapsed_sec": 68068.75352501869, "step_time_sec": 8.239736800896935, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8183, "loss": 4.090132713317871, "lr": 0.00016090030304934474, "elapsed_sec": 68076.9919347763, "step_time_sec": 8.238321184064262, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8184, "loss": 4.230189800262451, "lr": 0.00016083894999578224, "elapsed_sec": 68085.23042178154, "step_time_sec": 8.238269404042512, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8185, "loss": 4.182285308837891, "lr": 0.00016077756093782437, "elapsed_sec": 68093.4683368206, "step_time_sec": 8.237753004999831, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8186, "loss": 3.9215805530548096, "lr": 0.00016071613591335778, "elapsed_sec": 68101.70895886421, "step_time_sec": 8.2404996099649, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8187, "loss": 4.044232368469238, "lr": 0.00016065467496029155, "elapsed_sec": 68109.94841647148, "step_time_sec": 8.239263935014606, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8188, "loss": 3.9879848957061768, "lr": 0.00016059317811655685, "elapsed_sec": 68118.1864566803, "step_time_sec": 8.237871079007164, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8189, "loss": 4.012095928192139, "lr": 0.00016053164542010695, "elapsed_sec": 68126.4204390049, "step_time_sec": 8.233785295044072, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8190, "loss": 3.8981404304504395, "lr": 0.00016047007690891735, "elapsed_sec": 68134.65360069275, "step_time_sec": 8.232981028035283, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8191, "loss": 3.930332660675049, "lr": 0.00016040847262098552, "elapsed_sec": 68142.88745832443, "step_time_sec": 8.233674101065844, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8192, "loss": 4.074502944946289, "lr": 0.00016034683259433114, "elapsed_sec": 68151.120190382, "step_time_sec": 8.232586206053384, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8193, "loss": 3.9905049800872803, "lr": 0.00016028515686699593, "elapsed_sec": 68159.35306477547, "step_time_sec": 8.232735074940138, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8194, "loss": 4.041805744171143, "lr": 0.00016022344547704355, "elapsed_sec": 68167.59072208405, "step_time_sec": 8.237461999990046, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8195, "loss": 4.109440326690674, "lr": 0.00016016169846255974, "elapsed_sec": 68175.83025622368, "step_time_sec": 8.239405898959376, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8196, "loss": 4.069119453430176, "lr": 0.0001600999158616522, "elapsed_sec": 68184.06861567497, "step_time_sec": 8.238208212074824, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8197, "loss": 4.292605400085449, "lr": 0.0001600380977124506, "elapsed_sec": 68192.30932879448, "step_time_sec": 8.240464637987316, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8198, "loss": 3.997354745864868, "lr": 0.00015997624405310662, "elapsed_sec": 68200.54859495163, "step_time_sec": 8.239191117929295, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8199, "loss": 4.167558670043945, "lr": 0.0001599143549217937, "elapsed_sec": 68208.78740024567, "step_time_sec": 8.238560492056422, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8200, "loss": 4.06497049331665, "lr": 0.00015985243035670734, "elapsed_sec": 68217.02657318115, "step_time_sec": 8.239085208973847, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8201, "loss": 4.084473609924316, "lr": 0.00015979047039606478, "elapsed_sec": 68225.26538276672, "step_time_sec": 8.238570727990009, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8202, "loss": 4.099015712738037, "lr": 0.00015972847507810515, "elapsed_sec": 68233.50683093071, "step_time_sec": 8.241339911008254, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8203, "loss": 4.125532150268555, "lr": 0.00015966644444108947, "elapsed_sec": 68241.74852395058, "step_time_sec": 8.241483191028237, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8204, "loss": 3.9971728324890137, "lr": 0.0001596043785233004, "elapsed_sec": 68249.9860804081, "step_time_sec": 8.23740901995916, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8205, "loss": 4.0371832847595215, "lr": 0.0001595422773630425, "elapsed_sec": 68258.22575759888, "step_time_sec": 8.239505775040016, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8206, "loss": 4.039443492889404, "lr": 0.00015948014099864208, "elapsed_sec": 68266.45849227905, "step_time_sec": 8.232536796014756, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8207, "loss": 4.106745719909668, "lr": 0.0001594179694684471, "elapsed_sec": 68274.69253468513, "step_time_sec": 8.233854475896806, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8208, "loss": 4.078768730163574, "lr": 0.0001593557628108273, "elapsed_sec": 68282.92580938339, "step_time_sec": 8.233093505026773, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8209, "loss": 4.048995018005371, "lr": 0.00015929352106417406, "elapsed_sec": 68291.16579651833, "step_time_sec": 8.239899631007574, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8210, "loss": 4.197391510009766, "lr": 0.00015923124426690043, "elapsed_sec": 68299.40672540665, "step_time_sec": 8.240765577997081, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8211, "loss": 4.202535629272461, "lr": 0.00015916893245744102, "elapsed_sec": 68307.64470815659, "step_time_sec": 8.237796521978453, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8212, "loss": 4.063549995422363, "lr": 0.00015910658567425218, "elapsed_sec": 68315.8855202198, "step_time_sec": 8.240628356928937, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8213, "loss": 4.0707621574401855, "lr": 0.00015904420395581174, "elapsed_sec": 68324.12453341484, "step_time_sec": 8.238845994928852, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8214, "loss": 3.9575045108795166, "lr": 0.00015898178734061912, "elapsed_sec": 68332.3644182682, "step_time_sec": 8.239725696039386, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8215, "loss": 4.018185138702393, "lr": 0.00015891933586719534, "elapsed_sec": 68340.60297226906, "step_time_sec": 8.23839652701281, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8216, "loss": 3.9329802989959717, "lr": 0.0001588568495740828, "elapsed_sec": 68348.84184074402, "step_time_sec": 8.238683274947107, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8217, "loss": 4.00993013381958, "lr": 0.0001587943284998455, "elapsed_sec": 68357.07494282722, "step_time_sec": 8.232984254020266, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8218, "loss": 3.8856041431427, "lr": 0.00015873177268306888, "elapsed_sec": 68365.30795454979, "step_time_sec": 8.232778861070983, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8219, "loss": 4.066990375518799, "lr": 0.0001586691821623598, "elapsed_sec": 68373.54158186913, "step_time_sec": 8.233484874013811, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8220, "loss": 4.084809303283691, "lr": 0.00015860655697634652, "elapsed_sec": 68381.7809484005, "step_time_sec": 8.239163454039954, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8221, "loss": 4.064135551452637, "lr": 0.0001585438971636788, "elapsed_sec": 68390.02166199684, "step_time_sec": 8.24055752705317, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8222, "loss": 4.225200176239014, "lr": 0.0001584812027630276, "elapsed_sec": 68398.26123952866, "step_time_sec": 8.239454632042907, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8223, "loss": 4.033573627471924, "lr": 0.0001584184738130854, "elapsed_sec": 68406.49967980385, "step_time_sec": 8.238242366001941, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8224, "loss": 4.137914180755615, "lr": 0.00015835571035256588, "elapsed_sec": 68414.74042892456, "step_time_sec": 8.240588713088073, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8225, "loss": 4.144988059997559, "lr": 0.00015829291242020407, "elapsed_sec": 68422.98017191887, "step_time_sec": 8.239593064994551, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8226, "loss": 3.888219118118286, "lr": 0.00015823008005475626, "elapsed_sec": 68431.21986103058, "step_time_sec": 8.239504454075359, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8227, "loss": 4.094244956970215, "lr": 0.000158167213295, "elapsed_sec": 68439.46014046669, "step_time_sec": 8.24014467804227, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8228, "loss": 3.921241521835327, "lr": 0.00015810431217973408, "elapsed_sec": 68447.69246006012, "step_time_sec": 8.232172523043118, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8229, "loss": 3.9974026679992676, "lr": 0.00015804137674777844, "elapsed_sec": 68455.92536687851, "step_time_sec": 8.232655996922404, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8230, "loss": 3.93857479095459, "lr": 0.00015797840703797426, "elapsed_sec": 68464.15815281868, "step_time_sec": 8.232603506068699, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8231, "loss": 4.066547870635986, "lr": 0.00015791540308918387, "elapsed_sec": 68472.39846968651, "step_time_sec": 8.240158514934592, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8232, "loss": 3.97078537940979, "lr": 0.00015785236494029068, "elapsed_sec": 68480.63791513443, "step_time_sec": 8.239310899982229, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8233, "loss": 4.123042106628418, "lr": 0.0001577892926301992, "elapsed_sec": 68488.87748718262, "step_time_sec": 8.239354729070328, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8234, "loss": 4.028440952301025, "lr": 0.00015772618619783516, "elapsed_sec": 68497.11682510376, "step_time_sec": 8.239215341047384, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8235, "loss": 4.0909600257873535, "lr": 0.00015766304568214516, "elapsed_sec": 68505.34873628616, "step_time_sec": 8.231697358074598, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8236, "loss": 3.919024705886841, "lr": 0.00015759987112209698, "elapsed_sec": 68513.58272957802, "step_time_sec": 8.233864009031095, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8237, "loss": 4.074502944946289, "lr": 0.0001575366625566793, "elapsed_sec": 68521.81966090202, "step_time_sec": 8.236764044035226, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8238, "loss": 4.225866317749023, "lr": 0.00015747342002490186, "elapsed_sec": 68530.06021237373, "step_time_sec": 8.240407908917405, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8239, "loss": 4.325612545013428, "lr": 0.00015741014356579537, "elapsed_sec": 68538.30206775665, "step_time_sec": 8.241671712952666, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8240, "loss": 3.912898540496826, "lr": 0.0001573468332184115, "elapsed_sec": 68546.54128026962, "step_time_sec": 8.239031760022044, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8241, "loss": 4.2437357902526855, "lr": 0.00015728348902182266, "elapsed_sec": 68554.78118562698, "step_time_sec": 8.239780520903878, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8242, "loss": 4.168990612030029, "lr": 0.00015722011101512235, "elapsed_sec": 68563.02007555962, "step_time_sec": 8.238678095978685, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8243, "loss": 4.26850700378418, "lr": 0.0001571566992374249, "elapsed_sec": 68571.26063013077, "step_time_sec": 8.24039139400702, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8244, "loss": 4.078539848327637, "lr": 0.00015709325372786532, "elapsed_sec": 68579.50154399872, "step_time_sec": 8.240818853024393, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8245, "loss": 4.086050510406494, "lr": 0.0001570297745255997, "elapsed_sec": 68587.73367977142, "step_time_sec": 8.231929154018871, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8246, "loss": 4.013908386230469, "lr": 0.00015696626166980472, "elapsed_sec": 68595.96903371811, "step_time_sec": 8.235176748945378, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8247, "loss": 4.047311305999756, "lr": 0.00015690271519967793, "elapsed_sec": 68604.2010781765, "step_time_sec": 8.231897171004675, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8248, "loss": 4.277853488922119, "lr": 0.00015683913515443756, "elapsed_sec": 68612.44139194489, "step_time_sec": 8.24015532899648, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8249, "loss": 4.173303127288818, "lr": 0.00015677552157332262, "elapsed_sec": 68620.67908930779, "step_time_sec": 8.237520150956698, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8250, "loss": 4.0649495124816895, "lr": 0.00015671187449559275, "elapsed_sec": 68628.91856050491, "step_time_sec": 8.239318343927152, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8251, "loss": 4.184493541717529, "lr": 0.00015664819396052834, "elapsed_sec": 68637.15667843819, "step_time_sec": 8.2379853669554, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8252, "loss": 4.09155797958374, "lr": 0.0001565844800074304, "elapsed_sec": 68645.38952565193, "step_time_sec": 8.232600065995939, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8253, "loss": 4.11077880859375, "lr": 0.0001565207326756205, "elapsed_sec": 68653.62264156342, "step_time_sec": 8.233026791014709, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8254, "loss": 4.234591960906982, "lr": 0.0001564569520044409, "elapsed_sec": 68661.85692119598, "step_time_sec": 8.234067825018428, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8255, "loss": 4.0713396072387695, "lr": 0.00015639313803325446, "elapsed_sec": 68670.09749722481, "step_time_sec": 8.24033427296672, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8256, "loss": 4.025774002075195, "lr": 0.00015632929080144442, "elapsed_sec": 68678.33051347733, "step_time_sec": 8.232890406041406, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8257, "loss": 4.043542861938477, "lr": 0.00015626541034841475, "elapsed_sec": 68686.56457281113, "step_time_sec": 8.233875451958738, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8258, "loss": 4.177127838134766, "lr": 0.00015620149671358976, "elapsed_sec": 68694.79696297646, "step_time_sec": 8.232244477956556, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8259, "loss": 4.117354869842529, "lr": 0.00015613754993641437, "elapsed_sec": 68703.02981829643, "step_time_sec": 8.23263393889647, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8260, "loss": 4.109019756317139, "lr": 0.0001560735700563539, "elapsed_sec": 68711.26245641708, "step_time_sec": 8.232515979907475, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8261, "loss": 4.053259372711182, "lr": 0.00015600955711289404, "elapsed_sec": 68719.50198626518, "step_time_sec": 8.239335519028828, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8262, "loss": 4.20973014831543, "lr": 0.00015594551114554099, "elapsed_sec": 68727.74044704437, "step_time_sec": 8.238285482977517, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8263, "loss": 4.236515522003174, "lr": 0.00015588143219382125, "elapsed_sec": 68735.97971248627, "step_time_sec": 8.239092894014902, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8264, "loss": 4.024833679199219, "lr": 0.00015581732029728174, "elapsed_sec": 68744.21908092499, "step_time_sec": 8.239230955950916, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8265, "loss": 4.093031883239746, "lr": 0.00015575317549548965, "elapsed_sec": 68752.4583439827, "step_time_sec": 8.23907417501323, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8266, "loss": 4.165078639984131, "lr": 0.00015568899782803249, "elapsed_sec": 68760.69798588753, "step_time_sec": 8.239499703980982, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8267, "loss": 3.862856388092041, "lr": 0.00015562478733451815, "elapsed_sec": 68768.9369866848, "step_time_sec": 8.23884005099535, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8268, "loss": 4.0086259841918945, "lr": 0.00015556054405457462, "elapsed_sec": 68777.17676472664, "step_time_sec": 8.239654756034724, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8269, "loss": 4.047714710235596, "lr": 0.00015549626802785024, "elapsed_sec": 68785.4152648449, "step_time_sec": 8.238264465006068, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8270, "loss": 4.267788887023926, "lr": 0.0001554319592940135, "elapsed_sec": 68793.65355420113, "step_time_sec": 8.23817554290872, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8271, "loss": 4.216213226318359, "lr": 0.00015536761789275316, "elapsed_sec": 68801.8921341896, "step_time_sec": 8.238400414935313, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8272, "loss": 4.222320079803467, "lr": 0.000155303243863778, "elapsed_sec": 68810.12990450859, "step_time_sec": 8.237586067989469, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8273, "loss": 4.164577007293701, "lr": 0.00015523883724681708, "elapsed_sec": 68818.3631503582, "step_time_sec": 8.233110603992827, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8274, "loss": 4.136516571044922, "lr": 0.00015517439808161946, "elapsed_sec": 68826.59511327744, "step_time_sec": 8.231844483059831, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8275, "loss": 4.230520248413086, "lr": 0.0001551099264079544, "elapsed_sec": 68834.82986497879, "step_time_sec": 8.234572743065655, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8276, "loss": 4.066407203674316, "lr": 0.0001550454222656111, "elapsed_sec": 68843.06792616844, "step_time_sec": 8.237855775048956, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8277, "loss": 4.121504306793213, "lr": 0.0001549808856943989, "elapsed_sec": 68851.30818414688, "step_time_sec": 8.240136893931776, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8278, "loss": 4.017115592956543, "lr": 0.00015491631673414705, "elapsed_sec": 68859.5489730835, "step_time_sec": 8.24056358099915, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8279, "loss": 4.07236909866333, "lr": 0.00015485171542470495, "elapsed_sec": 68867.78769302368, "step_time_sec": 8.238593547022901, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8280, "loss": 4.141556262969971, "lr": 0.00015478708180594178, "elapsed_sec": 68876.02857494354, "step_time_sec": 8.240692234016024, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8281, "loss": 3.981466293334961, "lr": 0.0001547224159177468, "elapsed_sec": 68884.26806521416, "step_time_sec": 8.2393296220107, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8282, "loss": 4.037571907043457, "lr": 0.00015465771780002907, "elapsed_sec": 68892.50787687302, "step_time_sec": 8.239641014952213, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8283, "loss": 4.11475944519043, "lr": 0.00015459298749271764, "elapsed_sec": 68900.7424724102, "step_time_sec": 8.234475903911516, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8284, "loss": 4.135166645050049, "lr": 0.00015452822503576142, "elapsed_sec": 68908.97586965561, "step_time_sec": 8.233210197999142, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8285, "loss": 4.25974178314209, "lr": 0.00015446343046912907, "elapsed_sec": 68917.20946359634, "step_time_sec": 8.23345928103663, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8286, "loss": 4.093780517578125, "lr": 0.00015439860383280912, "elapsed_sec": 68925.44232082367, "step_time_sec": 8.232617784989998, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8287, "loss": 4.1377410888671875, "lr": 0.00015433374516681, "elapsed_sec": 68933.6759493351, "step_time_sec": 8.23350162699353, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8288, "loss": 3.9568657875061035, "lr": 0.00015426885451115971, "elapsed_sec": 68941.9089372158, "step_time_sec": 8.232801108038984, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8289, "loss": 3.9525399208068848, "lr": 0.00015420393190590607, "elapsed_sec": 68950.14954829216, "step_time_sec": 8.240490057971328, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8290, "loss": 4.246408939361572, "lr": 0.00015413897739111672, "elapsed_sec": 68958.38760995865, "step_time_sec": 8.237842605100013, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8291, "loss": 4.0320353507995605, "lr": 0.00015407399100687884, "elapsed_sec": 68966.62792944908, "step_time_sec": 8.240223321015947, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8292, "loss": 4.155069351196289, "lr": 0.0001540089727932994, "elapsed_sec": 68974.86788749695, "step_time_sec": 8.23980343400035, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8293, "loss": 4.277579307556152, "lr": 0.00015394392279050492, "elapsed_sec": 68983.10755372047, "step_time_sec": 8.23945862101391, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8294, "loss": 4.047975540161133, "lr": 0.00015387884103864162, "elapsed_sec": 68991.34165740013, "step_time_sec": 8.233930180082098, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8295, "loss": 4.039866924285889, "lr": 0.00015381372757787527, "elapsed_sec": 68999.57456493378, "step_time_sec": 8.232738967053592, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8296, "loss": 3.9591972827911377, "lr": 0.00015374858244839113, "elapsed_sec": 69007.81438064575, "step_time_sec": 8.239660823019221, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8297, "loss": 4.061697483062744, "lr": 0.0001536834056903942, "elapsed_sec": 69016.04741454124, "step_time_sec": 8.23286469199229, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8298, "loss": 3.9852283000946045, "lr": 0.00015361819734410886, "elapsed_sec": 69024.28083968163, "step_time_sec": 8.233266265015118, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8299, "loss": 4.117328643798828, "lr": 0.00015355295744977896, "elapsed_sec": 69032.51268982887, "step_time_sec": 8.231686142040417, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8300, "loss": 3.9315032958984375, "lr": 0.00015348768604766795, "elapsed_sec": 69040.74654769897, "step_time_sec": 8.233772578067146, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8301, "loss": 4.191080570220947, "lr": 0.00015342238317805855, "elapsed_sec": 69048.97993540764, "step_time_sec": 8.23318356310483, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8302, "loss": 4.087571620941162, "lr": 0.0001533570488812531, "elapsed_sec": 69057.22048783302, "step_time_sec": 8.24038675695192, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8303, "loss": 4.023418426513672, "lr": 0.0001532916831975732, "elapsed_sec": 69065.45863485336, "step_time_sec": 8.237923298031092, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8304, "loss": 4.053438663482666, "lr": 0.00015322628616735987, "elapsed_sec": 69073.6986951828, "step_time_sec": 8.239965353976004, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8305, "loss": 4.0266828536987305, "lr": 0.00015316085783097341, "elapsed_sec": 69081.93858456612, "step_time_sec": 8.239706129068509, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8306, "loss": 3.915318727493286, "lr": 0.00015309539822879354, "elapsed_sec": 69090.17736911774, "step_time_sec": 8.23861752008088, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8307, "loss": 4.19285249710083, "lr": 0.00015302990740121924, "elapsed_sec": 69098.417740345, "step_time_sec": 8.24016395304352, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8308, "loss": 4.057403087615967, "lr": 0.00015296438538866869, "elapsed_sec": 69106.65604567528, "step_time_sec": 8.238215417019092, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8309, "loss": 3.880587339401245, "lr": 0.0001528988322315794, "elapsed_sec": 69114.89556789398, "step_time_sec": 8.239336582948454, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8310, "loss": 3.819849729537964, "lr": 0.0001528332479704081, "elapsed_sec": 69123.13576817513, "step_time_sec": 8.239982531988062, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8311, "loss": 4.033807277679443, "lr": 0.00015276763264563073, "elapsed_sec": 69131.37415361404, "step_time_sec": 8.238274139002897, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8312, "loss": 3.8456766605377197, "lr": 0.00015270198629774225, "elapsed_sec": 69139.61446857452, "step_time_sec": 8.240145583054982, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8313, "loss": 4.0895795822143555, "lr": 0.00015263630896725694, "elapsed_sec": 69147.85368871689, "step_time_sec": 8.239051553071477, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8314, "loss": 3.9586637020111084, "lr": 0.0001525706006947082, "elapsed_sec": 69156.09333634377, "step_time_sec": 8.239458073978312, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8315, "loss": 3.9995718002319336, "lr": 0.0001525048615206484, "elapsed_sec": 69164.33446335793, "step_time_sec": 8.241033751983196, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8316, "loss": 4.026051044464111, "lr": 0.00015243909148564898, "elapsed_sec": 69172.57299876213, "step_time_sec": 8.23828061495442, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8317, "loss": 3.9495818614959717, "lr": 0.00015237329063030067, "elapsed_sec": 69180.81225085258, "step_time_sec": 8.2391122370027, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8318, "loss": 4.023978233337402, "lr": 0.00015230745899521292, "elapsed_sec": 69189.04499435425, "step_time_sec": 8.232607576064765, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8319, "loss": 3.941344738006592, "lr": 0.00015224159662101435, "elapsed_sec": 69197.27940607071, "step_time_sec": 8.234275992959738, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8320, "loss": 3.9232864379882812, "lr": 0.00015217570354835244, "elapsed_sec": 69205.51351356506, "step_time_sec": 8.23387920402456, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8321, "loss": 4.107198238372803, "lr": 0.0001521097798178938, "elapsed_sec": 69213.75229120255, "step_time_sec": 8.238609691965394, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8322, "loss": 3.898949146270752, "lr": 0.0001520438254703237, "elapsed_sec": 69221.99114608765, "step_time_sec": 8.238739717984572, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8323, "loss": 3.900632619857788, "lr": 0.00015197784054634653, "elapsed_sec": 69230.22882533073, "step_time_sec": 8.23750042100437, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8324, "loss": 4.121777534484863, "lr": 0.0001519118250866855, "elapsed_sec": 69238.46770262718, "step_time_sec": 8.238679072004743, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8325, "loss": 4.0236921310424805, "lr": 0.0001518457791320825, "elapsed_sec": 69246.70524072647, "step_time_sec": 8.237434015027247, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8326, "loss": 3.9605581760406494, "lr": 0.00015177970272329847, "elapsed_sec": 69254.94510102272, "step_time_sec": 8.239695752039552, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8327, "loss": 4.0208845138549805, "lr": 0.00015171359590111303, "elapsed_sec": 69263.18476009369, "step_time_sec": 8.239437735988759, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8328, "loss": 3.9862656593322754, "lr": 0.00015164745870632459, "elapsed_sec": 69271.42320346832, "step_time_sec": 8.23824480001349, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8329, "loss": 4.032589912414551, "lr": 0.00015158129117975023, "elapsed_sec": 69279.6576602459, "step_time_sec": 8.234288666048087, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8330, "loss": 4.074527740478516, "lr": 0.00015151509336222595, "elapsed_sec": 69287.89010834694, "step_time_sec": 8.23227993992623, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8331, "loss": 3.9948415756225586, "lr": 0.00015144886529460622, "elapsed_sec": 69296.12461805344, "step_time_sec": 8.234340531984344, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8332, "loss": 3.975517988204956, "lr": 0.0001513826070177643, "elapsed_sec": 69304.35692286491, "step_time_sec": 8.232156506041065, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8333, "loss": 4.119324684143066, "lr": 0.00015131631857259205, "elapsed_sec": 69312.5896320343, "step_time_sec": 8.232567596016452, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8334, "loss": 4.1082611083984375, "lr": 0.00015125000000000002, "elapsed_sec": 69320.8230035305, "step_time_sec": 8.233181430026889, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8335, "loss": 4.064830780029297, "lr": 0.00015118365134091723, "elapsed_sec": 69329.06393814087, "step_time_sec": 8.240780613967218, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8336, "loss": 4.040891647338867, "lr": 0.0001511172726362914, "elapsed_sec": 69337.30358862877, "step_time_sec": 8.239538108929992, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8337, "loss": 4.254116058349609, "lr": 0.00015105086392708868, "elapsed_sec": 69345.54295420647, "step_time_sec": 8.239131378941238, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8338, "loss": 3.888734817504883, "lr": 0.00015098442525429378, "elapsed_sec": 69353.78194570541, "step_time_sec": 8.238801916944794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8339, "loss": 4.075085639953613, "lr": 0.00015091795665891, "elapsed_sec": 69362.01918077469, "step_time_sec": 8.237105048960075, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8340, "loss": 4.044060230255127, "lr": 0.00015085145818195892, "elapsed_sec": 69370.2574326992, "step_time_sec": 8.23811872699298, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8341, "loss": 4.123946666717529, "lr": 0.0001507849298644807, "elapsed_sec": 69378.49736356735, "step_time_sec": 8.239762766985223, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8342, "loss": 4.133993625640869, "lr": 0.00015071837174753387, "elapsed_sec": 69386.7364461422, "step_time_sec": 8.238907975959592, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8343, "loss": 4.118891716003418, "lr": 0.00015065178387219533, "elapsed_sec": 69394.97563672066, "step_time_sec": 8.238923884928226, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8344, "loss": 3.969709873199463, "lr": 0.00015058516627956047, "elapsed_sec": 69403.21382808685, "step_time_sec": 8.238027689047158, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8345, "loss": 4.063067436218262, "lr": 0.00015051851901074281, "elapsed_sec": 69411.45306396484, "step_time_sec": 8.239084851928055, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8346, "loss": 4.008944034576416, "lr": 0.00015045184210687432, "elapsed_sec": 69419.68571472168, "step_time_sec": 8.232517171069048, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8347, "loss": 3.929692506790161, "lr": 0.00015038513560910528, "elapsed_sec": 69427.91888666153, "step_time_sec": 8.233023984008469, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8348, "loss": 4.103063583374023, "lr": 0.0001503183995586042, "elapsed_sec": 69436.15177965164, "step_time_sec": 8.232706005917862, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8349, "loss": 4.189934253692627, "lr": 0.0001502516339965577, "elapsed_sec": 69444.3921289444, "step_time_sec": 8.240207071066834, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8350, "loss": 4.064574241638184, "lr": 0.00015018483896417088, "elapsed_sec": 69452.63067984581, "step_time_sec": 8.238410706981085, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8351, "loss": 4.157996654510498, "lr": 0.00015011801450266684, "elapsed_sec": 69460.86979103088, "step_time_sec": 8.238905136939138, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8352, "loss": 4.051051139831543, "lr": 0.00015005116065328685, "elapsed_sec": 69469.1085870266, "step_time_sec": 8.238662433926947, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8353, "loss": 4.149816513061523, "lr": 0.00014998427745729036, "elapsed_sec": 69477.34636449814, "step_time_sec": 8.237646116991527, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8354, "loss": 3.9389448165893555, "lr": 0.00014991736495595492, "elapsed_sec": 69485.58630919456, "step_time_sec": 8.239717934047803, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8355, "loss": 3.9368090629577637, "lr": 0.0001498504231905762, "elapsed_sec": 69493.82539844513, "step_time_sec": 8.238970955018885, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8356, "loss": 4.0508341789245605, "lr": 0.00014978345220246787, "elapsed_sec": 69502.0638756752, "step_time_sec": 8.238237303099595, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8357, "loss": 4.084775924682617, "lr": 0.00014971645203296164, "elapsed_sec": 69510.2970302105, "step_time_sec": 8.233052258961834, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8358, "loss": 4.027275085449219, "lr": 0.00014964942272340727, "elapsed_sec": 69518.52938270569, "step_time_sec": 8.232189486036077, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8359, "loss": 4.0845441818237305, "lr": 0.00014958236431517256, "elapsed_sec": 69526.763671875, "step_time_sec": 8.234149695024826, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8360, "loss": 4.052775859832764, "lr": 0.00014951527684964309, "elapsed_sec": 69535.00227761269, "step_time_sec": 8.238365067983977, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8361, "loss": 3.935758113861084, "lr": 0.00014944816036822252, "elapsed_sec": 69543.24187803268, "step_time_sec": 8.239469654043205, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8362, "loss": 4.114671230316162, "lr": 0.00014938101491233238, "elapsed_sec": 69551.47978520393, "step_time_sec": 8.237747533945367, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8363, "loss": 3.9728336334228516, "lr": 0.00014931384052341209, "elapsed_sec": 69559.7198035717, "step_time_sec": 8.239827662007883, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8364, "loss": 4.003920555114746, "lr": 0.0001492466372429189, "elapsed_sec": 69567.95904636383, "step_time_sec": 8.239092630916275, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8365, "loss": 4.024953365325928, "lr": 0.0001491794051123279, "elapsed_sec": 69576.19778251648, "step_time_sec": 8.238558217999525, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8366, "loss": 4.0721235275268555, "lr": 0.000149112144173132, "elapsed_sec": 69584.4370071888, "step_time_sec": 8.239096197998151, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8367, "loss": 3.57171368598938, "lr": 0.00014904485446684188, "elapsed_sec": 69592.67583465576, "step_time_sec": 8.238629342988133, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8368, "loss": 3.9472949504852295, "lr": 0.00014897753603498596, "elapsed_sec": 69600.91605973244, "step_time_sec": 8.240051271975972, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8369, "loss": 4.007511615753174, "lr": 0.0001489101889191104, "elapsed_sec": 69609.15536904335, "step_time_sec": 8.23916738491971, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8370, "loss": 4.050489902496338, "lr": 0.00014884281316077908, "elapsed_sec": 69617.38868308067, "step_time_sec": 8.233204531949013, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8371, "loss": 4.036271572113037, "lr": 0.00014877540880157356, "elapsed_sec": 69625.62799572945, "step_time_sec": 8.239118637051433, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8372, "loss": 3.992954969406128, "lr": 0.00014870797588309293, "elapsed_sec": 69633.86761951447, "step_time_sec": 8.239405966945924, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8373, "loss": 4.094409942626953, "lr": 0.00014864051444695415, "elapsed_sec": 69642.1080904007, "step_time_sec": 8.240278529934585, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8374, "loss": 3.9445788860321045, "lr": 0.0001485730245347916, "elapsed_sec": 69650.34738516808, "step_time_sec": 8.239161927020177, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8375, "loss": 3.9760477542877197, "lr": 0.0001485055061882572, "elapsed_sec": 69658.58486151695, "step_time_sec": 8.237310010939837, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8376, "loss": 4.037055492401123, "lr": 0.00014843795944902055, "elapsed_sec": 69666.8233101368, "step_time_sec": 8.238289437955245, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8377, "loss": 3.978332281112671, "lr": 0.0001483703843587687, "elapsed_sec": 69675.0624268055, "step_time_sec": 8.238883621059358, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8378, "loss": 4.00816011428833, "lr": 0.00014830278095920622, "elapsed_sec": 69683.30129742622, "step_time_sec": 8.238742560963146, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8379, "loss": 3.9573421478271484, "lr": 0.0001482351492920551, "elapsed_sec": 69691.53976178169, "step_time_sec": 8.23828838497866, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8380, "loss": 3.950518846511841, "lr": 0.00014816748939905483, "elapsed_sec": 69699.7783112526, "step_time_sec": 8.238389723934233, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8381, "loss": 4.164855480194092, "lr": 0.00014809980132196238, "elapsed_sec": 69708.01822185516, "step_time_sec": 8.23975063196849, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8382, "loss": 4.041792869567871, "lr": 0.000148032085102552, "elapsed_sec": 69716.25066709518, "step_time_sec": 8.232293921988457, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8383, "loss": 4.036248207092285, "lr": 0.00014796434078261528, "elapsed_sec": 69724.48401212692, "step_time_sec": 8.233199583017267, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8384, "loss": 4.1308064460754395, "lr": 0.0001478965684039613, "elapsed_sec": 69732.71588253975, "step_time_sec": 8.231680119992234, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8385, "loss": 4.025559902191162, "lr": 0.00014782876800841638, "elapsed_sec": 69740.95722150803, "step_time_sec": 8.241233099019155, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8386, "loss": 3.8772096633911133, "lr": 0.0001477609396378241, "elapsed_sec": 69749.1961581707, "step_time_sec": 8.238718896987848, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8387, "loss": 3.9587879180908203, "lr": 0.00014769308333404532, "elapsed_sec": 69757.43423366547, "step_time_sec": 8.237966925953515, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8388, "loss": 4.005922317504883, "lr": 0.0001476251991389582, "elapsed_sec": 69765.67288303375, "step_time_sec": 8.238460437045433, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8389, "loss": 3.8436074256896973, "lr": 0.00014755728709445805, "elapsed_sec": 69773.91086912155, "step_time_sec": 8.237803044030443, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8390, "loss": 3.9920177459716797, "lr": 0.00014748934724245736, "elapsed_sec": 69782.15061450005, "step_time_sec": 8.239602403016761, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8391, "loss": 3.9735519886016846, "lr": 0.00014742137962488584, "elapsed_sec": 69790.38949155807, "step_time_sec": 8.238764293026179, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8392, "loss": 4.099327087402344, "lr": 0.00014735338428369026, "elapsed_sec": 69798.62808465958, "step_time_sec": 8.238426431897096, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8393, "loss": 4.1726508140563965, "lr": 0.0001472853612608346, "elapsed_sec": 69806.86231946945, "step_time_sec": 8.23405922902748, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8394, "loss": 3.989915132522583, "lr": 0.00014721731059829984, "elapsed_sec": 69815.09648537636, "step_time_sec": 8.23392650205642, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8395, "loss": 3.976860523223877, "lr": 0.00014714923233808403, "elapsed_sec": 69823.33041071892, "step_time_sec": 8.233805010095239, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8396, "loss": 3.9264330863952637, "lr": 0.00014708112652220227, "elapsed_sec": 69831.56319642067, "step_time_sec": 8.232582633965649, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8397, "loss": 4.045966148376465, "lr": 0.00014701299319268667, "elapsed_sec": 69839.79651165009, "step_time_sec": 8.233164872974157, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8398, "loss": 3.8057339191436768, "lr": 0.00014694483239158633, "elapsed_sec": 69848.03009486198, "step_time_sec": 8.233349688001908, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8399, "loss": 3.928269147872925, "lr": 0.00014687664416096724, "elapsed_sec": 69856.26564908028, "step_time_sec": 8.235413531889208, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8400, "loss": 3.8033621311187744, "lr": 0.00014680842854291243, "elapsed_sec": 69864.4991095066, "step_time_sec": 8.233355494914576, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8401, "loss": 4.266016006469727, "lr": 0.0001467401855795217, "elapsed_sec": 69872.73379659653, "step_time_sec": 8.23449297004845, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8402, "loss": 4.084186553955078, "lr": 0.00014667191531291183, "elapsed_sec": 69880.96681237221, "step_time_sec": 8.232867271988653, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8403, "loss": 3.9176251888275146, "lr": 0.0001466036177852164, "elapsed_sec": 69889.20074725151, "step_time_sec": 8.23375183402095, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8404, "loss": 3.950124740600586, "lr": 0.0001465352930385859, "elapsed_sec": 69897.43466830254, "step_time_sec": 8.233769521000795, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8405, "loss": 3.9626047611236572, "lr": 0.00014646694111518743, "elapsed_sec": 69905.66801691055, "step_time_sec": 8.233150757965632, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8406, "loss": 3.965149402618408, "lr": 0.00014639856205720503, "elapsed_sec": 69913.90131807327, "step_time_sec": 8.233143912046216, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8407, "loss": 4.08961820602417, "lr": 0.00014633015590683948, "elapsed_sec": 69922.141903162, "step_time_sec": 8.240404569078237, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8408, "loss": 3.9468605518341064, "lr": 0.0001462617227063082, "elapsed_sec": 69930.37382388115, "step_time_sec": 8.231748724007048, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8409, "loss": 4.015987873077393, "lr": 0.00014619326249784535, "elapsed_sec": 69938.6076824665, "step_time_sec": 8.233685459010303, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8410, "loss": 3.965832233428955, "lr": 0.00014612477532370167, "elapsed_sec": 69946.83987903595, "step_time_sec": 8.232080045039766, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8411, "loss": 3.9720823764801025, "lr": 0.00014605626122614476, "elapsed_sec": 69955.07474517822, "step_time_sec": 8.234670388977975, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8412, "loss": 3.9274332523345947, "lr": 0.0001459877202474586, "elapsed_sec": 69963.30903935432, "step_time_sec": 8.234081314993091, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8413, "loss": 3.8730709552764893, "lr": 0.0001459191524299439, "elapsed_sec": 69971.54116725922, "step_time_sec": 8.231996158952825, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8414, "loss": 4.164553165435791, "lr": 0.00014585055781591788, "elapsed_sec": 69979.77446246147, "step_time_sec": 8.233151244930923, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8415, "loss": 3.9294304847717285, "lr": 0.00014578193644771434, "elapsed_sec": 69988.00778365135, "step_time_sec": 8.233119216980413, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8416, "loss": 3.8735463619232178, "lr": 0.00014571328836768347, "elapsed_sec": 69996.24042248726, "step_time_sec": 8.232461279025301, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8417, "loss": 3.9549403190612793, "lr": 0.0001456446136181921, "elapsed_sec": 70004.47367429733, "step_time_sec": 8.233089137007482, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8418, "loss": 3.932623863220215, "lr": 0.00014557591224162347, "elapsed_sec": 70012.7059314251, "step_time_sec": 8.232085965923034, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8419, "loss": 3.933202028274536, "lr": 0.00014550718428037721, "elapsed_sec": 70020.94575428963, "step_time_sec": 8.239686507964507, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8420, "loss": 3.919262409210205, "lr": 0.00014543842977686935, "elapsed_sec": 70029.18414664268, "step_time_sec": 8.238226272980683, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8421, "loss": 3.850957155227661, "lr": 0.0001453696487735324, "elapsed_sec": 70037.42425990105, "step_time_sec": 8.239976437063888, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8422, "loss": 3.7186121940612793, "lr": 0.00014530084131281516, "elapsed_sec": 70045.66401982307, "step_time_sec": 8.239607442053966, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8423, "loss": 3.8812525272369385, "lr": 0.00014523200743718273, "elapsed_sec": 70053.9034819603, "step_time_sec": 8.239217704045586, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8424, "loss": 3.985614538192749, "lr": 0.0001451631471891165, "elapsed_sec": 70062.14235281944, "step_time_sec": 8.238692162092775, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8425, "loss": 3.9175355434417725, "lr": 0.00014509426061111425, "elapsed_sec": 70070.38137960434, "step_time_sec": 8.238879319978878, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8426, "loss": 3.8419039249420166, "lr": 0.00014502534774568992, "elapsed_sec": 70078.62190818787, "step_time_sec": 8.240380765986629, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8427, "loss": 3.938584804534912, "lr": 0.00014495640863537365, "elapsed_sec": 70086.86149144173, "step_time_sec": 8.239473798079416, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8428, "loss": 4.031978607177734, "lr": 0.00014488744332271186, "elapsed_sec": 70095.10158729553, "step_time_sec": 8.239900314016268, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8429, "loss": 3.9980127811431885, "lr": 0.0001448184518502671, "elapsed_sec": 70103.34172296524, "step_time_sec": 8.239963555010036, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8430, "loss": 3.8866400718688965, "lr": 0.00014474943426061806, "elapsed_sec": 70111.58092236519, "step_time_sec": 8.238999470020644, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8431, "loss": 4.021674633026123, "lr": 0.00014468039059635951, "elapsed_sec": 70119.82007408142, "step_time_sec": 8.239040210028179, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8432, "loss": 4.068121910095215, "lr": 0.00014461132090010236, "elapsed_sec": 70128.05953025818, "step_time_sec": 8.239314822014421, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8433, "loss": 3.921100616455078, "lr": 0.00014454222521447364, "elapsed_sec": 70136.29818511009, "step_time_sec": 8.238423479022458, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8434, "loss": 4.050067901611328, "lr": 0.00014447310358211635, "elapsed_sec": 70144.53928375244, "step_time_sec": 8.240960483904928, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8435, "loss": 3.84328031539917, "lr": 0.00014440395604568945, "elapsed_sec": 70152.77988672256, "step_time_sec": 8.240421811002307, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8436, "loss": 3.9075539112091064, "lr": 0.00014433478264786799, "elapsed_sec": 70161.01980495453, "step_time_sec": 8.239770118962042, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8437, "loss": 4.104640960693359, "lr": 0.00014426558343134293, "elapsed_sec": 70169.26097297668, "step_time_sec": 8.240978516987525, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8438, "loss": 4.115081310272217, "lr": 0.0001441963584388212, "elapsed_sec": 70177.50032639503, "step_time_sec": 8.23920467402786, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8439, "loss": 4.006830215454102, "lr": 0.00014412710771302556, "elapsed_sec": 70185.73966670036, "step_time_sec": 8.239183452911675, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8440, "loss": 4.055228233337402, "lr": 0.00014405783129669476, "elapsed_sec": 70193.98021149635, "step_time_sec": 8.240376573987305, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8441, "loss": 3.770993709564209, "lr": 0.00014398852923258333, "elapsed_sec": 70202.21905946732, "step_time_sec": 8.238691854989156, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8442, "loss": 3.9377970695495605, "lr": 0.00014391920156346162, "elapsed_sec": 70210.46092796326, "step_time_sec": 8.241700307000428, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8443, "loss": 4.025003433227539, "lr": 0.00014384984833211586, "elapsed_sec": 70218.69501852989, "step_time_sec": 8.233933530049399, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8444, "loss": 4.083452224731445, "lr": 0.0001437804695813479, "elapsed_sec": 70226.92903065681, "step_time_sec": 8.233846448012628, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8445, "loss": 3.9788362979888916, "lr": 0.00014371106535397563, "elapsed_sec": 70235.16351747513, "step_time_sec": 8.234272551024333, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8446, "loss": 3.980426073074341, "lr": 0.00014364163569283232, "elapsed_sec": 70243.3967256546, "step_time_sec": 8.233038404025137, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8447, "loss": 3.962052583694458, "lr": 0.00014357218064076714, "elapsed_sec": 70251.63033008575, "step_time_sec": 8.233437784016132, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8448, "loss": 3.9594812393188477, "lr": 0.00014350270024064492, "elapsed_sec": 70259.86432361603, "step_time_sec": 8.23384859203361, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8449, "loss": 4.038623332977295, "lr": 0.00014343319453534608, "elapsed_sec": 70268.10398030281, "step_time_sec": 8.239490646054037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8450, "loss": 3.8323521614074707, "lr": 0.00014336366356776663, "elapsed_sec": 70276.34396576881, "step_time_sec": 8.239800772047602, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8451, "loss": 4.125376224517822, "lr": 0.00014329410738081827, "elapsed_sec": 70284.5839343071, "step_time_sec": 8.23980733298231, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8452, "loss": 4.095417499542236, "lr": 0.00014322452601742817, "elapsed_sec": 70292.82284879684, "step_time_sec": 8.23874989990145, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8453, "loss": 3.9665422439575195, "lr": 0.00014315491952053916, "elapsed_sec": 70301.0618956089, "step_time_sec": 8.238908161991276, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8454, "loss": 3.855289936065674, "lr": 0.00014308528793310942, "elapsed_sec": 70309.30043578148, "step_time_sec": 8.23834733502008, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8455, "loss": 3.968147039413452, "lr": 0.00014301563129811272, "elapsed_sec": 70317.5389289856, "step_time_sec": 8.238286233041435, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8456, "loss": 3.933148145675659, "lr": 0.00014294594965853828, "elapsed_sec": 70325.77860856056, "step_time_sec": 8.239566246047616, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8457, "loss": 4.010982513427734, "lr": 0.00014287624305739073, "elapsed_sec": 70334.01853513718, "step_time_sec": 8.239802966010757, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8458, "loss": 4.023037433624268, "lr": 0.00014280651153769012, "elapsed_sec": 70342.25946521759, "step_time_sec": 8.240712922997773, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8459, "loss": 4.153665542602539, "lr": 0.00014273675514247182, "elapsed_sec": 70350.51561546326, "step_time_sec": 8.25597158295568, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8460, "loss": 3.921050786972046, "lr": 0.00014266697391478666, "elapsed_sec": 70358.75489068031, "step_time_sec": 8.239053667057306, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8461, "loss": 3.9197981357574463, "lr": 0.0001425971678977007, "elapsed_sec": 70366.986982584, "step_time_sec": 8.231956925941631, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8462, "loss": 3.9911422729492188, "lr": 0.00014252733713429535, "elapsed_sec": 70375.22111606598, "step_time_sec": 8.233945262036286, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8463, "loss": 3.9457027912139893, "lr": 0.00014245748166766728, "elapsed_sec": 70383.45314788818, "step_time_sec": 8.231846407055855, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8464, "loss": 4.0566229820251465, "lr": 0.00014238760154092843, "elapsed_sec": 70391.68757510185, "step_time_sec": 8.23428758000955, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8465, "loss": 4.073043346405029, "lr": 0.0001423176967972059, "elapsed_sec": 70399.92636394501, "step_time_sec": 8.238647267920896, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8466, "loss": 3.9559757709503174, "lr": 0.000142247767479642, "elapsed_sec": 70408.16675496101, "step_time_sec": 8.240208436967805, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8467, "loss": 3.944218873977661, "lr": 0.0001421778136313943, "elapsed_sec": 70416.4062564373, "step_time_sec": 8.239344901987351, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8468, "loss": 4.076536178588867, "lr": 0.00014210783529563535, "elapsed_sec": 70424.64574456215, "step_time_sec": 8.239315539947711, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8469, "loss": 3.884963274002075, "lr": 0.0001420378325155529, "elapsed_sec": 70432.88641905785, "step_time_sec": 8.240497104008682, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8470, "loss": 4.080539226531982, "lr": 0.00014196780533434986, "elapsed_sec": 70441.12515830994, "step_time_sec": 8.238585929968394, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8471, "loss": 4.033459186553955, "lr": 0.00014189775379524404, "elapsed_sec": 70449.36648726463, "step_time_sec": 8.241123689920641, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8472, "loss": 3.961164951324463, "lr": 0.0001418276779414684, "elapsed_sec": 70457.60727596283, "step_time_sec": 8.240610034088604, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8473, "loss": 4.017602443695068, "lr": 0.00014175757781627083, "elapsed_sec": 70465.84052848816, "step_time_sec": 8.233144544996321, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8474, "loss": 3.9374680519104004, "lr": 0.00014168745346291426, "elapsed_sec": 70474.07473754883, "step_time_sec": 8.234047911013477, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8475, "loss": 3.93697452545166, "lr": 0.00014161730492467662, "elapsed_sec": 70482.30748224258, "step_time_sec": 8.232532628928311, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8476, "loss": 4.0007147789001465, "lr": 0.00014154713224485057, "elapsed_sec": 70490.54111528397, "step_time_sec": 8.233491074992344, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8477, "loss": 4.210000991821289, "lr": 0.0001414769354667439, "elapsed_sec": 70498.77437448502, "step_time_sec": 8.23307402501814, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8478, "loss": 4.048171520233154, "lr": 0.0001414067146336791, "elapsed_sec": 70507.01397180557, "step_time_sec": 8.239485180005431, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8479, "loss": 3.8178470134735107, "lr": 0.0001413364697889936, "elapsed_sec": 70515.25292611122, "step_time_sec": 8.238773536053486, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8480, "loss": 4.056701183319092, "lr": 0.00014126620097603963, "elapsed_sec": 70523.49265408516, "step_time_sec": 8.239541033981368, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8481, "loss": 3.9646785259246826, "lr": 0.0001411959082381842, "elapsed_sec": 70531.7323524952, "step_time_sec": 8.239571734098718, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8482, "loss": 4.012657642364502, "lr": 0.0001411255916188091, "elapsed_sec": 70539.97144055367, "step_time_sec": 8.238900520955212, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8483, "loss": 3.914132595062256, "lr": 0.00014105525116131084, "elapsed_sec": 70548.21219778061, "step_time_sec": 8.240614362992346, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8484, "loss": 3.9639527797698975, "lr": 0.0001409848869091006, "elapsed_sec": 70556.45130896568, "step_time_sec": 8.238953104009852, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8485, "loss": 3.719233989715576, "lr": 0.0001409144989056044, "elapsed_sec": 70564.6912279129, "step_time_sec": 8.239723309059627, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8486, "loss": 3.9337682723999023, "lr": 0.00014084408719426276, "elapsed_sec": 70572.932087183, "step_time_sec": 8.240635773050599, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8487, "loss": 3.9419913291931152, "lr": 0.0001407736518185309, "elapsed_sec": 70581.17121839523, "step_time_sec": 8.239007016993128, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8488, "loss": 3.9664111137390137, "lr": 0.00014070319282187858, "elapsed_sec": 70589.4112124443, "step_time_sec": 8.239782329997979, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8489, "loss": 3.988401174545288, "lr": 0.0001406327102477902, "elapsed_sec": 70597.65042281151, "step_time_sec": 8.239027911913581, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8490, "loss": 3.807330846786499, "lr": 0.0001405622041397648, "elapsed_sec": 70605.88922333717, "step_time_sec": 8.238623464945704, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8491, "loss": 3.9333009719848633, "lr": 0.00014049167454131576, "elapsed_sec": 70614.12897849083, "step_time_sec": 8.23959141294472, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8492, "loss": 4.003763675689697, "lr": 0.000140421121495971, "elapsed_sec": 70622.36722397804, "step_time_sec": 8.238066553021781, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8493, "loss": 3.9908106327056885, "lr": 0.00014035054504727304, "elapsed_sec": 70630.60782527924, "step_time_sec": 8.240474810008891, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8494, "loss": 3.942307233810425, "lr": 0.0001402799452387787, "elapsed_sec": 70638.84816765785, "step_time_sec": 8.240197555045597, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8495, "loss": 3.9522531032562256, "lr": 0.0001402093221140593, "elapsed_sec": 70647.08679962158, "step_time_sec": 8.238477956969291, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8496, "loss": 3.852092742919922, "lr": 0.00014013867571670052, "elapsed_sec": 70655.32676148415, "step_time_sec": 8.239763343939558, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8497, "loss": 4.139208793640137, "lr": 0.0001400680060903024, "elapsed_sec": 70663.56566619873, "step_time_sec": 8.238717208965681, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8498, "loss": 4.175775051116943, "lr": 0.0001399973132784793, "elapsed_sec": 70671.80505371094, "step_time_sec": 8.239200116018765, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8499, "loss": 3.933176279067993, "lr": 0.00013992659732485992, "elapsed_sec": 70680.0437040329, "step_time_sec": 8.238525458029471, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8500, "loss": 3.9898996353149414, "lr": 0.00013985585827308727, "elapsed_sec": 70688.28264188766, "step_time_sec": 24.493630189914256, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": true, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8501, "loss": 4.1390380859375, "lr": 0.00013978509616681853, "elapsed_sec": 70712.7847931385, "step_time_sec": 8.246759126079269, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8502, "loss": 4.003602027893066, "lr": 0.00013971431104972519, "elapsed_sec": 70721.01758217812, "step_time_sec": 8.23266964405775, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8503, "loss": 4.047510147094727, "lr": 0.00013964350296549285, "elapsed_sec": 70729.24942350388, "step_time_sec": 8.2316393640358, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8504, "loss": 4.006047248840332, "lr": 0.00013957267195782141, "elapsed_sec": 70737.48151826859, "step_time_sec": 8.231879131053574, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8505, "loss": 3.9146955013275146, "lr": 0.00013950181807042484, "elapsed_sec": 70745.71341466904, "step_time_sec": 8.231702929944731, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8506, "loss": 3.8860018253326416, "lr": 0.00013943094134703118, "elapsed_sec": 70753.94814896584, "step_time_sec": 8.234602439915761, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8507, "loss": 3.947249412536621, "lr": 0.00013936004183138267, "elapsed_sec": 70762.18046474457, "step_time_sec": 8.232121899025515, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8508, "loss": 4.061490058898926, "lr": 0.00013928911956723555, "elapsed_sec": 70770.41405081749, "step_time_sec": 8.233402246027254, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8509, "loss": 4.141235828399658, "lr": 0.00013921817459836016, "elapsed_sec": 70778.64606261253, "step_time_sec": 8.231860903091729, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8510, "loss": 4.074986457824707, "lr": 0.00013914720696854072, "elapsed_sec": 70786.87881708145, "step_time_sec": 8.232609828934073, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8511, "loss": 4.0037407875061035, "lr": 0.00013907621672157563, "elapsed_sec": 70795.11153101921, "step_time_sec": 8.232481256010942, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8512, "loss": 3.8632872104644775, "lr": 0.0001390052039012771, "elapsed_sec": 70803.3524324894, "step_time_sec": 8.240768198040314, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8513, "loss": 3.918652057647705, "lr": 0.00013893416855147132, "elapsed_sec": 70811.59189653397, "step_time_sec": 8.239284005016088, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8514, "loss": 4.013270378112793, "lr": 0.00013886311071599836, "elapsed_sec": 70819.83292746544, "step_time_sec": 8.240883613936603, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8515, "loss": 3.86604642868042, "lr": 0.0001387920304387122, "elapsed_sec": 70828.07134771347, "step_time_sec": 8.238202237989753, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8516, "loss": 3.8063130378723145, "lr": 0.00013872092776348068, "elapsed_sec": 70836.30875205994, "step_time_sec": 8.23728940996807, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8517, "loss": 3.933867931365967, "lr": 0.00013864980273418538, "elapsed_sec": 70844.54692006111, "step_time_sec": 8.237956054974347, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8518, "loss": 3.9092090129852295, "lr": 0.0001385786553947218, "elapsed_sec": 70852.78629040718, "step_time_sec": 8.239220532006584, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8519, "loss": 4.040048599243164, "lr": 0.00013850748578899912, "elapsed_sec": 70861.02516198158, "step_time_sec": 8.238705707946792, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8520, "loss": 4.002757549285889, "lr": 0.00013843629396094028, "elapsed_sec": 70869.26490521431, "step_time_sec": 8.239577312022448, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8521, "loss": 4.230350017547607, "lr": 0.00013836507995448197, "elapsed_sec": 70877.50332188606, "step_time_sec": 8.238286657026038, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8522, "loss": 3.948920249938965, "lr": 0.00013829384381357451, "elapsed_sec": 70885.74478006363, "step_time_sec": 8.241325002978556, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8523, "loss": 3.98393177986145, "lr": 0.0001382225855821819, "elapsed_sec": 70893.98498511314, "step_time_sec": 8.240007344051264, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8524, "loss": 4.108191013336182, "lr": 0.00013815130530428187, "elapsed_sec": 70902.22381019592, "step_time_sec": 8.238695090054534, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8525, "loss": 3.948667526245117, "lr": 0.00013808000302386556, "elapsed_sec": 70910.4646654129, "step_time_sec": 8.240724449045956, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8526, "loss": 3.949575424194336, "lr": 0.00013800867878493783, "elapsed_sec": 70918.70330882072, "step_time_sec": 8.23848357796669, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8527, "loss": 3.9693188667297363, "lr": 0.00013793733263151712, "elapsed_sec": 70926.94264101982, "step_time_sec": 8.239197427988984, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8528, "loss": 3.9180781841278076, "lr": 0.00013786596460763526, "elapsed_sec": 70935.1830303669, "step_time_sec": 8.240141715039499, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8529, "loss": 3.9540011882781982, "lr": 0.00013779457475733765, "elapsed_sec": 70943.42107391357, "step_time_sec": 8.237891471013427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8530, "loss": 3.8872251510620117, "lr": 0.00013772316312468324, "elapsed_sec": 70951.66011977196, "step_time_sec": 8.238921026000753, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8531, "loss": 3.903197765350342, "lr": 0.00013765172975374428, "elapsed_sec": 70959.89856553078, "step_time_sec": 8.23826197499875, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8532, "loss": 3.9409477710723877, "lr": 0.00013758027468860654, "elapsed_sec": 70968.13811421394, "step_time_sec": 8.239397214027122, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8533, "loss": 3.8775441646575928, "lr": 0.0001375087979733691, "elapsed_sec": 70976.37706518173, "step_time_sec": 8.238836302072741, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8534, "loss": 3.9108898639678955, "lr": 0.00013743729965214448, "elapsed_sec": 70984.61562466621, "step_time_sec": 8.238372888998128, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8535, "loss": 3.8735594749450684, "lr": 0.00013736577976905854, "elapsed_sec": 70992.85604476929, "step_time_sec": 8.240257986006327, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8536, "loss": 4.009138584136963, "lr": 0.00013729423836825029, "elapsed_sec": 71001.0957968235, "step_time_sec": 8.239581239060499, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8537, "loss": 3.97377347946167, "lr": 0.00013722267549387218, "elapsed_sec": 71009.33469963074, "step_time_sec": 8.238744726986624, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8538, "loss": 3.856184244155884, "lr": 0.00013715109119008992, "elapsed_sec": 71017.5742828846, "step_time_sec": 8.239475214970298, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8539, "loss": 3.832453966140747, "lr": 0.00013707948550108234, "elapsed_sec": 71025.81325602531, "step_time_sec": 8.238761345972307, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8540, "loss": 3.9710779190063477, "lr": 0.0001370078584710415, "elapsed_sec": 71034.05408906937, "step_time_sec": 8.240662061958574, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8541, "loss": 3.9800643920898438, "lr": 0.0001369362101441727, "elapsed_sec": 71042.29426431656, "step_time_sec": 8.240005347994156, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8542, "loss": 3.9893383979797363, "lr": 0.00013686454056469432, "elapsed_sec": 71050.53157925606, "step_time_sec": 8.237214583903551, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8543, "loss": 3.872784376144409, "lr": 0.0001367928497768379, "elapsed_sec": 71058.77173924446, "step_time_sec": 8.240018912940286, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8544, "loss": 3.8511886596679688, "lr": 0.00013672113782484794, "elapsed_sec": 71067.01188158989, "step_time_sec": 8.239993416005746, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8545, "loss": 3.917339563369751, "lr": 0.00013664940475298221, "elapsed_sec": 71075.25067281723, "step_time_sec": 8.238589987973683, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8546, "loss": 4.038266658782959, "lr": 0.00013657765060551136, "elapsed_sec": 71083.49053263664, "step_time_sec": 8.239680608035997, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8547, "loss": 3.91483736038208, "lr": 0.00013650587542671908, "elapsed_sec": 71091.72854948044, "step_time_sec": 8.237883997964673, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8548, "loss": 3.8395228385925293, "lr": 0.0001364340792609021, "elapsed_sec": 71099.96854019165, "step_time_sec": 8.239821827039123, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8549, "loss": 3.9142260551452637, "lr": 0.00013636226215236997, "elapsed_sec": 71108.2093565464, "step_time_sec": 8.240659188013524, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8550, "loss": 4.073831558227539, "lr": 0.00013629042414544537, "elapsed_sec": 71116.44807887077, "step_time_sec": 8.238585670944303, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8551, "loss": 4.130488872528076, "lr": 0.00013621856528446362, "elapsed_sec": 71124.68765854836, "step_time_sec": 8.239399045938626, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8552, "loss": 3.9518144130706787, "lr": 0.00013614668561377316, "elapsed_sec": 71132.92586135864, "step_time_sec": 8.238027002895251, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8553, "loss": 3.8567557334899902, "lr": 0.00013607478517773508, "elapsed_sec": 71141.16573786736, "step_time_sec": 8.23971105995588, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8554, "loss": 3.8822247982025146, "lr": 0.00013600286402072343, "elapsed_sec": 71149.40451717377, "step_time_sec": 8.238612961024046, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8555, "loss": 3.9260621070861816, "lr": 0.00013593092218712496, "elapsed_sec": 71157.64208126068, "step_time_sec": 8.2374317079084, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8556, "loss": 3.804002285003662, "lr": 0.00013585895972133917, "elapsed_sec": 71165.88068389893, "step_time_sec": 8.238413136918098, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8557, "loss": 3.955944061279297, "lr": 0.00013578697666777838, "elapsed_sec": 71174.11970686913, "step_time_sec": 8.238831001915969, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8558, "loss": 3.9578685760498047, "lr": 0.00013571497307086753, "elapsed_sec": 71182.35952854156, "step_time_sec": 8.239657106925733, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8559, "loss": 3.819244623184204, "lr": 0.00013564294897504427, "elapsed_sec": 71190.59949302673, "step_time_sec": 8.239807663019747, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8560, "loss": 3.904184103012085, "lr": 0.00013557090442475894, "elapsed_sec": 71198.83835887909, "step_time_sec": 8.23869267909322, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8561, "loss": 4.099813461303711, "lr": 0.00013549883946447445, "elapsed_sec": 71207.07983279228, "step_time_sec": 8.241356080048718, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8562, "loss": 3.9381015300750732, "lr": 0.00013542675413866632, "elapsed_sec": 71215.31970739365, "step_time_sec": 8.23964539100416, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8563, "loss": 3.8168210983276367, "lr": 0.0001353546484918226, "elapsed_sec": 71223.56024503708, "step_time_sec": 8.240355027024634, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8564, "loss": 3.853024482727051, "lr": 0.000135282522568444, "elapsed_sec": 71231.80058908463, "step_time_sec": 8.240280281985179, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8565, "loss": 3.8852572441101074, "lr": 0.00013521037641304364, "elapsed_sec": 71240.03929519653, "step_time_sec": 8.238478978979401, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8566, "loss": 3.906550407409668, "lr": 0.0001351382100701471, "elapsed_sec": 71248.27923703194, "step_time_sec": 8.239767730003223, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8567, "loss": 3.931676149368286, "lr": 0.00013506602358429254, "elapsed_sec": 71256.51854348183, "step_time_sec": 8.23920615494717, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8568, "loss": 3.883884906768799, "lr": 0.00013499381700003045, "elapsed_sec": 71264.75702524185, "step_time_sec": 8.238279676996171, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8569, "loss": 4.0564656257629395, "lr": 0.00013492159036192377, "elapsed_sec": 71272.99688196182, "step_time_sec": 8.239681739010848, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8570, "loss": 3.8879287242889404, "lr": 0.0001348493437145478, "elapsed_sec": 71281.22886967659, "step_time_sec": 8.231819260981865, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8571, "loss": 3.8248374462127686, "lr": 0.0001347770771024902, "elapsed_sec": 71289.46205592155, "step_time_sec": 8.23302938498091, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8572, "loss": 3.9728007316589355, "lr": 0.00013470479057035094, "elapsed_sec": 71297.69493603706, "step_time_sec": 8.232728751027025, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8573, "loss": 4.057957649230957, "lr": 0.0001346324841627423, "elapsed_sec": 71305.93403601646, "step_time_sec": 8.238959742942825, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8574, "loss": 3.928272008895874, "lr": 0.00013456015792428878, "elapsed_sec": 71314.17344069481, "step_time_sec": 8.239215363981202, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8575, "loss": 3.875854730606079, "lr": 0.0001344878118996272, "elapsed_sec": 71322.41183638573, "step_time_sec": 8.238219299004413, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8576, "loss": 3.9179062843322754, "lr": 0.0001344154461334065, "elapsed_sec": 71330.65198087692, "step_time_sec": 8.239980293088593, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8577, "loss": 3.7768054008483887, "lr": 0.00013434306067028787, "elapsed_sec": 71338.89037680626, "step_time_sec": 8.238239265978336, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8578, "loss": 4.018414497375488, "lr": 0.0001342706555549446, "elapsed_sec": 71347.1307284832, "step_time_sec": 8.240223953966051, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8579, "loss": 4.047846794128418, "lr": 0.00013419823083206224, "elapsed_sec": 71355.37121844292, "step_time_sec": 8.240283273975365, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8580, "loss": 4.039827823638916, "lr": 0.00013412578654633825, "elapsed_sec": 71363.60897874832, "step_time_sec": 8.237585580092855, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8581, "loss": 3.9278628826141357, "lr": 0.00013405332274248226, "elapsed_sec": 71371.84857344627, "step_time_sec": 8.23948959808331, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8582, "loss": 4.052661895751953, "lr": 0.00013398083946521602, "elapsed_sec": 71380.08689045906, "step_time_sec": 8.238119621062651, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8583, "loss": 3.9759202003479004, "lr": 0.00013390833675927313, "elapsed_sec": 71388.32602405548, "step_time_sec": 8.238998707965948, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8584, "loss": 4.061026096343994, "lr": 0.00013383581466939936, "elapsed_sec": 71396.56464672089, "step_time_sec": 8.238446124014445, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8585, "loss": 4.000119686126709, "lr": 0.0001337632732403523, "elapsed_sec": 71404.80257201195, "step_time_sec": 8.2377579130698, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8586, "loss": 4.006286144256592, "lr": 0.00013369071251690147, "elapsed_sec": 71413.04228210449, "step_time_sec": 8.239576054969802, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8587, "loss": 4.072490692138672, "lr": 0.00013361813254382849, "elapsed_sec": 71421.28080630302, "step_time_sec": 8.238374599954113, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8588, "loss": 3.973963737487793, "lr": 0.00013354553336592664, "elapsed_sec": 71429.52049160004, "step_time_sec": 8.239448738982901, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8589, "loss": 4.127087593078613, "lr": 0.0001334729150280012, "elapsed_sec": 71437.75976419449, "step_time_sec": 8.239117279998027, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8590, "loss": 3.938967704772949, "lr": 0.00013340027757486914, "elapsed_sec": 71445.99841928482, "step_time_sec": 8.23850991600193, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8591, "loss": 4.059731483459473, "lr": 0.00013332762105135933, "elapsed_sec": 71454.23240065575, "step_time_sec": 8.233766932040453, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8592, "loss": 4.2402238845825195, "lr": 0.00013325494550231245, "elapsed_sec": 71462.47015094757, "step_time_sec": 8.237626429996453, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8593, "loss": 3.830359935760498, "lr": 0.00013318225097258072, "elapsed_sec": 71470.71018266678, "step_time_sec": 8.239826106000692, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8594, "loss": 3.9497783184051514, "lr": 0.00013310953750702832, "elapsed_sec": 71478.95043730736, "step_time_sec": 8.240129965939559, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8595, "loss": 3.942979097366333, "lr": 0.00013303680515053102, "elapsed_sec": 71487.18863844872, "step_time_sec": 8.238033532048576, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8596, "loss": 4.083684921264648, "lr": 0.00013296405394797611, "elapsed_sec": 71495.42909646034, "step_time_sec": 8.240236560930498, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8597, "loss": 4.106472015380859, "lr": 0.00013289128394426271, "elapsed_sec": 71503.66727542877, "step_time_sec": 8.238045454025269, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8598, "loss": 3.970365524291992, "lr": 0.00013281849518430148, "elapsed_sec": 71511.90729403496, "step_time_sec": 8.23984118399676, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8599, "loss": 3.9652397632598877, "lr": 0.00013274568771301466, "elapsed_sec": 71520.14655518532, "step_time_sec": 8.23905028810259, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8600, "loss": 3.995544672012329, "lr": 0.00013267286157533594, "elapsed_sec": 71528.38437390327, "step_time_sec": 8.237731114961207, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8601, "loss": 3.909146547317505, "lr": 0.00013260001681621067, "elapsed_sec": 71536.62583708763, "step_time_sec": 8.241284716990776, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8602, "loss": 3.9108262062072754, "lr": 0.0001325271534805956, "elapsed_sec": 71544.86505556107, "step_time_sec": 8.239076355006546, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8603, "loss": 4.04291296005249, "lr": 0.00013245427161345905, "elapsed_sec": 71553.11077785492, "step_time_sec": 8.24557113298215, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8604, "loss": 3.897059440612793, "lr": 0.00013238137125978056, "elapsed_sec": 71561.3456170559, "step_time_sec": 8.234611701103859, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8605, "loss": 4.064923286437988, "lr": 0.0001323084524645514, "elapsed_sec": 71569.57828712463, "step_time_sec": 8.232457314035855, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8606, "loss": 3.9958267211914062, "lr": 0.00013223551527277394, "elapsed_sec": 71577.81201958656, "step_time_sec": 8.233566742041148, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8607, "loss": 4.04480504989624, "lr": 0.00013216255972946198, "elapsed_sec": 71586.04361176491, "step_time_sec": 8.231467818026431, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8608, "loss": 3.9361395835876465, "lr": 0.00013208958587964076, "elapsed_sec": 71594.27794361115, "step_time_sec": 8.234157635015436, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8609, "loss": 3.985008478164673, "lr": 0.00013201659376834668, "elapsed_sec": 71602.50920438766, "step_time_sec": 8.231133293011226, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8610, "loss": 4.031867027282715, "lr": 0.00013194358344062752, "elapsed_sec": 71610.74283146858, "step_time_sec": 8.233448193990625, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8611, "loss": 3.789227247238159, "lr": 0.00013187055494154215, "elapsed_sec": 71618.97466087341, "step_time_sec": 8.23169421905186, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8612, "loss": 3.9901864528656006, "lr": 0.00013179750831616082, "elapsed_sec": 71627.20789885521, "step_time_sec": 8.233077444019727, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8613, "loss": 3.948279857635498, "lr": 0.00013172444360956485, "elapsed_sec": 71635.44036078453, "step_time_sec": 8.232280154014006, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8614, "loss": 3.9050025939941406, "lr": 0.00013165136086684687, "elapsed_sec": 71643.67296767235, "step_time_sec": 8.232468754053116, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8615, "loss": 3.8928866386413574, "lr": 0.0001315782601331104, "elapsed_sec": 71651.91202735901, "step_time_sec": 8.238860914018005, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8616, "loss": 3.8417482376098633, "lr": 0.00013150514145347027, "elapsed_sec": 71660.15123319626, "step_time_sec": 8.239008973934688, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8617, "loss": 3.8895797729492188, "lr": 0.00013143200487305233, "elapsed_sec": 71668.38818717003, "step_time_sec": 8.236802418949082, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8618, "loss": 4.161305904388428, "lr": 0.0001313588504369934, "elapsed_sec": 71676.62857317924, "step_time_sec": 8.240297183045186, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8619, "loss": 4.0371575355529785, "lr": 0.00013128567819044147, "elapsed_sec": 71684.86056137085, "step_time_sec": 8.23178358108271, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8620, "loss": 3.8369367122650146, "lr": 0.00013121248817855532, "elapsed_sec": 71693.09385848045, "step_time_sec": 8.233104955055751, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8621, "loss": 3.9476537704467773, "lr": 0.00013113928044650495, "elapsed_sec": 71701.32630634308, "step_time_sec": 8.232298303046264, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8622, "loss": 3.8578598499298096, "lr": 0.00013106605503947105, "elapsed_sec": 71709.56533360481, "step_time_sec": 8.238842954975553, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8623, "loss": 3.9568886756896973, "lr": 0.00013099281200264535, "elapsed_sec": 71717.80517053604, "step_time_sec": 8.239675526041538, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8624, "loss": 3.8782637119293213, "lr": 0.00013091955138123043, "elapsed_sec": 71726.0437271595, "step_time_sec": 8.238398420973681, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8625, "loss": 3.972552537918091, "lr": 0.00013084627322043974, "elapsed_sec": 71734.28230047226, "step_time_sec": 8.238393170991912, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8626, "loss": 3.909815788269043, "lr": 0.00013077297756549755, "elapsed_sec": 71742.51367640495, "step_time_sec": 8.23124417301733, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8627, "loss": 3.9226160049438477, "lr": 0.00013069966446163886, "elapsed_sec": 71750.74759054184, "step_time_sec": 8.23375663801562, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8628, "loss": 3.979234457015991, "lr": 0.00013062633395410957, "elapsed_sec": 71758.97928190231, "step_time_sec": 8.231554861995392, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8629, "loss": 4.019680976867676, "lr": 0.0001305529860881662, "elapsed_sec": 71767.21206879616, "step_time_sec": 8.232632428989746, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8630, "loss": 4.075453758239746, "lr": 0.00013047962090907602, "elapsed_sec": 71775.44526171684, "step_time_sec": 8.233026660978794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8631, "loss": 4.129173755645752, "lr": 0.00013040623846211706, "elapsed_sec": 71783.67887043953, "step_time_sec": 8.233450319967233, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8632, "loss": 3.913571357727051, "lr": 0.0001303328387925779, "elapsed_sec": 71791.91167783737, "step_time_sec": 8.232597841997631, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8633, "loss": 4.006766319274902, "lr": 0.00013025942194575777, "elapsed_sec": 71800.14459824562, "step_time_sec": 8.23274865897838, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8634, "loss": 4.184832572937012, "lr": 0.00013018598796696654, "elapsed_sec": 71808.3820669651, "step_time_sec": 8.237322676926851, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8635, "loss": 3.9896602630615234, "lr": 0.00013011253690152464, "elapsed_sec": 71816.6209013462, "step_time_sec": 8.23874156596139, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8636, "loss": 3.9366965293884277, "lr": 0.00013003906879476308, "elapsed_sec": 71824.85872268677, "step_time_sec": 8.237620730069466, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8637, "loss": 3.9322609901428223, "lr": 0.00012996558369202327, "elapsed_sec": 71833.09856319427, "step_time_sec": 8.239663616986945, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8638, "loss": 3.962249755859375, "lr": 0.0001298920816386572, "elapsed_sec": 71841.33830356598, "step_time_sec": 8.239586952026002, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8639, "loss": 4.025122165679932, "lr": 0.00012981856268002738, "elapsed_sec": 71849.57807540894, "step_time_sec": 8.239624109002762, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8640, "loss": 3.9143402576446533, "lr": 0.0001297450268615066, "elapsed_sec": 71857.81696367264, "step_time_sec": 8.238757257000543, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8641, "loss": 4.062652111053467, "lr": 0.00012967147422847817, "elapsed_sec": 71866.05343008041, "step_time_sec": 8.236260390025564, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8642, "loss": 4.02280330657959, "lr": 0.00012959790482633575, "elapsed_sec": 71874.29355430603, "step_time_sec": 8.239950458984822, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8643, "loss": 4.023840427398682, "lr": 0.00012952431870048332, "elapsed_sec": 71882.53321790695, "step_time_sec": 8.239496757043526, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8644, "loss": 3.9362876415252686, "lr": 0.00012945071589633524, "elapsed_sec": 71890.77216744423, "step_time_sec": 8.238789608003572, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8645, "loss": 4.068878173828125, "lr": 0.00012937709645931608, "elapsed_sec": 71899.01166081429, "step_time_sec": 8.239387099049054, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8646, "loss": 3.971407890319824, "lr": 0.00012930346043486078, "elapsed_sec": 71907.25054883957, "step_time_sec": 8.238733182079159, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8647, "loss": 3.9630215167999268, "lr": 0.00012922980786841438, "elapsed_sec": 71915.49162077904, "step_time_sec": 8.240859115961939, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8648, "loss": 3.964993476867676, "lr": 0.0001291561388054323, "elapsed_sec": 71923.73226118088, "step_time_sec": 8.24050533294212, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8649, "loss": 3.6700174808502197, "lr": 0.00012908245329137997, "elapsed_sec": 71931.97012472153, "step_time_sec": 8.23772276204545, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8650, "loss": 4.1293792724609375, "lr": 0.0001290087513717331, "elapsed_sec": 71940.20462703705, "step_time_sec": 8.234284196048975, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8651, "loss": 4.2211594581604, "lr": 0.00012893503309197748, "elapsed_sec": 71948.43682575226, "step_time_sec": 8.23210563894827, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8652, "loss": 4.147008419036865, "lr": 0.00012886129849760898, "elapsed_sec": 71956.67095398903, "step_time_sec": 8.233973787981085, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8653, "loss": 3.859262466430664, "lr": 0.00012878754763413351, "elapsed_sec": 71964.90395855904, "step_time_sec": 8.232785198022611, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8654, "loss": 4.179750442504883, "lr": 0.0001287137805470671, "elapsed_sec": 71973.13751196861, "step_time_sec": 8.233400878030807, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8655, "loss": 3.9730281829833984, "lr": 0.0001286399972819358, "elapsed_sec": 71981.37481212616, "step_time_sec": 8.237123406026512, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8656, "loss": 4.053314208984375, "lr": 0.00012856619788427558, "elapsed_sec": 71989.6137304306, "step_time_sec": 8.238793896976858, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8657, "loss": 3.997304677963257, "lr": 0.00012849238239963232, "elapsed_sec": 71997.85341954231, "step_time_sec": 8.239483483950607, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8658, "loss": 3.9360363483428955, "lr": 0.00012841855087356195, "elapsed_sec": 72006.09206318855, "step_time_sec": 8.238475176971406, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8659, "loss": 3.9384098052978516, "lr": 0.00012834470335163026, "elapsed_sec": 72014.33253288269, "step_time_sec": 8.240292098955251, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8660, "loss": 4.073970794677734, "lr": 0.00012827083987941283, "elapsed_sec": 72022.57087182999, "step_time_sec": 8.238188822986558, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8661, "loss": 3.9184539318084717, "lr": 0.0001281969605024952, "elapsed_sec": 72030.81063652039, "step_time_sec": 8.239613426034339, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8662, "loss": 3.768967628479004, "lr": 0.00012812306526647268, "elapsed_sec": 72039.05115747452, "step_time_sec": 8.240353722008877, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8663, "loss": 3.979391574859619, "lr": 0.0001280491542169504, "elapsed_sec": 72047.28907370567, "step_time_sec": 8.237786636920646, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8664, "loss": 3.8890926837921143, "lr": 0.0001279752273995431, "elapsed_sec": 72055.52210903168, "step_time_sec": 8.232825840939768, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8665, "loss": 3.9318301677703857, "lr": 0.00012790128485987541, "elapsed_sec": 72063.75513839722, "step_time_sec": 8.232843715930358, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8666, "loss": 3.8692831993103027, "lr": 0.00012782732664358174, "elapsed_sec": 72071.98901128769, "step_time_sec": 8.233732803957537, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8667, "loss": 3.9615001678466797, "lr": 0.0001277533527963059, "elapsed_sec": 72080.22756409645, "step_time_sec": 8.238376380992122, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8668, "loss": 3.79341197013855, "lr": 0.00012767936336370152, "elapsed_sec": 72088.46771931648, "step_time_sec": 8.240042107994668, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8669, "loss": 3.9868900775909424, "lr": 0.0001276053583914319, "elapsed_sec": 72096.70633220673, "step_time_sec": 8.238468161900528, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8670, "loss": 3.946413993835449, "lr": 0.00012753133792516984, "elapsed_sec": 72104.94583010674, "step_time_sec": 8.239280432928354, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8671, "loss": 3.912867784500122, "lr": 0.00012745730201059764, "elapsed_sec": 72113.18532681465, "step_time_sec": 8.239334040903486, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8672, "loss": 3.9021270275115967, "lr": 0.00012738325069340731, "elapsed_sec": 72121.42392063141, "step_time_sec": 8.238422976108268, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8673, "loss": 4.041886329650879, "lr": 0.00012730918401930024, "elapsed_sec": 72129.66396927834, "step_time_sec": 8.23989581107162, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8674, "loss": 4.088467597961426, "lr": 0.0001272351020339873, "elapsed_sec": 72137.89772677422, "step_time_sec": 8.233594268909656, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8675, "loss": 4.027468204498291, "lr": 0.00012716100478318888, "elapsed_sec": 72146.13134050369, "step_time_sec": 8.233462102944031, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8676, "loss": 3.990654230117798, "lr": 0.00012708689231263475, "elapsed_sec": 72154.36492109299, "step_time_sec": 8.233465416938998, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8677, "loss": 3.9906206130981445, "lr": 0.00012701276466806403, "elapsed_sec": 72162.59896349907, "step_time_sec": 8.233859514002688, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8678, "loss": 3.875169277191162, "lr": 0.00012693862189522525, "elapsed_sec": 72170.83188056946, "step_time_sec": 8.232721220934764, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8679, "loss": 3.9252264499664307, "lr": 0.00012686446403987635, "elapsed_sec": 72179.06674838066, "step_time_sec": 8.234740637009963, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8680, "loss": 3.848384141921997, "lr": 0.0001267902911477844, "elapsed_sec": 72187.29907011986, "step_time_sec": 8.23223273793701, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8681, "loss": 3.9527134895324707, "lr": 0.00012671610326472592, "elapsed_sec": 72195.53337025642, "step_time_sec": 8.234107349999249, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8682, "loss": 3.962679386138916, "lr": 0.00012664190043648656, "elapsed_sec": 72203.76636266708, "step_time_sec": 8.232824053964578, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8683, "loss": 3.8209683895111084, "lr": 0.00012656768270886132, "elapsed_sec": 72212.00668644905, "step_time_sec": 8.240127887926064, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8684, "loss": 3.706077814102173, "lr": 0.00012649345012765426, "elapsed_sec": 72220.24623942375, "step_time_sec": 8.239409981993958, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8685, "loss": 4.002861499786377, "lr": 0.0001264192027386787, "elapsed_sec": 72228.48401117325, "step_time_sec": 8.237570907920599, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8686, "loss": 4.002315044403076, "lr": 0.00012634494058775706, "elapsed_sec": 72236.72376322746, "step_time_sec": 8.239657192956656, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8687, "loss": 4.166804790496826, "lr": 0.0001262706637207209, "elapsed_sec": 72244.95577955246, "step_time_sec": 8.231805596034974, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8688, "loss": 3.987992525100708, "lr": 0.00012619637218341077, "elapsed_sec": 72253.18962693214, "step_time_sec": 8.233690545079298, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8689, "loss": 3.9238810539245605, "lr": 0.00012612206602167646, "elapsed_sec": 72261.42973446846, "step_time_sec": 8.239955096971244, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8690, "loss": 3.9295496940612793, "lr": 0.00012604774528137655, "elapsed_sec": 72269.6695394516, "step_time_sec": 8.239622293971479, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8691, "loss": 3.863377571105957, "lr": 0.0001259734100083788, "elapsed_sec": 72277.90860033035, "step_time_sec": 8.23890339396894, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8692, "loss": 3.8386380672454834, "lr": 0.00012589906024855987, "elapsed_sec": 72286.14713191986, "step_time_sec": 8.238370901090093, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8693, "loss": 3.933385133743286, "lr": 0.00012582469604780538, "elapsed_sec": 72294.38789248466, "step_time_sec": 8.240590558038093, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8694, "loss": 4.007864952087402, "lr": 0.00012575031745200982, "elapsed_sec": 72302.62675666809, "step_time_sec": 8.23871865705587, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8695, "loss": 3.920323133468628, "lr": 0.00012567592450707659, "elapsed_sec": 72310.86655592918, "step_time_sec": 8.239596333005466, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8696, "loss": 3.9521424770355225, "lr": 0.00012560151725891802, "elapsed_sec": 72319.1050772667, "step_time_sec": 8.238374748034403, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8697, "loss": 3.927902936935425, "lr": 0.00012552709575345514, "elapsed_sec": 72327.33718252182, "step_time_sec": 8.231947903055698, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8698, "loss": 3.8328471183776855, "lr": 0.00012545266003661783, "elapsed_sec": 72335.57058143616, "step_time_sec": 8.233184700948186, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8699, "loss": 4.158036708831787, "lr": 0.00012537821015434474, "elapsed_sec": 72343.80296421051, "step_time_sec": 8.23225318803452, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8700, "loss": 3.8313803672790527, "lr": 0.00012530374615258333, "elapsed_sec": 72352.04304599762, "step_time_sec": 8.23988129501231, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8701, "loss": 3.82159161567688, "lr": 0.00012522926807728962, "elapsed_sec": 72360.27514648438, "step_time_sec": 8.231963573023677, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8702, "loss": 4.001413822174072, "lr": 0.00012515477597442847, "elapsed_sec": 72368.50917458534, "step_time_sec": 8.233901167055592, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8703, "loss": 3.9901509284973145, "lr": 0.00012508026988997334, "elapsed_sec": 72376.74781751633, "step_time_sec": 8.2385180729907, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8704, "loss": 3.99412202835083, "lr": 0.0001250057498699063, "elapsed_sec": 72384.98037815094, "step_time_sec": 8.2323793170508, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8705, "loss": 4.119406700134277, "lr": 0.00012493121596021802, "elapsed_sec": 72393.21308016777, "step_time_sec": 8.232481168000959, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8706, "loss": 3.792166233062744, "lr": 0.0001248566682069078, "elapsed_sec": 72401.44677257538, "step_time_sec": 8.233590611955151, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8707, "loss": 3.9566879272460938, "lr": 0.00012478210665598337, "elapsed_sec": 72409.67903852463, "step_time_sec": 8.232045457931235, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8708, "loss": 3.874519109725952, "lr": 0.00012470753135346112, "elapsed_sec": 72417.91308021545, "step_time_sec": 8.233866741065867, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8709, "loss": 3.9632718563079834, "lr": 0.0001246329423453658, "elapsed_sec": 72426.14575219154, "step_time_sec": 8.232512610964477, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8710, "loss": 3.845919370651245, "lr": 0.0001245583396777307, "elapsed_sec": 72434.38577389717, "step_time_sec": 8.239839237066917, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8711, "loss": 3.854670524597168, "lr": 0.0001244837233965975, "elapsed_sec": 72442.69149398804, "step_time_sec": 8.242535094963387, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8712, "loss": 3.9531288146972656, "lr": 0.0001244090935480163, "elapsed_sec": 72450.93056440353, "step_time_sec": 8.238907762919553, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8713, "loss": 3.9774956703186035, "lr": 0.00012433445017804556, "elapsed_sec": 72459.16957473755, "step_time_sec": 8.238806514069438, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8714, "loss": 3.9043445587158203, "lr": 0.0001242597933327521, "elapsed_sec": 72467.40788674355, "step_time_sec": 8.238175284001045, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8715, "loss": 3.9653148651123047, "lr": 0.00012418512305821105, "elapsed_sec": 72475.6453590393, "step_time_sec": 8.2372525309911, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8716, "loss": 3.942800521850586, "lr": 0.0001241104394005058, "elapsed_sec": 72483.88369178772, "step_time_sec": 8.23819589894265, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8717, "loss": 3.841724157333374, "lr": 0.00012403574240572805, "elapsed_sec": 72492.12234950066, "step_time_sec": 8.238476113067009, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8718, "loss": 4.086141586303711, "lr": 0.00012396103211997773, "elapsed_sec": 72500.36479234695, "step_time_sec": 8.242294888012111, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8719, "loss": 3.965474843978882, "lr": 0.0001238863085893629, "elapsed_sec": 72508.60540056229, "step_time_sec": 8.240453316015191, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8720, "loss": 4.011312007904053, "lr": 0.00012381157185999985, "elapsed_sec": 72516.8435575962, "step_time_sec": 8.238039851072244, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8721, "loss": 4.119429111480713, "lr": 0.00012373682197801302, "elapsed_sec": 72525.07784628868, "step_time_sec": 8.234044192940928, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8722, "loss": 4.025001049041748, "lr": 0.000123662058989535, "elapsed_sec": 72533.31025671959, "step_time_sec": 8.232269420986995, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8723, "loss": 3.9836883544921875, "lr": 0.00012358728294070635, "elapsed_sec": 72541.55133461952, "step_time_sec": 8.240904637961648, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8724, "loss": 3.8261184692382812, "lr": 0.0001235124938776758, "elapsed_sec": 72549.78310799599, "step_time_sec": 8.231600490980782, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8725, "loss": 4.000496864318848, "lr": 0.00012343769184660006, "elapsed_sec": 72558.01794719696, "step_time_sec": 8.234670671983622, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8726, "loss": 4.009597301483154, "lr": 0.00012336287689364392, "elapsed_sec": 72566.25042271614, "step_time_sec": 8.232342628994957, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8727, "loss": 3.7797389030456543, "lr": 0.00012328804906498001, "elapsed_sec": 72574.48999404907, "step_time_sec": 8.239426125073805, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8728, "loss": 3.8477230072021484, "lr": 0.000123213208406789, "elapsed_sec": 72582.72865605354, "step_time_sec": 8.238510966068134, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8729, "loss": 3.942073106765747, "lr": 0.0001231383549652595, "elapsed_sec": 72590.96752786636, "step_time_sec": 8.238668423029594, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8730, "loss": 3.8976502418518066, "lr": 0.00012306348878658797, "elapsed_sec": 72599.20683550835, "step_time_sec": 8.239136913092807, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8731, "loss": 3.986353635787964, "lr": 0.00012298860991697867, "elapsed_sec": 72607.44466423988, "step_time_sec": 8.237689865985885, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8732, "loss": 4.071853160858154, "lr": 0.00012291371840264376, "elapsed_sec": 72615.68467020988, "step_time_sec": 8.239870521007106, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8733, "loss": 3.8755152225494385, "lr": 0.00012283881428980326, "elapsed_sec": 72623.92586255074, "step_time_sec": 8.241014305036515, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8734, "loss": 3.8605129718780518, "lr": 0.00012276389762468483, "elapsed_sec": 72632.16409826279, "step_time_sec": 8.238025645026937, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8735, "loss": 3.887563705444336, "lr": 0.000122688968453524, "elapsed_sec": 72640.40336680412, "step_time_sec": 8.239131670095958, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8736, "loss": 3.8567662239074707, "lr": 0.00012261402682256395, "elapsed_sec": 72648.64192938805, "step_time_sec": 8.238439470995218, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8737, "loss": 3.886251211166382, "lr": 0.00012253907277805557, "elapsed_sec": 72656.88253545761, "step_time_sec": 8.240456571918912, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8738, "loss": 3.9892773628234863, "lr": 0.00012246410636625738, "elapsed_sec": 72665.12192869186, "step_time_sec": 8.239147763000801, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8739, "loss": 3.7223947048187256, "lr": 0.00012238912763343554, "elapsed_sec": 72673.36060929298, "step_time_sec": 8.23855607106816, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8740, "loss": 3.8277924060821533, "lr": 0.00012231413662586394, "elapsed_sec": 72681.60012722015, "step_time_sec": 8.23936640797183, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8741, "loss": 3.8668265342712402, "lr": 0.00012223913338982388, "elapsed_sec": 72689.83856034279, "step_time_sec": 8.238265697960742, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8742, "loss": 3.791118621826172, "lr": 0.00012216411797160418, "elapsed_sec": 72698.07784080505, "step_time_sec": 8.239116533077322, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8743, "loss": 4.078054428100586, "lr": 0.0001220890904175014, "elapsed_sec": 72706.31745696068, "step_time_sec": 8.239445428014733, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8744, "loss": 3.982379674911499, "lr": 0.00012201405077381935, "elapsed_sec": 72714.55535626411, "step_time_sec": 8.237725722952746, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8745, "loss": 3.7077291011810303, "lr": 0.0001219389990868695, "elapsed_sec": 72722.79583287239, "step_time_sec": 8.24036036094185, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8746, "loss": 3.9822590351104736, "lr": 0.00012186393540297058, "elapsed_sec": 72731.03671884537, "step_time_sec": 8.240756222046912, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8747, "loss": 4.057362079620361, "lr": 0.00012178885976844883, "elapsed_sec": 72739.27617073059, "step_time_sec": 8.239279186003841, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8748, "loss": 3.932391405105591, "lr": 0.00012171377222963785, "elapsed_sec": 72747.53453493118, "step_time_sec": 8.258252477040514, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8749, "loss": 3.9440228939056396, "lr": 0.00012163867283287853, "elapsed_sec": 72755.78733897209, "step_time_sec": 8.252575653023086, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8750, "loss": 3.8822500705718994, "lr": 0.0001215635616245192, "elapsed_sec": 72764.02781271935, "step_time_sec": 8.240280885016546, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8751, "loss": 4.046899795532227, "lr": 0.00012148843865091531, "elapsed_sec": 72772.26693058014, "step_time_sec": 8.238961945986375, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8752, "loss": 3.956820011138916, "lr": 0.00012141330395842975, "elapsed_sec": 72780.50724816322, "step_time_sec": 8.240147024043836, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8753, "loss": 3.941972255706787, "lr": 0.00012133815759343251, "elapsed_sec": 72788.74722266197, "step_time_sec": 8.23981916194316, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8754, "loss": 4.0056610107421875, "lr": 0.0001212629996023008, "elapsed_sec": 72796.98677372932, "step_time_sec": 8.239428304950707, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8755, "loss": 4.005610942840576, "lr": 0.00012118783003141909, "elapsed_sec": 72805.22777342796, "step_time_sec": 8.240785127971321, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8756, "loss": 3.928964376449585, "lr": 0.00012111264892717893, "elapsed_sec": 72813.46764850616, "step_time_sec": 8.239740150980651, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8757, "loss": 3.8212738037109375, "lr": 0.00012103745633597897, "elapsed_sec": 72821.70645070076, "step_time_sec": 8.238668828038499, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8758, "loss": 3.9024438858032227, "lr": 0.00012096225230422498, "elapsed_sec": 72829.9450712204, "step_time_sec": 8.238428147975355, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8759, "loss": 3.77974271774292, "lr": 0.00012088703687832985, "elapsed_sec": 72838.18396878242, "step_time_sec": 8.23875738796778, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8760, "loss": 3.6299264430999756, "lr": 0.00012081181010471338, "elapsed_sec": 72846.42400026321, "step_time_sec": 8.239853876992129, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8761, "loss": 3.9921681880950928, "lr": 0.00012073657202980245, "elapsed_sec": 72854.66275334358, "step_time_sec": 8.238616119953804, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8762, "loss": 3.9570155143737793, "lr": 0.00012066132270003087, "elapsed_sec": 72862.90086174011, "step_time_sec": 8.237952601979487, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8763, "loss": 3.9019856452941895, "lr": 0.00012058606216183949, "elapsed_sec": 72871.13957166672, "step_time_sec": 8.238524141022936, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8764, "loss": 3.7624905109405518, "lr": 0.00012051079046167597, "elapsed_sec": 72879.37772703171, "step_time_sec": 8.237986426916905, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8765, "loss": 3.857753276824951, "lr": 0.00012043550764599488, "elapsed_sec": 72887.61758852005, "step_time_sec": 8.23973882291466, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8766, "loss": 3.9792988300323486, "lr": 0.00012036021376125772, "elapsed_sec": 72895.85826468468, "step_time_sec": 8.240478252992034, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8767, "loss": 3.933955669403076, "lr": 0.00012028490885393276, "elapsed_sec": 72904.09551000595, "step_time_sec": 8.237082673003897, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8768, "loss": 3.8943867683410645, "lr": 0.00012020959297049504, "elapsed_sec": 72912.32938933372, "step_time_sec": 8.233728818013333, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8769, "loss": 3.9320290088653564, "lr": 0.00012013426615742649, "elapsed_sec": 72920.56352972984, "step_time_sec": 8.234025741927326, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8770, "loss": 4.065852165222168, "lr": 0.00012005892846121567, "elapsed_sec": 72928.7986664772, "step_time_sec": 8.23493505094666, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8771, "loss": 4.0238566398620605, "lr": 0.0001199835799283579, "elapsed_sec": 72937.03743362427, "step_time_sec": 8.238636065041646, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8772, "loss": 3.967921018600464, "lr": 0.00011990822060535521, "elapsed_sec": 72945.2789940834, "step_time_sec": 8.241378276026808, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8773, "loss": 4.019485950469971, "lr": 0.00011983285053871624, "elapsed_sec": 72953.52021884918, "step_time_sec": 8.241066443035379, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8774, "loss": 3.8479344844818115, "lr": 0.0001197574697749563, "elapsed_sec": 72961.75810265541, "step_time_sec": 8.237673214054666, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8775, "loss": 3.933659315109253, "lr": 0.00011968207836059732, "elapsed_sec": 72969.99332094193, "step_time_sec": 8.23507861699909, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8776, "loss": 3.842491388320923, "lr": 0.00011960667634216774, "elapsed_sec": 72978.2257733345, "step_time_sec": 8.232287936960347, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8777, "loss": 3.8703091144561768, "lr": 0.00011953126376620261, "elapsed_sec": 72986.45979475975, "step_time_sec": 8.23389353393577, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8778, "loss": 3.9059293270111084, "lr": 0.00011945584067924342, "elapsed_sec": 72994.69188737869, "step_time_sec": 8.231938143027946, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8779, "loss": 3.8447234630584717, "lr": 0.0001193804071278382, "elapsed_sec": 73002.92612242699, "step_time_sec": 8.234064770047553, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8780, "loss": 3.994462251663208, "lr": 0.00011930496315854146, "elapsed_sec": 73011.16512179375, "step_time_sec": 8.23878674407024, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8781, "loss": 3.899709939956665, "lr": 0.00011922950881791406, "elapsed_sec": 73019.3983604908, "step_time_sec": 8.23309152701404, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8782, "loss": 4.006099700927734, "lr": 0.00011915404415252337, "elapsed_sec": 73027.63018798828, "step_time_sec": 8.231678218930028, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8783, "loss": 3.993448495864868, "lr": 0.00011907856920894304, "elapsed_sec": 73035.86508131027, "step_time_sec": 8.234718622989021, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8784, "loss": 3.9722158908843994, "lr": 0.00011900308403375307, "elapsed_sec": 73044.10249352455, "step_time_sec": 8.237294030026533, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8785, "loss": 3.8737263679504395, "lr": 0.00011892758867353984, "elapsed_sec": 73052.3431854248, "step_time_sec": 8.24050132301636, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8786, "loss": 3.94052791595459, "lr": 0.00011885208317489597, "elapsed_sec": 73060.5836801529, "step_time_sec": 8.240309515967965, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8787, "loss": 3.802802801132202, "lr": 0.0001187765675844203, "elapsed_sec": 73068.82212042809, "step_time_sec": 8.23833421105519, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8788, "loss": 3.9851977825164795, "lr": 0.00011870104194871799, "elapsed_sec": 73077.06230998039, "step_time_sec": 8.240046576946042, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8789, "loss": 3.937516212463379, "lr": 0.00011862550631440031, "elapsed_sec": 73085.30105757713, "step_time_sec": 8.238543785992078, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8790, "loss": 3.9115099906921387, "lr": 0.00011854996072808478, "elapsed_sec": 73093.54056572914, "step_time_sec": 8.239385123015381, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8791, "loss": 4.1842041015625, "lr": 0.00011847440523639497, "elapsed_sec": 73101.7805788517, "step_time_sec": 8.239841199945658, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8792, "loss": 3.9989161491394043, "lr": 0.00011839883988596068, "elapsed_sec": 73110.01854467392, "step_time_sec": 8.23783692496363, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8793, "loss": 3.945188522338867, "lr": 0.00011832326472341769, "elapsed_sec": 73118.25850653648, "step_time_sec": 8.239773017005064, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8794, "loss": 3.9023032188415527, "lr": 0.00011824767979540787, "elapsed_sec": 73126.49698019028, "step_time_sec": 8.238273352966644, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8795, "loss": 3.908543825149536, "lr": 0.00011817208514857914, "elapsed_sec": 73134.73605060577, "step_time_sec": 8.238967790035531, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8796, "loss": 4.024604320526123, "lr": 0.00011809648082958539, "elapsed_sec": 73142.97584700584, "step_time_sec": 8.239615557016805, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8797, "loss": 3.9191582202911377, "lr": 0.00011802086688508651, "elapsed_sec": 73151.21393799782, "step_time_sec": 8.237921325024217, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8798, "loss": 3.978985548019409, "lr": 0.00011794524336174828, "elapsed_sec": 73159.45442342758, "step_time_sec": 8.240370644954965, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8799, "loss": 3.8255233764648438, "lr": 0.00011786961030624241, "elapsed_sec": 73167.69248652458, "step_time_sec": 8.237876110011712, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8800, "loss": 3.977607250213623, "lr": 0.00011779396776524658, "elapsed_sec": 73175.93197584152, "step_time_sec": 8.239322329987772, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8801, "loss": 3.872056722640991, "lr": 0.0001177183157854442, "elapsed_sec": 73184.17143464088, "step_time_sec": 8.239324314985424, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8802, "loss": 4.016607284545898, "lr": 0.0001176426544135245, "elapsed_sec": 73192.4115922451, "step_time_sec": 8.239965526969172, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8803, "loss": 3.839252471923828, "lr": 0.00011756698369618266, "elapsed_sec": 73200.6529750824, "step_time_sec": 8.241158897057176, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8804, "loss": 3.9350500106811523, "lr": 0.00011749130368011945, "elapsed_sec": 73208.89298510551, "step_time_sec": 8.239865923998877, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8805, "loss": 3.9296882152557373, "lr": 0.00011741561441204152, "elapsed_sec": 73217.13421320915, "step_time_sec": 8.241064246976748, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8806, "loss": 4.079306602478027, "lr": 0.0001173399159386611, "elapsed_sec": 73225.3733279705, "step_time_sec": 8.23894821794238, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8807, "loss": 3.944411039352417, "lr": 0.00011726420830669623, "elapsed_sec": 73233.6122136116, "step_time_sec": 8.238732322002761, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8808, "loss": 3.909771203994751, "lr": 0.0001171884915628705, "elapsed_sec": 73241.85204648972, "step_time_sec": 8.23966720502358, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8809, "loss": 3.8566393852233887, "lr": 0.00011711276575391317, "elapsed_sec": 73250.09190177917, "step_time_sec": 8.239685515058227, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8810, "loss": 3.8879551887512207, "lr": 0.00011703703092655912, "elapsed_sec": 73258.33172988892, "step_time_sec": 8.2397150460165, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8811, "loss": 3.861783266067505, "lr": 0.0001169612871275487, "elapsed_sec": 73266.57123470306, "step_time_sec": 8.239244759082794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8812, "loss": 3.8249003887176514, "lr": 0.00011688553440362792, "elapsed_sec": 73274.80961561203, "step_time_sec": 8.23823513102252, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8813, "loss": 3.9655601978302, "lr": 0.00011680977280154823, "elapsed_sec": 73283.04792666435, "step_time_sec": 8.23813269904349, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8814, "loss": 3.8881664276123047, "lr": 0.00011673400236806653, "elapsed_sec": 73291.28939914703, "step_time_sec": 8.241315597086214, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8815, "loss": 3.9960062503814697, "lr": 0.00011665822314994522, "elapsed_sec": 73299.52830004692, "step_time_sec": 8.238738287007436, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8816, "loss": 3.858062982559204, "lr": 0.00011658243519395218, "elapsed_sec": 73307.76749944687, "step_time_sec": 8.239008094999008, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8817, "loss": 4.077389717102051, "lr": 0.00011650663854686054, "elapsed_sec": 73316.0057117939, "step_time_sec": 8.238067536032759, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8818, "loss": 3.923105239868164, "lr": 0.00011643083325544885, "elapsed_sec": 73324.23845148087, "step_time_sec": 8.232573186978698, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8819, "loss": 4.062121868133545, "lr": 0.00011635501936650109, "elapsed_sec": 73332.47202801704, "step_time_sec": 8.233395811053924, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8820, "loss": 3.7897536754608154, "lr": 0.00011627919692680641, "elapsed_sec": 73340.70542025566, "step_time_sec": 8.23325691302307, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8821, "loss": 3.841581106185913, "lr": 0.00011620336598315929, "elapsed_sec": 73348.93944764137, "step_time_sec": 8.233829458942637, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8822, "loss": 3.871138334274292, "lr": 0.00011612752658235947, "elapsed_sec": 73357.17342543602, "step_time_sec": 8.23385442502331, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8823, "loss": 3.8294155597686768, "lr": 0.00011605167877121194, "elapsed_sec": 73365.40662193298, "step_time_sec": 8.232980368891731, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8824, "loss": 3.8360466957092285, "lr": 0.00011597582259652682, "elapsed_sec": 73373.64507675171, "step_time_sec": 8.238299797987565, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8825, "loss": 4.078607082366943, "lr": 0.00011589995810511941, "elapsed_sec": 73381.87883210182, "step_time_sec": 8.233568420051597, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8826, "loss": 3.9828009605407715, "lr": 0.00011582408534381011, "elapsed_sec": 73390.1117515564, "step_time_sec": 8.23276549600996, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8827, "loss": 3.7950387001037598, "lr": 0.00011574820435942457, "elapsed_sec": 73398.34557294846, "step_time_sec": 8.23369712405838, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8828, "loss": 3.91113018989563, "lr": 0.00011567231519879327, "elapsed_sec": 73406.5823469162, "step_time_sec": 8.236556560033932, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8829, "loss": 3.8473093509674072, "lr": 0.00011559641790875195, "elapsed_sec": 73414.82071566582, "step_time_sec": 8.238249453017488, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8830, "loss": 3.8697750568389893, "lr": 0.0001155205125361413, "elapsed_sec": 73423.06078386307, "step_time_sec": 8.23989571200218, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8831, "loss": 3.8644773960113525, "lr": 0.00011544459912780696, "elapsed_sec": 73431.2991695404, "step_time_sec": 8.238205775036477, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8832, "loss": 3.9076552391052246, "lr": 0.00011536867773059954, "elapsed_sec": 73439.53834748268, "step_time_sec": 8.239004347007722, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8833, "loss": 4.121824741363525, "lr": 0.00011529274839137464, "elapsed_sec": 73447.77637982368, "step_time_sec": 8.237858926993795, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8834, "loss": 3.8084309101104736, "lr": 0.00011521681115699268, "elapsed_sec": 73456.01579356194, "step_time_sec": 8.239275535102934, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8835, "loss": 4.052783966064453, "lr": 0.00011514086607431903, "elapsed_sec": 73464.25525665283, "step_time_sec": 8.239274890045635, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8836, "loss": 4.1254167556762695, "lr": 0.0001150649131902238, "elapsed_sec": 73472.49509334564, "step_time_sec": 8.239635966951028, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8837, "loss": 3.9453723430633545, "lr": 0.00011498895255158206, "elapsed_sec": 73480.73511052132, "step_time_sec": 8.239854812971316, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8838, "loss": 3.9440526962280273, "lr": 0.00011491298420527354, "elapsed_sec": 73488.97361469269, "step_time_sec": 8.238356687012129, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8839, "loss": 3.819070816040039, "lr": 0.00011483700819818277, "elapsed_sec": 73497.2067861557, "step_time_sec": 8.233035854063928, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8840, "loss": 4.68446159362793, "lr": 0.00011476102457719905, "elapsed_sec": 73505.43897151947, "step_time_sec": 8.231968830921687, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8841, "loss": 3.911463975906372, "lr": 0.00011468503338921628, "elapsed_sec": 73513.67405796051, "step_time_sec": 8.234965842915699, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8842, "loss": 3.9401187896728516, "lr": 0.00011460903468113314, "elapsed_sec": 73521.90681934357, "step_time_sec": 8.232563689001836, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8843, "loss": 3.891446590423584, "lr": 0.00011453302849985285, "elapsed_sec": 73530.13975906372, "step_time_sec": 8.232824843027629, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8844, "loss": 3.953563690185547, "lr": 0.00011445701489228333, "elapsed_sec": 73538.37677454948, "step_time_sec": 8.236871076049283, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8845, "loss": 4.022687911987305, "lr": 0.00011438099390533703, "elapsed_sec": 73546.61037015915, "step_time_sec": 8.233375740004703, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8846, "loss": 3.8812472820281982, "lr": 0.00011430496558593099, "elapsed_sec": 73554.84232354164, "step_time_sec": 8.231789484038018, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8847, "loss": 4.022461414337158, "lr": 0.00011422892998098677, "elapsed_sec": 73563.07640695572, "step_time_sec": 8.233914861921221, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8848, "loss": 3.9458868503570557, "lr": 0.00011415288713743033, "elapsed_sec": 73571.3082857132, "step_time_sec": 8.231727543985471, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8849, "loss": 3.85831880569458, "lr": 0.00011407683710219228, "elapsed_sec": 73579.5422194004, "step_time_sec": 8.233750591054559, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8850, "loss": 3.8136706352233887, "lr": 0.0001140007799222075, "elapsed_sec": 73587.78109908104, "step_time_sec": 8.238708500983194, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8851, "loss": 3.9227004051208496, "lr": 0.00011392471564441535, "elapsed_sec": 73596.02046585083, "step_time_sec": 8.239193751011044, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8852, "loss": 3.911085605621338, "lr": 0.00011384864431575958, "elapsed_sec": 73604.25875473022, "step_time_sec": 8.238165838993154, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8853, "loss": 4.067852973937988, "lr": 0.00011377256598318832, "elapsed_sec": 73612.49714756012, "step_time_sec": 8.238189677940682, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8854, "loss": 4.137302398681641, "lr": 0.00011369648069365393, "elapsed_sec": 73620.73520755768, "step_time_sec": 8.237880485015921, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8855, "loss": 3.940220832824707, "lr": 0.0001136203884941131, "elapsed_sec": 73628.97245049477, "step_time_sec": 8.23710626293905, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8856, "loss": 3.9087836742401123, "lr": 0.00011354428943152684, "elapsed_sec": 73637.21241998672, "step_time_sec": 8.23983117705211, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8857, "loss": 3.7324578762054443, "lr": 0.00011346818355286038, "elapsed_sec": 73645.45127081871, "step_time_sec": 8.238644789089449, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8858, "loss": 3.973268508911133, "lr": 0.00011339207090508305, "elapsed_sec": 73653.69064188004, "step_time_sec": 8.239236045978032, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8859, "loss": 4.028547286987305, "lr": 0.00011331595153516848, "elapsed_sec": 73661.93135094643, "step_time_sec": 8.240510909003206, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8860, "loss": 3.9903411865234375, "lr": 0.00011323982549009446, "elapsed_sec": 73670.16848158836, "step_time_sec": 8.236969088902697, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8861, "loss": 3.8352878093719482, "lr": 0.00011316369281684277, "elapsed_sec": 73678.40808486938, "step_time_sec": 8.239485646947287, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8862, "loss": 3.9291346073150635, "lr": 0.00011308755356239937, "elapsed_sec": 73686.64761161804, "step_time_sec": 8.239371000090614, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8863, "loss": 4.0599541664123535, "lr": 0.00011301140777375428, "elapsed_sec": 73694.88669776917, "step_time_sec": 8.238880387041718, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8864, "loss": 3.962486982345581, "lr": 0.00011293525549790158, "elapsed_sec": 73703.12515163422, "step_time_sec": 8.23827948095277, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8865, "loss": 3.8512346744537354, "lr": 0.00011285909678183928, "elapsed_sec": 73711.36423182487, "step_time_sec": 8.238948430982418, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8866, "loss": 3.9701993465423584, "lr": 0.0001127829316725694, "elapsed_sec": 73719.60377883911, "step_time_sec": 8.239386021043174, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8867, "loss": 3.8515167236328125, "lr": 0.0001127067602170979, "elapsed_sec": 73727.84201073647, "step_time_sec": 8.238047348917462, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8868, "loss": 3.9167115688323975, "lr": 0.0001126305824624347, "elapsed_sec": 73736.07977557182, "step_time_sec": 8.237609145930037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8869, "loss": 3.887524366378784, "lr": 0.00011255439845559352, "elapsed_sec": 73744.31836247444, "step_time_sec": 8.238433665013872, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8870, "loss": 3.82182240486145, "lr": 0.00011247820824359201, "elapsed_sec": 73752.54988694191, "step_time_sec": 8.231370352092199, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8871, "loss": 3.942986249923706, "lr": 0.00011240201187345164, "elapsed_sec": 73760.78393292427, "step_time_sec": 8.233936254982837, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8872, "loss": 3.9162676334381104, "lr": 0.00011232580939219768, "elapsed_sec": 73769.02163672447, "step_time_sec": 8.2375701230485, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8873, "loss": 3.854189395904541, "lr": 0.00011224960084685911, "elapsed_sec": 73777.25540232658, "step_time_sec": 8.233592709060758, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8874, "loss": 3.8445143699645996, "lr": 0.00011217338628446877, "elapsed_sec": 73785.48789668083, "step_time_sec": 8.232271113083698, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8875, "loss": 3.764787197113037, "lr": 0.0001120971657520631, "elapsed_sec": 73793.72193956375, "step_time_sec": 8.233930928981863, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8876, "loss": 3.8693478107452393, "lr": 0.00011202093929668227, "elapsed_sec": 73801.9602572918, "step_time_sec": 8.23810534295626, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8877, "loss": 3.7859413623809814, "lr": 0.00011194470696537014, "elapsed_sec": 73810.20036554337, "step_time_sec": 8.239940536092035, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8878, "loss": 3.810612201690674, "lr": 0.00011186846880517412, "elapsed_sec": 73818.44226574898, "step_time_sec": 8.241791312932037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8879, "loss": 3.9839670658111572, "lr": 0.00011179222486314529, "elapsed_sec": 73826.68150877953, "step_time_sec": 8.239102969062515, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8880, "loss": 3.7544336318969727, "lr": 0.00011171597518633827, "elapsed_sec": 73834.9223253727, "step_time_sec": 8.240583167062141, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8881, "loss": 3.9333579540252686, "lr": 0.0001116397198218112, "elapsed_sec": 73843.16216945648, "step_time_sec": 8.239686877001077, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8882, "loss": 3.9459829330444336, "lr": 0.00011156345881662575, "elapsed_sec": 73851.40177106857, "step_time_sec": 8.239447467960417, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8883, "loss": 4.002512454986572, "lr": 0.00011148719221784709, "elapsed_sec": 73859.6402823925, "step_time_sec": 8.23835590807721, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8884, "loss": 3.755207061767578, "lr": 0.00011141092007254382, "elapsed_sec": 73867.87232995033, "step_time_sec": 8.231893839896657, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8885, "loss": 3.9587438106536865, "lr": 0.0001113346424277879, "elapsed_sec": 73876.10631537437, "step_time_sec": 8.233812483027577, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8886, "loss": 3.8984270095825195, "lr": 0.00011125835933065484, "elapsed_sec": 73884.33962202072, "step_time_sec": 8.233118405099958, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8887, "loss": 4.091635704040527, "lr": 0.00011118207082822342, "elapsed_sec": 73892.57382941246, "step_time_sec": 8.234052576939575, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8888, "loss": 3.8533833026885986, "lr": 0.00011110577696757567, "elapsed_sec": 73900.80767726898, "step_time_sec": 8.233683421975002, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8889, "loss": 3.931918144226074, "lr": 0.00011102947779579706, "elapsed_sec": 73909.04212665558, "step_time_sec": 8.234310481930152, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8890, "loss": 3.9320926666259766, "lr": 0.00011095317335997635, "elapsed_sec": 73917.28021001816, "step_time_sec": 8.237948747933842, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8891, "loss": 3.9332761764526367, "lr": 0.00011087686370720545, "elapsed_sec": 73925.52130198479, "step_time_sec": 8.240872164955363, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8892, "loss": 3.881359338760376, "lr": 0.00011080054888457948, "elapsed_sec": 73933.76225447655, "step_time_sec": 8.24085580301471, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8893, "loss": 3.847538709640503, "lr": 0.00011072422893919686, "elapsed_sec": 73942.00323176384, "step_time_sec": 8.24078172899317, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8894, "loss": 3.929023027420044, "lr": 0.00011064790391815913, "elapsed_sec": 73950.23770165443, "step_time_sec": 8.234382597962394, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8895, "loss": 3.853571891784668, "lr": 0.00011057157386857093, "elapsed_sec": 73958.48847770691, "step_time_sec": 8.247861991985701, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8896, "loss": 4.034526348114014, "lr": 0.00011049523883754, "elapsed_sec": 73966.72215890884, "step_time_sec": 8.233482102979906, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8897, "loss": 3.802516222000122, "lr": 0.0001104188988721772, "elapsed_sec": 73974.9557685852, "step_time_sec": 8.233373309019953, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8898, "loss": 3.8489274978637695, "lr": 0.0001103425540195964, "elapsed_sec": 73983.18986392021, "step_time_sec": 8.233922375016846, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8899, "loss": 3.9828453063964844, "lr": 0.00011026620432691448, "elapsed_sec": 73991.42293477058, "step_time_sec": 8.232953713042662, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8900, "loss": 3.9475109577178955, "lr": 0.00011018984984125135, "elapsed_sec": 73999.65657377243, "step_time_sec": 8.233463436015882, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8901, "loss": 3.9816694259643555, "lr": 0.00011011349060972985, "elapsed_sec": 74007.8886539936, "step_time_sec": 8.231901815976016, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8902, "loss": 3.9662749767303467, "lr": 0.00011003712667947575, "elapsed_sec": 74016.12268280983, "step_time_sec": 8.233867608942091, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8903, "loss": 3.9111101627349854, "lr": 0.00010996075809761766, "elapsed_sec": 74024.35626482964, "step_time_sec": 8.233424915000796, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8904, "loss": 3.7863452434539795, "lr": 0.00010988438491128722, "elapsed_sec": 74032.5889647007, "step_time_sec": 8.232564651058055, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8905, "loss": 3.8643221855163574, "lr": 0.00010980800716761877, "elapsed_sec": 74040.82123088837, "step_time_sec": 8.23213277803734, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8906, "loss": 3.8824644088745117, "lr": 0.00010973162491374949, "elapsed_sec": 74049.06016469002, "step_time_sec": 8.238753766054288, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8907, "loss": 3.9963245391845703, "lr": 0.00010965523819681937, "elapsed_sec": 74057.29891514778, "step_time_sec": 8.238619067938998, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8908, "loss": 3.8574023246765137, "lr": 0.00010957884706397114, "elapsed_sec": 74065.53647184372, "step_time_sec": 8.237329130992293, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8909, "loss": 4.00602388381958, "lr": 0.00010950245156235027, "elapsed_sec": 74073.7762811184, "step_time_sec": 8.239692590897903, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8910, "loss": 3.7531075477600098, "lr": 0.00010942605173910491, "elapsed_sec": 74082.01517915726, "step_time_sec": 8.238700083107688, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8911, "loss": 3.9062600135803223, "lr": 0.00010934964764138587, "elapsed_sec": 74090.25576424599, "step_time_sec": 8.240412206971087, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8912, "loss": 3.9251670837402344, "lr": 0.0001092732393163466, "elapsed_sec": 74098.4958331585, "step_time_sec": 8.239958138088696, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8913, "loss": 3.9525129795074463, "lr": 0.00010919682681114322, "elapsed_sec": 74106.73354482651, "step_time_sec": 8.237596567021683, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8914, "loss": 4.069599151611328, "lr": 0.00010912041017293434, "elapsed_sec": 74114.97340297699, "step_time_sec": 8.2396667819703, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8915, "loss": 3.8143343925476074, "lr": 0.00010904398944888116, "elapsed_sec": 74123.21277928352, "step_time_sec": 8.239131470094435, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8916, "loss": 3.8011538982391357, "lr": 0.0001089675646861474, "elapsed_sec": 74131.45258069038, "step_time_sec": 8.239694474032149, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8917, "loss": 3.958069086074829, "lr": 0.00010889113593189933, "elapsed_sec": 74139.69139170647, "step_time_sec": 8.23862366809044, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8918, "loss": 3.799896001815796, "lr": 0.00010881470323330552, "elapsed_sec": 74147.92300271988, "step_time_sec": 8.23143156094011, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8919, "loss": 3.780451774597168, "lr": 0.00010873826663753718, "elapsed_sec": 74156.15670776367, "step_time_sec": 8.2335627800785, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8920, "loss": 4.09577751159668, "lr": 0.00010866182619176779, "elapsed_sec": 74164.38908171654, "step_time_sec": 8.232189771952108, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8921, "loss": 3.8803670406341553, "lr": 0.0001085853819431732, "elapsed_sec": 74172.62210154533, "step_time_sec": 8.232870067004114, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8922, "loss": 4.179283142089844, "lr": 0.0001085089339389317, "elapsed_sec": 74192.26030015945, "step_time_sec": 19.63803463091608, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8923, "loss": 3.911233425140381, "lr": 0.00010843248222622386, "elapsed_sec": 74200.49194741249, "step_time_sec": 8.23147745395545, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8924, "loss": 3.9553229808807373, "lr": 0.0001083560268522325, "elapsed_sec": 74208.72570133209, "step_time_sec": 8.233654745970853, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8925, "loss": 3.9300968647003174, "lr": 0.00010827956786414271, "elapsed_sec": 74216.96365618706, "step_time_sec": 8.237810287973844, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8926, "loss": 3.9598734378814697, "lr": 0.00010820310530914183, "elapsed_sec": 74225.20268034935, "step_time_sec": 8.238797024008818, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8927, "loss": 3.9847049713134766, "lr": 0.00010812663923441945, "elapsed_sec": 74233.43489384651, "step_time_sec": 8.23206798103638, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8928, "loss": 3.817774534225464, "lr": 0.00010805016968716723, "elapsed_sec": 74241.6684653759, "step_time_sec": 8.2334116010461, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8929, "loss": 3.9495558738708496, "lr": 0.000107973696714579, "elapsed_sec": 74249.9002532959, "step_time_sec": 8.231656342046335, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8930, "loss": 3.9637229442596436, "lr": 0.0001078972203638508, "elapsed_sec": 74258.1347720623, "step_time_sec": 8.234403109992854, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8931, "loss": 3.920341730117798, "lr": 0.00010782074068218064, "elapsed_sec": 74266.36760139465, "step_time_sec": 8.232609481899999, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8932, "loss": 3.89770245552063, "lr": 0.00010774425771676862, "elapsed_sec": 74274.60101056099, "step_time_sec": 8.23328646295704, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8933, "loss": 3.9983999729156494, "lr": 0.00010766777151481686, "elapsed_sec": 74282.83552527428, "step_time_sec": 8.234308357932605, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8934, "loss": 4.072080612182617, "lr": 0.00010759128212352954, "elapsed_sec": 74291.06893706322, "step_time_sec": 8.233233061037026, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8935, "loss": 3.8967559337615967, "lr": 0.00010751478959011275, "elapsed_sec": 74299.30248904228, "step_time_sec": 8.233419291907921, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8936, "loss": 4.027994155883789, "lr": 0.00010743829396177449, "elapsed_sec": 74307.53555345535, "step_time_sec": 8.2328547189245, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8937, "loss": 3.9039950370788574, "lr": 0.00010736179528572476, "elapsed_sec": 74315.76947879791, "step_time_sec": 8.233748905011453, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8938, "loss": 3.8787593841552734, "lr": 0.00010728529360917533, "elapsed_sec": 74324.00268411636, "step_time_sec": 8.233075772994198, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8939, "loss": 4.0163679122924805, "lr": 0.00010720878897933993, "elapsed_sec": 74332.23591184616, "step_time_sec": 8.233044795924798, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8940, "loss": 3.8652143478393555, "lr": 0.00010713228144343402, "elapsed_sec": 74340.47107982635, "step_time_sec": 8.235005852999166, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8941, "loss": 3.925166368484497, "lr": 0.00010705577104867496, "elapsed_sec": 74348.70459032059, "step_time_sec": 8.233320428058505, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8942, "loss": 4.015835285186768, "lr": 0.00010697925784228177, "elapsed_sec": 74356.93780851364, "step_time_sec": 8.233099414035678, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8943, "loss": 3.813523769378662, "lr": 0.00010690274187147525, "elapsed_sec": 74365.17183756828, "step_time_sec": 8.233807830954902, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8944, "loss": 3.937248706817627, "lr": 0.00010682622318347793, "elapsed_sec": 74373.4052991867, "step_time_sec": 8.23328731500078, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8945, "loss": 3.872981071472168, "lr": 0.00010674970182551395, "elapsed_sec": 74381.6379404068, "step_time_sec": 8.23248121002689, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8946, "loss": 3.810739517211914, "lr": 0.00010667317784480915, "elapsed_sec": 74389.87248444557, "step_time_sec": 8.234384089009836, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8947, "loss": 3.993201971054077, "lr": 0.00010659665128859103, "elapsed_sec": 74398.11118674278, "step_time_sec": 8.238533313036896, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8948, "loss": 3.931309938430786, "lr": 0.00010652012220408854, "elapsed_sec": 74406.35125923157, "step_time_sec": 8.239923635032028, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8949, "loss": 3.9256198406219482, "lr": 0.00010644359063853233, "elapsed_sec": 74414.59001278877, "step_time_sec": 8.238569276989438, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8950, "loss": 3.876436710357666, "lr": 0.00010636705663915453, "elapsed_sec": 74422.82947134972, "step_time_sec": 8.239315357990563, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8951, "loss": 3.8351147174835205, "lr": 0.00010629052025318874, "elapsed_sec": 74431.06787109375, "step_time_sec": 8.238241387996823, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8952, "loss": 3.818528413772583, "lr": 0.00010621398152787007, "elapsed_sec": 74439.30767083168, "step_time_sec": 8.239652094081976, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8953, "loss": 3.9783644676208496, "lr": 0.00010613744051043506, "elapsed_sec": 74447.54847478867, "step_time_sec": 8.240607767016627, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8954, "loss": 3.8175227642059326, "lr": 0.00010606089724812176, "elapsed_sec": 74455.78771138191, "step_time_sec": 8.239092587959021, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8955, "loss": 3.770883321762085, "lr": 0.00010598435178816938, "elapsed_sec": 74464.02147102356, "step_time_sec": 8.23364527092781, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8956, "loss": 4.104328155517578, "lr": 0.0001059078041778187, "elapsed_sec": 74472.25481772423, "step_time_sec": 8.233140829019248, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8957, "loss": 3.9607486724853516, "lr": 0.00010583125446431176, "elapsed_sec": 74480.49019432068, "step_time_sec": 8.235215932014398, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8958, "loss": 4.215174198150635, "lr": 0.00010575470269489185, "elapsed_sec": 74488.72994875908, "step_time_sec": 8.239602204994299, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8959, "loss": 3.9246037006378174, "lr": 0.00010567814891680354, "elapsed_sec": 74496.96961522102, "step_time_sec": 8.239517605048604, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8960, "loss": 4.017494201660156, "lr": 0.00010560159317729276, "elapsed_sec": 74505.21043086052, "step_time_sec": 8.240610440960154, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8961, "loss": 3.849216938018799, "lr": 0.0001055250355236065, "elapsed_sec": 74513.44815254211, "step_time_sec": 8.23755712097045, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8962, "loss": 3.850921630859375, "lr": 0.000105448476002993, "elapsed_sec": 74521.68808746338, "step_time_sec": 8.23974268801976, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8963, "loss": 3.9302945137023926, "lr": 0.00010537191466270161, "elapsed_sec": 74529.92777729034, "step_time_sec": 8.239503527991474, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8964, "loss": 4.043667793273926, "lr": 0.00010529535154998288, "elapsed_sec": 74538.16616916656, "step_time_sec": 8.238222373998724, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8965, "loss": 3.98730206489563, "lr": 0.00010521878671208837, "elapsed_sec": 74546.40562653542, "step_time_sec": 8.23929615796078, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8966, "loss": 3.964440107345581, "lr": 0.00010514222019627077, "elapsed_sec": 74554.64453864098, "step_time_sec": 8.23876198893413, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8967, "loss": 3.9459590911865234, "lr": 0.00010506565204978377, "elapsed_sec": 74562.88454389572, "step_time_sec": 8.239832935039885, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8968, "loss": 3.9025731086730957, "lr": 0.00010498908231988204, "elapsed_sec": 74571.12399601936, "step_time_sec": 8.239290503086522, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8969, "loss": 3.9153099060058594, "lr": 0.00010491251105382132, "elapsed_sec": 74579.35606431961, "step_time_sec": 8.231936702970415, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8970, "loss": 4.085555553436279, "lr": 0.0001048359382988582, "elapsed_sec": 74587.58987689018, "step_time_sec": 8.23360076698009, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8971, "loss": 3.9710264205932617, "lr": 0.00010475936410225025, "elapsed_sec": 74595.8234269619, "step_time_sec": 8.233435702044517, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8972, "loss": 3.983879566192627, "lr": 0.00010468278851125588, "elapsed_sec": 74604.0628914833, "step_time_sec": 8.239280737936497, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8973, "loss": 3.8582041263580322, "lr": 0.00010460621157313443, "elapsed_sec": 74612.30172801018, "step_time_sec": 8.238643202930689, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8974, "loss": 3.9716835021972656, "lr": 0.00010452963333514601, "elapsed_sec": 74620.54102158546, "step_time_sec": 8.239111037924886, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8975, "loss": 3.868347406387329, "lr": 0.00010445305384455151, "elapsed_sec": 74628.78181695938, "step_time_sec": 8.240660396055318, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8976, "loss": 4.066086292266846, "lr": 0.0001043764731486127, "elapsed_sec": 74637.02008676529, "step_time_sec": 8.23812245100271, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8977, "loss": 4.0036396980285645, "lr": 0.00010429989129459199, "elapsed_sec": 74645.25367951393, "step_time_sec": 8.233397798030637, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8978, "loss": 3.8915019035339355, "lr": 0.0001042233083297526, "elapsed_sec": 74653.48689675331, "step_time_sec": 8.233128538005985, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8979, "loss": 3.9184117317199707, "lr": 0.00010414672430135827, "elapsed_sec": 74661.72544050217, "step_time_sec": 8.238325378973968, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8980, "loss": 3.8914601802825928, "lr": 0.00010407013925667361, "elapsed_sec": 74669.95845794678, "step_time_sec": 8.232840146985836, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8981, "loss": 3.9045159816741943, "lr": 0.00010399355324296373, "elapsed_sec": 74678.19296336174, "step_time_sec": 8.234338358975947, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8982, "loss": 3.887669801712036, "lr": 0.0001039169663074943, "elapsed_sec": 74686.42609810829, "step_time_sec": 8.232969349017367, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8983, "loss": 3.8860561847686768, "lr": 0.0001038403784975317, "elapsed_sec": 74694.66602754593, "step_time_sec": 8.239774949033745, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8984, "loss": 3.922358989715576, "lr": 0.00010376378986034277, "elapsed_sec": 74702.90660524368, "step_time_sec": 8.240402056952007, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8985, "loss": 3.8754971027374268, "lr": 0.00010368720044319477, "elapsed_sec": 74711.14569306374, "step_time_sec": 8.238938770024106, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8986, "loss": 4.127679347991943, "lr": 0.00010361061029335558, "elapsed_sec": 74719.38678073883, "step_time_sec": 8.240957561996765, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8987, "loss": 3.9639601707458496, "lr": 0.00010353401945809352, "elapsed_sec": 74727.61970543861, "step_time_sec": 8.232745034969412, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8988, "loss": 3.930051803588867, "lr": 0.0001034574279846773, "elapsed_sec": 74735.8539750576, "step_time_sec": 8.234124666079879, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8989, "loss": 3.8727526664733887, "lr": 0.00010338083592037593, "elapsed_sec": 74744.08713817596, "step_time_sec": 8.233020513085648, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8990, "loss": 3.8698153495788574, "lr": 0.00010330424331245895, "elapsed_sec": 74752.32222914696, "step_time_sec": 8.234917972935364, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8991, "loss": 3.9571380615234375, "lr": 0.00010322765020819617, "elapsed_sec": 74760.55540442467, "step_time_sec": 8.233049436006695, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8992, "loss": 3.8629956245422363, "lr": 0.00010315105665485766, "elapsed_sec": 74768.79049658775, "step_time_sec": 8.234894187073223, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8993, "loss": 3.9742321968078613, "lr": 0.0001030744626997138, "elapsed_sec": 74777.02318692207, "step_time_sec": 8.232524437946267, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8994, "loss": 3.936123847961426, "lr": 0.00010299786839003524, "elapsed_sec": 74785.25780916214, "step_time_sec": 8.234459967934527, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8995, "loss": 3.764385461807251, "lr": 0.00010292127377309287, "elapsed_sec": 74793.49091029167, "step_time_sec": 8.232914312044159, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8996, "loss": 3.901270627975464, "lr": 0.00010284467889615764, "elapsed_sec": 74801.73168373108, "step_time_sec": 8.240625391015783, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8997, "loss": 3.9881088733673096, "lr": 0.00010276808380650082, "elapsed_sec": 74809.97157740593, "step_time_sec": 8.23971034202259, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8998, "loss": 3.7766692638397217, "lr": 0.00010269148855139373, "elapsed_sec": 74818.210272789, "step_time_sec": 8.238542378065176, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8999, "loss": 3.9457428455352783, "lr": 0.00010261489317810781, "elapsed_sec": 74826.45008897781, "step_time_sec": 8.239648237009533, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9000, "loss": 3.8474369049072266, "lr": 0.00010253829773391451, "elapsed_sec": 74834.68965101242, "step_time_sec": 51.17239910806529, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": true, "validation_ran": true, "probe_ran": true, "probe_elapsed_sec": 1.0124574520159513, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9001, "loss": 3.9753782749176025, "lr": 0.00010246170226608546, "elapsed_sec": 74885.86803150177, "step_time_sec": 8.245243669953197, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9002, "loss": 4.08096170425415, "lr": 0.0001023851068218922, "elapsed_sec": 74894.10082626343, "step_time_sec": 8.23261810594704, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9003, "loss": 3.8886160850524902, "lr": 0.0001023085114486063, "elapsed_sec": 74902.33234024048, "step_time_sec": 8.231357382028364, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9004, "loss": 3.994823932647705, "lr": 0.00010223191619349916, "elapsed_sec": 74910.56675672531, "step_time_sec": 8.234278904041275, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9005, "loss": 4.0422258377075195, "lr": 0.00010215532110384237, "elapsed_sec": 74918.7990334034, "step_time_sec": 8.232076482032426, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9006, "loss": 4.095229148864746, "lr": 0.00010207872622690716, "elapsed_sec": 74927.03197693825, "step_time_sec": 8.232793779927306, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9007, "loss": 4.0197553634643555, "lr": 0.00010200213160996475, "elapsed_sec": 74935.26489257812, "step_time_sec": 8.23274413193576, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9008, "loss": 4.056230068206787, "lr": 0.00010192553730028621, "elapsed_sec": 74943.49868392944, "step_time_sec": 8.233632210991345, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9009, "loss": 3.96570086479187, "lr": 0.00010184894334514238, "elapsed_sec": 74951.73122215271, "step_time_sec": 8.232369001954794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9010, "loss": 3.9341039657592773, "lr": 0.00010177234979180385, "elapsed_sec": 74959.9656765461, "step_time_sec": 8.23428871599026, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9011, "loss": 3.980905055999756, "lr": 0.00010169575668754106, "elapsed_sec": 74968.19800782204, "step_time_sec": 8.232108455966227, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9012, "loss": 3.7909035682678223, "lr": 0.00010161916407962405, "elapsed_sec": 74976.43139433861, "step_time_sec": 8.233276925049722, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9013, "loss": 3.942711353302002, "lr": 0.00010154257201532272, "elapsed_sec": 74984.66369771957, "step_time_sec": 8.232139940024354, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9014, "loss": 3.9417150020599365, "lr": 0.00010146598054190649, "elapsed_sec": 74992.89746689796, "step_time_sec": 8.233572143944912, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9015, "loss": 3.950223922729492, "lr": 0.0001013893897066444, "elapsed_sec": 75001.13014698029, "step_time_sec": 8.2325157050509, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9016, "loss": 3.862018346786499, "lr": 0.00010131279955680524, "elapsed_sec": 75009.36306214333, "step_time_sec": 8.232797489035875, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9017, "loss": 3.8540151119232178, "lr": 0.00010123621013965727, "elapsed_sec": 75017.5956170559, "step_time_sec": 8.232340837013908, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9018, "loss": 4.0208330154418945, "lr": 0.00010115962150246829, "elapsed_sec": 75025.8274166584, "step_time_sec": 8.231638080906123, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9019, "loss": 3.8615946769714355, "lr": 0.00010108303369250569, "elapsed_sec": 75034.066365242, "step_time_sec": 8.23877212498337, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9020, "loss": 3.8297224044799805, "lr": 0.00010100644675703629, "elapsed_sec": 75042.30692005157, "step_time_sec": 8.240406655007973, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9021, "loss": 3.9179210662841797, "lr": 0.00010092986074332638, "elapsed_sec": 75050.54613757133, "step_time_sec": 8.239040129934438, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9022, "loss": 3.8241965770721436, "lr": 0.00010085327569864172, "elapsed_sec": 75058.78612732887, "step_time_sec": 8.239827805897221, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9023, "loss": 3.865318536758423, "lr": 0.00010077669167024744, "elapsed_sec": 75067.025411129, "step_time_sec": 8.23914116609376, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9024, "loss": 3.887427568435669, "lr": 0.00010070010870540799, "elapsed_sec": 75075.26455783844, "step_time_sec": 8.238921133917756, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9025, "loss": 3.8347058296203613, "lr": 0.00010062352685138732, "elapsed_sec": 75083.5049264431, "step_time_sec": 8.240204478963278, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9026, "loss": 3.7777459621429443, "lr": 0.00010054694615544851, "elapsed_sec": 75091.74218678474, "step_time_sec": 8.237133534974419, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9027, "loss": 3.832191228866577, "lr": 0.000100470366664854, "elapsed_sec": 75099.98191666603, "step_time_sec": 8.239615464000963, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9028, "loss": 3.8365800380706787, "lr": 0.00010039378842686558, "elapsed_sec": 75108.22173714638, "step_time_sec": 8.239677328965627, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9029, "loss": 3.8836987018585205, "lr": 0.00010031721148874413, "elapsed_sec": 75116.45997619629, "step_time_sec": 8.238018934964202, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9030, "loss": 3.8530073165893555, "lr": 0.00010024063589774976, "elapsed_sec": 75124.69418883324, "step_time_sec": 8.234053610009141, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9031, "loss": 3.831193685531616, "lr": 0.00010016406170114181, "elapsed_sec": 75132.92705368996, "step_time_sec": 8.23269859701395, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9032, "loss": 3.8232691287994385, "lr": 0.00010008748894617868, "elapsed_sec": 75141.1642036438, "step_time_sec": 8.237016301020049, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9033, "loss": 3.85369610786438, "lr": 0.00010001091768011797, "elapsed_sec": 75149.41497945786, "step_time_sec": 8.250569698982872, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9034, "loss": 3.9621734619140625, "lr": 9.993434795021627e-05, "elapsed_sec": 75157.65367650986, "step_time_sec": 8.2384940009797, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9035, "loss": 3.9265403747558594, "lr": 9.985777980372923e-05, "elapsed_sec": 75165.88755774498, "step_time_sec": 8.233720550895669, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9036, "loss": 3.9292352199554443, "lr": 9.978121328791165e-05, "elapsed_sec": 75174.12005448341, "step_time_sec": 8.23231711203698, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9037, "loss": 3.896132469177246, "lr": 9.970464845001716e-05, "elapsed_sec": 75182.3536093235, "step_time_sec": 8.23342581500765, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9038, "loss": 3.8874382972717285, "lr": 9.962808533729837e-05, "elapsed_sec": 75190.59146261215, "step_time_sec": 8.237706865067594, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9039, "loss": 3.846630573272705, "lr": 9.955152399700702e-05, "elapsed_sec": 75198.83200597763, "step_time_sec": 8.240389678976499, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9040, "loss": 3.772113084793091, "lr": 9.947496447639352e-05, "elapsed_sec": 75207.07175779343, "step_time_sec": 8.239590003038757, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9041, "loss": 3.904592275619507, "lr": 9.939840682270722e-05, "elapsed_sec": 75215.31135630608, "step_time_sec": 8.239403584972024, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9042, "loss": 3.7457287311553955, "lr": 9.932185108319645e-05, "elapsed_sec": 75223.5524199009, "step_time_sec": 8.240874364972115, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9043, "loss": 3.847200870513916, "lr": 9.924529730510819e-05, "elapsed_sec": 75231.79054093361, "step_time_sec": 8.237926386063918, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9044, "loss": 3.8499951362609863, "lr": 9.916874553568826e-05, "elapsed_sec": 75240.03093028069, "step_time_sec": 8.240238990983926, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9045, "loss": 3.959054946899414, "lr": 9.909219582218131e-05, "elapsed_sec": 75248.27099132538, "step_time_sec": 8.23991030396428, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9046, "loss": 3.8707938194274902, "lr": 9.901564821183065e-05, "elapsed_sec": 75256.51002216339, "step_time_sec": 8.238916623988189, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9047, "loss": 4.018327236175537, "lr": 9.893910275187825e-05, "elapsed_sec": 75264.74997901917, "step_time_sec": 8.239719163975678, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9048, "loss": 3.909493923187256, "lr": 9.886255948956494e-05, "elapsed_sec": 75272.98834609985, "step_time_sec": 8.238190824980848, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9049, "loss": 3.9748008251190186, "lr": 9.878601847212991e-05, "elapsed_sec": 75281.22807121277, "step_time_sec": 8.23956185404677, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9050, "loss": 3.756545066833496, "lr": 9.870947974681127e-05, "elapsed_sec": 75289.4668226242, "step_time_sec": 8.238572300993837, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9051, "loss": 3.849963665008545, "lr": 9.86329433608455e-05, "elapsed_sec": 75297.70445227623, "step_time_sec": 8.237457176088355, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9052, "loss": 3.922745943069458, "lr": 9.855640936146767e-05, "elapsed_sec": 75305.94386219978, "step_time_sec": 8.239189623040147, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9053, "loss": 3.8965771198272705, "lr": 9.847987779591146e-05, "elapsed_sec": 75314.17707014084, "step_time_sec": 8.233042384963483, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9054, "loss": 4.056118488311768, "lr": 9.840334871140899e-05, "elapsed_sec": 75322.41124463081, "step_time_sec": 8.234015961061232, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9055, "loss": 3.8565917015075684, "lr": 9.832682215519083e-05, "elapsed_sec": 75330.64420700073, "step_time_sec": 8.232778863050044, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9056, "loss": 3.856421709060669, "lr": 9.825029817448605e-05, "elapsed_sec": 75338.8857767582, "step_time_sec": 8.241419060970657, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9057, "loss": 3.886089324951172, "lr": 9.817377681652209e-05, "elapsed_sec": 75347.12421393394, "step_time_sec": 8.238251167000271, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9058, "loss": 3.9006266593933105, "lr": 9.809725812852472e-05, "elapsed_sec": 75355.36266040802, "step_time_sec": 8.23829483310692, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9059, "loss": 3.770085096359253, "lr": 9.802074215771824e-05, "elapsed_sec": 75363.60245919228, "step_time_sec": 8.23964404000435, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9060, "loss": 3.971256732940674, "lr": 9.794422895132506e-05, "elapsed_sec": 75371.83555078506, "step_time_sec": 8.232911013998091, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9061, "loss": 3.840352773666382, "lr": 9.786771855656595e-05, "elapsed_sec": 75380.06957960129, "step_time_sec": 8.233863575034775, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9062, "loss": 3.900864839553833, "lr": 9.779121102066008e-05, "elapsed_sec": 75388.3030591011, "step_time_sec": 8.233292149961926, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9063, "loss": 3.8418304920196533, "lr": 9.77147063908247e-05, "elapsed_sec": 75396.53789830208, "step_time_sec": 8.234682264039293, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9064, "loss": 3.8684582710266113, "lr": 9.763820471427525e-05, "elapsed_sec": 75404.77025794983, "step_time_sec": 8.23219255800359, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9065, "loss": 3.8432228565216064, "lr": 9.756170603822552e-05, "elapsed_sec": 75413.00352668762, "step_time_sec": 8.233102121972479, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9066, "loss": 3.9345076084136963, "lr": 9.748521040988727e-05, "elapsed_sec": 75421.2359611988, "step_time_sec": 8.232224825071171, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9067, "loss": 3.931938409805298, "lr": 9.740871787647045e-05, "elapsed_sec": 75429.468708992, "step_time_sec": 8.232576451031491, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9068, "loss": 3.7786197662353516, "lr": 9.733222848518313e-05, "elapsed_sec": 75437.70117354393, "step_time_sec": 8.23229456495028, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9069, "loss": 3.854006290435791, "lr": 9.725574228323137e-05, "elapsed_sec": 75445.94096684456, "step_time_sec": 8.239645962952636, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9070, "loss": 3.876596450805664, "lr": 9.717925931781937e-05, "elapsed_sec": 75454.17814731598, "step_time_sec": 8.23704747390002, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9071, "loss": 3.7484540939331055, "lr": 9.710277963614923e-05, "elapsed_sec": 75462.41105628014, "step_time_sec": 8.232668233918957, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9072, "loss": 3.904247522354126, "lr": 9.702630328542098e-05, "elapsed_sec": 75470.64952754974, "step_time_sec": 8.238317007897422, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9073, "loss": 4.1027421951293945, "lr": 9.694983031283278e-05, "elapsed_sec": 75478.88886165619, "step_time_sec": 8.239176311995834, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9074, "loss": 3.9504332542419434, "lr": 9.687336076558058e-05, "elapsed_sec": 75487.12831640244, "step_time_sec": 8.239316130056977, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9075, "loss": 3.863942861557007, "lr": 9.679689469085814e-05, "elapsed_sec": 75495.36711502075, "step_time_sec": 8.23860408400651, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9076, "loss": 3.870615243911743, "lr": 9.67204321358573e-05, "elapsed_sec": 75503.60025954247, "step_time_sec": 8.232951433979906, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9077, "loss": 3.9242324829101562, "lr": 9.664397314776753e-05, "elapsed_sec": 75511.8332130909, "step_time_sec": 8.232799481949769, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9078, "loss": 3.780261516571045, "lr": 9.656751777377614e-05, "elapsed_sec": 75520.06678557396, "step_time_sec": 8.233454916975461, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9079, "loss": 3.973396062850952, "lr": 9.64910660610683e-05, "elapsed_sec": 75528.30547380447, "step_time_sec": 8.23847108206246, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9080, "loss": 3.9423904418945312, "lr": 9.641461805682682e-05, "elapsed_sec": 75536.54507160187, "step_time_sec": 8.239483595942147, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9081, "loss": 3.7956459522247314, "lr": 9.633817380823224e-05, "elapsed_sec": 75544.78268051147, "step_time_sec": 8.23743209906388, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9082, "loss": 3.9292356967926025, "lr": 9.626173336246284e-05, "elapsed_sec": 75553.02175092697, "step_time_sec": 8.238875587005168, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9083, "loss": 3.9128894805908203, "lr": 9.618529676669449e-05, "elapsed_sec": 75561.26185941696, "step_time_sec": 8.239947570953518, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9084, "loss": 3.8466033935546875, "lr": 9.610886406810066e-05, "elapsed_sec": 75569.4999525547, "step_time_sec": 8.23790393397212, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9085, "loss": 3.748304843902588, "lr": 9.60324353138526e-05, "elapsed_sec": 75577.73976254463, "step_time_sec": 8.239654896897264, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9086, "loss": 3.760763645172119, "lr": 9.595601055111887e-05, "elapsed_sec": 75585.97894024849, "step_time_sec": 8.239023335976526, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9087, "loss": 3.828303098678589, "lr": 9.587958982706566e-05, "elapsed_sec": 75594.21801829338, "step_time_sec": 8.23889199597761, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9088, "loss": 3.8617990016937256, "lr": 9.580317318885677e-05, "elapsed_sec": 75602.45732593536, "step_time_sec": 8.2391522009857, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9089, "loss": 3.9923019409179688, "lr": 9.572676068365338e-05, "elapsed_sec": 75610.68972015381, "step_time_sec": 8.232250245055184, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9090, "loss": 3.8069007396698, "lr": 9.565035235861415e-05, "elapsed_sec": 75618.92343211174, "step_time_sec": 8.233568288967945, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9091, "loss": 3.8459837436676025, "lr": 9.557394826089511e-05, "elapsed_sec": 75627.15558481216, "step_time_sec": 8.231972138979472, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9092, "loss": 3.917792797088623, "lr": 9.549754843764973e-05, "elapsed_sec": 75635.38853359222, "step_time_sec": 8.232753798016347, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9093, "loss": 3.941192865371704, "lr": 9.542115293602888e-05, "elapsed_sec": 75643.62097597122, "step_time_sec": 8.232268577092327, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9094, "loss": 4.171342372894287, "lr": 9.534476180318065e-05, "elapsed_sec": 75651.85487341881, "step_time_sec": 8.23368861794006, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9095, "loss": 3.7961080074310303, "lr": 9.52683750862505e-05, "elapsed_sec": 75660.09212946892, "step_time_sec": 8.237139809993096, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9096, "loss": 3.982724905014038, "lr": 9.519199283238125e-05, "elapsed_sec": 75668.325507164, "step_time_sec": 8.233192029059865, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9097, "loss": 3.9860360622406006, "lr": 9.511561508871279e-05, "elapsed_sec": 75676.55831694603, "step_time_sec": 8.232625776086934, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9098, "loss": 3.8502156734466553, "lr": 9.503924190238232e-05, "elapsed_sec": 75684.79152536392, "step_time_sec": 8.233017006074078, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9099, "loss": 3.873872995376587, "lr": 9.496287332052427e-05, "elapsed_sec": 75693.02920651436, "step_time_sec": 8.237471895990893, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9100, "loss": 3.7705039978027344, "lr": 9.488650939027017e-05, "elapsed_sec": 75701.26901125908, "step_time_sec": 8.239666206063703, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9101, "loss": 3.7760884761810303, "lr": 9.481015015874864e-05, "elapsed_sec": 75709.50698757172, "step_time_sec": 8.237811101018451, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9102, "loss": 3.8029894828796387, "lr": 9.473379567308553e-05, "elapsed_sec": 75717.74073600769, "step_time_sec": 8.233603021944873, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9103, "loss": 3.8578717708587646, "lr": 9.465744598040363e-05, "elapsed_sec": 75725.97369813919, "step_time_sec": 8.23276327806525, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9104, "loss": 3.799126386642456, "lr": 9.458110112782281e-05, "elapsed_sec": 75734.21260380745, "step_time_sec": 8.238787612994201, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9105, "loss": 3.957010507583618, "lr": 9.450476116246001e-05, "elapsed_sec": 75742.44472193718, "step_time_sec": 8.231929049943574, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9106, "loss": 3.834505796432495, "lr": 9.442842613142905e-05, "elapsed_sec": 75750.67722415924, "step_time_sec": 8.232343427021988, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9107, "loss": 3.895578622817993, "lr": 9.435209608184087e-05, "elapsed_sec": 75758.91047358513, "step_time_sec": 8.23307748499792, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9108, "loss": 3.823594093322754, "lr": 9.427577106080314e-05, "elapsed_sec": 75767.14236688614, "step_time_sec": 8.231769706006162, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9109, "loss": 3.765638828277588, "lr": 9.419945111542052e-05, "elapsed_sec": 75775.37495017052, "step_time_sec": 8.232404873939231, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9110, "loss": 4.03483772277832, "lr": 9.412313629279458e-05, "elapsed_sec": 75783.6139266491, "step_time_sec": 8.238772703101858, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9111, "loss": 3.8941586017608643, "lr": 9.404682664002366e-05, "elapsed_sec": 75791.8518705368, "step_time_sec": 8.237780091003515, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9112, "loss": 3.9938788414001465, "lr": 9.39705222042029e-05, "elapsed_sec": 75800.09023666382, "step_time_sec": 8.238181727938354, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9113, "loss": 3.8337934017181396, "lr": 9.389422303242434e-05, "elapsed_sec": 75808.32885479927, "step_time_sec": 8.238485977053642, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9114, "loss": 3.9331812858581543, "lr": 9.381792917177662e-05, "elapsed_sec": 75816.56494188309, "step_time_sec": 8.23593272594735, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9115, "loss": 3.991071939468384, "lr": 9.374164066934514e-05, "elapsed_sec": 75824.79842853546, "step_time_sec": 8.233294134028256, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9116, "loss": 3.9596922397613525, "lr": 9.366535757221209e-05, "elapsed_sec": 75833.03079748154, "step_time_sec": 8.232188276015222, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9117, "loss": 4.007853031158447, "lr": 9.358907992745622e-05, "elapsed_sec": 75841.26513791084, "step_time_sec": 8.234221047023311, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9118, "loss": 3.9505884647369385, "lr": 9.351280778215289e-05, "elapsed_sec": 75849.49690365791, "step_time_sec": 8.231585953035392, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9119, "loss": 3.967242956161499, "lr": 9.343654118337427e-05, "elapsed_sec": 75857.73178243637, "step_time_sec": 8.234727039001882, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9120, "loss": 3.923976182937622, "lr": 9.336028017818882e-05, "elapsed_sec": 75865.96414089203, "step_time_sec": 8.232193166040815, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9121, "loss": 3.9445862770080566, "lr": 9.328402481366173e-05, "elapsed_sec": 75874.2048857212, "step_time_sec": 8.240600374992937, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9122, "loss": 3.820199728012085, "lr": 9.320777513685472e-05, "elapsed_sec": 75882.44330096245, "step_time_sec": 8.238248587003909, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9123, "loss": 3.917034149169922, "lr": 9.313153119482592e-05, "elapsed_sec": 75890.68304228783, "step_time_sec": 8.239562044036575, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9124, "loss": 3.8632850646972656, "lr": 9.305529303462988e-05, "elapsed_sec": 75898.91646647453, "step_time_sec": 8.233257619081996, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9125, "loss": 4.036222457885742, "lr": 9.297906070331774e-05, "elapsed_sec": 75907.14924883842, "step_time_sec": 8.232607112964615, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9126, "loss": 3.8288700580596924, "lr": 9.29028342479369e-05, "elapsed_sec": 75915.38320159912, "step_time_sec": 8.233775745029561, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9127, "loss": 3.9321370124816895, "lr": 9.282661371553124e-05, "elapsed_sec": 75923.61594414711, "step_time_sec": 8.232528530061245, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9128, "loss": 3.9229238033294678, "lr": 9.275039915314088e-05, "elapsed_sec": 75931.84744286537, "step_time_sec": 8.231335869058967, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9129, "loss": 3.737135887145996, "lr": 9.267419060780232e-05, "elapsed_sec": 75940.08102798462, "step_time_sec": 8.233417119947262, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9130, "loss": 3.884146213531494, "lr": 9.259798812654835e-05, "elapsed_sec": 75948.31358098984, "step_time_sec": 8.232379521010444, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9131, "loss": 3.9817802906036377, "lr": 9.2521791756408e-05, "elapsed_sec": 75956.54641771317, "step_time_sec": 8.232720319065265, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9132, "loss": 3.8742170333862305, "lr": 9.244560154440648e-05, "elapsed_sec": 75964.78454065323, "step_time_sec": 8.237912541953847, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9133, "loss": 3.98779034614563, "lr": 9.236941753756531e-05, "elapsed_sec": 75973.02440214157, "step_time_sec": 8.239691993105225, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9134, "loss": 3.9041383266448975, "lr": 9.229323978290213e-05, "elapsed_sec": 75981.2635204792, "step_time_sec": 8.238923989003524, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9135, "loss": 3.780804395675659, "lr": 9.22170683274306e-05, "elapsed_sec": 75989.50239300728, "step_time_sec": 8.238666479010135, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9136, "loss": 3.863301992416382, "lr": 9.214090321816073e-05, "elapsed_sec": 75997.74067306519, "step_time_sec": 8.238140688976273, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9137, "loss": 3.8521101474761963, "lr": 9.206474450209842e-05, "elapsed_sec": 76005.97330379486, "step_time_sec": 8.232462962041609, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9138, "loss": 3.9755215644836426, "lr": 9.19885922262457e-05, "elapsed_sec": 76014.20572924614, "step_time_sec": 8.232283064047806, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9139, "loss": 3.9687139987945557, "lr": 9.191244643760064e-05, "elapsed_sec": 76022.43790221214, "step_time_sec": 8.232006474980153, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9140, "loss": 3.836329221725464, "lr": 9.183630718315727e-05, "elapsed_sec": 76030.67197656631, "step_time_sec": 8.233919842052273, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9141, "loss": 3.8182129859924316, "lr": 9.176017450990557e-05, "elapsed_sec": 76038.90450644493, "step_time_sec": 8.232373159960844, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9142, "loss": 3.9962618350982666, "lr": 9.168404846483153e-05, "elapsed_sec": 76047.14323687553, "step_time_sec": 8.238548145978712, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9143, "loss": 3.8479809761047363, "lr": 9.160792909491697e-05, "elapsed_sec": 76055.37527632713, "step_time_sec": 8.23184945504181, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9144, "loss": 3.8532793521881104, "lr": 9.153181644713963e-05, "elapsed_sec": 76063.60851430893, "step_time_sec": 8.233060989994556, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9145, "loss": 4.1417975425720215, "lr": 9.145571056847317e-05, "elapsed_sec": 76071.84037470818, "step_time_sec": 8.23167807108257, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9146, "loss": 3.8588554859161377, "lr": 9.137961150588688e-05, "elapsed_sec": 76080.07362651825, "step_time_sec": 8.23308104800526, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9147, "loss": 4.0129313468933105, "lr": 9.130351930634608e-05, "elapsed_sec": 76088.30644845963, "step_time_sec": 8.232671404024586, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9148, "loss": 4.018311023712158, "lr": 9.122743401681169e-05, "elapsed_sec": 76096.53988671303, "step_time_sec": 8.233250221936032, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9149, "loss": 3.9002482891082764, "lr": 9.11513556842404e-05, "elapsed_sec": 76104.7783601284, "step_time_sec": 8.238353859982453, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9150, "loss": 3.9677531719207764, "lr": 9.107528435558466e-05, "elapsed_sec": 76113.01845169067, "step_time_sec": 8.239886589930393, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9151, "loss": 3.9324886798858643, "lr": 9.099922007779252e-05, "elapsed_sec": 76121.25717902184, "step_time_sec": 8.238559672026895, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9152, "loss": 3.8666563034057617, "lr": 9.092316289780773e-05, "elapsed_sec": 76129.48968148232, "step_time_sec": 8.232356317923404, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9153, "loss": 3.8081178665161133, "lr": 9.084711286256967e-05, "elapsed_sec": 76137.72260570526, "step_time_sec": 8.23276544502005, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9154, "loss": 3.7859654426574707, "lr": 9.077107001901327e-05, "elapsed_sec": 76145.95596671104, "step_time_sec": 8.233183417934924, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9155, "loss": 3.9531681537628174, "lr": 9.069503441406898e-05, "elapsed_sec": 76154.1965713501, "step_time_sec": 8.240425633033738, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9156, "loss": 4.183337688446045, "lr": 9.061900609466296e-05, "elapsed_sec": 76162.43563866615, "step_time_sec": 8.238964274059981, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9157, "loss": 3.738304376602173, "lr": 9.05429851077167e-05, "elapsed_sec": 76170.67471218109, "step_time_sec": 8.23887945106253, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9158, "loss": 4.042171478271484, "lr": 9.046697150014714e-05, "elapsed_sec": 76178.9075498581, "step_time_sec": 8.232656770967878, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9159, "loss": 3.8355112075805664, "lr": 9.039096531886688e-05, "elapsed_sec": 76187.14040899277, "step_time_sec": 8.232731611002237, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9160, "loss": 3.6879401206970215, "lr": 9.031496661078374e-05, "elapsed_sec": 76195.37324118614, "step_time_sec": 8.232650304911658, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9161, "loss": 3.925034284591675, "lr": 9.023897542280096e-05, "elapsed_sec": 76203.6068611145, "step_time_sec": 8.23341603204608, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9162, "loss": 3.910681962966919, "lr": 9.016299180181722e-05, "elapsed_sec": 76211.84454417229, "step_time_sec": 8.23754766001366, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9163, "loss": 3.8957786560058594, "lr": 9.008701579472645e-05, "elapsed_sec": 76220.08514904976, "step_time_sec": 8.240415132022463, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9164, "loss": 3.882291793823242, "lr": 9.001104744841795e-05, "elapsed_sec": 76228.32316660881, "step_time_sec": 8.237845742958598, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9165, "loss": 3.8930885791778564, "lr": 8.99350868097762e-05, "elapsed_sec": 76236.56101584435, "step_time_sec": 8.237683427985758, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9166, "loss": 3.891413450241089, "lr": 8.985913392568096e-05, "elapsed_sec": 76244.80022740364, "step_time_sec": 8.239064808003604, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9167, "loss": 4.104929447174072, "lr": 8.978318884300733e-05, "elapsed_sec": 76253.03745222092, "step_time_sec": 8.236988171935081, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9168, "loss": 3.852691173553467, "lr": 8.970725160862539e-05, "elapsed_sec": 76261.27133798599, "step_time_sec": 8.233745052944869, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9169, "loss": 3.956907033920288, "lr": 8.963132226940045e-05, "elapsed_sec": 76269.50311040878, "step_time_sec": 8.23162203503307, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9170, "loss": 3.7786402702331543, "lr": 8.955540087219304e-05, "elapsed_sec": 76277.73737168312, "step_time_sec": 8.23407347803004, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9171, "loss": 3.8204782009124756, "lr": 8.947948746385871e-05, "elapsed_sec": 76285.97013950348, "step_time_sec": 8.232603723066859, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9172, "loss": 3.710516929626465, "lr": 8.940358209124803e-05, "elapsed_sec": 76294.20386862755, "step_time_sec": 8.233555317041464, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9173, "loss": 3.877044200897217, "lr": 8.932768480120674e-05, "elapsed_sec": 76302.4373512268, "step_time_sec": 8.233317288919352, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9174, "loss": 3.8434317111968994, "lr": 8.925179564057547e-05, "elapsed_sec": 76310.67556834221, "step_time_sec": 8.238072495907545, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9175, "loss": 3.878220558166504, "lr": 8.917591465618987e-05, "elapsed_sec": 76318.90783429146, "step_time_sec": 8.232090143021196, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9176, "loss": 3.9395477771759033, "lr": 8.910004189488061e-05, "elapsed_sec": 76327.14178276062, "step_time_sec": 8.233806238044053, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9177, "loss": 3.9256629943847656, "lr": 8.90241774034732e-05, "elapsed_sec": 76335.37556743622, "step_time_sec": 8.233630606904626, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9178, "loss": 3.975872755050659, "lr": 8.894832122878807e-05, "elapsed_sec": 76343.6180729866, "step_time_sec": 8.242377783055417, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9179, "loss": 3.864924192428589, "lr": 8.887247341764054e-05, "elapsed_sec": 76351.88333654404, "step_time_sec": 8.26282797602471, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9180, "loss": 3.9226057529449463, "lr": 8.879663401684075e-05, "elapsed_sec": 76360.12169480324, "step_time_sec": 8.238191448966973, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9181, "loss": 3.839937210083008, "lr": 8.87208030731936e-05, "elapsed_sec": 76368.36068487167, "step_time_sec": 8.238827498978935, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9182, "loss": 4.0691423416137695, "lr": 8.864498063349892e-05, "elapsed_sec": 76376.59987592697, "step_time_sec": 8.239045417984016, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9183, "loss": 3.784132480621338, "lr": 8.856916674455112e-05, "elapsed_sec": 76384.83272719383, "step_time_sec": 8.232687214040197, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9184, "loss": 3.9049904346466064, "lr": 8.849336145313947e-05, "elapsed_sec": 76393.06507492065, "step_time_sec": 8.23213451798074, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9185, "loss": 3.837456226348877, "lr": 8.841756480604783e-05, "elapsed_sec": 76401.29828310013, "step_time_sec": 8.233038341975771, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9186, "loss": 3.9481201171875, "lr": 8.834177685005474e-05, "elapsed_sec": 76409.53333878517, "step_time_sec": 8.234894037013873, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9187, "loss": 3.7373178005218506, "lr": 8.826599763193348e-05, "elapsed_sec": 76417.76552510262, "step_time_sec": 8.232070047990419, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9188, "loss": 3.9361519813537598, "lr": 8.81902271984518e-05, "elapsed_sec": 76426.00436401367, "step_time_sec": 8.238669962040149, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9189, "loss": 3.869544267654419, "lr": 8.811446559637206e-05, "elapsed_sec": 76434.23697280884, "step_time_sec": 8.23242798098363, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9190, "loss": 3.763491630554199, "lr": 8.803871287245131e-05, "elapsed_sec": 76442.47027397156, "step_time_sec": 8.233107517007738, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9191, "loss": 4.040870189666748, "lr": 8.796296907344091e-05, "elapsed_sec": 76450.70352768898, "step_time_sec": 8.233173916931264, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9192, "loss": 3.9421908855438232, "lr": 8.788723424608682e-05, "elapsed_sec": 76458.9364938736, "step_time_sec": 8.232785853906535, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9193, "loss": 3.828773260116577, "lr": 8.781150843712949e-05, "elapsed_sec": 76467.16819524765, "step_time_sec": 8.231542089022696, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9194, "loss": 3.747121572494507, "lr": 8.77357916933038e-05, "elapsed_sec": 76475.40188503265, "step_time_sec": 8.233510283986107, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9195, "loss": 3.935534715652466, "lr": 8.76600840613389e-05, "elapsed_sec": 76483.63369226456, "step_time_sec": 8.231652313028462, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9196, "loss": 3.8254077434539795, "lr": 8.75843855879585e-05, "elapsed_sec": 76491.8670642376, "step_time_sec": 8.233164123026654, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9197, "loss": 3.8409340381622314, "lr": 8.750869631988056e-05, "elapsed_sec": 76500.09909296036, "step_time_sec": 8.23188073199708, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9198, "loss": 3.9180171489715576, "lr": 8.743301630381735e-05, "elapsed_sec": 76508.33191347122, "step_time_sec": 8.232678881962784, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9199, "loss": 3.9788904190063477, "lr": 8.735734558647549e-05, "elapsed_sec": 76516.56544184685, "step_time_sec": 8.233362889965065, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9200, "loss": 3.8551816940307617, "lr": 8.728168421455584e-05, "elapsed_sec": 76524.79798531532, "step_time_sec": 8.232367315911688, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9201, "loss": 3.8752293586730957, "lr": 8.720603223475342e-05, "elapsed_sec": 76533.03019404411, "step_time_sec": 8.232009980012663, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9202, "loss": 3.809380054473877, "lr": 8.713038969375758e-05, "elapsed_sec": 76541.26276278496, "step_time_sec": 8.232408515992574, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9203, "loss": 3.9198925495147705, "lr": 8.705475663825171e-05, "elapsed_sec": 76549.50173950195, "step_time_sec": 8.238822643063031, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9204, "loss": 3.8753621578216553, "lr": 8.69791331149135e-05, "elapsed_sec": 76557.74167609215, "step_time_sec": 8.239786240039393, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9205, "loss": 3.893941879272461, "lr": 8.690351917041463e-05, "elapsed_sec": 76565.98045039177, "step_time_sec": 8.238610212923959, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9206, "loss": 3.8863790035247803, "lr": 8.682791485142086e-05, "elapsed_sec": 76574.22144269943, "step_time_sec": 8.2407827420393, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9207, "loss": 3.978426694869995, "lr": 8.675232020459213e-05, "elapsed_sec": 76582.4611916542, "step_time_sec": 8.239577236003242, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9208, "loss": 3.894925355911255, "lr": 8.667673527658232e-05, "elapsed_sec": 76590.69970345497, "step_time_sec": 8.238381598028354, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9209, "loss": 3.966179132461548, "lr": 8.660116011403932e-05, "elapsed_sec": 76598.93871045113, "step_time_sec": 8.238811015035026, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9210, "loss": 3.917296886444092, "lr": 8.652559476360503e-05, "elapsed_sec": 76607.17591714859, "step_time_sec": 8.237098442972638, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9211, "loss": 3.877580165863037, "lr": 8.645003927191524e-05, "elapsed_sec": 76615.40883231163, "step_time_sec": 8.232750797993504, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9212, "loss": 3.862881898880005, "lr": 8.63744936855997e-05, "elapsed_sec": 76623.64109182358, "step_time_sec": 8.23205983499065, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9213, "loss": 3.767075538635254, "lr": 8.629895805128203e-05, "elapsed_sec": 76631.87504291534, "step_time_sec": 8.233775354921818, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9214, "loss": 3.8346750736236572, "lr": 8.622343241557972e-05, "elapsed_sec": 76640.11270999908, "step_time_sec": 8.23750388098415, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9215, "loss": 3.952960252761841, "lr": 8.614791682510402e-05, "elapsed_sec": 76648.35230350494, "step_time_sec": 8.239451191970147, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9216, "loss": 3.704771041870117, "lr": 8.607241132646017e-05, "elapsed_sec": 76656.59240078926, "step_time_sec": 8.239960258011706, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9217, "loss": 3.772357225418091, "lr": 8.599691596624695e-05, "elapsed_sec": 76664.83012747765, "step_time_sec": 8.237494925037026, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9218, "loss": 3.7880444526672363, "lr": 8.592143079105697e-05, "elapsed_sec": 76673.06349492073, "step_time_sec": 8.233227660064586, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9219, "loss": 3.860469102859497, "lr": 8.584595584747662e-05, "elapsed_sec": 76681.29615235329, "step_time_sec": 8.232460049097426, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9220, "loss": 3.772645950317383, "lr": 8.577049118208592e-05, "elapsed_sec": 76689.52965903282, "step_time_sec": 8.23332413402386, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9221, "loss": 3.7609496116638184, "lr": 8.569503684145855e-05, "elapsed_sec": 76697.76785349846, "step_time_sec": 8.238014428992756, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9222, "loss": 4.0383687019348145, "lr": 8.561959287216181e-05, "elapsed_sec": 76706.00774908066, "step_time_sec": 8.239786305930465, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9223, "loss": 3.9536962509155273, "lr": 8.554415932075659e-05, "elapsed_sec": 76714.2455868721, "step_time_sec": 8.237624049070291, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9224, "loss": 3.80688214302063, "lr": 8.546873623379741e-05, "elapsed_sec": 76722.48476672173, "step_time_sec": 8.23904048895929, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9225, "loss": 3.8160576820373535, "lr": 8.539332365783227e-05, "elapsed_sec": 76730.72265601158, "step_time_sec": 8.237721086014062, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9226, "loss": 3.838473320007324, "lr": 8.531792163940265e-05, "elapsed_sec": 76738.96043658257, "step_time_sec": 8.237611319986172, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9227, "loss": 3.825483798980713, "lr": 8.524253022504369e-05, "elapsed_sec": 76747.19894433022, "step_time_sec": 8.238355025998317, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9228, "loss": 4.03795862197876, "lr": 8.516714946128378e-05, "elapsed_sec": 76755.43166923523, "step_time_sec": 8.232606171048246, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9229, "loss": 3.6867573261260986, "lr": 8.509177939464479e-05, "elapsed_sec": 76763.66502451897, "step_time_sec": 8.233156085945666, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9230, "loss": 3.871316432952881, "lr": 8.501642007164209e-05, "elapsed_sec": 76771.89713287354, "step_time_sec": 8.231899562990293, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9231, "loss": 3.750870704650879, "lr": 8.494107153878435e-05, "elapsed_sec": 76780.13030076027, "step_time_sec": 8.232987268012948, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9232, "loss": 3.7786591053009033, "lr": 8.486573384257351e-05, "elapsed_sec": 76788.36281394958, "step_time_sec": 8.232355152023956, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9233, "loss": 4.043275356292725, "lr": 8.479040702950496e-05, "elapsed_sec": 76796.59595632553, "step_time_sec": 8.233030606992543, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9234, "loss": 3.8915700912475586, "lr": 8.471509114606726e-05, "elapsed_sec": 76804.83427858353, "step_time_sec": 8.238145379000343, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9235, "loss": 4.017185211181641, "lr": 8.463978623874227e-05, "elapsed_sec": 76813.07446050644, "step_time_sec": 8.240025002975017, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9236, "loss": 3.7997522354125977, "lr": 8.456449235400513e-05, "elapsed_sec": 76821.3131620884, "step_time_sec": 8.23850902193226, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9237, "loss": 3.8712589740753174, "lr": 8.448920953832407e-05, "elapsed_sec": 76829.55296516418, "step_time_sec": 8.23961909499485, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9238, "loss": 3.856642484664917, "lr": 8.441393783816052e-05, "elapsed_sec": 76837.7927942276, "step_time_sec": 8.239665069966577, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9239, "loss": 3.8883442878723145, "lr": 8.433867729996914e-05, "elapsed_sec": 76846.03132033348, "step_time_sec": 8.238380225026049, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9240, "loss": 3.9171581268310547, "lr": 8.426342797019755e-05, "elapsed_sec": 76854.27144265175, "step_time_sec": 8.239916348014958, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9241, "loss": 3.8481664657592773, "lr": 8.418818989528662e-05, "elapsed_sec": 76862.50987529755, "step_time_sec": 8.238269825931638, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9242, "loss": 3.8956189155578613, "lr": 8.411296312167014e-05, "elapsed_sec": 76870.74944472313, "step_time_sec": 8.23943972797133, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9243, "loss": 3.8576366901397705, "lr": 8.403774769577499e-05, "elapsed_sec": 76878.98927974701, "step_time_sec": 8.2396578110056, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9244, "loss": 3.807985544204712, "lr": 8.396254366402104e-05, "elapsed_sec": 76887.22215652466, "step_time_sec": 8.232719211955555, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9245, "loss": 3.7474803924560547, "lr": 8.388735107282109e-05, "elapsed_sec": 76895.45579504967, "step_time_sec": 8.233422065968625, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9246, "loss": 3.844542980194092, "lr": 8.381216996858091e-05, "elapsed_sec": 76903.6883893013, "step_time_sec": 8.232464718981646, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9247, "loss": 3.6503543853759766, "lr": 8.373700039769921e-05, "elapsed_sec": 76911.92140507698, "step_time_sec": 8.232821319950745, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9248, "loss": 3.9471709728240967, "lr": 8.366184240656753e-05, "elapsed_sec": 76920.15328216553, "step_time_sec": 8.231667913030833, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9249, "loss": 3.8672587871551514, "lr": 8.358669604157026e-05, "elapsed_sec": 76928.38653826714, "step_time_sec": 8.233120370074175, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9250, "loss": 3.808044672012329, "lr": 8.351156134908468e-05, "elapsed_sec": 76936.61899662018, "step_time_sec": 8.23230745200999, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9251, "loss": 3.7718801498413086, "lr": 8.343643837548083e-05, "elapsed_sec": 76944.85190033913, "step_time_sec": 8.232718767947517, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9252, "loss": 3.8543381690979004, "lr": 8.336132716712144e-05, "elapsed_sec": 76953.08512187004, "step_time_sec": 8.233030405943282, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9253, "loss": 4.01570987701416, "lr": 8.328622777036216e-05, "elapsed_sec": 76961.31857776642, "step_time_sec": 8.233342109946534, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9254, "loss": 3.9299569129943848, "lr": 8.32111402315512e-05, "elapsed_sec": 76969.55177426338, "step_time_sec": 8.2330026460113, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9255, "loss": 4.018810749053955, "lr": 8.313606459702942e-05, "elapsed_sec": 76977.78476929665, "step_time_sec": 8.232837065937929, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9256, "loss": 3.7859811782836914, "lr": 8.306100091313052e-05, "elapsed_sec": 76986.02391028404, "step_time_sec": 8.238962072995491, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9257, "loss": 3.7734172344207764, "lr": 8.298594922618066e-05, "elapsed_sec": 76994.2636885643, "step_time_sec": 8.239605032023974, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9258, "loss": 3.8808441162109375, "lr": 8.291090958249861e-05, "elapsed_sec": 77002.50124144554, "step_time_sec": 8.237361476989463, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9259, "loss": 3.8869385719299316, "lr": 8.283588202839583e-05, "elapsed_sec": 77010.74042701721, "step_time_sec": 8.239051779964939, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9260, "loss": 3.8951940536499023, "lr": 8.276086661017616e-05, "elapsed_sec": 77018.97791004181, "step_time_sec": 8.237299771979451, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9261, "loss": 3.9282264709472656, "lr": 8.268586337413607e-05, "elapsed_sec": 77027.21672773361, "step_time_sec": 8.238692132988945, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9262, "loss": 3.8701844215393066, "lr": 8.261087236656445e-05, "elapsed_sec": 77035.45608901978, "step_time_sec": 8.239174182992429, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9263, "loss": 3.987689971923828, "lr": 8.253589363374262e-05, "elapsed_sec": 77043.68801498413, "step_time_sec": 8.231740401010029, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9264, "loss": 3.8940932750701904, "lr": 8.246092722194444e-05, "elapsed_sec": 77051.92846727371, "step_time_sec": 8.240282232989557, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9265, "loss": 3.8633525371551514, "lr": 8.238597317743608e-05, "elapsed_sec": 77060.16662359238, "step_time_sec": 8.238003846025094, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9266, "loss": 3.9996695518493652, "lr": 8.231103154647598e-05, "elapsed_sec": 77068.40601563454, "step_time_sec": 8.239218365051784, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9267, "loss": 4.007564544677734, "lr": 8.223610237531516e-05, "elapsed_sec": 77076.64445352554, "step_time_sec": 8.238256449927576, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9268, "loss": 3.8598034381866455, "lr": 8.216118571019676e-05, "elapsed_sec": 77084.88384485245, "step_time_sec": 8.239217694033869, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9269, "loss": 3.9753334522247314, "lr": 8.208628159735623e-05, "elapsed_sec": 77093.12355852127, "step_time_sec": 8.239608194911852, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9270, "loss": 3.9025158882141113, "lr": 8.201139008302136e-05, "elapsed_sec": 77101.36285066605, "step_time_sec": 8.239127530017868, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9271, "loss": 3.897254228591919, "lr": 8.193651121341207e-05, "elapsed_sec": 77109.60283446312, "step_time_sec": 8.23980456602294, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9272, "loss": 3.8541207313537598, "lr": 8.18616450347405e-05, "elapsed_sec": 77117.84208154678, "step_time_sec": 8.239048646995798, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9273, "loss": 3.8087193965911865, "lr": 8.1786791593211e-05, "elapsed_sec": 77126.08105683327, "step_time_sec": 8.238789485068992, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9274, "loss": 3.8763115406036377, "lr": 8.171195093502002e-05, "elapsed_sec": 77134.32091784477, "step_time_sec": 8.239687626017258, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9275, "loss": 3.9080419540405273, "lr": 8.163712310635609e-05, "elapsed_sec": 77142.55399823189, "step_time_sec": 8.23290800000541, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9276, "loss": 3.8514208793640137, "lr": 8.156230815339995e-05, "elapsed_sec": 77150.78833436966, "step_time_sec": 8.234161333995871, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9277, "loss": 3.878201723098755, "lr": 8.14875061223242e-05, "elapsed_sec": 77159.02194190025, "step_time_sec": 8.233457885915413, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9278, "loss": 3.912977933883667, "lr": 8.141271705929367e-05, "elapsed_sec": 77167.26229381561, "step_time_sec": 8.240213711978868, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9279, "loss": 3.8385303020477295, "lr": 8.133794101046504e-05, "elapsed_sec": 77175.50237059593, "step_time_sec": 8.23990682605654, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9280, "loss": 3.797574758529663, "lr": 8.126317802198697e-05, "elapsed_sec": 77183.74071335793, "step_time_sec": 8.23816511395853, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9281, "loss": 3.867702007293701, "lr": 8.118842814000016e-05, "elapsed_sec": 77191.98032808304, "step_time_sec": 8.239405326079577, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9282, "loss": 4.019076347351074, "lr": 8.111369141063714e-05, "elapsed_sec": 77200.21335339546, "step_time_sec": 8.23288187594153, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9283, "loss": 3.9202237129211426, "lr": 8.103896788002228e-05, "elapsed_sec": 77208.44773030281, "step_time_sec": 8.234170343028381, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9284, "loss": 3.9268698692321777, "lr": 8.096425759427196e-05, "elapsed_sec": 77216.68016648293, "step_time_sec": 8.232318010996096, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9285, "loss": 3.897839307785034, "lr": 8.088956059949421e-05, "elapsed_sec": 77224.91985845566, "step_time_sec": 8.23946402198635, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9286, "loss": 3.8546855449676514, "lr": 8.081487694178895e-05, "elapsed_sec": 77233.15931534767, "step_time_sec": 8.239301819936372, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9287, "loss": 3.812417984008789, "lr": 8.074020666724792e-05, "elapsed_sec": 77241.39873862267, "step_time_sec": 8.239284742972814, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9288, "loss": 3.805424928665161, "lr": 8.066554982195446e-05, "elapsed_sec": 77249.63267493248, "step_time_sec": 8.233755410998128, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9289, "loss": 4.149288177490234, "lr": 8.05909064519837e-05, "elapsed_sec": 77257.870292902, "step_time_sec": 8.237463077064604, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9290, "loss": 4.021775722503662, "lr": 8.051627660340251e-05, "elapsed_sec": 77266.1085691452, "step_time_sec": 8.238097824039869, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9291, "loss": 3.947963237762451, "lr": 8.044166032226931e-05, "elapsed_sec": 77274.34643864632, "step_time_sec": 8.23768626106903, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9292, "loss": 3.932448387145996, "lr": 8.036705765463421e-05, "elapsed_sec": 77282.58631920815, "step_time_sec": 8.239728354965337, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9293, "loss": 4.14863920211792, "lr": 8.029246864653892e-05, "elapsed_sec": 77290.82625055313, "step_time_sec": 8.239783140015788, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9294, "loss": 4.044559478759766, "lr": 8.021789334401665e-05, "elapsed_sec": 77299.06605887413, "step_time_sec": 8.239628612995148, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9295, "loss": 4.01554012298584, "lr": 8.014333179309222e-05, "elapsed_sec": 77307.30673623085, "step_time_sec": 8.240548664005473, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9296, "loss": 3.943683385848999, "lr": 8.0068784039782e-05, "elapsed_sec": 77315.54604506493, "step_time_sec": 8.239070227951743, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9297, "loss": 3.934113025665283, "lr": 7.999425013009369e-05, "elapsed_sec": 77323.78533148766, "step_time_sec": 8.239169765030965, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9298, "loss": 3.8716840744018555, "lr": 7.991973011002667e-05, "elapsed_sec": 77332.02498054504, "step_time_sec": 8.239475316018797, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9299, "loss": 3.916398286819458, "lr": 7.984522402557155e-05, "elapsed_sec": 77340.26358485222, "step_time_sec": 8.238430637982674, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9300, "loss": 4.003186225891113, "lr": 7.977073192271038e-05, "elapsed_sec": 77348.5028898716, "step_time_sec": 8.239158037002198, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9301, "loss": 4.02300500869751, "lr": 7.96962538474167e-05, "elapsed_sec": 77356.74208140373, "step_time_sec": 8.239061995991506, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9302, "loss": 3.951612949371338, "lr": 7.962178984565527e-05, "elapsed_sec": 77364.98064017296, "step_time_sec": 8.238336736918427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9303, "loss": 3.905600070953369, "lr": 7.954733996338218e-05, "elapsed_sec": 77373.21990275383, "step_time_sec": 8.239114849013276, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9304, "loss": 3.862060546875, "lr": 7.947290424654487e-05, "elapsed_sec": 77381.45882701874, "step_time_sec": 8.238768592942506, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9305, "loss": 3.938598394393921, "lr": 7.939848274108199e-05, "elapsed_sec": 77389.70021510124, "step_time_sec": 8.241195348091424, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9306, "loss": 4.020549297332764, "lr": 7.932407549292338e-05, "elapsed_sec": 77397.93974399567, "step_time_sec": 8.239387967973016, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9307, "loss": 3.8999369144439697, "lr": 7.92496825479902e-05, "elapsed_sec": 77406.17222166061, "step_time_sec": 8.232280500000343, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9308, "loss": 3.7673721313476562, "lr": 7.917530395219466e-05, "elapsed_sec": 77414.41318750381, "step_time_sec": 8.240794076002203, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9309, "loss": 3.8870747089385986, "lr": 7.910093975144014e-05, "elapsed_sec": 77422.65260362625, "step_time_sec": 8.23923550802283, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9310, "loss": 4.021529197692871, "lr": 7.902658999162123e-05, "elapsed_sec": 77430.89281964302, "step_time_sec": 8.240072821965441, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9311, "loss": 3.8095035552978516, "lr": 7.89522547186235e-05, "elapsed_sec": 77439.13211226463, "step_time_sec": 8.239133505965583, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9312, "loss": 3.680063247680664, "lr": 7.887793397832357e-05, "elapsed_sec": 77447.36584162712, "step_time_sec": 8.233602614025585, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9313, "loss": 3.9713635444641113, "lr": 7.880362781658923e-05, "elapsed_sec": 77455.59939742088, "step_time_sec": 8.233339139958844, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9314, "loss": 3.9857845306396484, "lr": 7.872933627927916e-05, "elapsed_sec": 77463.83746147156, "step_time_sec": 8.237930078059435, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9315, "loss": 3.927177667617798, "lr": 7.865505941224293e-05, "elapsed_sec": 77472.0779337883, "step_time_sec": 8.240316071896814, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9316, "loss": 3.9276230335235596, "lr": 7.85807972613213e-05, "elapsed_sec": 77480.31707215309, "step_time_sec": 8.238970333011821, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9317, "loss": 3.9682981967926025, "lr": 7.850654987234573e-05, "elapsed_sec": 77488.55761241913, "step_time_sec": 8.240365014993586, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9318, "loss": 4.013240814208984, "lr": 7.843231729113869e-05, "elapsed_sec": 77496.79714870453, "step_time_sec": 8.239446755964309, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9319, "loss": 4.1348652839660645, "lr": 7.835809956351344e-05, "elapsed_sec": 77505.03451132774, "step_time_sec": 8.237125810002908, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9320, "loss": 3.9104344844818115, "lr": 7.82838967352741e-05, "elapsed_sec": 77513.27449917793, "step_time_sec": 8.239830399048515, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9321, "loss": 3.8635141849517822, "lr": 7.820970885221562e-05, "elapsed_sec": 77521.5119023323, "step_time_sec": 8.237251345999539, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9322, "loss": 3.921053409576416, "lr": 7.813553596012369e-05, "elapsed_sec": 77529.75099849701, "step_time_sec": 8.238938848022372, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9323, "loss": 3.9738399982452393, "lr": 7.806137810477474e-05, "elapsed_sec": 77537.99366593361, "step_time_sec": 8.242484076065011, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9324, "loss": 3.7165844440460205, "lr": 7.798723533193598e-05, "elapsed_sec": 77546.23272228241, "step_time_sec": 8.23889404302463, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9325, "loss": 4.038645267486572, "lr": 7.79131076873653e-05, "elapsed_sec": 77554.47247362137, "step_time_sec": 8.239562210044824, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9326, "loss": 3.846710205078125, "lr": 7.783899521681111e-05, "elapsed_sec": 77562.71198487282, "step_time_sec": 8.239327560993843, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9327, "loss": 3.950749635696411, "lr": 7.77648979660127e-05, "elapsed_sec": 77570.94999909401, "step_time_sec": 8.237841869005933, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9328, "loss": 4.015530586242676, "lr": 7.76908159806998e-05, "elapsed_sec": 77579.18907880783, "step_time_sec": 8.238946983939968, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9329, "loss": 3.8424911499023438, "lr": 7.76167493065927e-05, "elapsed_sec": 77587.42845153809, "step_time_sec": 8.239208717015572, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9330, "loss": 3.844071388244629, "lr": 7.754269798940237e-05, "elapsed_sec": 77595.66744351387, "step_time_sec": 8.23881095694378, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9331, "loss": 3.9152417182922363, "lr": 7.746866207483022e-05, "elapsed_sec": 77603.97434329987, "step_time_sec": 8.245678816107102, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9332, "loss": 3.890061855316162, "lr": 7.739464160856813e-05, "elapsed_sec": 77612.2124516964, "step_time_sec": 8.237914061057381, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9333, "loss": 3.9214727878570557, "lr": 7.73206366362985e-05, "elapsed_sec": 77620.44971013069, "step_time_sec": 8.237057952908799, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9334, "loss": 4.001599311828613, "lr": 7.724664720369412e-05, "elapsed_sec": 77628.68376302719, "step_time_sec": 8.233894534991123, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9335, "loss": 3.8800790309906006, "lr": 7.717267335641828e-05, "elapsed_sec": 77636.91659212112, "step_time_sec": 8.232715631951578, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9336, "loss": 4.04177188873291, "lr": 7.709871514012458e-05, "elapsed_sec": 77645.15043640137, "step_time_sec": 8.233584266970865, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9337, "loss": 4.086854934692383, "lr": 7.70247726004569e-05, "elapsed_sec": 77653.38280296326, "step_time_sec": 8.232245382969268, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9338, "loss": 3.7853152751922607, "lr": 7.695084578304965e-05, "elapsed_sec": 77661.61690831184, "step_time_sec": 8.233942816965282, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9339, "loss": 3.974500894546509, "lr": 7.687693473352733e-05, "elapsed_sec": 77669.84968495369, "step_time_sec": 8.232606356963515, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9340, "loss": 4.060367584228516, "lr": 7.68030394975048e-05, "elapsed_sec": 77678.08401656151, "step_time_sec": 8.234173908014782, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9341, "loss": 3.8508665561676025, "lr": 7.672916012058719e-05, "elapsed_sec": 77686.31752562523, "step_time_sec": 8.233317635953426, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9342, "loss": 3.8473305702209473, "lr": 7.66552966483698e-05, "elapsed_sec": 77694.55588173866, "step_time_sec": 8.238181688939221, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9343, "loss": 3.920701026916504, "lr": 7.658144912643807e-05, "elapsed_sec": 77702.78840756416, "step_time_sec": 8.232353807077743, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9344, "loss": 3.9517452716827393, "lr": 7.650761760036772e-05, "elapsed_sec": 77711.0223300457, "step_time_sec": 8.233743514982052, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9345, "loss": 3.9264729022979736, "lr": 7.643380211572447e-05, "elapsed_sec": 77719.2620484829, "step_time_sec": 8.23958308796864, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9346, "loss": 3.8301446437835693, "lr": 7.636000271806418e-05, "elapsed_sec": 77727.50187778473, "step_time_sec": 8.239710123976693, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9347, "loss": 3.837789535522461, "lr": 7.62862194529329e-05, "elapsed_sec": 77735.74051594734, "step_time_sec": 8.238434969913214, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9348, "loss": 3.8882522583007812, "lr": 7.621245236586653e-05, "elapsed_sec": 77743.97818160057, "step_time_sec": 8.237515153014101, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9349, "loss": 3.9588711261749268, "lr": 7.613870150239106e-05, "elapsed_sec": 77752.21768689156, "step_time_sec": 8.239354830002412, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9350, "loss": 3.961423397064209, "lr": 7.606496690802254e-05, "elapsed_sec": 77760.45063352585, "step_time_sec": 8.232811878900975, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9351, "loss": 3.88895583152771, "lr": 7.599124862826693e-05, "elapsed_sec": 77768.6842045784, "step_time_sec": 8.233388346037827, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9352, "loss": 3.883431911468506, "lr": 7.591754670862004e-05, "elapsed_sec": 77776.91713619232, "step_time_sec": 8.232792910072021, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9353, "loss": 3.853062391281128, "lr": 7.584386119456774e-05, "elapsed_sec": 77785.15182232857, "step_time_sec": 8.234459007973783, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9354, "loss": 3.8075270652770996, "lr": 7.577019213158563e-05, "elapsed_sec": 77793.3849644661, "step_time_sec": 8.233014381956309, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9355, "loss": 3.7833757400512695, "lr": 7.569653956513926e-05, "elapsed_sec": 77801.61918139458, "step_time_sec": 8.234043314005248, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9356, "loss": 3.892662763595581, "lr": 7.562290354068395e-05, "elapsed_sec": 77809.8510363102, "step_time_sec": 8.231717196060345, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9357, "loss": 3.777913808822632, "lr": 7.554928410366477e-05, "elapsed_sec": 77818.08416962624, "step_time_sec": 8.232904607080854, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9358, "loss": 3.916292667388916, "lr": 7.547568129951671e-05, "elapsed_sec": 77826.31698536873, "step_time_sec": 8.232682256028056, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9359, "loss": 3.903315544128418, "lr": 7.540209517366429e-05, "elapsed_sec": 77834.55186581612, "step_time_sec": 8.234719582018442, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9360, "loss": 3.761286497116089, "lr": 7.532852577152184e-05, "elapsed_sec": 77842.785861969, "step_time_sec": 8.233823242015205, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9361, "loss": 3.870300054550171, "lr": 7.525497313849342e-05, "elapsed_sec": 77851.0201075077, "step_time_sec": 8.234085288946517, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9362, "loss": 4.026121616363525, "lr": 7.518143731997266e-05, "elapsed_sec": 77859.25849246979, "step_time_sec": 8.238207914982922, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9363, "loss": 3.848517417907715, "lr": 7.51079183613428e-05, "elapsed_sec": 77867.49910879135, "step_time_sec": 8.24048392905388, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9364, "loss": 3.574307441711426, "lr": 7.503441630797677e-05, "elapsed_sec": 77875.73833370209, "step_time_sec": 8.239019014989026, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9365, "loss": 3.7991116046905518, "lr": 7.496093120523696e-05, "elapsed_sec": 77883.97645974159, "step_time_sec": 8.23795804893598, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9366, "loss": 3.840521812438965, "lr": 7.488746309847536e-05, "elapsed_sec": 77892.21540999413, "step_time_sec": 8.238812656956725, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9367, "loss": 3.886882781982422, "lr": 7.481401203303347e-05, "elapsed_sec": 77900.45528411865, "step_time_sec": 8.2396895440761, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9368, "loss": 3.935580015182495, "lr": 7.474057805424226e-05, "elapsed_sec": 77908.69458270073, "step_time_sec": 8.239136493997648, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9369, "loss": 3.868129014968872, "lr": 7.466716120742212e-05, "elapsed_sec": 77916.9345843792, "step_time_sec": 8.239861667971127, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9370, "loss": 4.0267791748046875, "lr": 7.459376153788297e-05, "elapsed_sec": 77925.17410731316, "step_time_sec": 8.239284396055155, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9371, "loss": 3.8893401622772217, "lr": 7.4520379090924e-05, "elapsed_sec": 77933.4120426178, "step_time_sec": 8.23778784496244, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9372, "loss": 3.858400583267212, "lr": 7.444701391183382e-05, "elapsed_sec": 77941.6499197483, "step_time_sec": 8.237708884989843, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9373, "loss": 4.063313961029053, "lr": 7.437366604589047e-05, "elapsed_sec": 77949.88936257362, "step_time_sec": 8.23927923804149, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9374, "loss": 3.8347177505493164, "lr": 7.430033553836115e-05, "elapsed_sec": 77958.12787008286, "step_time_sec": 8.23833941097837, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9375, "loss": 3.9597573280334473, "lr": 7.422702243450249e-05, "elapsed_sec": 77966.36618447304, "step_time_sec": 8.23816192499362, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9376, "loss": 3.9259166717529297, "lr": 7.41537267795603e-05, "elapsed_sec": 77974.60489106178, "step_time_sec": 8.238542160950601, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9377, "loss": 3.828295946121216, "lr": 7.40804486187696e-05, "elapsed_sec": 77982.83740663528, "step_time_sec": 8.23240433994215, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9378, "loss": 3.9999420642852783, "lr": 7.400718799735469e-05, "elapsed_sec": 77991.07193922997, "step_time_sec": 8.2343431240879, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9379, "loss": 3.999321699142456, "lr": 7.3933944960529e-05, "elapsed_sec": 77999.30952167511, "step_time_sec": 8.237409265944734, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9380, "loss": 4.020994186401367, "lr": 7.386071955349508e-05, "elapsed_sec": 78007.5432741642, "step_time_sec": 8.23355897795409, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9381, "loss": 3.8620963096618652, "lr": 7.378751182144467e-05, "elapsed_sec": 78015.77596068382, "step_time_sec": 8.232566684950143, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9382, "loss": 3.9190280437469482, "lr": 7.371432180955858e-05, "elapsed_sec": 78024.0110385418, "step_time_sec": 8.234911517007276, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9383, "loss": 3.7040677070617676, "lr": 7.364114956300659e-05, "elapsed_sec": 78032.24343371391, "step_time_sec": 8.232219038065523, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9384, "loss": 3.823948860168457, "lr": 7.356799512694769e-05, "elapsed_sec": 78040.4769449234, "step_time_sec": 8.233337108977139, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9385, "loss": 3.929203987121582, "lr": 7.349485854652977e-05, "elapsed_sec": 78048.7096209526, "step_time_sec": 8.232520963996649, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9386, "loss": 3.7451963424682617, "lr": 7.34217398668896e-05, "elapsed_sec": 78056.94292449951, "step_time_sec": 8.233159151975997, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9387, "loss": 3.9354970455169678, "lr": 7.334863913315316e-05, "elapsed_sec": 78065.17642521858, "step_time_sec": 8.233258870081045, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9388, "loss": 3.9211251735687256, "lr": 7.327555639043514e-05, "elapsed_sec": 78073.41594552994, "step_time_sec": 8.23940117598977, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9389, "loss": 3.890601634979248, "lr": 7.320249168383919e-05, "elapsed_sec": 78081.649777174, "step_time_sec": 8.233610340044834, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9390, "loss": 3.9321439266204834, "lr": 7.312944505845787e-05, "elapsed_sec": 78089.88435006142, "step_time_sec": 8.234454474993981, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9391, "loss": 3.9326043128967285, "lr": 7.305641655937251e-05, "elapsed_sec": 78098.11672854424, "step_time_sec": 8.232125817914493, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9392, "loss": 3.870546340942383, "lr": 7.298340623165333e-05, "elapsed_sec": 78106.35019016266, "step_time_sec": 8.233301865053363, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9393, "loss": 3.815469741821289, "lr": 7.291041412035927e-05, "elapsed_sec": 78114.58243489265, "step_time_sec": 8.232081548892893, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9394, "loss": 3.871474266052246, "lr": 7.283744027053801e-05, "elapsed_sec": 78122.81633663177, "step_time_sec": 8.233727807994, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9395, "loss": 3.885101318359375, "lr": 7.276448472722609e-05, "elapsed_sec": 78131.04771327972, "step_time_sec": 8.231202151975594, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9396, "loss": 3.8163769245147705, "lr": 7.269154753544865e-05, "elapsed_sec": 78139.28256464005, "step_time_sec": 8.234750402974896, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9397, "loss": 3.8208975791931152, "lr": 7.261862874021942e-05, "elapsed_sec": 78147.51525878906, "step_time_sec": 8.23251073306892, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9398, "loss": 3.852640390396118, "lr": 7.254572838654099e-05, "elapsed_sec": 78155.749158144, "step_time_sec": 8.233699701027945, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9399, "loss": 3.747166633605957, "lr": 7.24728465194044e-05, "elapsed_sec": 78163.98260045052, "step_time_sec": 8.233335795928724, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9400, "loss": 3.7451603412628174, "lr": 7.239998318378934e-05, "elapsed_sec": 78172.22218465805, "step_time_sec": 8.239397473051213, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9401, "loss": 3.9270756244659424, "lr": 7.232713842466408e-05, "elapsed_sec": 78180.45530653, "step_time_sec": 8.23291073704604, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9402, "loss": 3.876385450363159, "lr": 7.225431228698539e-05, "elapsed_sec": 78188.68863272667, "step_time_sec": 8.233179940027185, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9403, "loss": 3.864556074142456, "lr": 7.218150481569851e-05, "elapsed_sec": 78196.92159795761, "step_time_sec": 8.232849368010648, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9404, "loss": 3.918868064880371, "lr": 7.21087160557373e-05, "elapsed_sec": 78205.15604138374, "step_time_sec": 8.234189879964106, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9405, "loss": 3.874617576599121, "lr": 7.203594605202392e-05, "elapsed_sec": 78213.38872671127, "step_time_sec": 8.232547670952044, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9406, "loss": 3.6615984439849854, "lr": 7.196319484946901e-05, "elapsed_sec": 78221.62804865837, "step_time_sec": 8.23912244499661, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9407, "loss": 3.795517683029175, "lr": 7.18904624929717e-05, "elapsed_sec": 78229.85932350159, "step_time_sec": 8.23113931901753, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9408, "loss": 3.8800313472747803, "lr": 7.18177490274193e-05, "elapsed_sec": 78238.09292006493, "step_time_sec": 8.233420114964247, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9409, "loss": 3.887427806854248, "lr": 7.174505449768758e-05, "elapsed_sec": 78246.33130455017, "step_time_sec": 8.238230559043586, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9410, "loss": 3.7610607147216797, "lr": 7.167237894864068e-05, "elapsed_sec": 78254.56446051598, "step_time_sec": 8.233006824972108, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9411, "loss": 3.7978200912475586, "lr": 7.159972242513088e-05, "elapsed_sec": 78262.79688620567, "step_time_sec": 8.232227829983458, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9412, "loss": 3.8120763301849365, "lr": 7.152708497199885e-05, "elapsed_sec": 78271.02938914299, "step_time_sec": 8.232372127007693, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9413, "loss": 3.987670421600342, "lr": 7.145446663407338e-05, "elapsed_sec": 78279.26131129265, "step_time_sec": 8.231714165071025, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9414, "loss": 4.003869533538818, "lr": 7.138186745617152e-05, "elapsed_sec": 78287.49500584602, "step_time_sec": 8.233582120970823, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9415, "loss": 3.766892910003662, "lr": 7.130928748309853e-05, "elapsed_sec": 78295.72703409195, "step_time_sec": 8.23182273702696, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9416, "loss": 3.803581953048706, "lr": 7.123672675964776e-05, "elapsed_sec": 78303.96689796448, "step_time_sec": 8.239695789990947, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9417, "loss": 3.9400548934936523, "lr": 7.116418533060065e-05, "elapsed_sec": 78312.20539951324, "step_time_sec": 8.238343477947637, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9418, "loss": 3.803285837173462, "lr": 7.109166324072687e-05, "elapsed_sec": 78320.43945527077, "step_time_sec": 8.233886805013753, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9419, "loss": 3.822357654571533, "lr": 7.101916053478401e-05, "elapsed_sec": 78328.67738103867, "step_time_sec": 8.237704446073622, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9420, "loss": 3.7404112815856934, "lr": 7.094667725751772e-05, "elapsed_sec": 78336.91710042953, "step_time_sec": 8.2396270330064, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9421, "loss": 3.620678424835205, "lr": 7.087421345366175e-05, "elapsed_sec": 78345.15721035004, "step_time_sec": 8.239857507986017, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9422, "loss": 3.9525065422058105, "lr": 7.080176916793779e-05, "elapsed_sec": 78353.38932228088, "step_time_sec": 8.231964605976827, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9423, "loss": 3.898987293243408, "lr": 7.072934444505538e-05, "elapsed_sec": 78361.62246131897, "step_time_sec": 8.232964709983207, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9424, "loss": 3.765465497970581, "lr": 7.065693932971215e-05, "elapsed_sec": 78369.85574674606, "step_time_sec": 8.233150713029318, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9425, "loss": 3.818624258041382, "lr": 7.058455386659354e-05, "elapsed_sec": 78378.0898180008, "step_time_sec": 8.233858959982172, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9426, "loss": 3.8044497966766357, "lr": 7.051218810037283e-05, "elapsed_sec": 78386.32198882103, "step_time_sec": 8.231994837988168, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9427, "loss": 3.8338241577148438, "lr": 7.043984207571125e-05, "elapsed_sec": 78394.55586457253, "step_time_sec": 8.23372197302524, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9428, "loss": 3.7540433406829834, "lr": 7.036751583725774e-05, "elapsed_sec": 78402.787804842, "step_time_sec": 8.231793298036791, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9429, "loss": 3.855926990509033, "lr": 7.029520942964908e-05, "elapsed_sec": 78411.01995062828, "step_time_sec": 8.231985024060123, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9430, "loss": 3.8768553733825684, "lr": 7.022292289750982e-05, "elapsed_sec": 78419.25276446342, "step_time_sec": 8.232676757965237, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9431, "loss": 3.7489442825317383, "lr": 7.01506562854522e-05, "elapsed_sec": 78427.48607087135, "step_time_sec": 8.233128377003595, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9432, "loss": 3.954922914505005, "lr": 7.007840963807623e-05, "elapsed_sec": 78435.71996665001, "step_time_sec": 8.23372099804692, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9433, "loss": 3.7632479667663574, "lr": 7.000618299996957e-05, "elapsed_sec": 78443.9532032013, "step_time_sec": 8.233063968014903, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9434, "loss": 4.157113075256348, "lr": 6.993397641570747e-05, "elapsed_sec": 78452.18686175346, "step_time_sec": 8.233437392045744, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9435, "loss": 3.8349709510803223, "lr": 6.986178992985292e-05, "elapsed_sec": 78460.41980862617, "step_time_sec": 8.232779556070454, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9436, "loss": 3.8036439418792725, "lr": 6.978962358695641e-05, "elapsed_sec": 78468.65361952782, "step_time_sec": 8.233671543071978, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9437, "loss": 3.8073842525482178, "lr": 6.971747743155602e-05, "elapsed_sec": 78476.89400458336, "step_time_sec": 8.240251727052964, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9438, "loss": 3.90651798248291, "lr": 6.964535150817741e-05, "elapsed_sec": 78485.134329319, "step_time_sec": 8.240103074000217, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9439, "loss": 3.865147829055786, "lr": 6.957324586133374e-05, "elapsed_sec": 78493.37457847595, "step_time_sec": 8.240121950977482, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9440, "loss": 3.862309217453003, "lr": 6.950116053552557e-05, "elapsed_sec": 78501.61278128624, "step_time_sec": 8.237997226999141, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9441, "loss": 3.8238940238952637, "lr": 6.942909557524107e-05, "elapsed_sec": 78509.8533308506, "step_time_sec": 8.24037800298538, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9442, "loss": 3.9186134338378906, "lr": 6.935705102495575e-05, "elapsed_sec": 78518.09287071228, "step_time_sec": 8.239356784033589, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9443, "loss": 3.7455787658691406, "lr": 6.928502692913248e-05, "elapsed_sec": 78526.33158421516, "step_time_sec": 8.238575625000522, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9444, "loss": 3.7607474327087402, "lr": 6.921302333222164e-05, "elapsed_sec": 78534.57108926773, "step_time_sec": 8.239322531968355, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9445, "loss": 3.921020030975342, "lr": 6.914104027866088e-05, "elapsed_sec": 78542.80432534218, "step_time_sec": 8.233078692108393, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9446, "loss": 3.983161211013794, "lr": 6.906907781287507e-05, "elapsed_sec": 78551.03753733635, "step_time_sec": 8.233064410975203, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9447, "loss": 3.933307647705078, "lr": 6.899713597927659e-05, "elapsed_sec": 78559.26997947693, "step_time_sec": 8.232284578029066, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9448, "loss": 3.8829843997955322, "lr": 6.892521482226492e-05, "elapsed_sec": 78567.50395154953, "step_time_sec": 8.23383292707149, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9449, "loss": 3.9030635356903076, "lr": 6.885331438622686e-05, "elapsed_sec": 78575.73534846306, "step_time_sec": 8.231183558003977, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9450, "loss": 3.8960416316986084, "lr": 6.87814347155364e-05, "elapsed_sec": 78583.96911168098, "step_time_sec": 8.233621865976602, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9451, "loss": 4.003145694732666, "lr": 6.870957585455466e-05, "elapsed_sec": 78592.20060396194, "step_time_sec": 8.231311522889882, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9452, "loss": 4.024985313415527, "lr": 6.863773784763003e-05, "elapsed_sec": 78600.43378448486, "step_time_sec": 8.23302645701915, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9453, "loss": 3.8324263095855713, "lr": 6.856592073909794e-05, "elapsed_sec": 78608.6659271717, "step_time_sec": 8.23194376390893, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9454, "loss": 3.8185017108917236, "lr": 6.84941245732809e-05, "elapsed_sec": 78616.9003932476, "step_time_sec": 8.23437656601891, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9455, "loss": 3.866398572921753, "lr": 6.842234939448865e-05, "elapsed_sec": 78625.13364005089, "step_time_sec": 8.232965946081094, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9456, "loss": 3.948302984237671, "lr": 6.835059524701782e-05, "elapsed_sec": 78633.36685824394, "step_time_sec": 8.233053877949715, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9457, "loss": 3.9107906818389893, "lr": 6.827886217515204e-05, "elapsed_sec": 78641.59959864616, "step_time_sec": 8.232557351933792, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9458, "loss": 3.823928117752075, "lr": 6.820715022316214e-05, "elapsed_sec": 78649.83219981194, "step_time_sec": 8.232444141060114, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9459, "loss": 4.087196350097656, "lr": 6.813545943530569e-05, "elapsed_sec": 78658.06520342827, "step_time_sec": 8.232824680977501, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9460, "loss": 4.040773391723633, "lr": 6.806378985582731e-05, "elapsed_sec": 78666.29729604721, "step_time_sec": 8.231914234929718, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9461, "loss": 3.930940866470337, "lr": 6.799214152895852e-05, "elapsed_sec": 78674.5302567482, "step_time_sec": 8.232802970916964, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9462, "loss": 3.9447741508483887, "lr": 6.79205144989177e-05, "elapsed_sec": 78682.76265716553, "step_time_sec": 8.232252093963325, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9463, "loss": 3.922722101211548, "lr": 6.784890880991012e-05, "elapsed_sec": 78690.99529075623, "step_time_sec": 8.232496342039667, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9464, "loss": 3.8200266361236572, "lr": 6.777732450612785e-05, "elapsed_sec": 78699.2269911766, "step_time_sec": 8.231515064020641, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9465, "loss": 3.8896310329437256, "lr": 6.770576163174976e-05, "elapsed_sec": 78707.46060848236, "step_time_sec": 8.233493556035683, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9466, "loss": 3.934884786605835, "lr": 6.76342202309415e-05, "elapsed_sec": 78715.69263219833, "step_time_sec": 8.231802743976004, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9467, "loss": 3.7767608165740967, "lr": 6.756270034785553e-05, "elapsed_sec": 78723.92510652542, "step_time_sec": 8.232287561986595, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9468, "loss": 3.881669282913208, "lr": 6.749120202663088e-05, "elapsed_sec": 78732.15797066689, "step_time_sec": 8.232759643928148, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9469, "loss": 3.904984712600708, "lr": 6.741972531139348e-05, "elapsed_sec": 78740.39206790924, "step_time_sec": 8.233921707957052, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9470, "loss": 3.9091694355010986, "lr": 6.734827024625573e-05, "elapsed_sec": 78748.62630295753, "step_time_sec": 8.234077650937252, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9471, "loss": 3.7298550605773926, "lr": 6.727683687531678e-05, "elapsed_sec": 78756.86076235771, "step_time_sec": 8.234294360037893, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9472, "loss": 3.8731906414031982, "lr": 6.720542524266236e-05, "elapsed_sec": 78765.09360074997, "step_time_sec": 8.232654855004512, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9473, "loss": 4.0361151695251465, "lr": 6.713403539236479e-05, "elapsed_sec": 78773.33337211609, "step_time_sec": 8.239566790987737, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9474, "loss": 3.96585750579834, "lr": 6.706266736848292e-05, "elapsed_sec": 78781.57185411453, "step_time_sec": 8.238307452993467, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9475, "loss": 4.0125579833984375, "lr": 6.699132121506218e-05, "elapsed_sec": 78789.81202602386, "step_time_sec": 8.239979562931694, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9476, "loss": 3.93241548538208, "lr": 6.691999697613449e-05, "elapsed_sec": 78798.05154275894, "step_time_sec": 8.239342603017576, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9477, "loss": 3.916372776031494, "lr": 6.684869469571815e-05, "elapsed_sec": 78806.28358054161, "step_time_sec": 8.231889551971108, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9478, "loss": 3.876725196838379, "lr": 6.677741441781811e-05, "elapsed_sec": 78814.51661634445, "step_time_sec": 8.232892831088975, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9479, "loss": 3.9506568908691406, "lr": 6.670615618642552e-05, "elapsed_sec": 78822.74958896637, "step_time_sec": 8.232809102977626, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9480, "loss": 3.834362745285034, "lr": 6.663492004551804e-05, "elapsed_sec": 78830.98929691315, "step_time_sec": 8.23953989695292, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9481, "loss": 3.7548792362213135, "lr": 6.656370603905973e-05, "elapsed_sec": 78839.22202897072, "step_time_sec": 8.232564031961374, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9482, "loss": 3.9511935710906982, "lr": 6.649251421100091e-05, "elapsed_sec": 78847.45585036278, "step_time_sec": 8.23364967701491, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9483, "loss": 3.926860809326172, "lr": 6.642134460527823e-05, "elapsed_sec": 78855.68819475174, "step_time_sec": 8.232184085063636, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9484, "loss": 3.788616895675659, "lr": 6.635019726581465e-05, "elapsed_sec": 78863.92095255852, "step_time_sec": 8.232583009987138, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9485, "loss": 3.8813583850860596, "lr": 6.627907223651937e-05, "elapsed_sec": 78872.15277171135, "step_time_sec": 8.231604831991717, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9486, "loss": 3.831188440322876, "lr": 6.620796956128783e-05, "elapsed_sec": 78880.3913872242, "step_time_sec": 8.238450619974174, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9487, "loss": 3.8282155990600586, "lr": 6.613688928400168e-05, "elapsed_sec": 78888.62388658524, "step_time_sec": 8.232348962919787, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9488, "loss": 3.8974568843841553, "lr": 6.606583144852869e-05, "elapsed_sec": 78896.85710597038, "step_time_sec": 8.233088685083203, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9489, "loss": 3.9518322944641113, "lr": 6.599479609872292e-05, "elapsed_sec": 78905.09426546097, "step_time_sec": 8.236931646009907, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9490, "loss": 3.8960204124450684, "lr": 6.592378327842439e-05, "elapsed_sec": 78913.33465647697, "step_time_sec": 8.24026542797219, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9491, "loss": 3.7069883346557617, "lr": 6.585279303145926e-05, "elapsed_sec": 78921.57306361198, "step_time_sec": 8.238210552022792, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9492, "loss": 3.867156744003296, "lr": 6.578182540163987e-05, "elapsed_sec": 78929.8117403984, "step_time_sec": 8.238512088893913, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9493, "loss": 3.923299551010132, "lr": 6.571088043276446e-05, "elapsed_sec": 78938.05036902428, "step_time_sec": 8.238452784018591, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9494, "loss": 3.9681520462036133, "lr": 6.563995816861735e-05, "elapsed_sec": 78946.28273034096, "step_time_sec": 8.232252656947821, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9495, "loss": 3.8692893981933594, "lr": 6.556905865296884e-05, "elapsed_sec": 78954.51540589333, "step_time_sec": 8.232487768051215, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9496, "loss": 3.8972651958465576, "lr": 6.54981819295752e-05, "elapsed_sec": 78962.75262403488, "step_time_sec": 8.237042683991604, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9497, "loss": 3.935131072998047, "lr": 6.54273280421786e-05, "elapsed_sec": 78970.9911327362, "step_time_sec": 8.23828819906339, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9498, "loss": 3.8070733547210693, "lr": 6.535649703450716e-05, "elapsed_sec": 78979.23074579239, "step_time_sec": 8.239436601987109, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9499, "loss": 3.9250903129577637, "lr": 6.528568895027485e-05, "elapsed_sec": 78987.46957087517, "step_time_sec": 8.238708263961598, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9500, "loss": 3.766495704650879, "lr": 6.521490383318148e-05, "elapsed_sec": 78995.708122015, "step_time_sec": 35.66557171801105, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": true, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9501, "loss": 3.87225604057312, "lr": 6.514414172691277e-05, "elapsed_sec": 79031.378480196, "step_time_sec": 8.242621044977568, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9502, "loss": 4.009481430053711, "lr": 6.50734026751401e-05, "elapsed_sec": 79039.6115450859, "step_time_sec": 8.232897322974168, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9503, "loss": 3.9684224128723145, "lr": 6.500268672152071e-05, "elapsed_sec": 79047.84323644638, "step_time_sec": 8.231526748975739, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9504, "loss": 3.786667823791504, "lr": 6.493199390969763e-05, "elapsed_sec": 79056.07622408867, "step_time_sec": 8.232838923926465, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9505, "loss": 3.8338189125061035, "lr": 6.486132428329953e-05, "elapsed_sec": 79064.30919361115, "step_time_sec": 8.23282089305576, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9506, "loss": 3.850584030151367, "lr": 6.479067788594071e-05, "elapsed_sec": 79072.54149246216, "step_time_sec": 8.232093065977097, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9507, "loss": 3.996351480484009, "lr": 6.472005476122132e-05, "elapsed_sec": 79080.7998445034, "step_time_sec": 8.258240642026067, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9508, "loss": 3.952407121658325, "lr": 6.464945495272698e-05, "elapsed_sec": 79089.0371272564, "step_time_sec": 8.237070729024708, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9509, "loss": 3.993295192718506, "lr": 6.457887850402902e-05, "elapsed_sec": 79097.27209973335, "step_time_sec": 8.234819793025963, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9510, "loss": 3.997734785079956, "lr": 6.450832545868431e-05, "elapsed_sec": 79105.5050971508, "step_time_sec": 8.232764933956787, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9511, "loss": 3.853275775909424, "lr": 6.443779586023522e-05, "elapsed_sec": 79113.73788404465, "step_time_sec": 8.232649000943638, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9512, "loss": 3.7779979705810547, "lr": 6.43672897522098e-05, "elapsed_sec": 79121.97059845924, "step_time_sec": 8.232573210960254, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9513, "loss": 3.8089845180511475, "lr": 6.429680717812147e-05, "elapsed_sec": 79130.20357298851, "step_time_sec": 8.232766818022355, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9514, "loss": 3.9536216259002686, "lr": 6.422634818146913e-05, "elapsed_sec": 79138.43633508682, "step_time_sec": 8.232598530943505, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9515, "loss": 3.6940479278564453, "lr": 6.415591280573726e-05, "elapsed_sec": 79146.67059898376, "step_time_sec": 8.234103273018263, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9516, "loss": 3.953019857406616, "lr": 6.408550109439564e-05, "elapsed_sec": 79154.90286135674, "step_time_sec": 8.232092689955607, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9517, "loss": 3.857611894607544, "lr": 6.401511309089939e-05, "elapsed_sec": 79163.1352262497, "step_time_sec": 8.232214284013025, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9518, "loss": 3.92484712600708, "lr": 6.39447488386892e-05, "elapsed_sec": 79171.3682603836, "step_time_sec": 8.23284858907573, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9519, "loss": 3.7466018199920654, "lr": 6.387440838119093e-05, "elapsed_sec": 79179.60095000267, "step_time_sec": 8.232571679051034, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9520, "loss": 3.7275569438934326, "lr": 6.380409176181581e-05, "elapsed_sec": 79187.83423805237, "step_time_sec": 8.233136408030987, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9521, "loss": 3.9349911212921143, "lr": 6.37337990239604e-05, "elapsed_sec": 79196.06566047668, "step_time_sec": 8.231229142984375, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9522, "loss": 3.8377184867858887, "lr": 6.366353021100642e-05, "elapsed_sec": 79204.30686616898, "step_time_sec": 8.24101937096566, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9523, "loss": 3.765918493270874, "lr": 6.359328536632095e-05, "elapsed_sec": 79212.54708099365, "step_time_sec": 8.240064769051969, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9524, "loss": 4.002525806427002, "lr": 6.352306453325616e-05, "elapsed_sec": 79220.78667545319, "step_time_sec": 8.239465281949379, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9525, "loss": 3.9303557872772217, "lr": 6.345286775514942e-05, "elapsed_sec": 79229.02685928345, "step_time_sec": 8.23997077299282, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9526, "loss": 3.773127317428589, "lr": 6.338269507532341e-05, "elapsed_sec": 79237.25983524323, "step_time_sec": 8.232891796040349, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9527, "loss": 3.784682512283325, "lr": 6.331254653708576e-05, "elapsed_sec": 79245.4938864708, "step_time_sec": 8.233799632056616, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9528, "loss": 3.9197895526885986, "lr": 6.324242218372918e-05, "elapsed_sec": 79253.72680068016, "step_time_sec": 8.232771274982952, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9529, "loss": 3.851771354675293, "lr": 6.317232205853164e-05, "elapsed_sec": 79261.96043491364, "step_time_sec": 8.2335281369742, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9530, "loss": 3.8312549591064453, "lr": 6.3102246204756e-05, "elapsed_sec": 79270.1920568943, "step_time_sec": 8.231377121061087, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9531, "loss": 3.8909661769866943, "lr": 6.303219466565017e-05, "elapsed_sec": 79278.42671227455, "step_time_sec": 8.234494168893434, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9532, "loss": 3.949974536895752, "lr": 6.29621674844471e-05, "elapsed_sec": 79286.66396284103, "step_time_sec": 8.237090987968259, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9533, "loss": 3.8824973106384277, "lr": 6.289216470436468e-05, "elapsed_sec": 79294.90366721153, "step_time_sec": 8.239568924997002, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9534, "loss": 3.9002294540405273, "lr": 6.282218636860573e-05, "elapsed_sec": 79303.14256739616, "step_time_sec": 8.238724037073553, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9535, "loss": 3.858973741531372, "lr": 6.275223252035801e-05, "elapsed_sec": 79311.37554550171, "step_time_sec": 8.23278526996728, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9536, "loss": 3.815589427947998, "lr": 6.268230320279415e-05, "elapsed_sec": 79319.60793495178, "step_time_sec": 8.232212841976434, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9537, "loss": 3.817706346511841, "lr": 6.261239845907158e-05, "elapsed_sec": 79327.84116649628, "step_time_sec": 8.233103818027303, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9538, "loss": 3.7838687896728516, "lr": 6.254251833233274e-05, "elapsed_sec": 79336.08076930046, "step_time_sec": 8.239419320016168, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9539, "loss": 3.8363847732543945, "lr": 6.247266286570469e-05, "elapsed_sec": 79344.31969451904, "step_time_sec": 8.23872281506192, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9540, "loss": 3.987847328186035, "lr": 6.240283210229931e-05, "elapsed_sec": 79352.55987524986, "step_time_sec": 8.240042204968631, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9541, "loss": 3.8481574058532715, "lr": 6.233302608521338e-05, "elapsed_sec": 79360.79864740372, "step_time_sec": 8.23859025107231, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9542, "loss": 3.946739435195923, "lr": 6.226324485752821e-05, "elapsed_sec": 79369.03775429726, "step_time_sec": 8.238963982905261, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9543, "loss": 3.827136754989624, "lr": 6.219348846230993e-05, "elapsed_sec": 79377.27623939514, "step_time_sec": 8.238307763938792, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9544, "loss": 3.7889256477355957, "lr": 6.212375694260929e-05, "elapsed_sec": 79385.51527929306, "step_time_sec": 8.23880462499801, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9545, "loss": 3.937896490097046, "lr": 6.205405034146172e-05, "elapsed_sec": 79393.74850487709, "step_time_sec": 8.233124897931702, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9546, "loss": 3.800980806350708, "lr": 6.19843687018873e-05, "elapsed_sec": 79401.98068118095, "step_time_sec": 8.231991322012618, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9547, "loss": 3.9832093715667725, "lr": 6.19147120668906e-05, "elapsed_sec": 79410.21430253983, "step_time_sec": 8.233405363978818, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9548, "loss": 3.883985996246338, "lr": 6.184508047946084e-05, "elapsed_sec": 79418.45146346092, "step_time_sec": 8.237019762978889, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9549, "loss": 3.9873273372650146, "lr": 6.177547398257182e-05, "elapsed_sec": 79426.68624401093, "step_time_sec": 8.234611443942413, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9550, "loss": 3.9093003273010254, "lr": 6.170589261918175e-05, "elapsed_sec": 79434.9193572998, "step_time_sec": 8.232974998070858, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9551, "loss": 3.8701257705688477, "lr": 6.163633643223341e-05, "elapsed_sec": 79443.15359377861, "step_time_sec": 8.234090209938586, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9552, "loss": 3.7563464641571045, "lr": 6.156680546465399e-05, "elapsed_sec": 79451.38543987274, "step_time_sec": 8.231635851901956, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9553, "loss": 3.845146894454956, "lr": 6.149729975935513e-05, "elapsed_sec": 79459.61873698235, "step_time_sec": 8.233137332950719, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9554, "loss": 3.8202569484710693, "lr": 6.142781935923285e-05, "elapsed_sec": 79467.85022234917, "step_time_sec": 8.231369265005924, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9555, "loss": 3.7680277824401855, "lr": 6.135836430716772e-05, "elapsed_sec": 79476.08931946754, "step_time_sec": 8.238885856000707, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9556, "loss": 3.8636972904205322, "lr": 6.128893464602441e-05, "elapsed_sec": 79484.32872462273, "step_time_sec": 8.239229140919633, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9557, "loss": 3.9976892471313477, "lr": 6.121953041865206e-05, "elapsed_sec": 79492.56779313087, "step_time_sec": 8.238927486003377, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9558, "loss": 3.910914897918701, "lr": 6.115015166788417e-05, "elapsed_sec": 79500.80698943138, "step_time_sec": 8.2390443070326, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9559, "loss": 3.8524158000946045, "lr": 6.108079843653839e-05, "elapsed_sec": 79509.04676103592, "step_time_sec": 8.239577219937928, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9560, "loss": 3.9277451038360596, "lr": 6.101147076741671e-05, "elapsed_sec": 79517.28680872917, "step_time_sec": 8.23992322490085, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9561, "loss": 3.9498307704925537, "lr": 6.094216870330527e-05, "elapsed_sec": 79525.52583050728, "step_time_sec": 8.23881640296895, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9562, "loss": 3.8939690589904785, "lr": 6.087289228697448e-05, "elapsed_sec": 79533.7599811554, "step_time_sec": 8.233946728054434, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9563, "loss": 3.9761955738067627, "lr": 6.0803641561178823e-05, "elapsed_sec": 79541.99271774292, "step_time_sec": 8.232606010045856, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9564, "loss": 3.9977469444274902, "lr": 6.073441656865708e-05, "elapsed_sec": 79550.22632884979, "step_time_sec": 8.233446111902595, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9565, "loss": 3.8052351474761963, "lr": 6.0665217352131996e-05, "elapsed_sec": 79558.46541118622, "step_time_sec": 8.238932817941532, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9566, "loss": 3.8180696964263916, "lr": 6.059604395431058e-05, "elapsed_sec": 79566.7043337822, "step_time_sec": 8.238745677052066, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9567, "loss": 3.812209367752075, "lr": 6.0526896417883705e-05, "elapsed_sec": 79574.94318652153, "step_time_sec": 8.238665690994821, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9568, "loss": 3.717332601547241, "lr": 6.0457774785526346e-05, "elapsed_sec": 79583.17490363121, "step_time_sec": 8.231555712991394, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9569, "loss": 3.8886823654174805, "lr": 6.038867909989763e-05, "elapsed_sec": 79591.40905308723, "step_time_sec": 8.23394882294815, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9570, "loss": 3.9664459228515625, "lr": 6.031960940364052e-05, "elapsed_sec": 79599.64293003082, "step_time_sec": 8.233719139941968, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9571, "loss": 3.7956912517547607, "lr": 6.025056573938195e-05, "elapsed_sec": 79607.88314604759, "step_time_sec": 8.240102925919928, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9572, "loss": 3.8722167015075684, "lr": 6.01815481497329e-05, "elapsed_sec": 79616.11569213867, "step_time_sec": 8.232338516972959, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9573, "loss": 3.806419849395752, "lr": 6.011255667728814e-05, "elapsed_sec": 79624.34846735, "step_time_sec": 8.232595439068973, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9574, "loss": 3.6731653213500977, "lr": 6.004359136462636e-05, "elapsed_sec": 79632.5869216919, "step_time_sec": 8.238275623996742, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9575, "loss": 3.6491451263427734, "lr": 5.9974652254310125e-05, "elapsed_sec": 79640.82751250267, "step_time_sec": 8.240454777958803, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9576, "loss": 3.8035309314727783, "lr": 5.990573938888578e-05, "elapsed_sec": 79649.0667784214, "step_time_sec": 8.239076873054728, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9577, "loss": 3.9784841537475586, "lr": 5.9836852810883504e-05, "elapsed_sec": 79657.3043460846, "step_time_sec": 8.23751130909659, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9578, "loss": 3.7361562252044678, "lr": 5.9767992562817316e-05, "elapsed_sec": 79665.54439425468, "step_time_sec": 8.239739446085878, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9579, "loss": 3.8760650157928467, "lr": 5.9699158687184885e-05, "elapsed_sec": 79673.78235507011, "step_time_sec": 8.237836973974481, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9580, "loss": 3.826981782913208, "lr": 5.963035122646758e-05, "elapsed_sec": 79682.02330231667, "step_time_sec": 8.240766091970727, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9581, "loss": 3.8182153701782227, "lr": 5.9561570223130644e-05, "elapsed_sec": 79690.26312804222, "step_time_sec": 8.239671224029735, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9582, "loss": 3.9255313873291016, "lr": 5.949281571962283e-05, "elapsed_sec": 79698.50141739845, "step_time_sec": 8.238152213976718, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9583, "loss": 3.9762721061706543, "lr": 5.942408775837657e-05, "elapsed_sec": 79706.73521780968, "step_time_sec": 8.233583244029433, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9584, "loss": 3.725426197052002, "lr": 5.935538638180793e-05, "elapsed_sec": 79714.96750426292, "step_time_sec": 8.232122496003285, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9585, "loss": 3.776020050048828, "lr": 5.928671163231654e-05, "elapsed_sec": 79723.20207810402, "step_time_sec": 8.234421319095418, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9586, "loss": 3.9739432334899902, "lr": 5.9218063552285695e-05, "elapsed_sec": 79731.44009017944, "step_time_sec": 8.237797487992793, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9587, "loss": 3.9190096855163574, "lr": 5.914944218408212e-05, "elapsed_sec": 79739.67972421646, "step_time_sec": 8.239499810966663, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9588, "loss": 3.843440055847168, "lr": 5.9080847570056095e-05, "elapsed_sec": 79747.91830134392, "step_time_sec": 8.238419579924084, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9589, "loss": 3.977853536605835, "lr": 5.9012279752541414e-05, "elapsed_sec": 79756.15774011612, "step_time_sec": 8.23926867195405, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9590, "loss": 3.91554594039917, "lr": 5.894373877385527e-05, "elapsed_sec": 79764.39858531952, "step_time_sec": 8.240628371946514, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9591, "loss": 3.939011812210083, "lr": 5.8875224676298306e-05, "elapsed_sec": 79772.63721203804, "step_time_sec": 8.238442077999935, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9592, "loss": 3.9058096408843994, "lr": 5.8806737502154696e-05, "elapsed_sec": 79780.8771083355, "step_time_sec": 8.239747382001951, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9593, "loss": 3.8076131343841553, "lr": 5.873827729369184e-05, "elapsed_sec": 79789.11666464806, "step_time_sec": 8.239427231019363, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9594, "loss": 3.9005112648010254, "lr": 5.866984409316051e-05, "elapsed_sec": 79797.3558781147, "step_time_sec": 8.239072669995949, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9595, "loss": 3.7613048553466797, "lr": 5.860143794279496e-05, "elapsed_sec": 79805.59624099731, "step_time_sec": 8.240120912902057, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9596, "loss": 3.767982006072998, "lr": 5.853305888481259e-05, "elapsed_sec": 79813.83469176292, "step_time_sec": 8.238330075982958, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9597, "loss": 3.831575870513916, "lr": 5.846470696141415e-05, "elapsed_sec": 79822.07438516617, "step_time_sec": 8.239499513991177, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9598, "loss": 3.758209466934204, "lr": 5.8396382214783614e-05, "elapsed_sec": 79830.31464195251, "step_time_sec": 8.240107191959396, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9599, "loss": 3.9377281665802, "lr": 5.8328084687088215e-05, "elapsed_sec": 79838.55314850807, "step_time_sec": 8.238330246997066, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9600, "loss": 3.9323410987854004, "lr": 5.825981442047832e-05, "elapsed_sec": 79846.79450178146, "step_time_sec": 8.241144613944925, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9601, "loss": 3.8795154094696045, "lr": 5.819157145708761e-05, "elapsed_sec": 79855.03471684456, "step_time_sec": 8.240047962986864, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9602, "loss": 3.72541880607605, "lr": 5.812335583903275e-05, "elapsed_sec": 79863.27406859398, "step_time_sec": 8.23922205902636, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9603, "loss": 4.13469934463501, "lr": 5.8055167608413685e-05, "elapsed_sec": 79871.51370429993, "step_time_sec": 8.23942649003584, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9604, "loss": 3.91904354095459, "lr": 5.798700680731338e-05, "elapsed_sec": 79879.75181436539, "step_time_sec": 8.237943949992768, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9605, "loss": 3.803149461746216, "lr": 5.791887347779774e-05, "elapsed_sec": 79887.99189043045, "step_time_sec": 8.239945680019446, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9606, "loss": 3.902914524078369, "lr": 5.7850767661916006e-05, "elapsed_sec": 79896.23157668114, "step_time_sec": 8.239476756076328, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9607, "loss": 3.842244863510132, "lr": 5.778268940170019e-05, "elapsed_sec": 79904.47099399567, "step_time_sec": 8.239278631052002, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9608, "loss": 3.894700050354004, "lr": 5.7714638739165384e-05, "elapsed_sec": 79912.71100735664, "step_time_sec": 8.23984350496903, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9609, "loss": 3.8338351249694824, "lr": 5.7646615716309735e-05, "elapsed_sec": 79920.95053315163, "step_time_sec": 8.239318754989654, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9610, "loss": 3.7621965408325195, "lr": 5.7578620375114185e-05, "elapsed_sec": 79929.19003152847, "step_time_sec": 8.23939653497655, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9611, "loss": 4.090201377868652, "lr": 5.751065275754266e-05, "elapsed_sec": 79937.43158721924, "step_time_sec": 8.241400538012385, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9612, "loss": 3.9543027877807617, "lr": 5.744271290554199e-05, "elapsed_sec": 79945.68405842781, "step_time_sec": 8.252299312036484, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9613, "loss": 3.9023241996765137, "lr": 5.737480086104184e-05, "elapsed_sec": 79953.94330430031, "step_time_sec": 8.2590358419111, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9614, "loss": 3.899921417236328, "lr": 5.7306916665954687e-05, "elapsed_sec": 79962.18216276169, "step_time_sec": 8.238684730022214, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9615, "loss": 3.9696881771087646, "lr": 5.723906036217592e-05, "elapsed_sec": 79970.422498703, "step_time_sec": 8.240151913953014, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9616, "loss": 3.876357078552246, "lr": 5.717123199158366e-05, "elapsed_sec": 79978.66297674179, "step_time_sec": 8.24032277497463, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9617, "loss": 3.8593533039093018, "lr": 5.71034315960387e-05, "elapsed_sec": 79986.90183591843, "step_time_sec": 8.238670679042116, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9618, "loss": 3.8835816383361816, "lr": 5.7035659217384726e-05, "elapsed_sec": 79995.14222359657, "step_time_sec": 8.240207180962898, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9619, "loss": 3.7798354625701904, "lr": 5.696791489744805e-05, "elapsed_sec": 80003.38170671463, "step_time_sec": 8.239350958028808, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9620, "loss": 3.5978026390075684, "lr": 5.690019867803763e-05, "elapsed_sec": 80011.62250447273, "step_time_sec": 8.240628328057937, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9621, "loss": 3.8015987873077393, "lr": 5.683251060094518e-05, "elapsed_sec": 80019.86259460449, "step_time_sec": 8.239923493005335, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9622, "loss": 3.9203081130981445, "lr": 5.6764850707944925e-05, "elapsed_sec": 80028.10136198997, "step_time_sec": 8.238588289008476, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9623, "loss": 3.9766154289245605, "lr": 5.669721904079383e-05, "elapsed_sec": 80036.34105014801, "step_time_sec": 8.239530756953172, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9624, "loss": 3.7128210067749023, "lr": 5.6629615641231336e-05, "elapsed_sec": 80044.58200573921, "step_time_sec": 8.240784835070372, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9625, "loss": 3.9253346920013428, "lr": 5.656204055097945e-05, "elapsed_sec": 80052.82092165947, "step_time_sec": 8.238764804089442, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9626, "loss": 3.845240592956543, "lr": 5.649449381174285e-05, "elapsed_sec": 80061.06128692627, "step_time_sec": 8.240173193975352, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9627, "loss": 3.793057441711426, "lr": 5.642697546520845e-05, "elapsed_sec": 80069.30132842064, "step_time_sec": 8.23988655791618, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9628, "loss": 4.042472839355469, "lr": 5.635948555304585e-05, "elapsed_sec": 80077.54151654243, "step_time_sec": 8.240091016050428, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9629, "loss": 3.8885114192962646, "lr": 5.629202411690706e-05, "elapsed_sec": 80085.78169298172, "step_time_sec": 8.239946019952185, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9630, "loss": 4.002042293548584, "lr": 5.62245911984265e-05, "elapsed_sec": 80094.02080488205, "step_time_sec": 8.2389744350221, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9631, "loss": 3.935948610305786, "lr": 5.6157186839220925e-05, "elapsed_sec": 80102.2604637146, "step_time_sec": 8.239466817001812, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9632, "loss": 3.85931134223938, "lr": 5.608981108088962e-05, "elapsed_sec": 80110.5005171299, "step_time_sec": 8.239898444968276, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9633, "loss": 3.883291482925415, "lr": 5.602246396501408e-05, "elapsed_sec": 80118.74038338661, "step_time_sec": 8.239719945006073, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9634, "loss": 3.9021406173706055, "lr": 5.595514553315815e-05, "elapsed_sec": 80126.97990703583, "step_time_sec": 8.239269745885395, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9635, "loss": 3.834453582763672, "lr": 5.588785582686804e-05, "elapsed_sec": 80135.22013735771, "step_time_sec": 8.240129132987931, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9636, "loss": 3.950615882873535, "lr": 5.582059488767214e-05, "elapsed_sec": 80143.4607899189, "step_time_sec": 8.240449512028135, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9637, "loss": 3.9199211597442627, "lr": 5.575336275708111e-05, "elapsed_sec": 80151.7007932663, "step_time_sec": 8.239814669941552, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9638, "loss": 3.834036111831665, "lr": 5.568615947658792e-05, "elapsed_sec": 80159.940253973, "step_time_sec": 8.239283433998935, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9639, "loss": 3.836881637573242, "lr": 5.5618985087667595e-05, "elapsed_sec": 80168.18057084084, "step_time_sec": 8.240225658984855, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9640, "loss": 3.9199206829071045, "lr": 5.555183963177748e-05, "elapsed_sec": 80176.42129850388, "step_time_sec": 8.240528125897981, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9641, "loss": 3.9260287284851074, "lr": 5.5484723150356924e-05, "elapsed_sec": 80184.66095542908, "step_time_sec": 8.239435276016593, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9642, "loss": 3.7196061611175537, "lr": 5.541763568482747e-05, "elapsed_sec": 80192.90201663971, "step_time_sec": 8.240881462930702, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9643, "loss": 3.9351608753204346, "lr": 5.535057727659273e-05, "elapsed_sec": 80201.14053106308, "step_time_sec": 8.238353824010119, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9644, "loss": 3.987743854522705, "lr": 5.528354796703839e-05, "elapsed_sec": 80209.37864422798, "step_time_sec": 8.237967526074499, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9645, "loss": 3.9754223823547363, "lr": 5.521654779753215e-05, "elapsed_sec": 80217.61782288551, "step_time_sec": 8.239049987052567, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9646, "loss": 3.955033540725708, "lr": 5.514957680942382e-05, "elapsed_sec": 80225.85655355453, "step_time_sec": 8.238551210029982, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9647, "loss": 3.913656711578369, "lr": 5.508263504404508e-05, "elapsed_sec": 80234.0977537632, "step_time_sec": 8.241025303024799, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9648, "loss": 3.9786200523376465, "lr": 5.501572254270966e-05, "elapsed_sec": 80242.33805942535, "step_time_sec": 8.240116411005147, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9649, "loss": 3.8291141986846924, "lr": 5.4948839346713176e-05, "elapsed_sec": 80250.57743573189, "step_time_sec": 8.239222944015637, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9650, "loss": 3.8961610794067383, "lr": 5.488198549733319e-05, "elapsed_sec": 80258.81742954254, "step_time_sec": 8.239796337904409, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9651, "loss": 3.902545690536499, "lr": 5.4815161035829104e-05, "elapsed_sec": 80267.05630993843, "step_time_sec": 8.238678053952754, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9652, "loss": 3.804433822631836, "lr": 5.474836600344229e-05, "elapsed_sec": 80275.2969789505, "step_time_sec": 8.240535733988509, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9653, "loss": 3.8406295776367188, "lr": 5.468160044139586e-05, "elapsed_sec": 80283.53507781029, "step_time_sec": 8.23793671198655, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9654, "loss": 3.791823148727417, "lr": 5.461486439089471e-05, "elapsed_sec": 80291.77399659157, "step_time_sec": 8.238759923027828, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9655, "loss": 4.018931865692139, "lr": 5.4548157893125685e-05, "elapsed_sec": 80300.0156326294, "step_time_sec": 8.241451298934408, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9656, "loss": 3.930387020111084, "lr": 5.448148098925722e-05, "elapsed_sec": 80308.25638103485, "step_time_sec": 8.240584483952262, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9657, "loss": 4.055452823638916, "lr": 5.441483372043957e-05, "elapsed_sec": 80316.49514389038, "step_time_sec": 8.238631770014763, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9658, "loss": 3.8559975624084473, "lr": 5.434821612780468e-05, "elapsed_sec": 80324.7346110344, "step_time_sec": 8.239294196013361, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9659, "loss": 3.7120361328125, "lr": 5.428162825246614e-05, "elapsed_sec": 80332.973041296, "step_time_sec": 8.238229807000607, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9660, "loss": 4.013134479522705, "lr": 5.421507013551932e-05, "elapsed_sec": 80341.21346092224, "step_time_sec": 8.240255449083634, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9661, "loss": 3.7847330570220947, "lr": 5.4148541818041114e-05, "elapsed_sec": 80349.4529030323, "step_time_sec": 8.239266593009233, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9662, "loss": 3.8578848838806152, "lr": 5.408204334109e-05, "elapsed_sec": 80357.69189167023, "step_time_sec": 8.23886865307577, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9663, "loss": 3.738945245742798, "lr": 5.4015574745706205e-05, "elapsed_sec": 80365.93302869797, "step_time_sec": 8.2409513220191, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9664, "loss": 4.069569110870361, "lr": 5.394913607291138e-05, "elapsed_sec": 80374.17255926132, "step_time_sec": 8.239362754975446, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9665, "loss": 3.7218852043151855, "lr": 5.3882727363708635e-05, "elapsed_sec": 80382.41262578964, "step_time_sec": 8.239918158040382, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9666, "loss": 3.900175094604492, "lr": 5.381634865908279e-05, "elapsed_sec": 80390.65242052078, "step_time_sec": 8.239599036984146, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9667, "loss": 4.055240631103516, "lr": 5.375000000000002e-05, "elapsed_sec": 80398.89127421379, "step_time_sec": 8.238707859069109, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9668, "loss": 3.902198314666748, "lr": 5.3683681427407936e-05, "elapsed_sec": 80407.13174152374, "step_time_sec": 8.240318857948296, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9669, "loss": 3.864922046661377, "lr": 5.361739298223571e-05, "elapsed_sec": 80415.3727927208, "step_time_sec": 8.240854464937001, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9670, "loss": 3.847282648086548, "lr": 5.3551134705393795e-05, "elapsed_sec": 80423.61149144173, "step_time_sec": 8.238534023985267, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9671, "loss": 3.7717320919036865, "lr": 5.348490663777408e-05, "elapsed_sec": 80431.85222840309, "step_time_sec": 8.24055526999291, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9672, "loss": 3.7598602771759033, "lr": 5.341870882024978e-05, "elapsed_sec": 80440.09204506874, "step_time_sec": 8.239644038956612, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9673, "loss": 3.7561562061309814, "lr": 5.335254129367548e-05, "elapsed_sec": 80448.33099842072, "step_time_sec": 8.238804633030668, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9674, "loss": 3.877880811691284, "lr": 5.328640409888699e-05, "elapsed_sec": 80456.57168745995, "step_time_sec": 8.240585035993718, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9675, "loss": 3.9605143070220947, "lr": 5.322029727670156e-05, "elapsed_sec": 80464.81123280525, "step_time_sec": 8.239363178960048, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9676, "loss": 3.923847198486328, "lr": 5.315422086791754e-05, "elapsed_sec": 80473.05247235298, "step_time_sec": 8.241021391935647, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9677, "loss": 3.772738218307495, "lr": 5.308817491331453e-05, "elapsed_sec": 80481.29254198074, "step_time_sec": 8.23989339696709, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9678, "loss": 3.8218963146209717, "lr": 5.302215945365346e-05, "elapsed_sec": 80489.5307803154, "step_time_sec": 8.238120374036953, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9679, "loss": 3.97511887550354, "lr": 5.29561745296763e-05, "elapsed_sec": 80497.77114963531, "step_time_sec": 8.240223226021044, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9680, "loss": 3.885004997253418, "lr": 5.289022018210624e-05, "elapsed_sec": 80506.01123046875, "step_time_sec": 8.239881959045306, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9681, "loss": 3.931124448776245, "lr": 5.282429645164757e-05, "elapsed_sec": 80514.25010824203, "step_time_sec": 8.23872006603051, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9682, "loss": 4.056431770324707, "lr": 5.275840337898566e-05, "elapsed_sec": 80522.48993992805, "step_time_sec": 8.239651520038024, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9683, "loss": 3.8514723777770996, "lr": 5.269254100478709e-05, "elapsed_sec": 80530.72867679596, "step_time_sec": 8.238592314068228, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9684, "loss": 3.798386573791504, "lr": 5.262670936969935e-05, "elapsed_sec": 80538.96970200539, "step_time_sec": 8.240877727046609, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9685, "loss": 3.761315107345581, "lr": 5.2560908514350974e-05, "elapsed_sec": 80547.21069121361, "step_time_sec": 8.240837948047556, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9686, "loss": 3.777338743209839, "lr": 5.249513847935167e-05, "elapsed_sec": 80555.44929790497, "step_time_sec": 8.238378148991615, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9687, "loss": 3.8807289600372314, "lr": 5.242939930529185e-05, "elapsed_sec": 80563.68976044655, "step_time_sec": 8.240332882967778, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9688, "loss": 3.771305799484253, "lr": 5.236369103274305e-05, "elapsed_sec": 80571.92900800705, "step_time_sec": 8.239064470981248, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9689, "loss": 3.8802402019500732, "lr": 5.229801370225778e-05, "elapsed_sec": 80580.16805458069, "step_time_sec": 8.238882965990342, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9690, "loss": 3.7510225772857666, "lr": 5.2232367354369327e-05, "elapsed_sec": 80588.4072523117, "step_time_sec": 8.239052496035583, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9691, "loss": 3.848100185394287, "lr": 5.216675202959188e-05, "elapsed_sec": 80596.64579439163, "step_time_sec": 8.238389736041427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9692, "loss": 3.9217116832733154, "lr": 5.210116776842059e-05, "elapsed_sec": 80604.88681173325, "step_time_sec": 8.240848907036707, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9693, "loss": 3.8591041564941406, "lr": 5.203561461133133e-05, "elapsed_sec": 80613.12758541107, "step_time_sec": 8.24057040992193, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9694, "loss": 3.7823493480682373, "lr": 5.19700925987808e-05, "elapsed_sec": 80621.36552858353, "step_time_sec": 8.237763389013708, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9695, "loss": 3.795074224472046, "lr": 5.190460177120648e-05, "elapsed_sec": 80629.60584878922, "step_time_sec": 8.240135730011389, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9696, "loss": 3.937497854232788, "lr": 5.183914216902658e-05, "elapsed_sec": 80637.84534883499, "step_time_sec": 8.239404776017182, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9697, "loss": 3.862802743911743, "lr": 5.177371383264015e-05, "elapsed_sec": 80646.08538794518, "step_time_sec": 8.23984787100926, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9698, "loss": 3.8510935306549072, "lr": 5.170831680242681e-05, "elapsed_sec": 80654.32655858994, "step_time_sec": 8.241078966064379, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9699, "loss": 3.756175994873047, "lr": 5.1642951118746876e-05, "elapsed_sec": 80662.56594014168, "step_time_sec": 8.239121987018734, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9700, "loss": 3.9398553371429443, "lr": 5.1577616821941435e-05, "elapsed_sec": 80670.80717778206, "step_time_sec": 8.241106627043337, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9701, "loss": 3.969583749771118, "lr": 5.151231395233208e-05, "elapsed_sec": 80679.04706382751, "step_time_sec": 8.239737547002733, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9702, "loss": 3.8345859050750732, "lr": 5.144704255022106e-05, "elapsed_sec": 80687.28393030167, "step_time_sec": 8.236717610037886, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9703, "loss": 3.9804348945617676, "lr": 5.138180265589118e-05, "elapsed_sec": 80695.52391433716, "step_time_sec": 8.239741216995753, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9704, "loss": 3.870368480682373, "lr": 5.131659430960583e-05, "elapsed_sec": 80703.76326990128, "step_time_sec": 8.239210027037188, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9705, "loss": 3.9173836708068848, "lr": 5.125141755160887e-05, "elapsed_sec": 80712.00304722786, "step_time_sec": 8.23962697305251, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9706, "loss": 4.021983623504639, "lr": 5.118627242212476e-05, "elapsed_sec": 80720.24138903618, "step_time_sec": 8.238203121931292, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9707, "loss": 3.827453851699829, "lr": 5.11211589613584e-05, "elapsed_sec": 80728.47937178612, "step_time_sec": 8.23778712307103, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9708, "loss": 3.894085645675659, "lr": 5.10560772094951e-05, "elapsed_sec": 80736.71990323067, "step_time_sec": 8.240364202065393, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9709, "loss": 3.770545244216919, "lr": 5.099102720670063e-05, "elapsed_sec": 80744.95960855484, "step_time_sec": 8.239555668085814, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9710, "loss": 3.8069796562194824, "lr": 5.092600899312119e-05, "elapsed_sec": 80753.1983320713, "step_time_sec": 8.23854003206361, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9711, "loss": 3.8262879848480225, "lr": 5.086102260888329e-05, "elapsed_sec": 80761.43760561943, "step_time_sec": 8.23911320604384, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9712, "loss": 3.7683207988739014, "lr": 5.079606809409395e-05, "elapsed_sec": 80769.67493605614, "step_time_sec": 8.237144415033981, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9713, "loss": 3.773662567138672, "lr": 5.073114548884035e-05, "elapsed_sec": 80777.91623044014, "step_time_sec": 8.241180518991314, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9714, "loss": 4.047903060913086, "lr": 5.066625483319002e-05, "elapsed_sec": 80786.15564322472, "step_time_sec": 8.239214225090109, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9715, "loss": 3.804386854171753, "lr": 5.060139616719086e-05, "elapsed_sec": 80794.39388465881, "step_time_sec": 8.238082295050845, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9716, "loss": 3.8845152854919434, "lr": 5.0536569530870956e-05, "elapsed_sec": 80802.633040905, "step_time_sec": 8.23896769003477, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9717, "loss": 3.9627506732940674, "lr": 5.047177496423862e-05, "elapsed_sec": 80810.87230086327, "step_time_sec": 8.23907287302427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9718, "loss": 3.8052561283111572, "lr": 5.040701250728238e-05, "elapsed_sec": 80819.1122970581, "step_time_sec": 8.239900556975044, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9719, "loss": 3.976294755935669, "lr": 5.034228219997094e-05, "elapsed_sec": 80827.3505089283, "step_time_sec": 8.238067497033626, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9720, "loss": 3.758059501647949, "lr": 5.027758408225323e-05, "elapsed_sec": 80835.589884758, "step_time_sec": 8.239184182952158, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9721, "loss": 3.8536243438720703, "lr": 5.0212918194058234e-05, "elapsed_sec": 80843.83138942719, "step_time_sec": 8.241316367988475, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9722, "loss": 3.868256092071533, "lr": 5.0148284575295046e-05, "elapsed_sec": 80852.07020044327, "step_time_sec": 8.238563438062556, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9723, "loss": 3.825925350189209, "lr": 5.008368326585293e-05, "elapsed_sec": 80860.31007170677, "step_time_sec": 8.239723962033167, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9724, "loss": 3.879662036895752, "lr": 5.0019114305601154e-05, "elapsed_sec": 80868.54946160316, "step_time_sec": 8.239224147982895, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9725, "loss": 3.9112155437469482, "lr": 4.995457773438891e-05, "elapsed_sec": 80876.78678560257, "step_time_sec": 8.237151298904791, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9726, "loss": 4.013583183288574, "lr": 4.9890073592045616e-05, "elapsed_sec": 80885.02832651138, "step_time_sec": 8.241378805949353, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9727, "loss": 3.7611145973205566, "lr": 4.9825601918380554e-05, "elapsed_sec": 80893.26805758476, "step_time_sec": 8.23958405898884, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9728, "loss": 3.983644962310791, "lr": 4.976116275318292e-05, "elapsed_sec": 80901.50670838356, "step_time_sec": 8.238494822988287, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9729, "loss": 3.7425434589385986, "lr": 4.9696756136222e-05, "elapsed_sec": 80909.74636912346, "step_time_sec": 8.239463450037874, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9730, "loss": 3.87459659576416, "lr": 4.963238210724687e-05, "elapsed_sec": 80917.98544836044, "step_time_sec": 8.238961094990373, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9731, "loss": 3.916055202484131, "lr": 4.956804070598652e-05, "elapsed_sec": 80926.22558760643, "step_time_sec": 8.239914988982491, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9732, "loss": 3.950172185897827, "lr": 4.95037319721498e-05, "elapsed_sec": 80934.46651911736, "step_time_sec": 8.24077810603194, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9733, "loss": 3.821751117706299, "lr": 4.943945594542543e-05, "elapsed_sec": 80954.3218331337, "step_time_sec": 19.855145442066714, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9734, "loss": 3.9472904205322266, "lr": 4.937521266548187e-05, "elapsed_sec": 80962.55613732338, "step_time_sec": 8.23408947198186, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9735, "loss": 3.811418294906616, "lr": 4.931100217196752e-05, "elapsed_sec": 80970.78802347183, "step_time_sec": 8.231731462059543, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9736, "loss": 3.9278359413146973, "lr": 4.924682450451036e-05, "elapsed_sec": 80979.02143645287, "step_time_sec": 8.233238854096271, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9737, "loss": 3.8731069564819336, "lr": 4.918267970271827e-05, "elapsed_sec": 80987.25339150429, "step_time_sec": 8.231846809037961, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9738, "loss": 3.9955856800079346, "lr": 4.9118567806178757e-05, "elapsed_sec": 80995.48582005501, "step_time_sec": 8.232174939941615, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9739, "loss": 3.782226324081421, "lr": 4.905448885445903e-05, "elapsed_sec": 81003.71811628342, "step_time_sec": 8.23212414898444, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9740, "loss": 3.827287197113037, "lr": 4.899044288710598e-05, "elapsed_sec": 81011.95328688622, "step_time_sec": 8.235010891919956, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9741, "loss": 3.872931957244873, "lr": 4.892642994364614e-05, "elapsed_sec": 81020.19209504128, "step_time_sec": 8.238629103056155, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9742, "loss": 3.828523635864258, "lr": 4.8862450063585617e-05, "elapsed_sec": 81028.43155574799, "step_time_sec": 8.239303002948873, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9743, "loss": 3.960578203201294, "lr": 4.8798503286410244e-05, "elapsed_sec": 81036.67113637924, "step_time_sec": 8.23939776304178, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9744, "loss": 3.8111228942871094, "lr": 4.873458965158529e-05, "elapsed_sec": 81044.90977406502, "step_time_sec": 8.238483772031032, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9745, "loss": 3.820786952972412, "lr": 4.8670709198555574e-05, "elapsed_sec": 81053.14917397499, "step_time_sec": 8.239213300985284, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9746, "loss": 3.811495780944824, "lr": 4.86068619667456e-05, "elapsed_sec": 81061.38769459724, "step_time_sec": 8.238388963975012, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9747, "loss": 3.8191978931427, "lr": 4.854304799555913e-05, "elapsed_sec": 81069.62779378891, "step_time_sec": 8.23992054304108, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9748, "loss": 3.89123272895813, "lr": 4.8479267324379526e-05, "elapsed_sec": 81077.86845779419, "step_time_sec": 8.240517660975456, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9749, "loss": 3.7785181999206543, "lr": 4.841551999256964e-05, "elapsed_sec": 81086.10760641098, "step_time_sec": 8.238964819000103, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9750, "loss": 3.9112770557403564, "lr": 4.83518060394717e-05, "elapsed_sec": 81094.34758329391, "step_time_sec": 8.239807185018435, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9751, "loss": 3.7740776538848877, "lr": 4.828812550440725e-05, "elapsed_sec": 81102.58726167679, "step_time_sec": 8.239480491960421, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9752, "loss": 3.7636730670928955, "lr": 4.822447842667741e-05, "elapsed_sec": 81110.82814264297, "step_time_sec": 8.240701139904559, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9753, "loss": 3.925630569458008, "lr": 4.8160864845562454e-05, "elapsed_sec": 81119.06853318214, "step_time_sec": 8.240189422969706, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9754, "loss": 3.708604097366333, "lr": 4.8097284800322096e-05, "elapsed_sec": 81127.30667638779, "step_time_sec": 8.238037950941361, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9755, "loss": 3.793694019317627, "lr": 4.8033738330195294e-05, "elapsed_sec": 81135.54863667488, "step_time_sec": 8.241778419003822, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9756, "loss": 3.7835373878479004, "lr": 4.7970225474400296e-05, "elapsed_sec": 81143.79421520233, "step_time_sec": 8.245402240077965, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9757, "loss": 3.831794261932373, "lr": 4.7906746272134677e-05, "elapsed_sec": 81152.03585481644, "step_time_sec": 8.241478562005796, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9758, "loss": 3.921064853668213, "lr": 4.784330076257515e-05, "elapsed_sec": 81160.27835416794, "step_time_sec": 8.24228112702258, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9759, "loss": 3.722649574279785, "lr": 4.777988898487764e-05, "elapsed_sec": 81168.51925730705, "step_time_sec": 8.240743448026478, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9760, "loss": 3.8310680389404297, "lr": 4.771651097817735e-05, "elapsed_sec": 81176.75711894035, "step_time_sec": 8.237704936997034, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9761, "loss": 4.059420108795166, "lr": 4.765316678158853e-05, "elapsed_sec": 81184.99735832214, "step_time_sec": 8.240074415924028, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9762, "loss": 3.854227066040039, "lr": 4.758985643420462e-05, "elapsed_sec": 81193.23749256134, "step_time_sec": 8.239966891007498, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9763, "loss": 3.9247686862945557, "lr": 4.7526579975098145e-05, "elapsed_sec": 81201.47696661949, "step_time_sec": 8.239293275983073, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9764, "loss": 4.07949686050415, "lr": 4.7463337443320746e-05, "elapsed_sec": 81209.71804833412, "step_time_sec": 8.240940670017153, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9765, "loss": 3.717690944671631, "lr": 4.740012887790304e-05, "elapsed_sec": 81217.9584364891, "step_time_sec": 8.240225759916939, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9766, "loss": 3.8533782958984375, "lr": 4.733695431785486e-05, "elapsed_sec": 81226.19826197624, "step_time_sec": 8.239587894058786, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9767, "loss": 3.7986526489257812, "lr": 4.727381380216487e-05, "elapsed_sec": 81234.43866801262, "step_time_sec": 8.240293998038396, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9768, "loss": 3.829535961151123, "lr": 4.721070736980081e-05, "elapsed_sec": 81242.67728257179, "step_time_sec": 8.238407955970615, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9769, "loss": 3.8630270957946777, "lr": 4.714763505970937e-05, "elapsed_sec": 81250.91765069962, "step_time_sec": 8.24018814403098, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9770, "loss": 3.789641857147217, "lr": 4.708459691081618e-05, "elapsed_sec": 81259.15999555588, "step_time_sec": 8.242214398924261, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9771, "loss": 3.7321996688842773, "lr": 4.702159296202574e-05, "elapsed_sec": 81267.39824080467, "step_time_sec": 8.238085748977028, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9772, "loss": 3.8284637928009033, "lr": 4.695862325222158e-05, "elapsed_sec": 81275.6382727623, "step_time_sec": 8.239810177008621, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9773, "loss": 3.7836098670959473, "lr": 4.689568782026592e-05, "elapsed_sec": 81283.87796545029, "step_time_sec": 8.239533449988812, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9774, "loss": 3.8469595909118652, "lr": 4.6832786705e-05, "elapsed_sec": 81292.11764407158, "step_time_sec": 8.239489894011058, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9775, "loss": 3.8546829223632812, "lr": 4.676991994524376e-05, "elapsed_sec": 81300.35728502274, "step_time_sec": 8.239520579925738, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9776, "loss": 3.811854362487793, "lr": 4.670708757979595e-05, "elapsed_sec": 81308.59528565407, "step_time_sec": 8.237792460946366, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9777, "loss": 3.7835910320281982, "lr": 4.6644289647434146e-05, "elapsed_sec": 81316.83543848991, "step_time_sec": 8.24002887797542, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9778, "loss": 3.840843677520752, "lr": 4.6581526186914644e-05, "elapsed_sec": 81325.07583832741, "step_time_sec": 8.240168682998046, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9779, "loss": 3.8874523639678955, "lr": 4.65187972369724e-05, "elapsed_sec": 81333.31400489807, "step_time_sec": 8.238043374032713, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9780, "loss": 3.695387601852417, "lr": 4.645610283632123e-05, "elapsed_sec": 81341.5553855896, "step_time_sec": 8.241200143937021, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9781, "loss": 3.9458446502685547, "lr": 4.639344302365349e-05, "elapsed_sec": 81349.79292726517, "step_time_sec": 8.237373895011842, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9782, "loss": 3.9050493240356445, "lr": 4.6330817837640206e-05, "elapsed_sec": 81358.03251862526, "step_time_sec": 8.239393713069148, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9783, "loss": 3.878150224685669, "lr": 4.6268227316931125e-05, "elapsed_sec": 81366.27236914635, "step_time_sec": 8.239675319986418, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9784, "loss": 3.799210548400879, "lr": 4.620567150015454e-05, "elapsed_sec": 81374.51151275635, "step_time_sec": 8.239042994915508, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9785, "loss": 3.778841018676758, "lr": 4.6143150425917215e-05, "elapsed_sec": 81382.75162625313, "step_time_sec": 8.239926530979574, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9786, "loss": 3.7279083728790283, "lr": 4.608066413280469e-05, "elapsed_sec": 81390.99143147469, "step_time_sec": 8.23965904302895, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9787, "loss": 3.912025213241577, "lr": 4.6018212659380894e-05, "elapsed_sec": 81399.22970795631, "step_time_sec": 8.238057874026708, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9788, "loss": 3.7866361141204834, "lr": 4.5955796044188266e-05, "elapsed_sec": 81407.46958947182, "step_time_sec": 8.239777498994954, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9789, "loss": 3.7317490577697754, "lr": 4.589341432574783e-05, "elapsed_sec": 81415.7071120739, "step_time_sec": 8.237301027984358, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9790, "loss": 3.8426432609558105, "lr": 4.583106754255899e-05, "elapsed_sec": 81423.94643902779, "step_time_sec": 8.239145551924594, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9791, "loss": 3.9157423973083496, "lr": 4.5768755733099614e-05, "elapsed_sec": 81432.1856765747, "step_time_sec": 8.239071439020336, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9792, "loss": 3.8070898056030273, "lr": 4.570647893582597e-05, "elapsed_sec": 81440.42589497566, "step_time_sec": 8.240070696920156, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9793, "loss": 3.7362048625946045, "lr": 4.5644237189172686e-05, "elapsed_sec": 81448.66578245163, "step_time_sec": 8.2397302640602, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9794, "loss": 3.730788230895996, "lr": 4.5582030531552904e-05, "elapsed_sec": 81456.90468215942, "step_time_sec": 8.238755632890388, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9795, "loss": 3.7639083862304688, "lr": 4.551985900135794e-05, "elapsed_sec": 81465.14558339119, "step_time_sec": 8.240686979028396, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9796, "loss": 3.8812944889068604, "lr": 4.545772263695749e-05, "elapsed_sec": 81473.3859038353, "step_time_sec": 8.240164931979962, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9797, "loss": 3.859957695007324, "lr": 4.539562147669962e-05, "elapsed_sec": 81481.62535738945, "step_time_sec": 8.239302188972943, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9798, "loss": 3.8626246452331543, "lr": 4.533355555891056e-05, "elapsed_sec": 81489.86617803574, "step_time_sec": 8.240678254049271, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9799, "loss": 3.7276015281677246, "lr": 4.527152492189486e-05, "elapsed_sec": 81498.10660505295, "step_time_sec": 8.240190882002935, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9800, "loss": 3.7929112911224365, "lr": 4.5209529603935244e-05, "elapsed_sec": 81506.34730792046, "step_time_sec": 8.240546474000439, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9801, "loss": 3.7921369075775146, "lr": 4.5147569643292694e-05, "elapsed_sec": 81514.58858203888, "step_time_sec": 8.241164193954319, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9802, "loss": 3.781353712081909, "lr": 4.5085645078206285e-05, "elapsed_sec": 81522.82812833786, "step_time_sec": 8.239330858923495, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9803, "loss": 3.7732350826263428, "lr": 4.50237559468934e-05, "elapsed_sec": 81531.0667719841, "step_time_sec": 8.238486674963497, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9804, "loss": 3.924985408782959, "lr": 4.496190228754941e-05, "elapsed_sec": 81539.30681753159, "step_time_sec": 8.239889324060641, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9805, "loss": 3.8596928119659424, "lr": 4.49000841383478e-05, "elapsed_sec": 81547.5442507267, "step_time_sec": 8.237280996982008, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9806, "loss": 3.771496534347534, "lr": 4.483830153744032e-05, "elapsed_sec": 81555.7851793766, "step_time_sec": 8.240742641035467, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9807, "loss": 3.7631494998931885, "lr": 4.4776554522956504e-05, "elapsed_sec": 81564.02477097511, "step_time_sec": 8.239415063988417, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9808, "loss": 3.98242449760437, "lr": 4.47148431330041e-05, "elapsed_sec": 81572.26459646225, "step_time_sec": 8.23960565798916, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9809, "loss": 3.800337076187134, "lr": 4.465316740566887e-05, "elapsed_sec": 81580.50461769104, "step_time_sec": 8.239915251964703, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9810, "loss": 3.952719211578369, "lr": 4.459152737901448e-05, "elapsed_sec": 81588.74396085739, "step_time_sec": 8.239131318056025, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9811, "loss": 3.888658046722412, "lr": 4.452992309108269e-05, "elapsed_sec": 81596.98422670364, "step_time_sec": 8.24015387496911, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9812, "loss": 3.9153892993927, "lr": 4.446835457989307e-05, "elapsed_sec": 81605.2248468399, "step_time_sec": 8.240392350940965, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9813, "loss": 3.8220250606536865, "lr": 4.440682188344319e-05, "elapsed_sec": 81613.46393609047, "step_time_sec": 8.238937556976452, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9814, "loss": 4.0151824951171875, "lr": 4.4345325039708476e-05, "elapsed_sec": 81621.70597028732, "step_time_sec": 8.241891559911892, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9815, "loss": 3.7794642448425293, "lr": 4.4283864086642254e-05, "elapsed_sec": 81629.94656562805, "step_time_sec": 8.240421273978427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9816, "loss": 3.8640899658203125, "lr": 4.4222439062175666e-05, "elapsed_sec": 81638.18611359596, "step_time_sec": 8.239391810027882, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9817, "loss": 3.943798780441284, "lr": 4.416105000421776e-05, "elapsed_sec": 81646.42740988731, "step_time_sec": 8.241069104056805, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9818, "loss": 3.8768866062164307, "lr": 4.40996969506553e-05, "elapsed_sec": 81654.66615128517, "step_time_sec": 8.238598445081152, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9819, "loss": 3.998091697692871, "lr": 4.4038379939352834e-05, "elapsed_sec": 81662.90576791763, "step_time_sec": 8.239453825983219, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9820, "loss": 3.799481153488159, "lr": 4.397709900815275e-05, "elapsed_sec": 81671.14628434181, "step_time_sec": 8.240322296973318, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9821, "loss": 3.9398512840270996, "lr": 4.3915854194875107e-05, "elapsed_sec": 81679.38593769073, "step_time_sec": 8.23949394095689, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9822, "loss": 4.013787269592285, "lr": 4.385464553731764e-05, "elapsed_sec": 81687.62655639648, "step_time_sec": 8.24050219496712, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9823, "loss": 3.8344552516937256, "lr": 4.379347307325584e-05, "elapsed_sec": 81695.86597752571, "step_time_sec": 8.239241818082519, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9824, "loss": 3.86348032951355, "lr": 4.373233684044283e-05, "elapsed_sec": 81704.10602068901, "step_time_sec": 8.23985856201034, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9825, "loss": 4.013617992401123, "lr": 4.3671236876609306e-05, "elapsed_sec": 81712.34716272354, "step_time_sec": 8.240940276999027, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9826, "loss": 3.9773318767547607, "lr": 4.3610173219463754e-05, "elapsed_sec": 81720.5859310627, "step_time_sec": 8.238587955012918, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9827, "loss": 3.9190151691436768, "lr": 4.3549145906692096e-05, "elapsed_sec": 81728.82615232468, "step_time_sec": 8.240055078058504, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9828, "loss": 3.9172468185424805, "lr": 4.3488154975957876e-05, "elapsed_sec": 81737.06774282455, "step_time_sec": 8.241454000002705, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9829, "loss": 3.9972150325775146, "lr": 4.342720046490218e-05, "elapsed_sec": 81745.3063929081, "step_time_sec": 8.238479251042008, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9830, "loss": 3.8582725524902344, "lr": 4.336628241114359e-05, "elapsed_sec": 81753.54763913155, "step_time_sec": 8.241099640959874, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9831, "loss": 3.9105403423309326, "lr": 4.330540085227829e-05, "elapsed_sec": 81761.78812980652, "step_time_sec": 8.240292675094679, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9832, "loss": 4.053820610046387, "lr": 4.324455582587982e-05, "elapsed_sec": 81770.02687358856, "step_time_sec": 8.23859184095636, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9833, "loss": 3.931243658065796, "lr": 4.3183747369499204e-05, "elapsed_sec": 81778.26715373993, "step_time_sec": 8.240049315965734, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9834, "loss": 4.09175968170166, "lr": 4.3122975520664965e-05, "elapsed_sec": 81786.50534248352, "step_time_sec": 8.238052236963995, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9835, "loss": 3.8753952980041504, "lr": 4.306224031688297e-05, "elapsed_sec": 81794.74520421028, "step_time_sec": 8.239714749972336, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9836, "loss": 3.945483446121216, "lr": 4.300154179563647e-05, "elapsed_sec": 81802.9854888916, "step_time_sec": 8.240074689965695, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9837, "loss": 3.917668104171753, "lr": 4.2940879994386095e-05, "elapsed_sec": 81811.22420573235, "step_time_sec": 8.238536736927927, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9838, "loss": 3.9750123023986816, "lr": 4.288025495056982e-05, "elapsed_sec": 81819.46536421776, "step_time_sec": 8.24102482793387, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9839, "loss": 3.8617353439331055, "lr": 4.2819666701602865e-05, "elapsed_sec": 81827.70480751991, "step_time_sec": 8.239342267042957, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9840, "loss": 3.9793620109558105, "lr": 4.275911528487789e-05, "elapsed_sec": 81835.94293475151, "step_time_sec": 8.237893227022141, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9841, "loss": 3.816819429397583, "lr": 4.2698600737764685e-05, "elapsed_sec": 81844.18112826347, "step_time_sec": 8.238023945945315, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9842, "loss": 3.8518543243408203, "lr": 4.263812309761031e-05, "elapsed_sec": 81852.42034721375, "step_time_sec": 8.239045473048463, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9843, "loss": 3.8028225898742676, "lr": 4.257768240173913e-05, "elapsed_sec": 81860.6615896225, "step_time_sec": 8.24110608396586, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9844, "loss": 4.021744728088379, "lr": 4.251727868745266e-05, "elapsed_sec": 81868.90201830864, "step_time_sec": 8.240245596971363, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9845, "loss": 3.70422625541687, "lr": 4.2456911992029484e-05, "elapsed_sec": 81877.14079141617, "step_time_sec": 8.238617570023052, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9846, "loss": 3.833724021911621, "lr": 4.239658235272552e-05, "elapsed_sec": 81885.38075733185, "step_time_sec": 8.239780214964412, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9847, "loss": 3.8248844146728516, "lr": 4.233628980677372e-05, "elapsed_sec": 81893.61798000336, "step_time_sec": 8.237030684016645, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9848, "loss": 3.907167673110962, "lr": 4.227603439138412e-05, "elapsed_sec": 81901.85740017891, "step_time_sec": 8.239295081002638, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9849, "loss": 3.8176143169403076, "lr": 4.2215816143743965e-05, "elapsed_sec": 81910.0969069004, "step_time_sec": 8.23933254298754, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9850, "loss": 3.9485068321228027, "lr": 4.2155635101017396e-05, "elapsed_sec": 81918.33504295349, "step_time_sec": 8.237929382012226, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9851, "loss": 3.7862956523895264, "lr": 4.209549130034571e-05, "elapsed_sec": 81926.57665348053, "step_time_sec": 8.241441370919347, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9852, "loss": 3.91341495513916, "lr": 4.203538477884718e-05, "elapsed_sec": 81934.81628847122, "step_time_sec": 8.239484795019962, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9853, "loss": 3.938736915588379, "lr": 4.197531557361702e-05, "elapsed_sec": 81943.05578303337, "step_time_sec": 8.23932184302248, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9854, "loss": 3.7677247524261475, "lr": 4.191528372172755e-05, "elapsed_sec": 81951.29593658447, "step_time_sec": 8.239979846053757, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9855, "loss": 3.80836820602417, "lr": 4.185528926022792e-05, "elapsed_sec": 81959.5348815918, "step_time_sec": 8.23876681108959, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9856, "loss": 3.851731061935425, "lr": 4.1795332226144194e-05, "elapsed_sec": 81967.77488017082, "step_time_sec": 8.239862933987752, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9857, "loss": 3.9591269493103027, "lr": 4.173541265647946e-05, "elapsed_sec": 81976.01516962051, "step_time_sec": 8.240077215014026, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9858, "loss": 3.734860897064209, "lr": 4.167553058821357e-05, "elapsed_sec": 81984.25343084335, "step_time_sec": 8.238080241018906, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9859, "loss": 3.815228223800659, "lr": 4.161568605830328e-05, "elapsed_sec": 81992.4931731224, "step_time_sec": 8.239583346992731, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9860, "loss": 3.9371237754821777, "lr": 4.1555879103682154e-05, "elapsed_sec": 82000.7323691845, "step_time_sec": 8.239016099949367, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9861, "loss": 3.9350759983062744, "lr": 4.149610976126058e-05, "elapsed_sec": 82008.97176146507, "step_time_sec": 8.239290575962514, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9862, "loss": 3.8336381912231445, "lr": 4.1436378067925695e-05, "elapsed_sec": 82017.21255898476, "step_time_sec": 8.240659655071795, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9863, "loss": 3.9861674308776855, "lr": 4.137668406054151e-05, "elapsed_sec": 82025.4511525631, "step_time_sec": 8.23838287696708, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9864, "loss": 3.858731508255005, "lr": 4.131702777594869e-05, "elapsed_sec": 82033.69078421593, "step_time_sec": 8.23947136302013, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9865, "loss": 3.9466567039489746, "lr": 4.125740925096457e-05, "elapsed_sec": 82041.93128156662, "step_time_sec": 8.240274231997319, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9866, "loss": 3.840414524078369, "lr": 4.1197828522383376e-05, "elapsed_sec": 82050.16991400719, "step_time_sec": 8.238506025983952, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9867, "loss": 3.9409492015838623, "lr": 4.113828562697572e-05, "elapsed_sec": 82058.41085600853, "step_time_sec": 8.240772739052773, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9868, "loss": 3.7798268795013428, "lr": 4.107878060148914e-05, "elapsed_sec": 82066.65125989914, "step_time_sec": 8.240194650017656, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9869, "loss": 3.8358795642852783, "lr": 4.101931348264764e-05, "elapsed_sec": 82074.89144706726, "step_time_sec": 8.240036136005074, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9870, "loss": 3.85092830657959, "lr": 4.095988430715185e-05, "elapsed_sec": 82083.13119411469, "step_time_sec": 8.239571225945838, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9871, "loss": 3.912555456161499, "lr": 4.090049311167908e-05, "elapsed_sec": 82091.36412668228, "step_time_sec": 8.232769440975972, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9872, "loss": 3.90069580078125, "lr": 4.0841139932883075e-05, "elapsed_sec": 82099.59848594666, "step_time_sec": 8.23419922508765, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9873, "loss": 3.7982802391052246, "lr": 4.0781824807394186e-05, "elapsed_sec": 82107.83137249947, "step_time_sec": 8.232756519922987, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9874, "loss": 3.925732374191284, "lr": 4.072254777181926e-05, "elapsed_sec": 82116.07157301903, "step_time_sec": 8.239979233941995, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9875, "loss": 3.7835018634796143, "lr": 4.0663308862741644e-05, "elapsed_sec": 82124.31116628647, "step_time_sec": 8.239484107936732, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9876, "loss": 3.9494829177856445, "lr": 4.060410811672111e-05, "elapsed_sec": 82132.55072641373, "step_time_sec": 8.239332903991453, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9877, "loss": 3.9181010723114014, "lr": 4.054494557029399e-05, "elapsed_sec": 82140.79154729843, "step_time_sec": 8.240683462005109, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9878, "loss": 4.005282402038574, "lr": 4.0485821259972926e-05, "elapsed_sec": 82149.02984428406, "step_time_sec": 8.23809620400425, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9879, "loss": 3.8082022666931152, "lr": 4.0426735222246984e-05, "elapsed_sec": 82157.26986980438, "step_time_sec": 8.239918334991671, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9880, "loss": 3.8187789916992188, "lr": 4.0367687493581705e-05, "elapsed_sec": 82165.51049923897, "step_time_sec": 8.240434657898732, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9881, "loss": 3.708258867263794, "lr": 4.030867811041886e-05, "elapsed_sec": 82173.74864625931, "step_time_sec": 8.237947526969947, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9882, "loss": 4.091879367828369, "lr": 4.024970710917662e-05, "elapsed_sec": 82181.98918199539, "step_time_sec": 8.240352443070151, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9883, "loss": 3.8313405513763428, "lr": 4.019077452624945e-05, "elapsed_sec": 82190.22919654846, "step_time_sec": 8.239891962031834, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9884, "loss": 3.877760410308838, "lr": 4.0131880398008124e-05, "elapsed_sec": 82198.46918058395, "step_time_sec": 8.239820651011541, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9885, "loss": 3.839761257171631, "lr": 4.0073024760799626e-05, "elapsed_sec": 82206.70993685722, "step_time_sec": 8.240569352987222, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9886, "loss": 3.704981565475464, "lr": 4.001420765094729e-05, "elapsed_sec": 82214.94226408005, "step_time_sec": 8.232197402976453, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9887, "loss": 3.74519681930542, "lr": 3.995542910475057e-05, "elapsed_sec": 82223.17621231079, "step_time_sec": 8.233722698059864, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9888, "loss": 3.8542537689208984, "lr": 3.989668915848524e-05, "elapsed_sec": 82231.40892314911, "step_time_sec": 8.232537955977023, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9889, "loss": 3.857023000717163, "lr": 3.983798784840307e-05, "elapsed_sec": 82239.64954829216, "step_time_sec": 8.240454804035835, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9890, "loss": 3.7923569679260254, "lr": 3.977932521073211e-05, "elapsed_sec": 82247.88889884949, "step_time_sec": 8.239225336001255, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9891, "loss": 3.8107173442840576, "lr": 3.972070128167658e-05, "elapsed_sec": 82256.12866687775, "step_time_sec": 8.239549495978281, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9892, "loss": 3.7655081748962402, "lr": 3.9662116097416715e-05, "elapsed_sec": 82264.367456913, "step_time_sec": 8.238632308901288, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9893, "loss": 3.883694648742676, "lr": 3.9603569694108836e-05, "elapsed_sec": 82272.60552358627, "step_time_sec": 8.2378863140475, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9894, "loss": 3.776841163635254, "lr": 3.9545062107885466e-05, "elapsed_sec": 82280.84560728073, "step_time_sec": 8.239958444028161, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9895, "loss": 4.016205310821533, "lr": 3.948659337485501e-05, "elapsed_sec": 82289.08564209938, "step_time_sec": 8.239858053973876, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9896, "loss": 3.773930311203003, "lr": 3.942816353110197e-05, "elapsed_sec": 82297.32555794716, "step_time_sec": 8.239780287025496, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9897, "loss": 3.8105034828186035, "lr": 3.936977261268684e-05, "elapsed_sec": 82305.56614875793, "step_time_sec": 8.240386973018758, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9898, "loss": 3.7597849369049072, "lr": 3.9311420655646096e-05, "elapsed_sec": 82313.80506944656, "step_time_sec": 8.238712580990978, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9899, "loss": 3.8339812755584717, "lr": 3.925310769599211e-05, "elapsed_sec": 82322.0456237793, "step_time_sec": 8.24040045402944, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9900, "loss": 3.78623366355896, "lr": 3.919483376971332e-05, "elapsed_sec": 82330.28524065018, "step_time_sec": 8.239502819022164, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9901, "loss": 3.97202205657959, "lr": 3.9136598912773954e-05, "elapsed_sec": 82338.5256266594, "step_time_sec": 8.240178785054013, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9902, "loss": 3.9831345081329346, "lr": 3.9078403161114133e-05, "elapsed_sec": 82346.76575279236, "step_time_sec": 8.239995942916721, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9903, "loss": 3.7684903144836426, "lr": 3.902024655064993e-05, "elapsed_sec": 82355.0164437294, "step_time_sec": 8.2505663089687, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9904, "loss": 3.862222909927368, "lr": 3.896212911727322e-05, "elapsed_sec": 82363.26589035988, "step_time_sec": 8.24919999903068, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9905, "loss": 3.750441789627075, "lr": 3.890405089685159e-05, "elapsed_sec": 82371.5049712658, "step_time_sec": 8.238851558999158, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9906, "loss": 3.8961730003356934, "lr": 3.884601192522862e-05, "elapsed_sec": 82379.74445033073, "step_time_sec": 8.239303836016916, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9907, "loss": 3.914957046508789, "lr": 3.87880122382235e-05, "elapsed_sec": 82387.9835486412, "step_time_sec": 8.2389813689515, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9908, "loss": 3.892754554748535, "lr": 3.873005187163133e-05, "elapsed_sec": 82396.22412443161, "step_time_sec": 8.240335952024907, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9909, "loss": 3.858922004699707, "lr": 3.8672130861222796e-05, "elapsed_sec": 82404.46526527405, "step_time_sec": 8.24097520892974, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9910, "loss": 3.766422748565674, "lr": 3.8614249242744384e-05, "elapsed_sec": 82412.70386123657, "step_time_sec": 8.23845506098587, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9911, "loss": 3.873093366622925, "lr": 3.855640705191822e-05, "elapsed_sec": 82420.94420075417, "step_time_sec": 8.24014816293493, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9912, "loss": 3.7230517864227295, "lr": 3.849860432444215e-05, "elapsed_sec": 82429.1843612194, "step_time_sec": 8.240056730923243, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9913, "loss": 3.7156410217285156, "lr": 3.8440841095989594e-05, "elapsed_sec": 82437.42355012894, "step_time_sec": 8.239006173913367, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9914, "loss": 3.8545238971710205, "lr": 3.838311740220967e-05, "elapsed_sec": 82445.66374468803, "step_time_sec": 8.23998357495293, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9915, "loss": 3.695570468902588, "lr": 3.832543327872708e-05, "elapsed_sec": 82453.90412306786, "step_time_sec": 8.240251338924281, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9916, "loss": 4.0637736320495605, "lr": 3.826778876114202e-05, "elapsed_sec": 82462.14367055893, "step_time_sec": 8.239291956997477, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9917, "loss": 3.921757221221924, "lr": 3.821018388503038e-05, "elapsed_sec": 82470.38257241249, "step_time_sec": 8.238763593952172, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9918, "loss": 3.8728504180908203, "lr": 3.815261868594349e-05, "elapsed_sec": 82478.61489152908, "step_time_sec": 8.232143769972026, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9919, "loss": 3.833939790725708, "lr": 3.809509319940822e-05, "elapsed_sec": 82486.84828472137, "step_time_sec": 8.233212592895143, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9920, "loss": 3.8431601524353027, "lr": 3.803760746092691e-05, "elapsed_sec": 82495.08060145378, "step_time_sec": 8.232149739051238, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9921, "loss": 3.8248634338378906, "lr": 3.798016150597742e-05, "elapsed_sec": 82503.31544589996, "step_time_sec": 8.234676001942717, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9922, "loss": 3.9698431491851807, "lr": 3.792275537001298e-05, "elapsed_sec": 82511.5484392643, "step_time_sec": 8.232809378998354, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9923, "loss": 3.8805201053619385, "lr": 3.786538908846234e-05, "elapsed_sec": 82519.7830979824, "step_time_sec": 8.23448233900126, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9924, "loss": 3.723508358001709, "lr": 3.7808062696729584e-05, "elapsed_sec": 82528.01633000374, "step_time_sec": 8.233109938912094, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9925, "loss": 3.7505366802215576, "lr": 3.775077623019418e-05, "elapsed_sec": 82536.25046491623, "step_time_sec": 8.233912150957622, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9926, "loss": 3.8909459114074707, "lr": 3.769352972421105e-05, "elapsed_sec": 82544.48364686966, "step_time_sec": 8.233057272969745, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9927, "loss": 3.72642183303833, "lr": 3.763632321411026e-05, "elapsed_sec": 82552.71644711494, "step_time_sec": 8.23262614395935, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9928, "loss": 3.781092882156372, "lr": 3.75791567351974e-05, "elapsed_sec": 82560.95033168793, "step_time_sec": 8.233670916990377, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9929, "loss": 3.874668836593628, "lr": 3.752203032275324e-05, "elapsed_sec": 82569.18426203728, "step_time_sec": 8.23380581906531, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9930, "loss": 3.8144009113311768, "lr": 3.746494401203381e-05, "elapsed_sec": 82577.41845607758, "step_time_sec": 8.234081361908466, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9931, "loss": 3.845092535018921, "lr": 3.740789783827049e-05, "elapsed_sec": 82585.65258383751, "step_time_sec": 8.233939775964245, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9932, "loss": 3.736560344696045, "lr": 3.73508918366698e-05, "elapsed_sec": 82593.88678050041, "step_time_sec": 8.23397419997491, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9933, "loss": 3.7746739387512207, "lr": 3.7293926042413504e-05, "elapsed_sec": 82602.11914277077, "step_time_sec": 8.232186594977975, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9934, "loss": 3.757688522338867, "lr": 3.7237000490658515e-05, "elapsed_sec": 82610.35330843925, "step_time_sec": 8.234013816923834, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9935, "loss": 3.8686256408691406, "lr": 3.718011521653697e-05, "elapsed_sec": 82618.58717656136, "step_time_sec": 8.233705357997678, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9936, "loss": 3.916041851043701, "lr": 3.7123270255156047e-05, "elapsed_sec": 82626.82110524178, "step_time_sec": 8.233771105995402, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9937, "loss": 3.9304606914520264, "lr": 3.706646564159822e-05, "elapsed_sec": 82635.05527019501, "step_time_sec": 8.234051106963307, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9938, "loss": 3.8887786865234375, "lr": 3.7009701410920885e-05, "elapsed_sec": 82643.29485344887, "step_time_sec": 8.239352112985216, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9939, "loss": 3.7968943119049072, "lr": 3.695297759815658e-05, "elapsed_sec": 82651.53446435928, "step_time_sec": 8.23945144098252, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9940, "loss": 3.681570291519165, "lr": 3.689629423831295e-05, "elapsed_sec": 82659.77281403542, "step_time_sec": 8.238203417975456, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9941, "loss": 3.7045843601226807, "lr": 3.6839651366372614e-05, "elapsed_sec": 82668.01358246803, "step_time_sec": 8.240617304923944, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9942, "loss": 3.7972323894500732, "lr": 3.67830490172932e-05, "elapsed_sec": 82676.25290608406, "step_time_sec": 8.239087746944278, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9943, "loss": 3.8484177589416504, "lr": 3.6726487226007356e-05, "elapsed_sec": 82684.49170517921, "step_time_sec": 8.23867289593909, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9944, "loss": 3.8621604442596436, "lr": 3.666996602742266e-05, "elapsed_sec": 82692.73024249077, "step_time_sec": 8.238365923985839, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9945, "loss": 3.7501792907714844, "lr": 3.661348545642171e-05, "elapsed_sec": 82700.97094345093, "step_time_sec": 8.240534843993373, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9946, "loss": 3.6635680198669434, "lr": 3.655704554786197e-05, "elapsed_sec": 82709.21131062508, "step_time_sec": 8.240184176014736, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9947, "loss": 3.9514098167419434, "lr": 3.6500646336575806e-05, "elapsed_sec": 82717.45005917549, "step_time_sec": 8.238612675922923, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9948, "loss": 3.88387131690979, "lr": 3.644428785737056e-05, "elapsed_sec": 82725.68897628784, "step_time_sec": 8.238719081971794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9949, "loss": 3.7710959911346436, "lr": 3.638797014502827e-05, "elapsed_sec": 82733.92906117439, "step_time_sec": 8.239867727039382, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9950, "loss": 3.7644224166870117, "lr": 3.633169323430592e-05, "elapsed_sec": 82742.16884207726, "step_time_sec": 8.239595563034527, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9951, "loss": 3.7114319801330566, "lr": 3.627545715993536e-05, "elapsed_sec": 82750.4091656208, "step_time_sec": 8.240180923021398, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9952, "loss": 3.851067304611206, "lr": 3.621926195662313e-05, "elapsed_sec": 82758.64908885956, "step_time_sec": 8.23979598004371, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9953, "loss": 3.781255006790161, "lr": 3.6163107659050585e-05, "elapsed_sec": 82766.88857531548, "step_time_sec": 8.2392681799829, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9954, "loss": 3.590757131576538, "lr": 3.610699430187388e-05, "elapsed_sec": 82775.128510952, "step_time_sec": 8.239765352918766, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9955, "loss": 3.852830171585083, "lr": 3.605092191972385e-05, "elapsed_sec": 82783.36677193642, "step_time_sec": 8.238074515014887, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9956, "loss": 3.895784378051758, "lr": 3.599489054720607e-05, "elapsed_sec": 82791.67569613457, "step_time_sec": 8.246124735102057, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9957, "loss": 3.75881290435791, "lr": 3.593890021890077e-05, "elapsed_sec": 82799.91480994225, "step_time_sec": 8.238955189008266, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9958, "loss": 3.8585638999938965, "lr": 3.588295096936288e-05, "elapsed_sec": 82808.15544462204, "step_time_sec": 8.240456387051381, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9959, "loss": 3.7715132236480713, "lr": 3.5827042833121945e-05, "elapsed_sec": 82816.39470243454, "step_time_sec": 8.239089875016361, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9960, "loss": 3.6167783737182617, "lr": 3.577117584468223e-05, "elapsed_sec": 82824.63386964798, "step_time_sec": 8.238988616038114, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9961, "loss": 3.704648494720459, "lr": 3.5715350038522486e-05, "elapsed_sec": 82832.87503528595, "step_time_sec": 8.241020823013969, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9962, "loss": 3.626296281814575, "lr": 3.5659565449096094e-05, "elapsed_sec": 82841.11527180672, "step_time_sec": 8.240074191009626, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9963, "loss": 3.729855537414551, "lr": 3.5603822110831056e-05, "elapsed_sec": 82849.35415673256, "step_time_sec": 8.238714226987213, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9964, "loss": 3.884904384613037, "lr": 3.554812005812982e-05, "elapsed_sec": 82857.59291529655, "step_time_sec": 8.238629395025782, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9965, "loss": 3.858316659927368, "lr": 3.549245932536941e-05, "elapsed_sec": 82865.8316321373, "step_time_sec": 8.238532611983828, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9966, "loss": 3.8432154655456543, "lr": 3.5436839946901336e-05, "elapsed_sec": 82874.07297515869, "step_time_sec": 8.24112504790537, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9967, "loss": 3.925887107849121, "lr": 3.5381261957051556e-05, "elapsed_sec": 82882.31372213364, "step_time_sec": 8.24058696301654, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9968, "loss": 3.8660175800323486, "lr": 3.5325725390120554e-05, "elapsed_sec": 82890.55236411095, "step_time_sec": 8.2384736899985, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9969, "loss": 3.9028730392456055, "lr": 3.527023028038321e-05, "elapsed_sec": 82898.79217147827, "step_time_sec": 8.239638547063805, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9970, "loss": 3.7858424186706543, "lr": 3.521477666208881e-05, "elapsed_sec": 82907.03012800217, "step_time_sec": 8.23777266801335, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9971, "loss": 3.7363197803497314, "lr": 3.5159364569461025e-05, "elapsed_sec": 82915.27149915695, "step_time_sec": 8.241200039978139, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9972, "loss": 3.879350185394287, "lr": 3.510399403669793e-05, "elapsed_sec": 82923.51196432114, "step_time_sec": 8.240290665999055, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9973, "loss": 3.9430184364318848, "lr": 3.504866509797191e-05, "elapsed_sec": 82931.7511177063, "step_time_sec": 8.239003552007489, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9974, "loss": 3.832031726837158, "lr": 3.499337778742975e-05, "elapsed_sec": 82939.99211621284, "step_time_sec": 8.240812645992264, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9975, "loss": 3.71754789352417, "lr": 3.493813213919248e-05, "elapsed_sec": 82948.23119902611, "step_time_sec": 8.23889905505348, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9976, "loss": 3.9682116508483887, "lr": 3.488292818735541e-05, "elapsed_sec": 82956.47031235695, "step_time_sec": 8.238974767969921, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9977, "loss": 3.92775297164917, "lr": 3.482776596598819e-05, "elapsed_sec": 82964.71083903313, "step_time_sec": 8.240367935039103, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9978, "loss": 3.6042380332946777, "lr": 3.4772645509134655e-05, "elapsed_sec": 82972.95022010803, "step_time_sec": 8.239205604069866, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9979, "loss": 3.8364572525024414, "lr": 3.471756685081287e-05, "elapsed_sec": 82981.19036746025, "step_time_sec": 8.239984681014903, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9980, "loss": 3.8317618370056152, "lr": 3.4662530025015123e-05, "elapsed_sec": 82989.43027472496, "step_time_sec": 8.239699100959115, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9981, "loss": 3.8472299575805664, "lr": 3.4607535065707855e-05, "elapsed_sec": 82997.66902923584, "step_time_sec": 8.238616799004376, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9982, "loss": 3.8903911113739014, "lr": 3.455258200683167e-05, "elapsed_sec": 83005.91026687622, "step_time_sec": 8.241081448970363, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9983, "loss": 3.85433292388916, "lr": 3.44976708823014e-05, "elapsed_sec": 83014.14989829063, "step_time_sec": 8.239408659981564, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9984, "loss": 3.7396724224090576, "lr": 3.444280172600585e-05, "elapsed_sec": 83022.38885712624, "step_time_sec": 8.238794240052812, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9985, "loss": 3.9490582942962646, "lr": 3.438797457180807e-05, "elapsed_sec": 83030.62955594063, "step_time_sec": 8.240568764973432, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9986, "loss": 3.8823156356811523, "lr": 3.433318945354512e-05, "elapsed_sec": 83038.86893677711, "step_time_sec": 8.239164902013727, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9987, "loss": 3.758732318878174, "lr": 3.4278446405028034e-05, "elapsed_sec": 83047.1098792553, "step_time_sec": 8.240763367968611, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9988, "loss": 3.862408399581909, "lr": 3.4223745460042044e-05, "elapsed_sec": 83055.35023593903, "step_time_sec": 8.24023172201123, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9989, "loss": 3.810037612915039, "lr": 3.416908665234629e-05, "elapsed_sec": 83063.58905625343, "step_time_sec": 8.238587217987515, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9990, "loss": 3.8020031452178955, "lr": 3.4114470015673916e-05, "elapsed_sec": 83071.82945203781, "step_time_sec": 8.240241528954357, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9991, "loss": 3.9636175632476807, "lr": 3.40598955837321e-05, "elapsed_sec": 83080.06806564331, "step_time_sec": 8.238436155952513, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9992, "loss": 3.929258108139038, "lr": 3.4005363390201926e-05, "elapsed_sec": 83088.30724740028, "step_time_sec": 8.239024730050005, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9993, "loss": 3.795940399169922, "lr": 3.3950873468738404e-05, "elapsed_sec": 83096.5479888916, "step_time_sec": 8.240605397033505, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9994, "loss": 3.8148229122161865, "lr": 3.389642585297046e-05, "elapsed_sec": 83104.7863330841, "step_time_sec": 8.23815983789973, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9995, "loss": 3.8951680660247803, "lr": 3.384202057650093e-05, "elapsed_sec": 83113.02817630768, "step_time_sec": 8.241705785039812, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9996, "loss": 3.8979523181915283, "lr": 3.378765767290647e-05, "elapsed_sec": 83121.26914739609, "step_time_sec": 8.2407866519643, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9997, "loss": 3.9553921222686768, "lr": 3.3733337175737685e-05, "elapsed_sec": 83129.50833129883, "step_time_sec": 8.238991194986738, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9998, "loss": 3.9742982387542725, "lr": 3.367905911851893e-05, "elapsed_sec": 83137.74882650375, "step_time_sec": 8.24036956904456, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9999, "loss": 3.8286893367767334, "lr": 3.362482353474833e-05, "elapsed_sec": 83145.98755598068, "step_time_sec": 8.238551147980615, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10000, "loss": 3.985189914703369, "lr": 3.3570630457897936e-05, "elapsed_sec": 83154.2269961834, "step_time_sec": 56.53314789896831, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": true, "validation_ran": true, "probe_ran": true, "probe_elapsed_sec": 0.9772892689798027, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10001, "loss": 3.790560722351074, "lr": 3.3516479921413444e-05, "elapsed_sec": 83210.76587486267, "step_time_sec": 8.244858268997632, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10002, "loss": 3.940976858139038, "lr": 3.346237195871432e-05, "elapsed_sec": 83218.99915194511, "step_time_sec": 8.233132419991307, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10003, "loss": 3.824387550354004, "lr": 3.34083066031938e-05, "elapsed_sec": 83227.23205947876, "step_time_sec": 8.232793848961592, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10004, "loss": 3.845874786376953, "lr": 3.3354283888218725e-05, "elapsed_sec": 83235.46538949013, "step_time_sec": 8.233150470070541, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10005, "loss": 3.8974592685699463, "lr": 3.330030384712977e-05, "elapsed_sec": 83243.70294475555, "step_time_sec": 8.237376376986504, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10006, "loss": 3.865811347961426, "lr": 3.324636651324116e-05, "elapsed_sec": 83251.94280147552, "step_time_sec": 8.239635574980639, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10007, "loss": 3.7058656215667725, "lr": 3.319247191984076e-05, "elapsed_sec": 83260.18083810806, "step_time_sec": 8.237887957016937, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10008, "loss": 3.787797451019287, "lr": 3.3138620100190184e-05, "elapsed_sec": 83268.4211177826, "step_time_sec": 8.240120226982981, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10009, "loss": 3.896048069000244, "lr": 3.308481108752446e-05, "elapsed_sec": 83276.66033005714, "step_time_sec": 8.239070595940575, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10010, "loss": 3.8518996238708496, "lr": 3.30310449150523e-05, "elapsed_sec": 83284.89868307114, "step_time_sec": 8.238160568056628, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10011, "loss": 3.6520657539367676, "lr": 3.297732161595603e-05, "elapsed_sec": 83293.13899922371, "step_time_sec": 8.240144969080575, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10012, "loss": 4.2923784255981445, "lr": 3.2923641223391424e-05, "elapsed_sec": 83301.37874341011, "step_time_sec": 8.239616980077699, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10013, "loss": 3.8714845180511475, "lr": 3.287000377048777e-05, "elapsed_sec": 83309.61824893951, "step_time_sec": 8.239281444926746, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10014, "loss": 3.879317283630371, "lr": 3.281640929034797e-05, "elapsed_sec": 83317.85758757591, "step_time_sec": 8.239223439944908, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10015, "loss": 3.8505172729492188, "lr": 3.276285781604827e-05, "elapsed_sec": 83326.09573817253, "step_time_sec": 8.23796725505963, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10016, "loss": 3.8371994495391846, "lr": 3.270934938063847e-05, "elapsed_sec": 83334.33708286285, "step_time_sec": 8.241131178918295, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10017, "loss": 3.773134231567383, "lr": 3.2655884017141754e-05, "elapsed_sec": 83342.57555508614, "step_time_sec": 8.238295881077647, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10018, "loss": 3.776902914047241, "lr": 3.260246175855472e-05, "elapsed_sec": 83350.81508040428, "step_time_sec": 8.239363366970792, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10019, "loss": 3.859815835952759, "lr": 3.25490826378474e-05, "elapsed_sec": 83359.05564856529, "step_time_sec": 8.24044279602822, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10020, "loss": 3.8925576210021973, "lr": 3.2495746687963196e-05, "elapsed_sec": 83367.29400753975, "step_time_sec": 8.238192082033493, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10021, "loss": 3.930626153945923, "lr": 3.244245394181883e-05, "elapsed_sec": 83375.53487062454, "step_time_sec": 8.240632127039135, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10022, "loss": 3.8232250213623047, "lr": 3.238920443230444e-05, "elapsed_sec": 83383.77429485321, "step_time_sec": 8.239272337057628, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10023, "loss": 3.9464163780212402, "lr": 3.23359981922834e-05, "elapsed_sec": 83392.01268076897, "step_time_sec": 8.238224980072118, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10024, "loss": 3.83516001701355, "lr": 3.228283525459241e-05, "elapsed_sec": 83400.25292825699, "step_time_sec": 8.240065822959878, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10025, "loss": 3.534651279449463, "lr": 3.222971565204145e-05, "elapsed_sec": 83408.49213862419, "step_time_sec": 8.23906639998313, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10026, "loss": 3.7593371868133545, "lr": 3.217663941741372e-05, "elapsed_sec": 83416.73232102394, "step_time_sec": 8.240014289971441, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10027, "loss": 3.8918514251708984, "lr": 3.21236065834657e-05, "elapsed_sec": 83424.97196626663, "step_time_sec": 8.23945485195145, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10028, "loss": 3.852936267852783, "lr": 3.207061718292709e-05, "elapsed_sec": 83433.20830655098, "step_time_sec": 8.236180467996746, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10029, "loss": 3.711822509765625, "lr": 3.2017671248500755e-05, "elapsed_sec": 83441.44910597801, "step_time_sec": 8.240673437016085, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10030, "loss": 3.796755790710449, "lr": 3.1964768812862725e-05, "elapsed_sec": 83449.68871688843, "step_time_sec": 8.23941493500024, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10031, "loss": 3.8152410984039307, "lr": 3.1911909908662224e-05, "elapsed_sec": 83457.92824912071, "step_time_sec": 8.239407151006162, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10032, "loss": 3.7994532585144043, "lr": 3.185909456852158e-05, "elapsed_sec": 83466.16764831543, "step_time_sec": 8.239176547038369, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10033, "loss": 3.769474506378174, "lr": 3.180632282503621e-05, "elapsed_sec": 83474.40616583824, "step_time_sec": 8.238361106952652, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10034, "loss": 3.821367025375366, "lr": 3.175359471077473e-05, "elapsed_sec": 83482.64699316025, "step_time_sec": 8.2406686010072, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10035, "loss": 3.9703543186187744, "lr": 3.1700910258278724e-05, "elapsed_sec": 83490.88784694672, "step_time_sec": 8.240717503940687, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10036, "loss": 3.788328170776367, "lr": 3.164826950006283e-05, "elapsed_sec": 83499.12618160248, "step_time_sec": 8.238150759949349, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10037, "loss": 3.765454053878784, "lr": 3.159567246861481e-05, "elapsed_sec": 83507.36734199524, "step_time_sec": 8.24105461104773, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10038, "loss": 3.8556745052337646, "lr": 3.154311919639536e-05, "elapsed_sec": 83515.60749006271, "step_time_sec": 8.239909384050407, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10039, "loss": 3.780393362045288, "lr": 3.14906097158382e-05, "elapsed_sec": 83523.84592652321, "step_time_sec": 8.238222423009574, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10040, "loss": 3.968395471572876, "lr": 3.143814405935002e-05, "elapsed_sec": 83532.08480763435, "step_time_sec": 8.23878076102119, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10041, "loss": 3.9691500663757324, "lr": 3.1385722259310416e-05, "elapsed_sec": 83540.32302069664, "step_time_sec": 8.238032896071672, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10042, "loss": 3.8843085765838623, "lr": 3.133334434807203e-05, "elapsed_sec": 83548.5747077465, "step_time_sec": 8.251519475947134, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10043, "loss": 3.8146209716796875, "lr": 3.128101035796031e-05, "elapsed_sec": 83556.81497883797, "step_time_sec": 8.240102540003136, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10044, "loss": 3.7849013805389404, "lr": 3.122872032127363e-05, "elapsed_sec": 83565.05453705788, "step_time_sec": 8.239388876012526, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10045, "loss": 3.8928279876708984, "lr": 3.117647427028329e-05, "elapsed_sec": 83573.29510307312, "step_time_sec": 8.240390292019583, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10046, "loss": 3.840622901916504, "lr": 3.112427223723339e-05, "elapsed_sec": 83581.53504705429, "step_time_sec": 8.23978211497888, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10047, "loss": 3.8761796951293945, "lr": 3.1072114254340814e-05, "elapsed_sec": 83589.77355122566, "step_time_sec": 8.238285565981641, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10048, "loss": 3.731428861618042, "lr": 3.1020000353795373e-05, "elapsed_sec": 83598.0150051117, "step_time_sec": 8.241327118012123, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10049, "loss": 3.8245725631713867, "lr": 3.096793056775961e-05, "elapsed_sec": 83606.25490283966, "step_time_sec": 8.239692394970916, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10050, "loss": 4.065815448760986, "lr": 3.091590492836882e-05, "elapsed_sec": 83614.49407935143, "step_time_sec": 8.239016254083253, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10051, "loss": 3.911656379699707, "lr": 3.0863923467731144e-05, "elapsed_sec": 83622.73477888107, "step_time_sec": 8.24055245204363, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10052, "loss": 3.7931132316589355, "lr": 3.081198621792736e-05, "elapsed_sec": 83630.97306084633, "step_time_sec": 8.238069835002534, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10053, "loss": 3.756659984588623, "lr": 3.0760093211010996e-05, "elapsed_sec": 83639.2121989727, "step_time_sec": 8.238991670892574, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10054, "loss": 3.8052518367767334, "lr": 3.070824447900828e-05, "elapsed_sec": 83647.45066404343, "step_time_sec": 8.238358987029642, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10055, "loss": 3.8945531845092773, "lr": 3.0656440053918116e-05, "elapsed_sec": 83655.68949890137, "step_time_sec": 8.238641194999218, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10056, "loss": 3.816622734069824, "lr": 3.060467996771204e-05, "elapsed_sec": 83663.93056321144, "step_time_sec": 8.24088279006537, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10057, "loss": 3.875727891921997, "lr": 3.055296425233428e-05, "elapsed_sec": 83672.1708035469, "step_time_sec": 8.2400144940475, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10058, "loss": 3.88080096244812, "lr": 3.0501292939701594e-05, "elapsed_sec": 83680.41031241417, "step_time_sec": 8.23937334108632, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10059, "loss": 3.803995132446289, "lr": 3.0449666061703455e-05, "elapsed_sec": 83688.64951014519, "step_time_sec": 8.239061048952863, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10060, "loss": 3.9134507179260254, "lr": 3.0398083650201806e-05, "elapsed_sec": 83696.88923811913, "step_time_sec": 8.239514546934515, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10061, "loss": 3.715710401535034, "lr": 3.0346545737031192e-05, "elapsed_sec": 83705.12946510315, "step_time_sec": 8.240061630960554, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10062, "loss": 3.700343608856201, "lr": 3.0295052353998677e-05, "elapsed_sec": 83713.36800193787, "step_time_sec": 8.23838811297901, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10063, "loss": 3.797982692718506, "lr": 3.0243603532883868e-05, "elapsed_sec": 83721.60756444931, "step_time_sec": 8.239415945019573, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10064, "loss": 3.9433887004852295, "lr": 3.0192199305438816e-05, "elapsed_sec": 83729.84837532043, "step_time_sec": 8.24058770108968, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10065, "loss": 3.9458820819854736, "lr": 3.0140839703388158e-05, "elapsed_sec": 83738.08792495728, "step_time_sec": 8.239389332942665, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10066, "loss": 3.895613431930542, "lr": 3.008952475842889e-05, "elapsed_sec": 83746.32708072662, "step_time_sec": 8.239021959947422, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10067, "loss": 3.7742669582366943, "lr": 3.0038254502230457e-05, "elapsed_sec": 83754.56595635414, "step_time_sec": 8.238709781086072, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10068, "loss": 4.0132036209106445, "lr": 2.998702896643483e-05, "elapsed_sec": 83762.80450439453, "step_time_sec": 8.238437952008098, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10069, "loss": 3.7770113945007324, "lr": 2.9935848182656204e-05, "elapsed_sec": 83771.04518842697, "step_time_sec": 8.240464938106015, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10070, "loss": 3.923947811126709, "lr": 2.9884712182481266e-05, "elapsed_sec": 83779.28552031517, "step_time_sec": 8.240186457056552, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10071, "loss": 3.859200954437256, "lr": 2.9833620997469102e-05, "elapsed_sec": 83787.5231795311, "step_time_sec": 8.237546230084263, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10072, "loss": 3.892446994781494, "lr": 2.978257465915105e-05, "elapsed_sec": 83795.76277589798, "step_time_sec": 8.239374177064747, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10073, "loss": 4.083146572113037, "lr": 2.9731573199030782e-05, "elapsed_sec": 83804.00137090683, "step_time_sec": 8.23842963099014, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10074, "loss": 3.660261392593384, "lr": 2.9680616648584348e-05, "elapsed_sec": 83812.24088525772, "step_time_sec": 8.239320304011926, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10075, "loss": 3.8762760162353516, "lr": 2.962970503926001e-05, "elapsed_sec": 83820.48152112961, "step_time_sec": 8.240501841995865, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10076, "loss": 3.7943410873413086, "lr": 2.9578838402478313e-05, "elapsed_sec": 83828.71884441376, "step_time_sec": 8.237188774975948, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10077, "loss": 3.847306728363037, "lr": 2.952801676963205e-05, "elapsed_sec": 83836.95866179466, "step_time_sec": 8.239603873924352, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10078, "loss": 3.942606210708618, "lr": 2.9477240172086198e-05, "elapsed_sec": 83845.19944429398, "step_time_sec": 8.240616658004, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10079, "loss": 3.9041576385498047, "lr": 2.942650864117805e-05, "elapsed_sec": 83853.43881011009, "step_time_sec": 8.23922564892564, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10080, "loss": 3.8601269721984863, "lr": 2.9375822208216983e-05, "elapsed_sec": 83861.67832565308, "step_time_sec": 8.239349821931683, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10081, "loss": 3.8437294960021973, "lr": 2.9325180904484553e-05, "elapsed_sec": 83869.91610193253, "step_time_sec": 8.23756932001561, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10082, "loss": 3.7526161670684814, "lr": 2.9274584761234526e-05, "elapsed_sec": 83878.15673279762, "step_time_sec": 8.240444398950785, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10083, "loss": 3.7783854007720947, "lr": 2.9224033809692744e-05, "elapsed_sec": 83886.39705491066, "step_time_sec": 8.24016592593398, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10084, "loss": 3.712519884109497, "lr": 2.9173528081057154e-05, "elapsed_sec": 83894.63450622559, "step_time_sec": 8.237287405063398, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10085, "loss": 3.5936200618743896, "lr": 2.912306760649783e-05, "elapsed_sec": 83902.87478375435, "step_time_sec": 8.24013157794252, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10086, "loss": 4.000773906707764, "lr": 2.907265241715689e-05, "elapsed_sec": 83911.11295890808, "step_time_sec": 8.238046701997519, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10087, "loss": 3.756300449371338, "lr": 2.902228254414848e-05, "elapsed_sec": 83919.35213494301, "step_time_sec": 8.238973535015248, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10088, "loss": 3.835012197494507, "lr": 2.8971958018558866e-05, "elapsed_sec": 83927.59269189835, "step_time_sec": 8.240455375052989, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10089, "loss": 4.065596580505371, "lr": 2.8921678871446234e-05, "elapsed_sec": 83935.83144688606, "step_time_sec": 8.23853315494489, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10090, "loss": 3.8410067558288574, "lr": 2.8871445133840817e-05, "elapsed_sec": 83944.07252907753, "step_time_sec": 8.240902319084853, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10091, "loss": 3.917525291442871, "lr": 2.8821256836744805e-05, "elapsed_sec": 83952.31194019318, "step_time_sec": 8.239242480020039, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10092, "loss": 3.9290966987609863, "lr": 2.8771114011132332e-05, "elapsed_sec": 83960.55113172531, "step_time_sec": 8.23905995010864, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10093, "loss": 3.7948875427246094, "lr": 2.8721016687949466e-05, "elapsed_sec": 83968.7914443016, "step_time_sec": 8.240134453983046, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10094, "loss": 3.802220344543457, "lr": 2.867096489811428e-05, "elapsed_sec": 83977.02816963196, "step_time_sec": 8.236568749998696, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10095, "loss": 3.8384554386138916, "lr": 2.8620958672516622e-05, "elapsed_sec": 83985.26876997948, "step_time_sec": 8.24040290794801, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10096, "loss": 3.904468059539795, "lr": 2.8570998042018258e-05, "elapsed_sec": 83993.50801515579, "step_time_sec": 8.239086974062957, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10097, "loss": 3.7648026943206787, "lr": 2.852108303745289e-05, "elapsed_sec": 84001.74674654007, "step_time_sec": 8.238594982074574, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10098, "loss": 3.8665568828582764, "lr": 2.847121368962596e-05, "elapsed_sec": 84009.9855837822, "step_time_sec": 8.23863088095095, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10099, "loss": 3.894320249557495, "lr": 2.842139002931479e-05, "elapsed_sec": 84018.22696137428, "step_time_sec": 8.241249470040202, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10100, "loss": 3.7808279991149902, "lr": 2.8371612087268474e-05, "elapsed_sec": 84026.4676296711, "step_time_sec": 8.240460656001233, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10101, "loss": 3.944148063659668, "lr": 2.8321879894207888e-05, "elapsed_sec": 84034.70717573166, "step_time_sec": 8.23937321593985, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10102, "loss": 3.988328218460083, "lr": 2.8272193480825758e-05, "elapsed_sec": 84042.94575715065, "step_time_sec": 8.23847686802037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10103, "loss": 3.6925530433654785, "lr": 2.822255287778644e-05, "elapsed_sec": 84051.18664312363, "step_time_sec": 8.240658252965659, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10104, "loss": 3.844163656234741, "lr": 2.817295811572605e-05, "elapsed_sec": 84059.4256310463, "step_time_sec": 8.238881647004746, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10105, "loss": 3.90152907371521, "lr": 2.8123409225252485e-05, "elapsed_sec": 84067.66538596153, "step_time_sec": 8.239570003934205, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10106, "loss": 3.9086201190948486, "lr": 2.8073906236945292e-05, "elapsed_sec": 84075.90534067154, "step_time_sec": 8.239726527011953, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10107, "loss": 3.8434765338897705, "lr": 2.8024449181355574e-05, "elapsed_sec": 84084.14441895485, "step_time_sec": 8.238935500965454, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10108, "loss": 3.8801045417785645, "lr": 2.7975038089006273e-05, "elapsed_sec": 84092.3847014904, "step_time_sec": 8.240074458997697, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10109, "loss": 3.766641139984131, "lr": 2.792567299039186e-05, "elapsed_sec": 84100.62407970428, "step_time_sec": 8.239203677047044, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10110, "loss": 3.6938095092773438, "lr": 2.78763539159784e-05, "elapsed_sec": 84108.86237978935, "step_time_sec": 8.238182058907114, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10111, "loss": 3.7503767013549805, "lr": 2.782708089620366e-05, "elapsed_sec": 84117.10139465332, "step_time_sec": 8.238832887960598, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10112, "loss": 3.78633713722229, "lr": 2.777785396147689e-05, "elapsed_sec": 84125.34012556076, "step_time_sec": 8.238538040895946, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10113, "loss": 3.8195858001708984, "lr": 2.7728673142178906e-05, "elapsed_sec": 84133.57959961891, "step_time_sec": 8.239334076060914, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10114, "loss": 3.847663640975952, "lr": 2.76795384686621e-05, "elapsed_sec": 84141.82047367096, "step_time_sec": 8.240693727973849, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10115, "loss": 3.858447313308716, "lr": 2.763044997125034e-05, "elapsed_sec": 84150.05889678001, "step_time_sec": 8.238229493028484, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10116, "loss": 3.88873028755188, "lr": 2.7581407680239077e-05, "elapsed_sec": 84158.29946422577, "step_time_sec": 8.240441540954635, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10117, "loss": 3.7036259174346924, "lr": 2.753241162589515e-05, "elapsed_sec": 84166.53898906708, "step_time_sec": 8.239348871982656, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10118, "loss": 3.6501381397247314, "lr": 2.74834618384569e-05, "elapsed_sec": 84174.77773427963, "step_time_sec": 8.238547640969045, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10119, "loss": 3.8146181106567383, "lr": 2.743455834813415e-05, "elapsed_sec": 84183.01778268814, "step_time_sec": 8.239888910087757, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10120, "loss": 3.784392833709717, "lr": 2.738570118510811e-05, "elapsed_sec": 84191.25550103188, "step_time_sec": 8.23753152799327, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10121, "loss": 3.8213894367218018, "lr": 2.7336890379531403e-05, "elapsed_sec": 84199.49611473083, "step_time_sec": 8.240450268960558, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10122, "loss": 3.9512834548950195, "lr": 2.7288125961528036e-05, "elapsed_sec": 84207.73687171936, "step_time_sec": 8.2406495900359, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10123, "loss": 3.8373494148254395, "lr": 2.7239407961193406e-05, "elapsed_sec": 84215.97439956665, "step_time_sec": 8.237336170044728, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10124, "loss": 3.8427231311798096, "lr": 2.7190736408594225e-05, "elapsed_sec": 84224.21364688873, "step_time_sec": 8.239099988015369, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10125, "loss": 3.7712881565093994, "lr": 2.7142111333768632e-05, "elapsed_sec": 84232.45430850983, "step_time_sec": 8.240452714962885, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10126, "loss": 3.7464511394500732, "lr": 2.7093532766725982e-05, "elapsed_sec": 84240.69358682632, "step_time_sec": 8.239111612085253, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10127, "loss": 3.8179116249084473, "lr": 2.704500073744694e-05, "elapsed_sec": 84248.93314409256, "step_time_sec": 8.239412720082328, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10128, "loss": 3.930434226989746, "lr": 2.6996515275883563e-05, "elapsed_sec": 84257.17005372047, "step_time_sec": 8.236743993940763, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10129, "loss": 3.9839093685150146, "lr": 2.6948076411958978e-05, "elapsed_sec": 84265.4098072052, "step_time_sec": 8.239567459910177, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10130, "loss": 3.8433127403259277, "lr": 2.6899684175567676e-05, "elapsed_sec": 84273.64751887321, "step_time_sec": 8.237545454059727, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10131, "loss": 3.7309648990631104, "lr": 2.68513385965754e-05, "elapsed_sec": 84281.8862016201, "step_time_sec": 8.238552435068414, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10132, "loss": 3.807722806930542, "lr": 2.6803039704819013e-05, "elapsed_sec": 84290.12706446648, "step_time_sec": 8.240655383095145, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10133, "loss": 3.954495906829834, "lr": 2.6754787530106596e-05, "elapsed_sec": 84298.36479091644, "step_time_sec": 8.237551263999194, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10134, "loss": 3.868319272994995, "lr": 2.6706582102217415e-05, "elapsed_sec": 84306.6054110527, "step_time_sec": 8.240438850945793, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10135, "loss": 3.7855334281921387, "lr": 2.6658423450901887e-05, "elapsed_sec": 84314.8456170559, "step_time_sec": 8.240045016049407, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10136, "loss": 3.944913864135742, "lr": 2.661031160588153e-05, "elapsed_sec": 84323.08387613297, "step_time_sec": 8.238086738972925, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10137, "loss": 3.702956199645996, "lr": 2.6562246596848993e-05, "elapsed_sec": 84331.32261371613, "step_time_sec": 8.238592238980345, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10138, "loss": 3.6876790523529053, "lr": 2.6514228453467993e-05, "elapsed_sec": 84339.56109428406, "step_time_sec": 8.238327175029553, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10139, "loss": 3.717111349105835, "lr": 2.6466257205373397e-05, "elapsed_sec": 84347.80157923698, "step_time_sec": 8.240355447982438, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10140, "loss": 3.7083740234375, "lr": 2.6418332882171057e-05, "elapsed_sec": 84356.04181265831, "step_time_sec": 8.240013743983582, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10141, "loss": 3.701772689819336, "lr": 2.6370455513437865e-05, "elapsed_sec": 84364.28054237366, "step_time_sec": 8.238598771975376, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10142, "loss": 3.931013584136963, "lr": 2.6322625128721824e-05, "elapsed_sec": 84372.51863741875, "step_time_sec": 8.237886762944981, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10143, "loss": 3.9745726585388184, "lr": 2.6274841757541836e-05, "elapsed_sec": 84380.75833845139, "step_time_sec": 8.239559444016777, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10144, "loss": 3.8184237480163574, "lr": 2.6227105429387827e-05, "elapsed_sec": 84388.99784564972, "step_time_sec": 8.239339085994288, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10145, "loss": 3.770230531692505, "lr": 2.6179416173720693e-05, "elapsed_sec": 84397.23765277863, "step_time_sec": 8.239690201007761, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10146, "loss": 3.8196399211883545, "lr": 2.613177401997229e-05, "elapsed_sec": 84405.47592902184, "step_time_sec": 8.23804604599718, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10147, "loss": 3.838482618331909, "lr": 2.6084178997545352e-05, "elapsed_sec": 84413.71706867218, "step_time_sec": 8.240989962010644, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10148, "loss": 3.9114091396331787, "lr": 2.603663113581361e-05, "elapsed_sec": 84421.95658349991, "step_time_sec": 8.239378898986615, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10149, "loss": 3.9019815921783447, "lr": 2.5989130464121644e-05, "elapsed_sec": 84430.1952240467, "step_time_sec": 8.238456270075403, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10150, "loss": 3.7796640396118164, "lr": 2.5941677011784888e-05, "elapsed_sec": 84438.43365240097, "step_time_sec": 8.238260483019985, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10151, "loss": 3.8392271995544434, "lr": 2.589427080808966e-05, "elapsed_sec": 84446.67167949677, "step_time_sec": 8.237844657967798, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10152, "loss": 3.5088181495666504, "lr": 2.5846911882293128e-05, "elapsed_sec": 84454.91169834137, "step_time_sec": 8.239885532064363, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10153, "loss": 3.81303071975708, "lr": 2.579960026362324e-05, "elapsed_sec": 84463.15141630173, "step_time_sec": 8.239528049947694, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10154, "loss": 3.8780579566955566, "lr": 2.575233598127882e-05, "elapsed_sec": 84471.38891220093, "step_time_sec": 8.237333443015814, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10155, "loss": 3.932856559753418, "lr": 2.5705119064429386e-05, "elapsed_sec": 84479.62982225418, "step_time_sec": 8.240727205993608, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10156, "loss": 3.9822821617126465, "lr": 2.5657949542215325e-05, "elapsed_sec": 84487.86835026741, "step_time_sec": 8.238414880004711, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10157, "loss": 3.876542806625366, "lr": 2.56108274437477e-05, "elapsed_sec": 84496.10843801498, "step_time_sec": 8.239869761979207, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10158, "loss": 3.8455450534820557, "lr": 2.5563752798108317e-05, "elapsed_sec": 84504.34924960136, "step_time_sec": 8.24066104798112, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10159, "loss": 3.9303739070892334, "lr": 2.5516725634349724e-05, "elapsed_sec": 84512.58699941635, "step_time_sec": 8.23754670098424, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10160, "loss": 3.7972607612609863, "lr": 2.546974598149514e-05, "elapsed_sec": 84520.8267967701, "step_time_sec": 8.239627770031802, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10161, "loss": 3.790585517883301, "lr": 2.5422813868538435e-05, "elapsed_sec": 84529.06481361389, "step_time_sec": 8.237897632992826, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10162, "loss": 3.9490015506744385, "lr": 2.537592932444424e-05, "elapsed_sec": 84537.30305027962, "step_time_sec": 8.238097353023477, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10163, "loss": 3.8282999992370605, "lr": 2.532909237814773e-05, "elapsed_sec": 84545.54202389717, "step_time_sec": 8.238740148954093, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10164, "loss": 3.8532423973083496, "lr": 2.5282303058554725e-05, "elapsed_sec": 84553.78081560135, "step_time_sec": 8.238665413926356, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10165, "loss": 3.8653018474578857, "lr": 2.5235561394541702e-05, "elapsed_sec": 84562.0213675499, "step_time_sec": 8.240332536981441, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10166, "loss": 3.9569926261901855, "lr": 2.5188867414955683e-05, "elapsed_sec": 84570.2609539032, "step_time_sec": 8.239424192928709, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10167, "loss": 3.91062068939209, "lr": 2.5142221148614206e-05, "elapsed_sec": 84578.49886727333, "step_time_sec": 8.237772744963877, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10168, "loss": 3.747807741165161, "lr": 2.5095622624305506e-05, "elapsed_sec": 84586.73692202568, "step_time_sec": 8.237835129024461, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10169, "loss": 3.8768184185028076, "lr": 2.5049071870788225e-05, "elapsed_sec": 84594.97518086433, "step_time_sec": 8.238106006057933, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10170, "loss": 3.92946720123291, "lr": 2.5002568916791565e-05, "elapsed_sec": 84603.21479582787, "step_time_sec": 8.239466328988783, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10171, "loss": 4.0107741355896, "lr": 2.495611379101526e-05, "elapsed_sec": 84611.45433592796, "step_time_sec": 8.239397070952691, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10172, "loss": 3.8542187213897705, "lr": 2.4909706522129506e-05, "elapsed_sec": 84619.69401311874, "step_time_sec": 8.2395350409206, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10173, "loss": 4.065855979919434, "lr": 2.4863347138774925e-05, "elapsed_sec": 84627.9332678318, "step_time_sec": 8.239106357912533, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10174, "loss": 3.91021990776062, "lr": 2.4817035669562627e-05, "elapsed_sec": 84636.17051529884, "step_time_sec": 8.237043299013749, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10175, "loss": 3.8534810543060303, "lr": 2.4770772143074128e-05, "elapsed_sec": 84644.41058731079, "step_time_sec": 8.239933552918956, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10176, "loss": 3.875983238220215, "lr": 2.472455658786141e-05, "elapsed_sec": 84652.65068650246, "step_time_sec": 8.239960791077465, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10177, "loss": 3.735781192779541, "lr": 2.4678389032446777e-05, "elapsed_sec": 84660.88952445984, "step_time_sec": 8.23866472102236, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10178, "loss": 3.795685052871704, "lr": 2.4632269505322936e-05, "elapsed_sec": 84669.12997555733, "step_time_sec": 8.24024182697758, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10179, "loss": 3.8866004943847656, "lr": 2.4586198034952994e-05, "elapsed_sec": 84677.36909627914, "step_time_sec": 8.239020087989047, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10180, "loss": 3.7640233039855957, "lr": 2.4540174649770343e-05, "elapsed_sec": 84685.60760784149, "step_time_sec": 8.238290683017112, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10181, "loss": 3.680305004119873, "lr": 2.4494199378178717e-05, "elapsed_sec": 84693.84752058983, "step_time_sec": 8.23972540802788, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10182, "loss": 3.958233118057251, "lr": 2.4448272248552167e-05, "elapsed_sec": 84702.08594179153, "step_time_sec": 8.238257899996825, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10183, "loss": 3.8498342037200928, "lr": 2.4402393289235022e-05, "elapsed_sec": 84710.32627153397, "step_time_sec": 8.240178200998344, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10184, "loss": 3.828172206878662, "lr": 2.435656252854184e-05, "elapsed_sec": 84718.56520152092, "step_time_sec": 8.238771628937684, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10185, "loss": 3.902045249938965, "lr": 2.431077999475755e-05, "elapsed_sec": 84726.80391907692, "step_time_sec": 8.238543339073658, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10186, "loss": 3.8157246112823486, "lr": 2.426504571613721e-05, "elapsed_sec": 84735.04672431946, "step_time_sec": 8.242688393918797, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10187, "loss": 3.7782862186431885, "lr": 2.4219359720906114e-05, "elapsed_sec": 84743.2854707241, "step_time_sec": 8.23862092604395, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10188, "loss": 3.872401475906372, "lr": 2.417372203725985e-05, "elapsed_sec": 84751.5449924469, "step_time_sec": 8.259281344944611, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10189, "loss": 3.7461860179901123, "lr": 2.4128132693364038e-05, "elapsed_sec": 84759.78572797775, "step_time_sec": 8.240514423931018, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10190, "loss": 3.861825466156006, "lr": 2.4082591717354525e-05, "elapsed_sec": 84768.02631402016, "step_time_sec": 8.240469336975366, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10191, "loss": 3.9789984226226807, "lr": 2.403709913733741e-05, "elapsed_sec": 84776.2643020153, "step_time_sec": 8.237823648029007, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10192, "loss": 3.802935838699341, "lr": 2.3991654981388755e-05, "elapsed_sec": 84784.5040667057, "step_time_sec": 8.239578748936765, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10193, "loss": 3.9337053298950195, "lr": 2.3946259277554885e-05, "elapsed_sec": 84792.74505519867, "step_time_sec": 8.24082438298501, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10194, "loss": 3.8781960010528564, "lr": 2.3900912053852132e-05, "elapsed_sec": 84800.98357915878, "step_time_sec": 8.238389350939542, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10195, "loss": 3.8535258769989014, "lr": 2.3855613338266925e-05, "elapsed_sec": 84809.22418308258, "step_time_sec": 8.240373783977702, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10196, "loss": 3.851069927215576, "lr": 2.3810363158755784e-05, "elapsed_sec": 84817.46244812012, "step_time_sec": 8.23815932997968, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10197, "loss": 3.7168641090393066, "lr": 2.3765161543245232e-05, "elapsed_sec": 84825.70158147812, "step_time_sec": 8.23891208798159, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10198, "loss": 3.85669207572937, "lr": 2.372000851963184e-05, "elapsed_sec": 84833.94222068787, "step_time_sec": 8.24047573597636, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10199, "loss": 3.893256187438965, "lr": 2.3674904115782228e-05, "elapsed_sec": 84842.17970848083, "step_time_sec": 8.237326888018288, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10200, "loss": 3.862039089202881, "lr": 2.3629848359532946e-05, "elapsed_sec": 84850.41945290565, "step_time_sec": 8.239583264105022, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10201, "loss": 3.8885791301727295, "lr": 2.3584841278690533e-05, "elapsed_sec": 84858.6587998867, "step_time_sec": 8.239200611948036, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10202, "loss": 3.7685365676879883, "lr": 2.3539882901031564e-05, "elapsed_sec": 84866.89788269997, "step_time_sec": 8.238855089992285, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10203, "loss": 3.8523457050323486, "lr": 2.3494973254302446e-05, "elapsed_sec": 84875.13659095764, "step_time_sec": 8.238562608952634, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10204, "loss": 3.8360321521759033, "lr": 2.3450112366219588e-05, "elapsed_sec": 84883.37490200996, "step_time_sec": 8.238142192014493, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10205, "loss": 3.8363401889801025, "lr": 2.3405300264469268e-05, "elapsed_sec": 84891.61463093758, "step_time_sec": 8.23956481704954, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10206, "loss": 3.6707370281219482, "lr": 2.3360536976707667e-05, "elapsed_sec": 84899.85328245163, "step_time_sec": 8.23848412989173, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10207, "loss": 3.90610408782959, "lr": 2.3315822530560823e-05, "elapsed_sec": 84908.09168052673, "step_time_sec": 8.238292666967027, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10208, "loss": 3.817610740661621, "lr": 2.3271156953624696e-05, "elapsed_sec": 84916.33185482025, "step_time_sec": 8.240001378930174, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10209, "loss": 3.845673084259033, "lr": 2.3226540273465015e-05, "elapsed_sec": 84924.56948447227, "step_time_sec": 8.23745166696608, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10210, "loss": 3.7670671939849854, "lr": 2.318197251761736e-05, "elapsed_sec": 84932.81015276909, "step_time_sec": 8.240511595038697, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10211, "loss": 3.8419606685638428, "lr": 2.3137453713587096e-05, "elapsed_sec": 84941.05072999, "step_time_sec": 8.240357349975966, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10212, "loss": 3.8822851181030273, "lr": 2.309298388884939e-05, "elapsed_sec": 84949.2889752388, "step_time_sec": 8.238128764904104, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10213, "loss": 3.8445706367492676, "lr": 2.304856307084922e-05, "elapsed_sec": 84957.52873325348, "step_time_sec": 8.239614231977612, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10214, "loss": 3.840804100036621, "lr": 2.3004191287001266e-05, "elapsed_sec": 84965.76636648178, "step_time_sec": 8.237417672062293, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10215, "loss": 4.068812370300293, "lr": 2.2959868564689943e-05, "elapsed_sec": 84974.0082783699, "step_time_sec": 8.241739097982645, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10216, "loss": 3.7089762687683105, "lr": 2.2915594931269457e-05, "elapsed_sec": 84982.24797558784, "step_time_sec": 8.23949060193263, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10217, "loss": 3.8438878059387207, "lr": 2.287137041406364e-05, "elapsed_sec": 84990.4866540432, "step_time_sec": 8.23855485394597, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10218, "loss": 4.019757270812988, "lr": 2.282719504036603e-05, "elapsed_sec": 84998.72736501694, "step_time_sec": 8.240571714937687, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10219, "loss": 3.7793519496917725, "lr": 2.278306883743987e-05, "elapsed_sec": 85006.96693515778, "step_time_sec": 8.23940997896716, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10220, "loss": 3.8670687675476074, "lr": 2.2738991832518027e-05, "elapsed_sec": 85015.20597219467, "step_time_sec": 8.238822872983292, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10221, "loss": 3.7309720516204834, "lr": 2.2694964052802974e-05, "elapsed_sec": 85023.446144104, "step_time_sec": 8.240011977031827, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10222, "loss": 3.76859450340271, "lr": 2.2650985525466917e-05, "elapsed_sec": 85031.68438005447, "step_time_sec": 8.238102501025423, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10223, "loss": 3.9662985801696777, "lr": 2.2607056277651534e-05, "elapsed_sec": 85039.92463040352, "step_time_sec": 8.240108479978517, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10224, "loss": 3.7632052898406982, "lr": 2.2563176336468143e-05, "elapsed_sec": 85048.16433358192, "step_time_sec": 8.239520155009814, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10225, "loss": 3.8725619316101074, "lr": 2.2519345728997682e-05, "elapsed_sec": 85056.40088510513, "step_time_sec": 8.236380483023822, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10226, "loss": 3.801544427871704, "lr": 2.2475564482290582e-05, "elapsed_sec": 85064.64031887054, "step_time_sec": 8.239283322007395, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10227, "loss": 3.855290651321411, "lr": 2.2431832623366764e-05, "elapsed_sec": 85072.87758231163, "step_time_sec": 8.237118044984527, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10228, "loss": 3.828580141067505, "lr": 2.2388150179215804e-05, "elapsed_sec": 85081.1174235344, "step_time_sec": 8.239628189010546, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10229, "loss": 3.7452023029327393, "lr": 2.2344517176796646e-05, "elapsed_sec": 85089.357213974, "step_time_sec": 8.239634118042886, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10230, "loss": 3.9481422901153564, "lr": 2.2300933643037844e-05, "elapsed_sec": 85097.59561419487, "step_time_sec": 8.238287634099834, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10231, "loss": 3.888554811477661, "lr": 2.2257399604837315e-05, "elapsed_sec": 85105.83714485168, "step_time_sec": 8.241295404033735, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10232, "loss": 3.7825229167938232, "lr": 2.2213915089062494e-05, "elapsed_sec": 85114.07643938065, "step_time_sec": 8.239143733982928, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10233, "loss": 3.848196506500244, "lr": 2.2170480122550213e-05, "elapsed_sec": 85122.31559824944, "step_time_sec": 8.238979253917933, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10234, "loss": 3.787646770477295, "lr": 2.2127094732106765e-05, "elapsed_sec": 85130.55627608299, "step_time_sec": 8.240544295986183, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10235, "loss": 3.8040215969085693, "lr": 2.2083758944507776e-05, "elapsed_sec": 85138.79481649399, "step_time_sec": 8.238389992970042, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10236, "loss": 4.030694484710693, "lr": 2.2040472786498366e-05, "elapsed_sec": 85147.03609395027, "step_time_sec": 8.241105186985806, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10237, "loss": 3.855565071105957, "lr": 2.1997236284792953e-05, "elapsed_sec": 85155.27519798279, "step_time_sec": 8.23888999701012, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10238, "loss": 3.7460830211639404, "lr": 2.195404946607529e-05, "elapsed_sec": 85163.51430892944, "step_time_sec": 8.2389552120585, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10239, "loss": 3.7004265785217285, "lr": 2.191091235699855e-05, "elapsed_sec": 85171.75252652168, "step_time_sec": 8.238063963013701, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10240, "loss": 3.926008701324463, "lr": 2.1867824984185156e-05, "elapsed_sec": 85179.99051761627, "step_time_sec": 8.237837946973741, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10241, "loss": 3.9463560581207275, "lr": 2.1824787374226863e-05, "elapsed_sec": 85188.23020482063, "step_time_sec": 8.23950688005425, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10242, "loss": 3.8467376232147217, "lr": 2.17817995536847e-05, "elapsed_sec": 85196.4696753025, "step_time_sec": 8.23934214306064, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10243, "loss": 3.860395908355713, "lr": 2.1738861549088992e-05, "elapsed_sec": 85204.70847153664, "step_time_sec": 8.238635667948984, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10244, "loss": 4.006292819976807, "lr": 2.1695973386939277e-05, "elapsed_sec": 85212.94613909721, "step_time_sec": 8.237508354941383, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10245, "loss": 3.7714555263519287, "lr": 2.1653135093704402e-05, "elapsed_sec": 85221.18528556824, "step_time_sec": 8.238912341068499, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10246, "loss": 4.039280414581299, "lr": 2.1610346695822365e-05, "elapsed_sec": 85229.42588067055, "step_time_sec": 8.240455708932132, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10247, "loss": 3.765078544616699, "lr": 2.1567608219700396e-05, "elapsed_sec": 85237.66550064087, "step_time_sec": 8.239493957953528, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10248, "loss": 3.8076021671295166, "lr": 2.1524919691714985e-05, "elapsed_sec": 85245.90363574028, "step_time_sec": 8.2379199169809, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10249, "loss": 3.8016843795776367, "lr": 2.1482281138211644e-05, "elapsed_sec": 85254.1449136734, "step_time_sec": 8.241111095063388, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10250, "loss": 3.8574776649475098, "lr": 2.143969258550519e-05, "elapsed_sec": 85262.38352274895, "step_time_sec": 8.238391894032247, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10251, "loss": 3.969027280807495, "lr": 2.139715405987951e-05, "elapsed_sec": 85270.62203550339, "step_time_sec": 8.238336602924392, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10252, "loss": 3.821978807449341, "lr": 2.1354665587587587e-05, "elapsed_sec": 85278.86250400543, "step_time_sec": 8.240354101057164, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10253, "loss": 3.9711458683013916, "lr": 2.1312227194851618e-05, "elapsed_sec": 85287.09472370148, "step_time_sec": 8.232031102990732, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10254, "loss": 3.8324012756347656, "lr": 2.1269838907862806e-05, "elapsed_sec": 85295.32897400856, "step_time_sec": 8.23404415906407, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10255, "loss": 3.6729464530944824, "lr": 2.1227500752781446e-05, "elapsed_sec": 85303.56177568436, "step_time_sec": 8.23264956602361, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10256, "loss": 3.9200568199157715, "lr": 2.1185212755736912e-05, "elapsed_sec": 85311.80205154419, "step_time_sec": 8.240109512931667, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10257, "loss": 3.9948129653930664, "lr": 2.1142974942827606e-05, "elapsed_sec": 85320.04065775871, "step_time_sec": 8.238451418001205, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10258, "loss": 3.807896852493286, "lr": 2.1100787340120944e-05, "elapsed_sec": 85328.28019189835, "step_time_sec": 8.23933768097777, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10259, "loss": 3.7940213680267334, "lr": 2.105864997365343e-05, "elapsed_sec": 85336.52111196518, "step_time_sec": 8.240770102944225, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10260, "loss": 3.817685127258301, "lr": 2.1016562869430464e-05, "elapsed_sec": 85344.7588365078, "step_time_sec": 8.237553560989909, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10261, "loss": 3.768345832824707, "lr": 2.0974526053426476e-05, "elapsed_sec": 85352.99744319916, "step_time_sec": 8.238455720013008, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10262, "loss": 3.8591787815093994, "lr": 2.0932539551584884e-05, "elapsed_sec": 85361.23667621613, "step_time_sec": 8.239025280927308, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10263, "loss": 3.8337645530700684, "lr": 2.089060338981801e-05, "elapsed_sec": 85369.47670936584, "step_time_sec": 8.23991991695948, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10264, "loss": 3.9499740600585938, "lr": 2.0848717594007125e-05, "elapsed_sec": 85377.71593785286, "step_time_sec": 8.239096916047856, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10265, "loss": 3.796722412109375, "lr": 2.080688219000242e-05, "elapsed_sec": 85385.95488333702, "step_time_sec": 8.23872378806118, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10266, "loss": 3.9786269664764404, "lr": 2.076509720362297e-05, "elapsed_sec": 85394.19563484192, "step_time_sec": 8.240609974018298, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10267, "loss": 3.841057300567627, "lr": 2.072336266065674e-05, "elapsed_sec": 85402.43538284302, "step_time_sec": 8.239587999065407, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10268, "loss": 3.753117322921753, "lr": 2.0681678586860596e-05, "elapsed_sec": 85410.67450523376, "step_time_sec": 8.238930135965347, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10269, "loss": 3.820413827896118, "lr": 2.0640045007960197e-05, "elapsed_sec": 85418.91357827187, "step_time_sec": 8.238925441983156, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10270, "loss": 3.957650661468506, "lr": 2.059846194965014e-05, "elapsed_sec": 85427.15261340141, "step_time_sec": 8.238869015011005, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10271, "loss": 3.759493112564087, "lr": 2.0556929437593703e-05, "elapsed_sec": 85435.39286756516, "step_time_sec": 8.24008747690823, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10272, "loss": 3.898723602294922, "lr": 2.0515447497423045e-05, "elapsed_sec": 85443.63251185417, "step_time_sec": 8.239470312022604, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10273, "loss": 3.8284080028533936, "lr": 2.0474016154739145e-05, "elapsed_sec": 85451.87123918533, "step_time_sec": 8.238580378005281, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10274, "loss": 3.665809392929077, "lr": 2.0432635435111708e-05, "elapsed_sec": 85460.11320066452, "step_time_sec": 8.241814858978614, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10275, "loss": 3.562548875808716, "lr": 2.039130536407919e-05, "elapsed_sec": 85468.35275864601, "step_time_sec": 8.239374445052817, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10276, "loss": 3.8606512546539307, "lr": 2.0350025967148845e-05, "elapsed_sec": 85476.59142160416, "step_time_sec": 8.23852705792524, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10277, "loss": 3.7438056468963623, "lr": 2.0308797269796585e-05, "elapsed_sec": 85484.83043169975, "step_time_sec": 8.238795476965606, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10278, "loss": 3.869239330291748, "lr": 2.026761929746707e-05, "elapsed_sec": 85493.06904935837, "step_time_sec": 8.23850637592841, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10279, "loss": 3.7687172889709473, "lr": 2.0226492075573657e-05, "elapsed_sec": 85501.30962657928, "step_time_sec": 8.240368961938657, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10280, "loss": 3.902927875518799, "lr": 2.0185415629498362e-05, "elapsed_sec": 85509.5490603447, "step_time_sec": 8.23923949105665, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10281, "loss": 3.780923843383789, "lr": 2.0144389984591843e-05, "elapsed_sec": 85517.78737974167, "step_time_sec": 8.238212599069811, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10282, "loss": 3.680387258529663, "lr": 2.01034151661735e-05, "elapsed_sec": 85526.02523970604, "step_time_sec": 8.237628755974583, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10283, "loss": 3.7855312824249268, "lr": 2.0062491199531274e-05, "elapsed_sec": 85534.26243710518, "step_time_sec": 8.237053486984223, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10284, "loss": 3.7195262908935547, "lr": 2.002161810992173e-05, "elapsed_sec": 85542.50185251236, "step_time_sec": 8.239219336071983, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10285, "loss": 3.7306406497955322, "lr": 1.998079592257009e-05, "elapsed_sec": 85550.7419834137, "step_time_sec": 8.239965804037638, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10286, "loss": 3.837447166442871, "lr": 1.9940024662670105e-05, "elapsed_sec": 85558.98112463951, "step_time_sec": 8.238979835994542, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10287, "loss": 3.7961230278015137, "lr": 1.989930435538412e-05, "elapsed_sec": 85567.21467852592, "step_time_sec": 8.233447089907713, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10288, "loss": 3.790689706802368, "lr": 1.985863502584304e-05, "elapsed_sec": 85575.44871592522, "step_time_sec": 8.233841849956661, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10289, "loss": 3.670336961746216, "lr": 1.9818016699146263e-05, "elapsed_sec": 85583.68294453621, "step_time_sec": 8.234058171045035, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10290, "loss": 3.81502628326416, "lr": 1.9777449400361787e-05, "elapsed_sec": 85591.92070508003, "step_time_sec": 8.237548275967129, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10291, "loss": 3.740809679031372, "lr": 1.973693315452606e-05, "elapsed_sec": 85600.16072320938, "step_time_sec": 8.239890455035493, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10292, "loss": 3.727315902709961, "lr": 1.969646798664406e-05, "elapsed_sec": 85608.39887809753, "step_time_sec": 8.237988941022195, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10293, "loss": 3.898550033569336, "lr": 1.965605392168918e-05, "elapsed_sec": 85616.6379814148, "step_time_sec": 8.23895296594128, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10294, "loss": 3.688170909881592, "lr": 1.9615690984603348e-05, "elapsed_sec": 85624.87733721733, "step_time_sec": 8.239168061991222, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10295, "loss": 3.6937999725341797, "lr": 1.9575379200296866e-05, "elapsed_sec": 85633.11555123329, "step_time_sec": 8.238083282019943, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10296, "loss": 3.9719767570495605, "lr": 1.953511859364856e-05, "elapsed_sec": 85641.35654449463, "step_time_sec": 8.240799977094866, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10297, "loss": 3.7261011600494385, "lr": 1.9494909189505594e-05, "elapsed_sec": 85649.59629535675, "step_time_sec": 8.239574939012527, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10298, "loss": 3.8334031105041504, "lr": 1.9454751012683524e-05, "elapsed_sec": 85657.83592605591, "step_time_sec": 8.239522554911673, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10299, "loss": 3.8541860580444336, "lr": 1.9414644087966383e-05, "elapsed_sec": 85666.0756342411, "step_time_sec": 8.239506741985679, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10300, "loss": 3.8515279293060303, "lr": 1.9374588440106467e-05, "elapsed_sec": 85674.31436634064, "step_time_sec": 8.238570044981316, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10301, "loss": 3.753615617752075, "lr": 1.9334584093824487e-05, "elapsed_sec": 85682.55492115021, "step_time_sec": 8.240403414005414, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10302, "loss": 3.737344980239868, "lr": 1.9294631073809466e-05, "elapsed_sec": 85690.79367017746, "step_time_sec": 8.238551749964245, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10303, "loss": 3.777941942214966, "lr": 1.9254729404718773e-05, "elapsed_sec": 85699.0324857235, "step_time_sec": 8.238688852987252, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10304, "loss": 3.8431711196899414, "lr": 1.921487911117805e-05, "elapsed_sec": 85707.2728126049, "step_time_sec": 8.240171230980195, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10305, "loss": 3.9673807621002197, "lr": 1.9175080217781292e-05, "elapsed_sec": 85715.5111489296, "step_time_sec": 8.238119735033251, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10306, "loss": 3.876387357711792, "lr": 1.9135332749090687e-05, "elapsed_sec": 85723.7494149208, "step_time_sec": 8.238104338990524, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10307, "loss": 3.753009080886841, "lr": 1.9095636729636792e-05, "elapsed_sec": 85731.98911166191, "step_time_sec": 8.239485569065437, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10308, "loss": 3.8771612644195557, "lr": 1.9055992183918354e-05, "elapsed_sec": 85740.2254626751, "step_time_sec": 8.236222416046076, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10309, "loss": 3.8660130500793457, "lr": 1.901639913640228e-05, "elapsed_sec": 85748.46618461609, "step_time_sec": 8.240536142024212, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10310, "loss": 3.9455647468566895, "lr": 1.8976857611523836e-05, "elapsed_sec": 85756.70607733727, "step_time_sec": 8.239694808027707, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10311, "loss": 3.7177786827087402, "lr": 1.8937367633686394e-05, "elapsed_sec": 85764.94632291794, "step_time_sec": 8.24007406597957, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10312, "loss": 3.8140251636505127, "lr": 1.8897929227261518e-05, "elapsed_sec": 85773.18674874306, "step_time_sec": 8.240273723960854, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10313, "loss": 3.804971933364868, "lr": 1.8858542416589022e-05, "elapsed_sec": 85781.42441558838, "step_time_sec": 8.237478779978119, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10314, "loss": 3.7592899799346924, "lr": 1.8819207225976793e-05, "elapsed_sec": 85789.66480183601, "step_time_sec": 8.240236650104634, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10315, "loss": 4.059289455413818, "lr": 1.877992367970089e-05, "elapsed_sec": 85797.90555596352, "step_time_sec": 8.240608262014575, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10316, "loss": 3.8902299404144287, "lr": 1.8740691802005503e-05, "elapsed_sec": 85806.14436841011, "step_time_sec": 8.23857678892091, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10317, "loss": 3.8912885189056396, "lr": 1.870151161710292e-05, "elapsed_sec": 85814.3847129345, "step_time_sec": 8.240204112953506, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10318, "loss": 3.9045767784118652, "lr": 1.8662383149173513e-05, "elapsed_sec": 85822.62318181992, "step_time_sec": 8.238295461982489, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10319, "loss": 3.8952724933624268, "lr": 1.8623306422365818e-05, "elapsed_sec": 85830.86343073845, "step_time_sec": 8.240060452953912, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10320, "loss": 3.8179473876953125, "lr": 1.8584281460796358e-05, "elapsed_sec": 85839.10323238373, "step_time_sec": 8.239634172990918, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10321, "loss": 3.825491428375244, "lr": 1.85453082885497e-05, "elapsed_sec": 85847.34090995789, "step_time_sec": 8.237569093005732, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10322, "loss": 3.818047285079956, "lr": 1.850638692967852e-05, "elapsed_sec": 85855.58075404167, "step_time_sec": 8.239614790072665, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10323, "loss": 3.7869417667388916, "lr": 1.8467517408203465e-05, "elapsed_sec": 85863.81948828697, "step_time_sec": 8.238548650988378, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10324, "loss": 3.917088270187378, "lr": 1.842869974811319e-05, "elapsed_sec": 85872.05780792236, "step_time_sec": 8.238140919944271, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10325, "loss": 3.655111312866211, "lr": 1.838993397336436e-05, "elapsed_sec": 85880.29708194733, "step_time_sec": 8.239144462975673, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10326, "loss": 3.895949363708496, "lr": 1.83512201078816e-05, "elapsed_sec": 85888.53431296349, "step_time_sec": 8.237069027032703, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10327, "loss": 3.945918560028076, "lr": 1.831255817555755e-05, "elapsed_sec": 85896.77397847176, "step_time_sec": 8.239473994006403, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10328, "loss": 3.7888505458831787, "lr": 1.827394820025273e-05, "elapsed_sec": 85905.01395440102, "step_time_sec": 8.23982487397734, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10329, "loss": 3.7123863697052, "lr": 1.823539020579562e-05, "elapsed_sec": 85913.25270056725, "step_time_sec": 8.238593325950205, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10330, "loss": 3.960967779159546, "lr": 1.8196884215982673e-05, "elapsed_sec": 85921.49250817299, "step_time_sec": 8.239616460981779, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10331, "loss": 3.7023465633392334, "lr": 1.8158430254578155e-05, "elapsed_sec": 85929.73218750954, "step_time_sec": 8.239572972990572, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10332, "loss": 3.910012722015381, "lr": 1.812002834531424e-05, "elapsed_sec": 85937.97367191315, "step_time_sec": 8.241301613976248, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10333, "loss": 3.8410661220550537, "lr": 1.8081678511891073e-05, "elapsed_sec": 85946.21346712112, "step_time_sec": 8.239628287032247, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10334, "loss": 3.798354148864746, "lr": 1.8043380777976556e-05, "elapsed_sec": 85954.45281600952, "step_time_sec": 8.239161074045114, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10335, "loss": 3.892519950866699, "lr": 1.8005135167206442e-05, "elapsed_sec": 85962.69328761101, "step_time_sec": 8.240274359006435, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10336, "loss": 3.8313958644866943, "lr": 1.79669417031844e-05, "elapsed_sec": 85970.93265247345, "step_time_sec": 8.239201419055462, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10337, "loss": 3.7616524696350098, "lr": 1.792880040948184e-05, "elapsed_sec": 85979.1728489399, "step_time_sec": 8.240053969086148, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10338, "loss": 3.8990633487701416, "lr": 1.7890711309638e-05, "elapsed_sec": 85987.41205358505, "step_time_sec": 8.239068488008343, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10339, "loss": 3.8545162677764893, "lr": 1.7852674427159886e-05, "elapsed_sec": 85995.65212583542, "step_time_sec": 8.239887916948646, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10340, "loss": 3.837921142578125, "lr": 1.781468978552231e-05, "elapsed_sec": 86003.89295625687, "step_time_sec": 8.240646001999266, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10341, "loss": 3.790292263031006, "lr": 1.7776757408167804e-05, "elapsed_sec": 86012.1317486763, "step_time_sec": 8.238612691988237, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10342, "loss": 3.9431450366973877, "lr": 1.7738877318506714e-05, "elapsed_sec": 86020.37035179138, "step_time_sec": 8.238423969014548, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10343, "loss": 3.6649696826934814, "lr": 1.7701049539917028e-05, "elapsed_sec": 86028.60950183868, "step_time_sec": 8.239001226029359, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10344, "loss": 3.9426443576812744, "lr": 1.7663274095744498e-05, "elapsed_sec": 86036.84732580185, "step_time_sec": 8.237599584972486, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10345, "loss": 3.6130149364471436, "lr": 1.7625551009302585e-05, "elapsed_sec": 86045.08656597137, "step_time_sec": 8.239082287997007, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10346, "loss": 3.8898680210113525, "lr": 1.7587880303872436e-05, "elapsed_sec": 86053.32635951042, "step_time_sec": 8.239631105097942, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10347, "loss": 3.7972798347473145, "lr": 1.755026200270282e-05, "elapsed_sec": 86061.56376576424, "step_time_sec": 8.237232959014364, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10348, "loss": 3.8310163021087646, "lr": 1.7512696129010238e-05, "elapsed_sec": 86069.80370664597, "step_time_sec": 8.23976710694842, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10349, "loss": 3.816427707672119, "lr": 1.747518270597875e-05, "elapsed_sec": 86078.04321098328, "step_time_sec": 8.23938048700802, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10350, "loss": 3.8443069458007812, "lr": 1.743772175676015e-05, "elapsed_sec": 86086.28370165825, "step_time_sec": 8.240280142053962, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10351, "loss": 4.093235969543457, "lr": 1.7400313304473778e-05, "elapsed_sec": 86094.52352905273, "step_time_sec": 8.239712978014722, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10352, "loss": 3.818986415863037, "lr": 1.7362957372206577e-05, "elapsed_sec": 86102.76123428345, "step_time_sec": 8.237505924073048, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10353, "loss": 3.9071037769317627, "lr": 1.7325653983013086e-05, "elapsed_sec": 86111.00212788582, "step_time_sec": 8.240720200003125, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10354, "loss": 3.8073959350585938, "lr": 1.7288403159915426e-05, "elapsed_sec": 86119.2424595356, "step_time_sec": 8.240202400018461, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10355, "loss": 3.753704786300659, "lr": 1.7251204925903247e-05, "elapsed_sec": 86127.4805366993, "step_time_sec": 8.23792784102261, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10356, "loss": 3.9443507194519043, "lr": 1.721405930393381e-05, "elapsed_sec": 86135.72034287453, "step_time_sec": 8.23959482589271, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10357, "loss": 3.7529172897338867, "lr": 1.7176966316931836e-05, "elapsed_sec": 86143.95929932594, "step_time_sec": 8.238778489991091, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10358, "loss": 3.896470069885254, "lr": 1.713992598778957e-05, "elapsed_sec": 86152.19894123077, "step_time_sec": 8.239439908997156, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10359, "loss": 3.625800132751465, "lr": 1.7102938339366822e-05, "elapsed_sec": 86160.43982839584, "step_time_sec": 8.240732784965076, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10360, "loss": 3.8386828899383545, "lr": 1.7066003394490823e-05, "elapsed_sec": 86168.6776638031, "step_time_sec": 8.237669775960967, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10361, "loss": 3.8345954418182373, "lr": 1.702912117595629e-05, "elapsed_sec": 86176.91605353355, "step_time_sec": 8.238198295934126, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10362, "loss": 3.7886435985565186, "lr": 1.699229170652541e-05, "elapsed_sec": 86185.1553080082, "step_time_sec": 8.239075500983745, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10363, "loss": 3.85307240486145, "lr": 1.6955515008927797e-05, "elapsed_sec": 86193.39347624779, "step_time_sec": 8.23804576008115, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10364, "loss": 3.776024580001831, "lr": 1.6918791105860558e-05, "elapsed_sec": 86201.63243198395, "step_time_sec": 8.238742011948489, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10365, "loss": 3.710493564605713, "lr": 1.688212001998815e-05, "elapsed_sec": 86209.87000131607, "step_time_sec": 8.237420673016459, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10366, "loss": 3.782947540283203, "lr": 1.6845501773942426e-05, "elapsed_sec": 86218.10989928246, "step_time_sec": 8.239701576996595, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10367, "loss": 3.835601329803467, "lr": 1.6808936390322705e-05, "elapsed_sec": 86226.34930253029, "step_time_sec": 8.239214867004193, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10368, "loss": 3.914611577987671, "lr": 1.677242389169564e-05, "elapsed_sec": 86234.5898630619, "step_time_sec": 8.240402871975675, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10369, "loss": 3.861236572265625, "lr": 1.6735964300595176e-05, "elapsed_sec": 86242.82859063148, "step_time_sec": 8.23856661398895, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10370, "loss": 3.7824411392211914, "lr": 1.6699557639522724e-05, "elapsed_sec": 86251.06683063507, "step_time_sec": 8.238050075015053, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10371, "loss": 3.889880895614624, "lr": 1.6663203930946965e-05, "elapsed_sec": 86259.30738449097, "step_time_sec": 8.240419568028301, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10372, "loss": 3.8469736576080322, "lr": 1.6626903197303884e-05, "elapsed_sec": 86267.54746890068, "step_time_sec": 8.239904958987609, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10373, "loss": 3.8160643577575684, "lr": 1.659065546099684e-05, "elapsed_sec": 86275.78606700897, "step_time_sec": 8.238398776971735, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10374, "loss": 3.9062600135803223, "lr": 1.655446074439641e-05, "elapsed_sec": 86284.02395915985, "step_time_sec": 8.237721672048792, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10375, "loss": 3.803394079208374, "lr": 1.6518319069840495e-05, "elapsed_sec": 86292.26232933998, "step_time_sec": 8.238191591924988, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10376, "loss": 3.7760772705078125, "lr": 1.6482230459634248e-05, "elapsed_sec": 86300.50277233124, "step_time_sec": 8.240312259993516, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10377, "loss": 3.843329429626465, "lr": 1.6446194936050064e-05, "elapsed_sec": 86308.74323034286, "step_time_sec": 8.240298628923483, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10378, "loss": 3.7432217597961426, "lr": 1.641021252132757e-05, "elapsed_sec": 86316.98081302643, "step_time_sec": 8.237441690987907, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10379, "loss": 3.902284860610962, "lr": 1.637428323767366e-05, "elapsed_sec": 86325.22195506096, "step_time_sec": 8.240926266997121, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10380, "loss": 3.7058322429656982, "lr": 1.6338407107262393e-05, "elapsed_sec": 86333.46208333969, "step_time_sec": 8.239990856964141, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10381, "loss": 3.8088736534118652, "lr": 1.630258415223502e-05, "elapsed_sec": 86341.70165801048, "step_time_sec": 8.239413990988396, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10382, "loss": 3.9654433727264404, "lr": 1.626681439470003e-05, "elapsed_sec": 86349.94094848633, "step_time_sec": 8.239100300008431, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10383, "loss": 3.9059274196624756, "lr": 1.6231097856733012e-05, "elapsed_sec": 86358.18030190468, "step_time_sec": 8.23913845710922, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10384, "loss": 3.7651238441467285, "lr": 1.6195434560376764e-05, "elapsed_sec": 86366.4202837944, "step_time_sec": 8.239816506044008, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10385, "loss": 3.782827138900757, "lr": 1.6159824527641183e-05, "elapsed_sec": 86374.66057252884, "step_time_sec": 8.240150124998763, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10386, "loss": 3.9047656059265137, "lr": 1.61242677805033e-05, "elapsed_sec": 86382.89903998375, "step_time_sec": 8.238340217038058, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10387, "loss": 3.824033260345459, "lr": 1.608876434090732e-05, "elapsed_sec": 86391.13864326477, "step_time_sec": 8.239408995956182, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10388, "loss": 3.82966685295105, "lr": 1.6053314230764483e-05, "elapsed_sec": 86399.37842655182, "step_time_sec": 8.23959721904248, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10389, "loss": 3.93108868598938, "lr": 1.6017917471953102e-05, "elapsed_sec": 86407.61869001389, "step_time_sec": 8.240165599971078, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10390, "loss": 3.7694051265716553, "lr": 1.5982574086318665e-05, "elapsed_sec": 86415.8588385582, "step_time_sec": 8.239963565021753, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10391, "loss": 3.8039090633392334, "lr": 1.59472840956736e-05, "elapsed_sec": 86424.09669733047, "step_time_sec": 8.237681456026621, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10392, "loss": 3.826864004135132, "lr": 1.591204752179743e-05, "elapsed_sec": 86432.33725738525, "step_time_sec": 8.240327455918305, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10393, "loss": 3.8382174968719482, "lr": 1.5876864386436744e-05, "elapsed_sec": 86440.57715392113, "step_time_sec": 8.239781174925156, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10394, "loss": 3.8620009422302246, "lr": 1.5841734711305124e-05, "elapsed_sec": 86448.81618523598, "step_time_sec": 8.238848702050745, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10395, "loss": 3.7907845973968506, "lr": 1.5806658518083115e-05, "elapsed_sec": 86457.05646061897, "step_time_sec": 8.240093652973883, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10396, "loss": 3.8944249153137207, "lr": 1.577163582841836e-05, "elapsed_sec": 86465.29500174522, "step_time_sec": 8.238336927024648, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10397, "loss": 3.7650086879730225, "lr": 1.5736666663925388e-05, "elapsed_sec": 86473.53549575806, "step_time_sec": 8.240370371029712, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10398, "loss": 3.788069725036621, "lr": 1.5701751046185727e-05, "elapsed_sec": 86481.7761182785, "step_time_sec": 8.240408421959728, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10399, "loss": 3.7274601459503174, "lr": 1.566688899674786e-05, "elapsed_sec": 86490.01461553574, "step_time_sec": 8.238333937013522, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10400, "loss": 3.902702808380127, "lr": 1.56320805371272e-05, "elapsed_sec": 86498.25509691238, "step_time_sec": 8.240279855090193, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10401, "loss": 3.7581968307495117, "lr": 1.5597325688806087e-05, "elapsed_sec": 86506.49305534363, "step_time_sec": 8.2377907440532, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10402, "loss": 3.8494038581848145, "lr": 1.5562624473233802e-05, "elapsed_sec": 86514.73229932785, "step_time_sec": 8.239085434004664, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10403, "loss": 3.587099313735962, "lr": 1.552797691182648e-05, "elapsed_sec": 86522.97207903862, "step_time_sec": 8.239623055094853, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10404, "loss": 3.779694080352783, "lr": 1.5493383025967208e-05, "elapsed_sec": 86531.21028494835, "step_time_sec": 8.238070201012306, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10405, "loss": 3.825345754623413, "lr": 1.5458842837005873e-05, "elapsed_sec": 86539.4507715702, "step_time_sec": 8.240288399043493, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10406, "loss": 3.863513469696045, "lr": 1.5424356366259276e-05, "elapsed_sec": 86547.68995976448, "step_time_sec": 8.239073673961684, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10407, "loss": 3.7402420043945312, "lr": 1.538992363501103e-05, "elapsed_sec": 86555.9296233654, "step_time_sec": 8.239485210971907, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10408, "loss": 3.8527820110321045, "lr": 1.535554466451159e-05, "elapsed_sec": 86564.17077207565, "step_time_sec": 8.240962759009562, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10409, "loss": 3.74838924407959, "lr": 1.5321219475978242e-05, "elapsed_sec": 86572.41045546532, "step_time_sec": 8.239498459966853, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10410, "loss": 3.726205348968506, "lr": 1.5286948090595106e-05, "elapsed_sec": 86580.65023493767, "step_time_sec": 8.23961813095957, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10411, "loss": 3.9210939407348633, "lr": 1.5252730529513048e-05, "elapsed_sec": 86588.88962125778, "step_time_sec": 8.239251357968897, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10412, "loss": 3.9226412773132324, "lr": 1.5218566813849744e-05, "elapsed_sec": 86597.1278591156, "step_time_sec": 8.23798838199582, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10413, "loss": 3.8671059608459473, "lr": 1.5184456964689615e-05, "elapsed_sec": 86605.36902332306, "step_time_sec": 8.24102447507903, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10414, "loss": 3.7328176498413086, "lr": 1.5150401003083867e-05, "elapsed_sec": 86613.60706567764, "step_time_sec": 8.237859716988169, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10415, "loss": 3.890738010406494, "lr": 1.5116398950050417e-05, "elapsed_sec": 86621.84540319443, "step_time_sec": 8.238186503062025, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10416, "loss": 3.8055644035339355, "lr": 1.5082450826573967e-05, "elapsed_sec": 86630.0848107338, "step_time_sec": 8.23925338103436, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10417, "loss": 3.8463950157165527, "lr": 1.5048556653605882e-05, "elapsed_sec": 86638.32278084755, "step_time_sec": 8.237794945947826, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10418, "loss": 3.9125754833221436, "lr": 1.5014716452064216e-05, "elapsed_sec": 86646.56308412552, "step_time_sec": 8.240125378943048, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10419, "loss": 3.795700788497925, "lr": 1.4980930242833802e-05, "elapsed_sec": 86654.80344843864, "step_time_sec": 8.24018247902859, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10420, "loss": 3.795395612716675, "lr": 1.494719804676607e-05, "elapsed_sec": 86663.04337716103, "step_time_sec": 8.239738827105612, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10421, "loss": 3.9001598358154297, "lr": 1.4913519884679144e-05, "elapsed_sec": 86671.28364300728, "step_time_sec": 8.240127132972702, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10422, "loss": 3.808976173400879, "lr": 1.4879895777357798e-05, "elapsed_sec": 86679.52223134041, "step_time_sec": 8.238399278954603, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10423, "loss": 3.6500470638275146, "lr": 1.4846325745553415e-05, "elapsed_sec": 86687.76268148422, "step_time_sec": 8.240315238013864, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10424, "loss": 3.852344274520874, "lr": 1.4812809809984088e-05, "elapsed_sec": 86696.00189781189, "step_time_sec": 8.239002230926417, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10425, "loss": 3.8024892807006836, "lr": 1.477934799133444e-05, "elapsed_sec": 86704.23979783058, "step_time_sec": 8.237732162000611, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10426, "loss": 3.9689784049987793, "lr": 1.474594031025572e-05, "elapsed_sec": 86712.47996711731, "step_time_sec": 8.240018518059514, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10427, "loss": 3.9121408462524414, "lr": 1.4712586787365792e-05, "elapsed_sec": 86720.7207596302, "step_time_sec": 8.24062576808501, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10428, "loss": 3.6665029525756836, "lr": 1.4679287443249093e-05, "elapsed_sec": 86728.95991039276, "step_time_sec": 8.238990063080564, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10429, "loss": 3.684426784515381, "lr": 1.4646042298456535e-05, "elapsed_sec": 86737.19962382317, "step_time_sec": 8.239522909047082, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10430, "loss": 3.8728904724121094, "lr": 1.461285137350573e-05, "elapsed_sec": 86745.4373998642, "step_time_sec": 8.237618006998673, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10431, "loss": 3.8631784915924072, "lr": 1.4579714688880695e-05, "elapsed_sec": 86753.6784286499, "step_time_sec": 8.240847356966697, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10432, "loss": 3.898865222930908, "lr": 1.454663226503204e-05, "elapsed_sec": 86761.9192302227, "step_time_sec": 8.24060823500622, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10433, "loss": 3.847879409790039, "lr": 1.4513604122376895e-05, "elapsed_sec": 86770.15576148033, "step_time_sec": 8.236373018007725, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10434, "loss": 3.8431460857391357, "lr": 1.4480630281298853e-05, "elapsed_sec": 86778.39567542076, "step_time_sec": 8.239770187996328, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10435, "loss": 3.8076510429382324, "lr": 1.444771076214802e-05, "elapsed_sec": 86786.63522934914, "step_time_sec": 8.239349244977348, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10436, "loss": 3.8414721488952637, "lr": 1.4414845585240934e-05, "elapsed_sec": 86794.87475967407, "step_time_sec": 8.239362954976968, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10437, "loss": 3.798293352127075, "lr": 1.4382034770860662e-05, "elapsed_sec": 86803.11439919472, "step_time_sec": 8.23952320299577, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10438, "loss": 3.9122507572174072, "lr": 1.4349278339256653e-05, "elapsed_sec": 86811.35190534592, "step_time_sec": 8.237295549013652, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10439, "loss": 3.854909658432007, "lr": 1.4316576310644867e-05, "elapsed_sec": 86819.59201598167, "step_time_sec": 8.239929767907597, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10440, "loss": 3.81870174407959, "lr": 1.4283928705207611e-05, "elapsed_sec": 86827.83154559135, "step_time_sec": 8.239401794970036, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10441, "loss": 3.763701915740967, "lr": 1.4251335543093676e-05, "elapsed_sec": 86836.0706038475, "step_time_sec": 8.238881586003117, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10442, "loss": 3.82938551902771, "lr": 1.4218796844418203e-05, "elapsed_sec": 86844.31112742424, "step_time_sec": 8.240331786102615, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10443, "loss": 3.572094678878784, "lr": 1.4186312629262734e-05, "elapsed_sec": 86852.5495800972, "step_time_sec": 8.238281166995876, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10444, "loss": 3.7541165351867676, "lr": 1.4153882917675188e-05, "elapsed_sec": 86860.78984975815, "step_time_sec": 8.24009986105375, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10445, "loss": 3.7718982696533203, "lr": 1.412150772966984e-05, "elapsed_sec": 86869.02926683426, "step_time_sec": 8.239284791983664, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10446, "loss": 3.9615917205810547, "lr": 1.4089187085227315e-05, "elapsed_sec": 86877.266715765, "step_time_sec": 8.237316959071904, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10447, "loss": 3.7221877574920654, "lr": 1.4056921004294611e-05, "elapsed_sec": 86885.50600600243, "step_time_sec": 8.239061382017098, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10448, "loss": 3.831533193588257, "lr": 1.4024709506785016e-05, "elapsed_sec": 86893.7442855835, "step_time_sec": 8.238142358022742, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10449, "loss": 3.9378294944763184, "lr": 1.3992552612578098e-05, "elapsed_sec": 86901.98303055763, "step_time_sec": 8.238525848952122, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10450, "loss": 3.6746621131896973, "lr": 1.3960450341519848e-05, "elapsed_sec": 86910.22236227989, "step_time_sec": 8.239197788992897, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10451, "loss": 3.8799874782562256, "lr": 1.3928402713422383e-05, "elapsed_sec": 86918.4608874321, "step_time_sec": 8.238398563000374, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10452, "loss": 3.876819133758545, "lr": 1.3896409748064207e-05, "elapsed_sec": 86926.70060014725, "step_time_sec": 8.239539834088646, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10453, "loss": 3.84304141998291, "lr": 1.3864471465190074e-05, "elapsed_sec": 86934.93897032738, "step_time_sec": 8.238140871981159, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10454, "loss": 3.81427264213562, "lr": 1.3832587884510975e-05, "elapsed_sec": 86943.17913746834, "step_time_sec": 8.240058172959834, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10455, "loss": 3.8012962341308594, "lr": 1.3800759025704105e-05, "elapsed_sec": 86951.41915154457, "step_time_sec": 8.239818110014312, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10456, "loss": 3.844633102416992, "lr": 1.3768984908412962e-05, "elapsed_sec": 86959.65770816803, "step_time_sec": 8.23840236698743, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10457, "loss": 3.8375515937805176, "lr": 1.3737265552247219e-05, "elapsed_sec": 86967.89747214317, "step_time_sec": 8.239625567919575, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10458, "loss": 3.9712953567504883, "lr": 1.3705600976782751e-05, "elapsed_sec": 86976.13724732399, "step_time_sec": 8.2395721799694, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10459, "loss": 3.7819101810455322, "lr": 1.3673991201561617e-05, "elapsed_sec": 86984.37722206116, "step_time_sec": 8.239821088965982, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10460, "loss": 3.917334794998169, "lr": 1.3642436246092059e-05, "elapsed_sec": 86992.61713671684, "step_time_sec": 8.239727071020752, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10461, "loss": 3.8293886184692383, "lr": 1.3610936129848535e-05, "elapsed_sec": 87000.85471630096, "step_time_sec": 8.237421309924684, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10462, "loss": 3.867587089538574, "lr": 1.3579490872271592e-05, "elapsed_sec": 87009.09508824348, "step_time_sec": 8.240241251885891, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10463, "loss": 3.959892988204956, "lr": 1.3548100492767926e-05, "elapsed_sec": 87017.33746933937, "step_time_sec": 8.242244324996136, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10464, "loss": 3.673438310623169, "lr": 1.351676501071043e-05, "elapsed_sec": 87025.57618927956, "step_time_sec": 8.238483579014428, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10465, "loss": 3.7556650638580322, "lr": 1.3485484445438053e-05, "elapsed_sec": 87033.81635951996, "step_time_sec": 8.239990705973469, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10466, "loss": 3.641779661178589, "lr": 1.3454258816255869e-05, "elapsed_sec": 87042.05670142174, "step_time_sec": 8.240179761894979, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10467, "loss": 3.751000165939331, "lr": 1.3423088142435042e-05, "elapsed_sec": 87050.2953851223, "step_time_sec": 8.238547688000835, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10468, "loss": 3.7683000564575195, "lr": 1.3391972443212839e-05, "elapsed_sec": 87058.53388953209, "step_time_sec": 8.238367764046416, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10469, "loss": 3.8291890621185303, "lr": 1.3360911737792558e-05, "elapsed_sec": 87066.77132248878, "step_time_sec": 8.237247623037547, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10470, "loss": 3.795677661895752, "lr": 1.3329906045343638e-05, "elapsed_sec": 87075.01227903366, "step_time_sec": 8.240775713929906, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10471, "loss": 3.8826801776885986, "lr": 1.3298955385001482e-05, "elapsed_sec": 87083.25293397903, "step_time_sec": 8.240484319045208, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10472, "loss": 3.6877036094665527, "lr": 1.3268059775867567e-05, "elapsed_sec": 87091.49246454239, "step_time_sec": 8.239404764957726, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10473, "loss": 3.861848831176758, "lr": 1.3237219237009391e-05, "elapsed_sec": 87099.73231506348, "step_time_sec": 8.239660330931656, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10474, "loss": 3.9077670574188232, "lr": 1.3206433787460463e-05, "elapsed_sec": 87107.97218060493, "step_time_sec": 8.23975591501221, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10475, "loss": 3.835883140563965, "lr": 1.3175703446220268e-05, "elapsed_sec": 87116.21228194237, "step_time_sec": 8.239892138983123, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10476, "loss": 3.8462677001953125, "lr": 1.314502823225435e-05, "elapsed_sec": 87124.45223093033, "step_time_sec": 8.239803503965959, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10477, "loss": 3.7154810428619385, "lr": 1.3114408164494157e-05, "elapsed_sec": 87132.69064617157, "step_time_sec": 8.238277020049281, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10478, "loss": 3.9580767154693604, "lr": 1.3083843261837152e-05, "elapsed_sec": 87140.93526721, "step_time_sec": 8.244377754046582, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10479, "loss": 3.8418679237365723, "lr": 1.305333354314672e-05, "elapsed_sec": 87149.17619419098, "step_time_sec": 8.240805762005039, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10480, "loss": 3.7702136039733887, "lr": 1.3022879027252217e-05, "elapsed_sec": 87157.41986393929, "step_time_sec": 8.243511324981228, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10481, "loss": 3.7739040851593018, "lr": 1.2992479732948903e-05, "elapsed_sec": 87165.67221522331, "step_time_sec": 8.252101814025082, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10482, "loss": 3.998915910720825, "lr": 1.2962135678997967e-05, "elapsed_sec": 87173.91215157509, "step_time_sec": 8.239761115983129, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10483, "loss": 3.6920721530914307, "lr": 1.2931846884126491e-05, "elapsed_sec": 87182.15286874771, "step_time_sec": 8.24054063402582, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10484, "loss": 3.7387616634368896, "lr": 1.2901613367027515e-05, "elapsed_sec": 87190.39247202873, "step_time_sec": 8.239474880043417, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10485, "loss": 3.72627854347229, "lr": 1.2871435146359893e-05, "elapsed_sec": 87198.63245129585, "step_time_sec": 8.239789385930635, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10486, "loss": 3.8613815307617188, "lr": 1.2841312240748364e-05, "elapsed_sec": 87206.87151741982, "step_time_sec": 8.238904013996944, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10487, "loss": 3.7591631412506104, "lr": 1.2811244668783589e-05, "elapsed_sec": 87215.11088585854, "step_time_sec": 8.239197334973142, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10488, "loss": 3.9866607189178467, "lr": 1.2781232449022014e-05, "elapsed_sec": 87223.35128998756, "step_time_sec": 8.240251147071831, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10489, "loss": 3.7512104511260986, "lr": 1.2751275599985908e-05, "elapsed_sec": 87231.58888697624, "step_time_sec": 8.237460246076807, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10490, "loss": 3.7383460998535156, "lr": 1.2721374140163443e-05, "elapsed_sec": 87239.8288128376, "step_time_sec": 8.23974126810208, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10491, "loss": 3.9981675148010254, "lr": 1.2691528088008568e-05, "elapsed_sec": 87248.06759262085, "step_time_sec": 8.238675724947825, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10492, "loss": 3.611105442047119, "lr": 1.2661737461941e-05, "elapsed_sec": 87256.30678606033, "step_time_sec": 8.238976895925589, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10493, "loss": 3.8821189403533936, "lr": 1.263200228034634e-05, "elapsed_sec": 87264.54636859894, "step_time_sec": 8.239455408998765, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10494, "loss": 3.7653019428253174, "lr": 1.2602322561575874e-05, "elapsed_sec": 87272.78393149376, "step_time_sec": 8.237377019948326, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10495, "loss": 3.5175445079803467, "lr": 1.2572698323946721e-05, "elapsed_sec": 87281.02345252037, "step_time_sec": 8.239339528023265, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10496, "loss": 3.7915027141571045, "lr": 1.2543129585741736e-05, "elapsed_sec": 87289.26202559471, "step_time_sec": 8.238438665983267, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10497, "loss": 3.846958637237549, "lr": 1.2513616365209499e-05, "elapsed_sec": 87297.50017046928, "step_time_sec": 8.237996671930887, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10498, "loss": 3.763956308364868, "lr": 1.24841586805644e-05, "elapsed_sec": 87305.73989653587, "step_time_sec": 8.239502605982125, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10499, "loss": 3.798855781555176, "lr": 1.2454756549986488e-05, "elapsed_sec": 87313.97963285446, "step_time_sec": 8.239585211966187, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10500, "loss": 3.7800583839416504, "lr": 1.2425409991621523e-05, "elapsed_sec": 87322.22125840187, "step_time_sec": 25.513449987047352, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": true, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10501, "loss": 3.759272575378418, "lr": 1.2396119023581026e-05, "elapsed_sec": 87347.73800206184, "step_time_sec": 8.244276647921652, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10502, "loss": 3.8069050312042236, "lr": 1.236688366394216e-05, "elapsed_sec": 87355.96993041039, "step_time_sec": 8.231662471080199, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10503, "loss": 3.84687876701355, "lr": 1.2337703930747785e-05, "elapsed_sec": 87364.2035381794, "step_time_sec": 8.233464226010256, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10504, "loss": 3.8926939964294434, "lr": 1.2308579842006435e-05, "elapsed_sec": 87372.44199585915, "step_time_sec": 8.238277311087586, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10505, "loss": 3.7946743965148926, "lr": 1.2279511415692288e-05, "elapsed_sec": 87380.68283343315, "step_time_sec": 8.240699497051537, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10506, "loss": 3.786769390106201, "lr": 1.2250498669745162e-05, "elapsed_sec": 87388.92264962196, "step_time_sec": 8.239652892923914, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10507, "loss": 3.7511277198791504, "lr": 1.2221541622070575e-05, "elapsed_sec": 87397.16085910797, "step_time_sec": 8.23803139606025, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10508, "loss": 3.8158774375915527, "lr": 1.2192640290539587e-05, "elapsed_sec": 87405.40073084831, "step_time_sec": 8.239691672031768, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10509, "loss": 3.9384169578552246, "lr": 1.2163794692988903e-05, "elapsed_sec": 87413.63914847374, "step_time_sec": 8.238270792062394, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10510, "loss": 3.7607414722442627, "lr": 1.2135004847220891e-05, "elapsed_sec": 87421.87677502632, "step_time_sec": 8.237466422026046, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10511, "loss": 3.7655069828033447, "lr": 1.2106270771003398e-05, "elapsed_sec": 87430.11504220963, "step_time_sec": 8.2380869690096, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10512, "loss": 3.7498466968536377, "lr": 1.2077592482069914e-05, "elapsed_sec": 87438.35311841965, "step_time_sec": 8.237902757013217, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10513, "loss": 3.7366905212402344, "lr": 1.204896999811953e-05, "elapsed_sec": 87446.59250497818, "step_time_sec": 8.239229604951106, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10514, "loss": 3.860393524169922, "lr": 1.2020403336816859e-05, "elapsed_sec": 87454.8324496746, "step_time_sec": 8.23977580992505, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10515, "loss": 3.8558170795440674, "lr": 1.1991892515792025e-05, "elapsed_sec": 87463.07049250603, "step_time_sec": 8.237858486943878, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10516, "loss": 3.8114116191864014, "lr": 1.1963437552640788e-05, "elapsed_sec": 87471.3092944622, "step_time_sec": 8.23861278698314, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10517, "loss": 3.7690443992614746, "lr": 1.193503846492435e-05, "elapsed_sec": 87479.54671168327, "step_time_sec": 8.23730419203639, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10518, "loss": 3.973936080932617, "lr": 1.1906695270169465e-05, "elapsed_sec": 87487.78613638878, "step_time_sec": 8.239278930006549, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10519, "loss": 3.8868343830108643, "lr": 1.1878407985868383e-05, "elapsed_sec": 87496.02651548386, "step_time_sec": 8.240159162902273, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10520, "loss": 3.634230136871338, "lr": 1.1850176629478821e-05, "elapsed_sec": 87504.26576375961, "step_time_sec": 8.239088176982477, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10521, "loss": 3.8380613327026367, "lr": 1.1822001218424053e-05, "elapsed_sec": 87512.50424480438, "step_time_sec": 8.238320257049054, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10522, "loss": 3.8879952430725098, "lr": 1.1793881770092774e-05, "elapsed_sec": 87520.74161839485, "step_time_sec": 8.23722927691415, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10523, "loss": 3.812452793121338, "lr": 1.176581830183911e-05, "elapsed_sec": 87528.9820702076, "step_time_sec": 8.240309342043474, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10524, "loss": 3.7949023246765137, "lr": 1.1737810830982729e-05, "elapsed_sec": 87537.21969270706, "step_time_sec": 8.237460308941081, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10525, "loss": 3.953756093978882, "lr": 1.1709859374808656e-05, "elapsed_sec": 87545.45847797394, "step_time_sec": 8.238619068986736, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10526, "loss": 3.711995840072632, "lr": 1.1681963950567387e-05, "elapsed_sec": 87553.69886136055, "step_time_sec": 8.24024479801301, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10527, "loss": 3.774503707885742, "lr": 1.1654124575474828e-05, "elapsed_sec": 87561.93765592575, "step_time_sec": 8.238584792939946, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10528, "loss": 3.8258869647979736, "lr": 1.1626341266712294e-05, "elapsed_sec": 87570.17899465561, "step_time_sec": 8.241193415946327, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10529, "loss": 3.979901075363159, "lr": 1.1598614041426482e-05, "elapsed_sec": 87578.4193470478, "step_time_sec": 8.240152162965387, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10530, "loss": 3.7607991695404053, "lr": 1.1570942916729521e-05, "elapsed_sec": 87586.6586587429, "step_time_sec": 8.239153854898177, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10531, "loss": 3.7256152629852295, "lr": 1.1543327909698875e-05, "elapsed_sec": 87594.8995001316, "step_time_sec": 8.240713455947116, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10532, "loss": 3.7475473880767822, "lr": 1.1515769037377386e-05, "elapsed_sec": 87603.13879227638, "step_time_sec": 8.239109574002214, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10533, "loss": 3.9308533668518066, "lr": 1.148826631677326e-05, "elapsed_sec": 87611.37729668617, "step_time_sec": 8.238323697005399, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10534, "loss": 3.8777334690093994, "lr": 1.1460819764860017e-05, "elapsed_sec": 87619.61603403091, "step_time_sec": 8.238605463993736, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10535, "loss": 3.9734420776367188, "lr": 1.1433429398576579e-05, "elapsed_sec": 87627.85389900208, "step_time_sec": 8.237709087901749, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10536, "loss": 3.6844096183776855, "lr": 1.1406095234827119e-05, "elapsed_sec": 87636.09514856339, "step_time_sec": 8.241037516039796, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10537, "loss": 3.6960132122039795, "lr": 1.137881729048117e-05, "elapsed_sec": 87644.33442091942, "step_time_sec": 8.239192503970116, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10538, "loss": 3.798309087753296, "lr": 1.1351595582373559e-05, "elapsed_sec": 87652.57185459137, "step_time_sec": 8.23718086106237, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10539, "loss": 3.78912353515625, "lr": 1.1324430127304405e-05, "elapsed_sec": 87660.81181740761, "step_time_sec": 8.239821754978038, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10540, "loss": 3.923449754714966, "lr": 1.1297320942039087e-05, "elapsed_sec": 87669.05022192001, "step_time_sec": 8.238230755086988, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10541, "loss": 3.8377137184143066, "lr": 1.1270268043308295e-05, "elapsed_sec": 87677.29221582413, "step_time_sec": 8.24188728898298, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10542, "loss": 3.8821237087249756, "lr": 1.1243271447807964e-05, "elapsed_sec": 87685.53222894669, "step_time_sec": 8.239797222078778, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10543, "loss": 3.8329451084136963, "lr": 1.1216331172199251e-05, "elapsed_sec": 87693.77060103416, "step_time_sec": 8.238202737062238, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10544, "loss": 3.842653274536133, "lr": 1.1189447233108624e-05, "elapsed_sec": 87713.48044490814, "step_time_sec": 19.709688703995198, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10545, "loss": 3.824313163757324, "lr": 1.1162619647127716e-05, "elapsed_sec": 87721.71222996712, "step_time_sec": 8.231604588916525, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10546, "loss": 3.7627837657928467, "lr": 1.1135848430813406e-05, "elapsed_sec": 87729.94592571259, "step_time_sec": 8.233550532022491, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10547, "loss": 3.825305700302124, "lr": 1.1109133600687806e-05, "elapsed_sec": 87738.17874383926, "step_time_sec": 8.232610232080333, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10548, "loss": 3.8153767585754395, "lr": 1.1082475173238213e-05, "elapsed_sec": 87746.41290903091, "step_time_sec": 8.233999673044309, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10549, "loss": 3.707090139389038, "lr": 1.1055873164917066e-05, "elapsed_sec": 87754.64487934113, "step_time_sec": 8.231857210979797, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10550, "loss": 3.8690028190612793, "lr": 1.1029327592142068e-05, "elapsed_sec": 87762.87819981575, "step_time_sec": 8.233123884070665, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10551, "loss": 3.8323330879211426, "lr": 1.1002838471296043e-05, "elapsed_sec": 87771.11139917374, "step_time_sec": 8.233013929100707, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10552, "loss": 3.7690024375915527, "lr": 1.0976405818726965e-05, "elapsed_sec": 87779.3515625, "step_time_sec": 8.239989183959551, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10553, "loss": 3.7459795475006104, "lr": 1.0950029650748008e-05, "elapsed_sec": 87787.59010696411, "step_time_sec": 8.238403244060464, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10554, "loss": 3.7951529026031494, "lr": 1.0923709983637447e-05, "elapsed_sec": 87795.83083343506, "step_time_sec": 8.240538198966533, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10555, "loss": 3.8620948791503906, "lr": 1.089744683363869e-05, "elapsed_sec": 87804.07094192505, "step_time_sec": 8.239941156003624, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10556, "loss": 3.8918139934539795, "lr": 1.0871240216960264e-05, "elapsed_sec": 87812.30995035172, "step_time_sec": 8.238782204920426, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10557, "loss": 3.748560905456543, "lr": 1.0845090149775806e-05, "elapsed_sec": 87820.55031847954, "step_time_sec": 8.240211080992594, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10558, "loss": 3.736375331878662, "lr": 1.0818996648224096e-05, "elapsed_sec": 87828.79031634331, "step_time_sec": 8.23985996900592, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10559, "loss": 3.9136621952056885, "lr": 1.0792959728408938e-05, "elapsed_sec": 87837.028819561, "step_time_sec": 8.238306299084798, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10560, "loss": 3.9703454971313477, "lr": 1.0766979406399238e-05, "elapsed_sec": 87845.26824712753, "step_time_sec": 8.239251257968135, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10561, "loss": 3.8838233947753906, "lr": 1.0741055698229021e-05, "elapsed_sec": 87853.50665354729, "step_time_sec": 8.238265567924827, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10562, "loss": 3.997178554534912, "lr": 1.0715188619897297e-05, "elapsed_sec": 87861.74759531021, "step_time_sec": 8.240729570039548, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10563, "loss": 3.871023178100586, "lr": 1.0689378187368168e-05, "elapsed_sec": 87869.98761796951, "step_time_sec": 8.239889568998478, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10564, "loss": 3.8804116249084473, "lr": 1.066362441657077e-05, "elapsed_sec": 87878.22735643387, "step_time_sec": 8.239521868992597, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10565, "loss": 3.999459743499756, "lr": 1.0637927323399272e-05, "elapsed_sec": 87886.46672058105, "step_time_sec": 8.239203513017856, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10566, "loss": 3.7891805171966553, "lr": 1.0612286923712843e-05, "elapsed_sec": 87894.70573544502, "step_time_sec": 8.238860931945965, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10567, "loss": 3.8961422443389893, "lr": 1.0586703233335706e-05, "elapsed_sec": 87902.94632530212, "step_time_sec": 8.240392819046974, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10568, "loss": 3.873832941055298, "lr": 1.0561176268057055e-05, "elapsed_sec": 87911.18647432327, "step_time_sec": 8.240031906054355, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10569, "loss": 3.9430723190307617, "lr": 1.0535706043631058e-05, "elapsed_sec": 87919.42556524277, "step_time_sec": 8.23884194297716, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10570, "loss": 3.801964044570923, "lr": 1.0510292575776936e-05, "elapsed_sec": 87927.66463160515, "step_time_sec": 8.238973813015036, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10571, "loss": 3.3914337158203125, "lr": 1.0484935880178792e-05, "elapsed_sec": 87935.90428972244, "step_time_sec": 8.239425264997408, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10572, "loss": 3.733712673187256, "lr": 1.0459635972485726e-05, "elapsed_sec": 87944.14396762848, "step_time_sec": 8.239515084889717, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10573, "loss": 3.9185678958892822, "lr": 1.043439286831183e-05, "elapsed_sec": 87952.3825764656, "step_time_sec": 8.23845830897335, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10574, "loss": 3.910426378250122, "lr": 1.0409206583236088e-05, "elapsed_sec": 87960.6214325428, "step_time_sec": 8.238663828000426, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10575, "loss": 3.8268203735351562, "lr": 1.038407713280246e-05, "elapsed_sec": 87968.86157679558, "step_time_sec": 8.239959823898971, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10576, "loss": 3.6550509929656982, "lr": 1.0359004532519783e-05, "elapsed_sec": 87977.17455291748, "step_time_sec": 8.249975256039761, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10577, "loss": 3.7577476501464844, "lr": 1.0333988797861841e-05, "elapsed_sec": 87985.41385698318, "step_time_sec": 8.239122448023409, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10578, "loss": 3.901383876800537, "lr": 1.0309029944267315e-05, "elapsed_sec": 87993.6524105072, "step_time_sec": 8.238354405970313, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10579, "loss": 3.84366512298584, "lr": 1.0284127987139772e-05, "elapsed_sec": 88001.89299941063, "step_time_sec": 8.240422622999176, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10580, "loss": 3.7098376750946045, "lr": 1.0259282941847673e-05, "elapsed_sec": 88010.13226556778, "step_time_sec": 8.23913086799439, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10581, "loss": 3.9829046726226807, "lr": 1.0234494823724378e-05, "elapsed_sec": 88018.37065720558, "step_time_sec": 8.23819583200384, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10582, "loss": 3.945575475692749, "lr": 1.020976364806808e-05, "elapsed_sec": 88026.6108572483, "step_time_sec": 8.240018540993333, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10583, "loss": 3.621398448944092, "lr": 1.0185089430141825e-05, "elapsed_sec": 88034.85191655159, "step_time_sec": 8.240925787948072, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10584, "loss": 3.8923261165618896, "lr": 1.0160472185173554e-05, "elapsed_sec": 88043.09200763702, "step_time_sec": 8.239911499083973, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10585, "loss": 3.835066318511963, "lr": 1.0135911928355993e-05, "elapsed_sec": 88051.3315038681, "step_time_sec": 8.239334956044331, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10586, "loss": 3.9140045642852783, "lr": 1.0111408674846726e-05, "elapsed_sec": 88059.57104086876, "step_time_sec": 8.23935230798088, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10587, "loss": 3.8269574642181396, "lr": 1.0086962439768162e-05, "elapsed_sec": 88067.81064200401, "step_time_sec": 8.239430715097114, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10588, "loss": 3.7876341342926025, "lr": 1.0062573238207491e-05, "elapsed_sec": 88076.04973435402, "step_time_sec": 8.238897489965893, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10589, "loss": 3.7986502647399902, "lr": 1.0038241085216716e-05, "elapsed_sec": 88084.28891158104, "step_time_sec": 8.239002271089703, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10590, "loss": 3.8461296558380127, "lr": 1.0013965995812672e-05, "elapsed_sec": 88092.52873206139, "step_time_sec": 8.239666124107316, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10591, "loss": 3.9411728382110596, "lr": 9.98974798497691e-06, "elapsed_sec": 88100.76770377159, "step_time_sec": 8.23881072900258, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10592, "loss": 3.7658839225769043, "lr": 9.965587067655831e-06, "elapsed_sec": 88109.00732898712, "step_time_sec": 8.23940371198114, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10593, "loss": 3.845914363861084, "lr": 9.941483258760518e-06, "elapsed_sec": 88117.24627017975, "step_time_sec": 8.238864085054956, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10594, "loss": 3.8316071033477783, "lr": 9.917436573166849e-06, "elapsed_sec": 88125.48483300209, "step_time_sec": 8.238360109040514, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10595, "loss": 5.076749801635742, "lr": 9.893447025715469e-06, "elapsed_sec": 88133.72574543953, "step_time_sec": 8.240692066028714, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10596, "loss": 3.784254789352417, "lr": 9.869514631211733e-06, "elapsed_sec": 88141.96494531631, "step_time_sec": 8.239043044974096, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10597, "loss": 4.044383525848389, "lr": 9.845639404425712e-06, "elapsed_sec": 88150.20333743095, "step_time_sec": 8.238212583004497, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10598, "loss": 3.752251386642456, "lr": 9.821821360092238e-06, "elapsed_sec": 88158.44272732735, "step_time_sec": 8.239173135021701, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10599, "loss": 3.697774648666382, "lr": 9.798060512910804e-06, "elapsed_sec": 88166.6805768013, "step_time_sec": 8.237647991045378, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10600, "loss": 3.8237290382385254, "lr": 9.77435687754563e-06, "elapsed_sec": 88174.92115569115, "step_time_sec": 8.240443467046134, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10601, "loss": 3.892974853515625, "lr": 9.750710468625639e-06, "elapsed_sec": 88183.16241407394, "step_time_sec": 8.241062468965538, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10602, "loss": 3.66833758354187, "lr": 9.727121300744402e-06, "elapsed_sec": 88191.40070724487, "step_time_sec": 8.238142944057472, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10603, "loss": 3.9421586990356445, "lr": 9.703589388460168e-06, "elapsed_sec": 88199.64100646973, "step_time_sec": 8.240107996971346, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10604, "loss": 3.731839656829834, "lr": 9.680114746295906e-06, "elapsed_sec": 88207.87990903854, "step_time_sec": 8.238763850997202, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10605, "loss": 3.6550328731536865, "lr": 9.656697388739161e-06, "elapsed_sec": 88216.11912226677, "step_time_sec": 8.239033702993765, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10606, "loss": 3.7996749877929688, "lr": 9.633337330242169e-06, "elapsed_sec": 88224.35969281197, "step_time_sec": 8.240368991042487, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10607, "loss": 3.9079482555389404, "lr": 9.61003458522181e-06, "elapsed_sec": 88232.59706664085, "step_time_sec": 8.23719468608033, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10608, "loss": 3.9018826484680176, "lr": 9.586789168059573e-06, "elapsed_sec": 88240.83653378487, "step_time_sec": 8.239261399023235, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10609, "loss": 3.909233570098877, "lr": 9.563601093101554e-06, "elapsed_sec": 88249.07604241371, "step_time_sec": 8.2393981810892, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10610, "loss": 3.824056386947632, "lr": 9.540470374658491e-06, "elapsed_sec": 88257.3160750866, "step_time_sec": 8.239885984105058, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10611, "loss": 3.762014865875244, "lr": 9.517397027005694e-06, "elapsed_sec": 88265.55699825287, "step_time_sec": 8.240718798013404, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10612, "loss": 3.7257485389709473, "lr": 9.49438106438312e-06, "elapsed_sec": 88273.79546880722, "step_time_sec": 8.238335717003793, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10613, "loss": 3.8529176712036133, "lr": 9.47142250099524e-06, "elapsed_sec": 88282.03605747223, "step_time_sec": 8.240398169029504, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10614, "loss": 3.7796850204467773, "lr": 9.448521351011145e-06, "elapsed_sec": 88290.27637290955, "step_time_sec": 8.240147350006737, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10615, "loss": 3.856006622314453, "lr": 9.425677628564472e-06, "elapsed_sec": 88298.51465058327, "step_time_sec": 8.238046407932416, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10616, "loss": 3.8364744186401367, "lr": 9.402891347753435e-06, "elapsed_sec": 88306.7518620491, "step_time_sec": 8.237084058928303, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10617, "loss": 3.6141624450683594, "lr": 9.380162522640773e-06, "elapsed_sec": 88314.99169397354, "step_time_sec": 8.239578621927649, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10618, "loss": 3.8793723583221436, "lr": 9.357491167253803e-06, "elapsed_sec": 88323.23081088066, "step_time_sec": 8.238970519043505, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10619, "loss": 3.8362984657287598, "lr": 9.334877295584336e-06, "elapsed_sec": 88331.47061681747, "step_time_sec": 8.239702222053893, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10620, "loss": 4.077843189239502, "lr": 9.31232092158871e-06, "elapsed_sec": 88339.70991373062, "step_time_sec": 8.239108511945233, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10621, "loss": 3.8018388748168945, "lr": 9.289822059187815e-06, "elapsed_sec": 88347.95194149017, "step_time_sec": 8.241868016077206, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10622, "loss": 3.7198405265808105, "lr": 9.267380722267002e-06, "elapsed_sec": 88356.19212770462, "step_time_sec": 8.240011613001116, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10623, "loss": 3.7976574897766113, "lr": 9.24499692467616e-06, "elapsed_sec": 88364.43088412285, "step_time_sec": 8.238541613915004, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10624, "loss": 3.8400473594665527, "lr": 9.222670680229633e-06, "elapsed_sec": 88372.67101836205, "step_time_sec": 8.239985457039438, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10625, "loss": 3.8857626914978027, "lr": 9.200402002706251e-06, "elapsed_sec": 88380.91095924377, "step_time_sec": 8.239767294959165, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10626, "loss": 3.745027542114258, "lr": 9.17819090584932e-06, "elapsed_sec": 88389.15059947968, "step_time_sec": 8.23944598599337, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10627, "loss": 3.9071242809295654, "lr": 9.156037403366642e-06, "elapsed_sec": 88397.39051961899, "step_time_sec": 8.239825433003716, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10628, "loss": 3.851463556289673, "lr": 9.13394150893043e-06, "elapsed_sec": 88405.62823653221, "step_time_sec": 8.23745993396733, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10629, "loss": 3.612954616546631, "lr": 9.11190323617734e-06, "elapsed_sec": 88413.86751461029, "step_time_sec": 8.239079307066277, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10630, "loss": 3.846050977706909, "lr": 9.08992259870853e-06, "elapsed_sec": 88422.10718798637, "step_time_sec": 8.239510918036103, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10631, "loss": 3.707660436630249, "lr": 9.067999610089493e-06, "elapsed_sec": 88430.34675264359, "step_time_sec": 8.23947228398174, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10632, "loss": 3.83758544921875, "lr": 9.04613428385024e-06, "elapsed_sec": 88438.58669424057, "step_time_sec": 8.239754489972256, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10633, "loss": 3.781318426132202, "lr": 9.024326633485113e-06, "elapsed_sec": 88446.82547664642, "step_time_sec": 8.238594468915835, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10634, "loss": 3.6165966987609863, "lr": 9.002576672452906e-06, "elapsed_sec": 88455.06688928604, "step_time_sec": 8.241206740960479, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10635, "loss": 3.7850027084350586, "lr": 8.980884414176821e-06, "elapsed_sec": 88463.30584716797, "step_time_sec": 8.238785416004248, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10636, "loss": 3.701502561569214, "lr": 8.959249872044395e-06, "elapsed_sec": 88471.54416012764, "step_time_sec": 8.238155485014431, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10637, "loss": 3.9060256481170654, "lr": 8.937673059407591e-06, "elapsed_sec": 88479.78455901146, "step_time_sec": 8.24021255900152, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10638, "loss": 3.621737480163574, "lr": 8.91615398958273e-06, "elapsed_sec": 88488.02314090729, "step_time_sec": 8.23846175998915, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10639, "loss": 3.67750883102417, "lr": 8.894692675850487e-06, "elapsed_sec": 88496.26314306259, "step_time_sec": 8.23976062098518, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10640, "loss": 3.769315004348755, "lr": 8.873289131455887e-06, "elapsed_sec": 88504.50349760056, "step_time_sec": 8.240196540951729, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10641, "loss": 3.8365700244903564, "lr": 8.85194336960835e-06, "elapsed_sec": 88512.7409722805, "step_time_sec": 8.237357791978866, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10642, "loss": 3.7972073554992676, "lr": 8.830655403481587e-06, "elapsed_sec": 88520.98060178757, "step_time_sec": 8.23940380802378, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10643, "loss": 3.7155518531799316, "lr": 8.809425246213625e-06, "elapsed_sec": 88529.22122573853, "step_time_sec": 8.240463146939874, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10644, "loss": 3.9392855167388916, "lr": 8.788252910906882e-06, "elapsed_sec": 88537.46004462242, "step_time_sec": 8.238703147973865, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10645, "loss": 3.556124448776245, "lr": 8.767138410628034e-06, "elapsed_sec": 88545.69953227043, "step_time_sec": 8.23923107702285, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10646, "loss": 3.861882209777832, "lr": 8.746081758408085e-06, "elapsed_sec": 88553.93769741058, "step_time_sec": 8.238033107016236, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10647, "loss": 3.8423991203308105, "lr": 8.72508296724232e-06, "elapsed_sec": 88562.17934775352, "step_time_sec": 8.241441202000715, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10648, "loss": 3.66652512550354, "lr": 8.704142050090327e-06, "elapsed_sec": 88570.41968345642, "step_time_sec": 8.240218522958457, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10649, "loss": 3.809455633163452, "lr": 8.683259019876e-06, "elapsed_sec": 88578.6583533287, "step_time_sec": 8.23851041099988, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10650, "loss": 3.754080295562744, "lr": 8.662433889487469e-06, "elapsed_sec": 88586.89841294289, "step_time_sec": 8.23982508003246, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10651, "loss": 3.7593038082122803, "lr": 8.641666671777139e-06, "elapsed_sec": 88595.13759374619, "step_time_sec": 8.238999931956641, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10652, "loss": 3.699252128601074, "lr": 8.620957379561716e-06, "elapsed_sec": 88603.37733411789, "step_time_sec": 8.239619996049441, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10653, "loss": 3.766075611114502, "lr": 8.600306025622068e-06, "elapsed_sec": 88611.61717486382, "step_time_sec": 8.23962185299024, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10654, "loss": 3.908733606338501, "lr": 8.579712622703396e-06, "elapsed_sec": 88619.85591578484, "step_time_sec": 8.238584874081425, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10655, "loss": 3.7889857292175293, "lr": 8.559177183515087e-06, "elapsed_sec": 88628.09652686119, "step_time_sec": 8.240477990009822, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10656, "loss": 3.7229340076446533, "lr": 8.538699720730767e-06, "elapsed_sec": 88636.33741688728, "step_time_sec": 8.240658187074587, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10657, "loss": 3.8604962825775146, "lr": 8.518280246988292e-06, "elapsed_sec": 88644.57503795624, "step_time_sec": 8.237417718046345, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10658, "loss": 3.7139387130737305, "lr": 8.49791877488971e-06, "elapsed_sec": 88652.8154001236, "step_time_sec": 8.24021078797523, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10659, "loss": 3.7695188522338867, "lr": 8.477615317001292e-06, "elapsed_sec": 88661.05485224724, "step_time_sec": 8.239254667889327, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10660, "loss": 3.793287515640259, "lr": 8.457369885853484e-06, "elapsed_sec": 88669.29461932182, "step_time_sec": 8.239617862040177, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10661, "loss": 3.812565326690674, "lr": 8.437182493940938e-06, "elapsed_sec": 88677.53502440453, "step_time_sec": 8.240277350996621, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10662, "loss": 3.823394536972046, "lr": 8.417053153722486e-06, "elapsed_sec": 88685.77302956581, "step_time_sec": 8.23777923209127, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10663, "loss": 3.7778918743133545, "lr": 8.396981877621098e-06, "elapsed_sec": 88694.01598072052, "step_time_sec": 8.242827326990664, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10664, "loss": 3.7029876708984375, "lr": 8.376968678023983e-06, "elapsed_sec": 88702.25611543655, "step_time_sec": 8.239914270001464, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10665, "loss": 3.880495309829712, "lr": 8.357013567282434e-06, "elapsed_sec": 88710.49434161186, "step_time_sec": 8.23809082689695, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10666, "loss": 3.828826427459717, "lr": 8.337116557711916e-06, "elapsed_sec": 88718.73403310776, "step_time_sec": 8.239524859003723, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10667, "loss": 3.785940170288086, "lr": 8.317277661592068e-06, "elapsed_sec": 88726.97335147858, "step_time_sec": 8.239147359039634, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10668, "loss": 3.8737313747406006, "lr": 8.297496891166636e-06, "elapsed_sec": 88735.21389055252, "step_time_sec": 8.240315039991401, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10669, "loss": 3.7337594032287598, "lr": 8.277774258643482e-06, "elapsed_sec": 88743.45518708229, "step_time_sec": 8.24117847206071, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10670, "loss": 3.783888101577759, "lr": 8.258109776194607e-06, "elapsed_sec": 88751.69307804108, "step_time_sec": 8.237679084995762, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10671, "loss": 3.7039756774902344, "lr": 8.238503455956105e-06, "elapsed_sec": 88759.93260526657, "step_time_sec": 8.239372383919545, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10672, "loss": 3.8446805477142334, "lr": 8.218955310028218e-06, "elapsed_sec": 88768.17155790329, "step_time_sec": 8.238795318990014, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10673, "loss": 3.9295716285705566, "lr": 8.199465350475241e-06, "elapsed_sec": 88776.41008400917, "step_time_sec": 8.238337139948271, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10674, "loss": 3.8456571102142334, "lr": 8.180033589325577e-06, "elapsed_sec": 88784.6510746479, "step_time_sec": 8.240827656001784, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10675, "loss": 3.5997040271759033, "lr": 8.160660038571705e-06, "elapsed_sec": 88792.88989305496, "step_time_sec": 8.238686425960623, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10676, "loss": 3.7205207347869873, "lr": 8.141344710170187e-06, "elapsed_sec": 88801.13144183159, "step_time_sec": 8.2413536880631, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10677, "loss": 3.837815761566162, "lr": 8.122087616041625e-06, "elapsed_sec": 88809.36997056007, "step_time_sec": 8.238347581936978, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10678, "loss": 3.9823665618896484, "lr": 8.102888768070737e-06, "elapsed_sec": 88817.60860776901, "step_time_sec": 8.238537551020272, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10679, "loss": 3.832564115524292, "lr": 8.083748178106241e-06, "elapsed_sec": 88825.84711813927, "step_time_sec": 8.238295792019926, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10680, "loss": 3.9157214164733887, "lr": 8.06466585796091e-06, "elapsed_sec": 88834.08627676964, "step_time_sec": 8.238981044967659, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10681, "loss": 3.8724026679992676, "lr": 8.045641819411603e-06, "elapsed_sec": 88842.3250989914, "step_time_sec": 8.23864474100992, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10682, "loss": 3.8577311038970947, "lr": 8.026676074199137e-06, "elapsed_sec": 88850.56487870216, "step_time_sec": 8.23965452099219, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10683, "loss": 3.8537940979003906, "lr": 8.007768634028409e-06, "elapsed_sec": 88858.80363035202, "step_time_sec": 8.238591118948534, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10684, "loss": 3.8629069328308105, "lr": 7.988919510568305e-06, "elapsed_sec": 88867.04366159439, "step_time_sec": 8.239779615076259, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10685, "loss": 3.858987331390381, "lr": 7.970128715451719e-06, "elapsed_sec": 88875.28308677673, "step_time_sec": 8.239297852967866, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10686, "loss": 3.788991928100586, "lr": 7.951396260275554e-06, "elapsed_sec": 88883.52155733109, "step_time_sec": 8.2383031839272, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10687, "loss": 3.685289144515991, "lr": 7.932722156600725e-06, "elapsed_sec": 88891.76029801369, "step_time_sec": 8.238550234003924, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10688, "loss": 3.861971855163574, "lr": 7.914106415952113e-06, "elapsed_sec": 88899.99803566933, "step_time_sec": 8.237579112988897, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10689, "loss": 3.826291561126709, "lr": 7.895549049818583e-06, "elapsed_sec": 88908.23772716522, "step_time_sec": 8.23953293391969, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10690, "loss": 3.7804417610168457, "lr": 7.877050069652986e-06, "elapsed_sec": 88916.47689032555, "step_time_sec": 8.239006391027942, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10691, "loss": 3.909897804260254, "lr": 7.858609486872106e-06, "elapsed_sec": 88924.71574997902, "step_time_sec": 8.238676527049392, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10692, "loss": 3.884244203567505, "lr": 7.840227312856739e-06, "elapsed_sec": 88932.95525717735, "step_time_sec": 8.239333406905644, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10693, "loss": 3.8664958477020264, "lr": 7.821903558951592e-06, "elapsed_sec": 88941.19315385818, "step_time_sec": 8.237697536940686, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10694, "loss": 3.8344926834106445, "lr": 7.803638236465327e-06, "elapsed_sec": 88949.43350601196, "step_time_sec": 8.240201796987094, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10695, "loss": 3.9939777851104736, "lr": 7.785431356670553e-06, "elapsed_sec": 88957.6743247509, "step_time_sec": 8.240686462959275, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10696, "loss": 3.6942036151885986, "lr": 7.767282930803811e-06, "elapsed_sec": 88965.91159653664, "step_time_sec": 8.23705712298397, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10697, "loss": 3.7358381748199463, "lr": 7.749192970065555e-06, "elapsed_sec": 88974.15143942833, "step_time_sec": 8.2396896739956, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10698, "loss": 3.9649643898010254, "lr": 7.73116148562016e-06, "elapsed_sec": 88982.39057254791, "step_time_sec": 8.238947571953759, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10699, "loss": 3.848393678665161, "lr": 7.71318848859592e-06, "elapsed_sec": 88990.63091874123, "step_time_sec": 8.240185952046886, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10700, "loss": 3.704488754272461, "lr": 7.695273990085021e-06, "elapsed_sec": 88998.8715441227, "step_time_sec": 8.24046012503095, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10701, "loss": 3.787339687347412, "lr": 7.67741800114357e-06, "elapsed_sec": 89007.10997009277, "step_time_sec": 8.238203190034255, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10702, "loss": 3.986358642578125, "lr": 7.659620532791528e-06, "elapsed_sec": 89015.35006070137, "step_time_sec": 8.239935406949371, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10703, "loss": 3.831805467605591, "lr": 7.641881596012768e-06, "elapsed_sec": 89023.58961677551, "step_time_sec": 8.239398814039305, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10704, "loss": 3.825972080230713, "lr": 7.624201201755032e-06, "elapsed_sec": 89031.82743883133, "step_time_sec": 8.237608546041884, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10705, "loss": 3.8860580921173096, "lr": 7.606579360929934e-06, "elapsed_sec": 89040.06797361374, "step_time_sec": 8.240346431965008, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10706, "loss": 3.6968536376953125, "lr": 7.589016084412944e-06, "elapsed_sec": 89048.30531835556, "step_time_sec": 8.237243063980713, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10707, "loss": 3.8285419940948486, "lr": 7.571511383043387e-06, "elapsed_sec": 89056.54559588432, "step_time_sec": 8.240088866092265, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10708, "loss": 3.7774970531463623, "lr": 7.554065267624444e-06, "elapsed_sec": 89064.78482604027, "step_time_sec": 8.239021133980714, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10709, "loss": 3.892608642578125, "lr": 7.536677748923146e-06, "elapsed_sec": 89073.02472162247, "step_time_sec": 8.239748019957915, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10710, "loss": 3.8658833503723145, "lr": 7.51934883767034e-06, "elapsed_sec": 89081.26437234879, "step_time_sec": 8.239452193956822, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10711, "loss": 3.789484739303589, "lr": 7.502078544560728e-06, "elapsed_sec": 89089.50274968147, "step_time_sec": 8.23822029097937, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10712, "loss": 3.868293523788452, "lr": 7.484866880252838e-06, "elapsed_sec": 89097.74283480644, "step_time_sec": 8.23995976394508, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10713, "loss": 3.7857272624969482, "lr": 7.467713855368965e-06, "elapsed_sec": 89105.98305153847, "step_time_sec": 8.240011165034957, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10714, "loss": 3.7728898525238037, "lr": 7.450619480495262e-06, "elapsed_sec": 89114.22180652618, "step_time_sec": 8.238591153989546, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10715, "loss": 3.9290013313293457, "lr": 7.4335837661816835e-06, "elapsed_sec": 89122.46119642258, "step_time_sec": 8.23919170408044, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10716, "loss": 3.9092366695404053, "lr": 7.416606722941956e-06, "elapsed_sec": 89130.7018532753, "step_time_sec": 8.240498648025095, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10717, "loss": 3.7424397468566895, "lr": 7.399688361253616e-06, "elapsed_sec": 89138.94124889374, "step_time_sec": 8.239177439012565, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10718, "loss": 3.8647985458374023, "lr": 7.382828691557983e-06, "elapsed_sec": 89147.18099093437, "step_time_sec": 8.239570789039135, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10719, "loss": 3.7816450595855713, "lr": 7.366027724260156e-06, "elapsed_sec": 89155.41956067085, "step_time_sec": 8.238452216959558, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10720, "loss": 3.83528733253479, "lr": 7.349285469728982e-06, "elapsed_sec": 89163.65948867798, "step_time_sec": 8.239725868916139, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10721, "loss": 3.861816644668579, "lr": 7.332601938297111e-06, "elapsed_sec": 89171.89920282364, "step_time_sec": 8.239543418982066, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10722, "loss": 3.9157233238220215, "lr": 7.31597714026091e-06, "elapsed_sec": 89180.13655948639, "step_time_sec": 8.237195755005814, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10723, "loss": 3.727553606033325, "lr": 7.299411085880523e-06, "elapsed_sec": 89188.37556505203, "step_time_sec": 8.238871315959841, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10724, "loss": 3.754133462905884, "lr": 7.282903785379856e-06, "elapsed_sec": 89196.61458706856, "step_time_sec": 8.2388612650102, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10725, "loss": 3.791381597518921, "lr": 7.266455248946508e-06, "elapsed_sec": 89204.85558652878, "step_time_sec": 8.240803627995774, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10726, "loss": 3.7934439182281494, "lr": 7.250065486731858e-06, "elapsed_sec": 89213.09559440613, "step_time_sec": 8.239833594067022, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10727, "loss": 3.9884402751922607, "lr": 7.23373450885098e-06, "elapsed_sec": 89221.33325147629, "step_time_sec": 8.237497579073533, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10728, "loss": 3.7736897468566895, "lr": 7.2174623253826865e-06, "elapsed_sec": 89229.57297420502, "step_time_sec": 8.239600830012932, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10729, "loss": 3.899061679840088, "lr": 7.201248946369503e-06, "elapsed_sec": 89237.81250786781, "step_time_sec": 8.239378780010156, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10730, "loss": 3.8035292625427246, "lr": 7.18509438181764e-06, "elapsed_sec": 89246.05106163025, "step_time_sec": 8.238329784013331, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10731, "loss": 3.807110071182251, "lr": 7.1689986416970445e-06, "elapsed_sec": 89254.29097557068, "step_time_sec": 8.239710952038877, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10732, "loss": 3.864290952682495, "lr": 7.152961735941336e-06, "elapsed_sec": 89262.5292084217, "step_time_sec": 8.238039236981422, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10733, "loss": 3.7275383472442627, "lr": 7.136983674447851e-06, "elapsed_sec": 89270.77024555206, "step_time_sec": 8.240897039999254, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10734, "loss": 3.822284698486328, "lr": 7.121064467077564e-06, "elapsed_sec": 89279.0110476017, "step_time_sec": 8.240578236989677, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10735, "loss": 3.8808836936950684, "lr": 7.105204123655167e-06, "elapsed_sec": 89287.24815273285, "step_time_sec": 8.237003048998304, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10736, "loss": 3.7846078872680664, "lr": 7.089402653969011e-06, "elapsed_sec": 89295.48876738548, "step_time_sec": 8.240434712031856, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10737, "loss": 3.8575944900512695, "lr": 7.073660067771101e-06, "elapsed_sec": 89303.727268219, "step_time_sec": 8.238318501971662, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10738, "loss": 3.7857768535614014, "lr": 7.057976374777134e-06, "elapsed_sec": 89311.96631669998, "step_time_sec": 8.238813256029971, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10739, "loss": 3.8488693237304688, "lr": 7.0423515846664144e-06, "elapsed_sec": 89320.20679616928, "step_time_sec": 8.240310972905718, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10740, "loss": 3.818681001663208, "lr": 7.026785707081929e-06, "elapsed_sec": 89328.44584989548, "step_time_sec": 8.238924902980216, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10741, "loss": 3.7744836807250977, "lr": 7.01127875163031e-06, "elapsed_sec": 89336.68625688553, "step_time_sec": 8.24021689593792, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10742, "loss": 3.8160455226898193, "lr": 6.995830727881789e-06, "elapsed_sec": 89344.92594027519, "step_time_sec": 8.239513695938513, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10743, "loss": 3.871825695037842, "lr": 6.980441645370263e-06, "elapsed_sec": 89353.16330909729, "step_time_sec": 8.23713698098436, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10744, "loss": 3.866926431655884, "lr": 6.965111513593227e-06, "elapsed_sec": 89361.4045214653, "step_time_sec": 8.241056478931569, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10745, "loss": 3.7232184410095215, "lr": 6.949840342011815e-06, "elapsed_sec": 89369.64236235619, "step_time_sec": 8.237672324059531, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10746, "loss": 3.8852365016937256, "lr": 6.934628140050767e-06, "elapsed_sec": 89377.88106918335, "step_time_sec": 8.23858180094976, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10747, "loss": 3.7997539043426514, "lr": 6.919474917098433e-06, "elapsed_sec": 89386.12028646469, "step_time_sec": 8.239036103943363, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10748, "loss": 3.715224266052246, "lr": 6.904380682506733e-06, "elapsed_sec": 89394.35992717743, "step_time_sec": 8.239437289070338, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10749, "loss": 3.8217215538024902, "lr": 6.889345445591233e-06, "elapsed_sec": 89402.59981417656, "step_time_sec": 8.239710529102013, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10750, "loss": 3.8765757083892822, "lr": 6.874369215631061e-06, "elapsed_sec": 89410.83962607384, "step_time_sec": 8.239646242931485, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10751, "loss": 3.843728542327881, "lr": 6.859452001868896e-06, "elapsed_sec": 89419.07992649078, "step_time_sec": 8.24014351889491, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10752, "loss": 3.7399139404296875, "lr": 6.844593813511059e-06, "elapsed_sec": 89427.31890010834, "step_time_sec": 8.238808622932993, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10753, "loss": 3.779806137084961, "lr": 6.829794659727414e-06, "elapsed_sec": 89435.55755639076, "step_time_sec": 8.238442907924764, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10754, "loss": 3.7663211822509766, "lr": 6.815054549651355e-06, "elapsed_sec": 89443.7965939045, "step_time_sec": 8.238887083949521, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10755, "loss": 3.8732688426971436, "lr": 6.800373492379907e-06, "elapsed_sec": 89452.03490662575, "step_time_sec": 8.23810892389156, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10756, "loss": 3.8406472206115723, "lr": 6.785751496973604e-06, "elapsed_sec": 89460.27494454384, "step_time_sec": 8.23989609000273, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10757, "loss": 3.75374698638916, "lr": 6.77118857245652e-06, "elapsed_sec": 89468.5155518055, "step_time_sec": 8.24048220703844, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10758, "loss": 3.938042402267456, "lr": 6.7566847278163214e-06, "elapsed_sec": 89476.75449419022, "step_time_sec": 8.238742970046587, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10759, "loss": 3.8386614322662354, "lr": 6.742239972004166e-06, "elapsed_sec": 89484.99528336525, "step_time_sec": 8.24060009594541, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10760, "loss": 3.8353867530822754, "lr": 6.727854313934749e-06, "elapsed_sec": 89493.23357248306, "step_time_sec": 8.238120967987925, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10761, "loss": 3.9476184844970703, "lr": 6.713527762486327e-06, "elapsed_sec": 89501.47240257263, "step_time_sec": 8.238673276966438, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10762, "loss": 3.790417194366455, "lr": 6.699260326500646e-06, "elapsed_sec": 89509.71123099327, "step_time_sec": 8.238651580060832, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10763, "loss": 3.7169418334960938, "lr": 6.685052014782993e-06, "elapsed_sec": 89517.95018792152, "step_time_sec": 8.238833916955628, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10764, "loss": 3.8055648803710938, "lr": 6.670902836102147e-06, "elapsed_sec": 89526.1905579567, "step_time_sec": 8.240186859969981, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10765, "loss": 3.7716245651245117, "lr": 6.6568127991903926e-06, "elapsed_sec": 89534.43184423447, "step_time_sec": 8.241075354046188, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10766, "loss": 3.7708024978637695, "lr": 6.642781912743515e-06, "elapsed_sec": 89542.66930961609, "step_time_sec": 8.237319045001641, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10767, "loss": 3.8888180255889893, "lr": 6.628810185420802e-06, "elapsed_sec": 89550.91030526161, "step_time_sec": 8.24080667400267, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10768, "loss": 3.849691867828369, "lr": 6.614897625845034e-06, "elapsed_sec": 89559.15015816689, "step_time_sec": 8.23973873606883, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10769, "loss": 3.6988978385925293, "lr": 6.601044242602475e-06, "elapsed_sec": 89567.38933825493, "step_time_sec": 8.239020257955417, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10770, "loss": 3.7600345611572266, "lr": 6.5872500442428436e-06, "elapsed_sec": 89575.63026332855, "step_time_sec": 8.240721832029521, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10771, "loss": 3.8099026679992676, "lr": 6.573515039279365e-06, "elapsed_sec": 89583.86877155304, "step_time_sec": 8.238323168945499, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10772, "loss": 4.084328651428223, "lr": 6.559839236188713e-06, "elapsed_sec": 89592.1092312336, "step_time_sec": 8.240251576993614, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10773, "loss": 3.7753801345825195, "lr": 6.546222643411037e-06, "elapsed_sec": 89600.34991908073, "step_time_sec": 8.240523176966235, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10774, "loss": 3.9108197689056396, "lr": 6.532665269349923e-06, "elapsed_sec": 89608.58857750893, "step_time_sec": 8.23856592096854, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10775, "loss": 3.9235827922821045, "lr": 6.519167122372433e-06, "elapsed_sec": 89616.829226017, "step_time_sec": 8.240428659017198, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10776, "loss": 3.824000120162964, "lr": 6.505728210809062e-06, "elapsed_sec": 89625.06850194931, "step_time_sec": 8.239083488006145, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10777, "loss": 3.774127960205078, "lr": 6.492348542953752e-06, "elapsed_sec": 89633.30799770355, "step_time_sec": 8.239319446031004, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10778, "loss": 3.789004325866699, "lr": 6.479028127063889e-06, "elapsed_sec": 89641.5484354496, "step_time_sec": 8.240280339960009, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10779, "loss": 3.7542996406555176, "lr": 6.465766971360284e-06, "elapsed_sec": 89649.78665757179, "step_time_sec": 8.238033280940726, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10780, "loss": 3.961369514465332, "lr": 6.45256508402717e-06, "elapsed_sec": 89658.02592682838, "step_time_sec": 8.23914609698113, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10781, "loss": 3.7711868286132812, "lr": 6.439422473212213e-06, "elapsed_sec": 89666.26719903946, "step_time_sec": 8.241075413068756, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10782, "loss": 3.9401376247406006, "lr": 6.426339147026484e-06, "elapsed_sec": 89674.50463199615, "step_time_sec": 8.237240615999326, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10783, "loss": 3.9636893272399902, "lr": 6.413315113544486e-06, "elapsed_sec": 89682.74528217316, "step_time_sec": 8.240500982967205, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10784, "loss": 3.924116611480713, "lr": 6.4003503808041025e-06, "elapsed_sec": 89690.98420214653, "step_time_sec": 8.238731992081739, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10785, "loss": 3.8863513469696045, "lr": 6.3874449568066295e-06, "elapsed_sec": 89699.22343039513, "step_time_sec": 8.239070247043855, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10786, "loss": 3.9537949562072754, "lr": 6.374598849516778e-06, "elapsed_sec": 89707.46292233467, "step_time_sec": 8.239364149048924, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10787, "loss": 3.868499994277954, "lr": 6.361812066862623e-06, "elapsed_sec": 89715.70246648788, "step_time_sec": 8.239364225999452, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10788, "loss": 3.822216749191284, "lr": 6.3490846167356574e-06, "elapsed_sec": 89723.94407010078, "step_time_sec": 8.24142687593121, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10789, "loss": 3.8466110229492188, "lr": 6.336416506990708e-06, "elapsed_sec": 89732.18431162834, "step_time_sec": 8.240035161958076, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10790, "loss": 3.822087049484253, "lr": 6.323807745446037e-06, "elapsed_sec": 89740.42380690575, "step_time_sec": 8.23932445794344, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10791, "loss": 3.984405279159546, "lr": 6.311258339883244e-06, "elapsed_sec": 89748.66370844841, "step_time_sec": 8.239809012971818, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10792, "loss": 3.5847415924072266, "lr": 6.2987682980472955e-06, "elapsed_sec": 89756.9035692215, "step_time_sec": 8.239620193024166, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10793, "loss": 3.8401763439178467, "lr": 6.286337627646551e-06, "elapsed_sec": 89765.14303398132, "step_time_sec": 8.239286664058454, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10794, "loss": 3.772186279296875, "lr": 6.2739663363526955e-06, "elapsed_sec": 89773.38351678848, "step_time_sec": 8.240324415965006, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10795, "loss": 3.805539131164551, "lr": 6.261654431800772e-06, "elapsed_sec": 89781.62242937088, "step_time_sec": 8.23879043397028, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10796, "loss": 3.825327157974243, "lr": 6.249401921589194e-06, "elapsed_sec": 89789.86344766617, "step_time_sec": 8.240776424994692, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10797, "loss": 3.831955671310425, "lr": 6.2372088132796995e-06, "elapsed_sec": 89798.10407710075, "step_time_sec": 8.240488193929195, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10798, "loss": 3.8484363555908203, "lr": 6.225075114397375e-06, "elapsed_sec": 89806.3429415226, "step_time_sec": 8.238706771982834, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10799, "loss": 3.612531900405884, "lr": 6.213000832430634e-06, "elapsed_sec": 89814.5837237835, "step_time_sec": 8.240552098955959, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10800, "loss": 3.809760332107544, "lr": 6.2009859748312355e-06, "elapsed_sec": 89822.82271909714, "step_time_sec": 8.23881218105089, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10801, "loss": 3.702406406402588, "lr": 6.189030549014244e-06, "elapsed_sec": 89831.06267356873, "step_time_sec": 8.23980602005031, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10802, "loss": 3.8950252532958984, "lr": 6.177134562358072e-06, "elapsed_sec": 89839.30235934258, "step_time_sec": 8.239490902982652, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10803, "loss": 3.8722310066223145, "lr": 6.1652980222044135e-06, "elapsed_sec": 89847.54005551338, "step_time_sec": 8.237576586077921, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10804, "loss": 3.817152500152588, "lr": 6.1535209358583e-06, "elapsed_sec": 89855.78073215485, "step_time_sec": 8.240453043021262, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10805, "loss": 3.7753257751464844, "lr": 6.1418033105880574e-06, "elapsed_sec": 89864.02075648308, "step_time_sec": 8.239882685942575, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10806, "loss": 3.8764102458953857, "lr": 6.130145153625337e-06, "elapsed_sec": 89872.25974345207, "step_time_sec": 8.238789247930981, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10807, "loss": 3.6634459495544434, "lr": 6.1185464721650604e-06, "elapsed_sec": 89880.5003361702, "step_time_sec": 8.240448941010982, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10808, "loss": 3.830211877822876, "lr": 6.107007273365456e-06, "elapsed_sec": 89888.73866915703, "step_time_sec": 8.238233787938952, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10809, "loss": 3.7161600589752197, "lr": 6.095527564348046e-06, "elapsed_sec": 89896.97973680496, "step_time_sec": 8.240840958897024, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10810, "loss": 3.6952061653137207, "lr": 6.084107352197642e-06, "elapsed_sec": 89905.22005224228, "step_time_sec": 8.240158090018667, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10811, "loss": 3.74290132522583, "lr": 6.072746643962298e-06, "elapsed_sec": 89913.45846748352, "step_time_sec": 8.238256467040628, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10812, "loss": 3.7820332050323486, "lr": 6.061445446653418e-06, "elapsed_sec": 89921.6986694336, "step_time_sec": 8.240047914907336, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10813, "loss": 3.603884696960449, "lr": 6.050203767245598e-06, "elapsed_sec": 89929.93803429604, "step_time_sec": 8.239199894946069, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10814, "loss": 3.641089677810669, "lr": 6.039021612676751e-06, "elapsed_sec": 89938.17697739601, "step_time_sec": 8.238748912001029, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10815, "loss": 3.880148410797119, "lr": 6.027898989848061e-06, "elapsed_sec": 89946.41786289215, "step_time_sec": 8.240760963992216, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10816, "loss": 3.8323147296905518, "lr": 6.01683590562392e-06, "elapsed_sec": 89954.65671181679, "step_time_sec": 8.238654769957066, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10817, "loss": 3.8756744861602783, "lr": 6.0058323668320354e-06, "elapsed_sec": 89962.89719200134, "step_time_sec": 8.24032045295462, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10818, "loss": 3.769944906234741, "lr": 5.9948883802633194e-06, "elapsed_sec": 89971.13546466827, "step_time_sec": 8.238107816083357, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10819, "loss": 3.9383456707000732, "lr": 5.9840039526719594e-06, "elapsed_sec": 89979.37393307686, "step_time_sec": 8.23829971300438, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10820, "loss": 3.683614492416382, "lr": 5.973179090775368e-06, "elapsed_sec": 89987.61414408684, "step_time_sec": 8.240105679025874, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10821, "loss": 3.8975839614868164, "lr": 5.962413801254209e-06, "elapsed_sec": 89995.85356903076, "step_time_sec": 8.23920893506147, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10822, "loss": 3.7767279148101807, "lr": 5.951708090752378e-06, "elapsed_sec": 90004.09173989296, "step_time_sec": 8.238023548969068, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10823, "loss": 3.8147377967834473, "lr": 5.941061965876995e-06, "elapsed_sec": 90012.33002352715, "step_time_sec": 8.238069095998071, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10824, "loss": 3.783808946609497, "lr": 5.930475433198409e-06, "elapsed_sec": 90020.56814837456, "step_time_sec": 8.237994321039878, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10825, "loss": 3.782042980194092, "lr": 5.9199484992501895e-06, "elapsed_sec": 90028.80835175514, "step_time_sec": 8.239999352954328, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10826, "loss": 3.7344470024108887, "lr": 5.909481170529131e-06, "elapsed_sec": 90037.04639530182, "step_time_sec": 8.237917617079802, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10827, "loss": 3.6770458221435547, "lr": 5.89907345349524e-06, "elapsed_sec": 90045.28797602654, "step_time_sec": 8.241385661996901, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10828, "loss": 3.818092107772827, "lr": 5.888725354571714e-06, "elapsed_sec": 90053.52816677094, "step_time_sec": 8.240031262044795, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10829, "loss": 3.736768960952759, "lr": 5.878436880144991e-06, "elapsed_sec": 90061.76673793793, "step_time_sec": 8.238456057966687, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10830, "loss": 3.886118173599243, "lr": 5.8682080365646854e-06, "elapsed_sec": 90070.00784015656, "step_time_sec": 8.240903957979754, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10831, "loss": 3.882174491882324, "lr": 5.858038830143607e-06, "elapsed_sec": 90078.24787187576, "step_time_sec": 8.239816386951134, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10832, "loss": 3.8818955421447754, "lr": 5.847929267157805e-06, "elapsed_sec": 90086.48756575584, "step_time_sec": 8.239566660951823, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10833, "loss": 3.752185583114624, "lr": 5.837879353846457e-06, "elapsed_sec": 90094.7281923294, "step_time_sec": 8.240404567914084, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10834, "loss": 3.946866512298584, "lr": 5.827889096411953e-06, "elapsed_sec": 90102.96754169464, "step_time_sec": 8.239204343990423, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10835, "loss": 3.7802653312683105, "lr": 5.817958501019884e-06, "elapsed_sec": 90111.2078742981, "step_time_sec": 8.240156162064523, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10836, "loss": 3.750107765197754, "lr": 5.8080875737989995e-06, "elapsed_sec": 90119.44788503647, "step_time_sec": 8.239816286950372, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10837, "loss": 3.773667812347412, "lr": 5.7982763208412194e-06, "elapsed_sec": 90127.68568372726, "step_time_sec": 8.237713658018038, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10838, "loss": 3.7393639087677, "lr": 5.78852474820166e-06, "elapsed_sec": 90135.92483377457, "step_time_sec": 8.238919812021777, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10839, "loss": 3.7263481616973877, "lr": 5.778832861898571e-06, "elapsed_sec": 90144.16525650024, "step_time_sec": 8.240280886064284, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10840, "loss": 3.9062600135803223, "lr": 5.769200667913406e-06, "elapsed_sec": 90152.40518975258, "step_time_sec": 8.239735755021684, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10841, "loss": 3.9321556091308594, "lr": 5.759628172190743e-06, "elapsed_sec": 90160.64475798607, "step_time_sec": 8.239421613980085, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10842, "loss": 3.898683547973633, "lr": 5.750115380638331e-06, "elapsed_sec": 90168.88336277008, "step_time_sec": 8.238460008054972, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10843, "loss": 3.8173587322235107, "lr": 5.740662299127084e-06, "elapsed_sec": 90177.12288284302, "step_time_sec": 8.239351411000825, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10844, "loss": 3.903473377227783, "lr": 5.7312689334910575e-06, "elapsed_sec": 90185.3637778759, "step_time_sec": 8.240729379002005, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10845, "loss": 3.678114414215088, "lr": 5.721935289527427e-06, "elapsed_sec": 90193.60268950462, "step_time_sec": 8.238809886039235, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10846, "loss": 3.782341241836548, "lr": 5.712661372996555e-06, "elapsed_sec": 90201.8430159092, "step_time_sec": 8.240145240910351, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10847, "loss": 3.8855369091033936, "lr": 5.703447189621911e-06, "elapsed_sec": 90210.08301115036, "step_time_sec": 8.239781006006524, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10848, "loss": 3.7682759761810303, "lr": 5.69429274509011e-06, "elapsed_sec": 90218.3217446804, "step_time_sec": 8.238543722894974, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10849, "loss": 3.964602470397949, "lr": 5.685198045050894e-06, "elapsed_sec": 90226.56070995331, "step_time_sec": 8.23883911804296, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10850, "loss": 3.9141292572021484, "lr": 5.676163095117149e-06, "elapsed_sec": 90234.79826283455, "step_time_sec": 8.237354769953527, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10851, "loss": 3.780900478363037, "lr": 5.667187900864861e-06, "elapsed_sec": 90243.03893637657, "step_time_sec": 8.240510466974229, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10852, "loss": 3.7168805599212646, "lr": 5.658272467833156e-06, "elapsed_sec": 90251.27899575233, "step_time_sec": 8.239872797974385, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10853, "loss": 3.823604106903076, "lr": 5.64941680152427e-06, "elapsed_sec": 90259.51835393906, "step_time_sec": 8.239184828009456, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10854, "loss": 3.7890374660491943, "lr": 5.64062090740356e-06, "elapsed_sec": 90267.75850629807, "step_time_sec": 8.240027450956404, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10855, "loss": 3.802874803543091, "lr": 5.631884790899491e-06, "elapsed_sec": 90275.9983549118, "step_time_sec": 8.239634552970529, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10856, "loss": 3.8515686988830566, "lr": 5.623208457403618e-06, "elapsed_sec": 90284.2383890152, "step_time_sec": 8.239887239062227, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10857, "loss": 3.755357265472412, "lr": 5.614591912270625e-06, "elapsed_sec": 90292.47861790657, "step_time_sec": 8.240035490947776, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10858, "loss": 3.7441914081573486, "lr": 5.606035160818297e-06, "elapsed_sec": 90300.71668028831, "step_time_sec": 8.237899373983964, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10859, "loss": 3.66532826423645, "lr": 5.597538208327505e-06, "elapsed_sec": 90308.95573735237, "step_time_sec": 8.238901145057753, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10860, "loss": 3.933408737182617, "lr": 5.589101060042208e-06, "elapsed_sec": 90317.19503831863, "step_time_sec": 8.239119653007947, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10861, "loss": 3.9186837673187256, "lr": 5.580723721169475e-06, "elapsed_sec": 90325.43411588669, "step_time_sec": 8.238927043043077, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10862, "loss": 3.7575113773345947, "lr": 5.572406196879452e-06, "elapsed_sec": 90333.67442512512, "step_time_sec": 8.240121489972807, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10863, "loss": 3.872683525085449, "lr": 5.564148492305384e-06, "elapsed_sec": 90341.91287636757, "step_time_sec": 8.238326488994062, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10864, "loss": 3.7477309703826904, "lr": 5.55595061254356e-06, "elapsed_sec": 90350.15322947502, "step_time_sec": 8.240125868003815, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10865, "loss": 3.7621665000915527, "lr": 5.54781256265339e-06, "elapsed_sec": 90358.39170312881, "step_time_sec": 8.23831650102511, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10866, "loss": 3.8850111961364746, "lr": 5.539734347657339e-06, "elapsed_sec": 90366.62926197052, "step_time_sec": 8.237402368918993, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10867, "loss": 3.8614017963409424, "lr": 5.531715972540959e-06, "elapsed_sec": 90374.87031149864, "step_time_sec": 8.240820250008255, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10868, "loss": 3.789422035217285, "lr": 5.523757442252837e-06, "elapsed_sec": 90383.10938858986, "step_time_sec": 8.238938543014228, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10869, "loss": 3.8243300914764404, "lr": 5.5158587617046795e-06, "elapsed_sec": 90391.34865307808, "step_time_sec": 8.239106336026452, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10870, "loss": 3.6950175762176514, "lr": 5.508019935771216e-06, "elapsed_sec": 90399.58771586418, "step_time_sec": 8.238889692001976, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10871, "loss": 3.7356863021850586, "lr": 5.5002409692902224e-06, "elapsed_sec": 90407.82571077347, "step_time_sec": 8.23788439494092, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10872, "loss": 3.8099257946014404, "lr": 5.4925218670625925e-06, "elapsed_sec": 90416.0655412674, "step_time_sec": 8.239592898986302, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10873, "loss": 3.8267111778259277, "lr": 5.484862633852211e-06, "elapsed_sec": 90424.29822039604, "step_time_sec": 8.232532006921247, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10874, "loss": 3.7114548683166504, "lr": 5.477263274386062e-06, "elapsed_sec": 90432.53296899796, "step_time_sec": 8.234546668943949, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10875, "loss": 3.912719488143921, "lr": 5.469723793354142e-06, "elapsed_sec": 90440.76593184471, "step_time_sec": 8.23282745690085, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10876, "loss": 3.9627058506011963, "lr": 5.462244195409512e-06, "elapsed_sec": 90449.00576043129, "step_time_sec": 8.23963271791581, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10877, "loss": 3.764779806137085, "lr": 5.454824485168269e-06, "elapsed_sec": 90457.24398136139, "step_time_sec": 8.238099862006493, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10878, "loss": 3.9002420902252197, "lr": 5.4474646672095625e-06, "elapsed_sec": 90465.48288083076, "step_time_sec": 8.238717047963291, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10879, "loss": 3.8414177894592285, "lr": 5.440164746075558e-06, "elapsed_sec": 90473.72145962715, "step_time_sec": 8.238390199025162, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10880, "loss": 3.843980073928833, "lr": 5.432924726271473e-06, "elapsed_sec": 90481.95995569229, "step_time_sec": 8.23838270700071, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10881, "loss": 3.8189456462860107, "lr": 5.425744612265537e-06, "elapsed_sec": 90490.20083832741, "step_time_sec": 8.240669874940068, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10882, "loss": 3.7902419567108154, "lr": 5.418624408489039e-06, "elapsed_sec": 90498.44119668007, "step_time_sec": 8.240213384036906, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10883, "loss": 3.876483678817749, "lr": 5.411564119336264e-06, "elapsed_sec": 90506.67923688889, "step_time_sec": 8.237835936946794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10884, "loss": 3.8942155838012695, "lr": 5.404563749164536e-06, "elapsed_sec": 90514.91301679611, "step_time_sec": 8.233595710014924, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10885, "loss": 3.799121379852295, "lr": 5.3976233022941845e-06, "elapsed_sec": 90523.1467268467, "step_time_sec": 8.233549072989263, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10886, "loss": 3.805570125579834, "lr": 5.390742783008579e-06, "elapsed_sec": 90531.38153195381, "step_time_sec": 8.234629938961007, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10887, "loss": 3.870168924331665, "lr": 5.383922195554088e-06, "elapsed_sec": 90539.61977982521, "step_time_sec": 8.238064415985718, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10888, "loss": 3.7372543811798096, "lr": 5.377161544140101e-06, "elapsed_sec": 90547.85907697678, "step_time_sec": 8.239169772015885, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10889, "loss": 3.9910776615142822, "lr": 5.370460832939011e-06, "elapsed_sec": 90556.09765005112, "step_time_sec": 8.238348320010118, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10890, "loss": 3.8225226402282715, "lr": 5.3638200660862135e-06, "elapsed_sec": 90564.33685207367, "step_time_sec": 8.239088286063634, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10891, "loss": 3.7881648540496826, "lr": 5.3572392476801225e-06, "elapsed_sec": 90572.57726597786, "step_time_sec": 8.240184215945192, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10892, "loss": 3.7878477573394775, "lr": 5.350718381782157e-06, "elapsed_sec": 90580.81516337395, "step_time_sec": 8.237727295025252, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10893, "loss": 3.8727645874023438, "lr": 5.3442574724167145e-06, "elapsed_sec": 90589.05538511276, "step_time_sec": 8.240076469955966, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10894, "loss": 3.8671553134918213, "lr": 5.337856523571193e-06, "elapsed_sec": 90597.29489970207, "step_time_sec": 8.239359920960851, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10895, "loss": 3.8076319694519043, "lr": 5.331515539196016e-06, "elapsed_sec": 90605.53448605537, "step_time_sec": 8.239347651950084, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10896, "loss": 4.047199249267578, "lr": 5.325234523204562e-06, "elapsed_sec": 90613.77438592911, "step_time_sec": 8.239764243015088, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10897, "loss": 3.714759111404419, "lr": 5.319013479473222e-06, "elapsed_sec": 90622.01426768303, "step_time_sec": 8.239703326951712, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10898, "loss": 3.9211783409118652, "lr": 5.312852411841379e-06, "elapsed_sec": 90630.2533788681, "step_time_sec": 8.238910731044598, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10899, "loss": 3.9013285636901855, "lr": 5.306751324111363e-06, "elapsed_sec": 90638.49345064163, "step_time_sec": 8.239934487035498, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10900, "loss": 3.9580202102661133, "lr": 5.300710220048534e-06, "elapsed_sec": 90646.73199987411, "step_time_sec": 8.238348674029112, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10901, "loss": 3.922398567199707, "lr": 5.294729103381217e-06, "elapsed_sec": 90654.97138667107, "step_time_sec": 8.239206787897274, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10902, "loss": 3.8311948776245117, "lr": 5.288807977800688e-06, "elapsed_sec": 90663.20484638214, "step_time_sec": 8.23329012200702, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10903, "loss": 3.7858736515045166, "lr": 5.282946846961229e-06, "elapsed_sec": 90671.43845653534, "step_time_sec": 8.233467801008373, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10904, "loss": 3.982666254043579, "lr": 5.2771457144801096e-06, "elapsed_sec": 90679.67168951035, "step_time_sec": 8.23309795698151, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10905, "loss": 3.7785801887512207, "lr": 5.271404583937526e-06, "elapsed_sec": 90687.91047167778, "step_time_sec": 8.238609993946739, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10906, "loss": 3.9365060329437256, "lr": 5.265723458876676e-06, "elapsed_sec": 90696.14993238449, "step_time_sec": 8.23922925395891, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10907, "loss": 3.8435232639312744, "lr": 5.260102342803715e-06, "elapsed_sec": 90704.38769960403, "step_time_sec": 8.237603919929825, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10908, "loss": 3.8445253372192383, "lr": 5.254541239187767e-06, "elapsed_sec": 90712.62734508514, "step_time_sec": 8.239479169948027, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10909, "loss": 3.826693296432495, "lr": 5.249040151460903e-06, "elapsed_sec": 90720.85958623886, "step_time_sec": 8.232071568956599, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10910, "loss": 3.836085796356201, "lr": 5.243599083018185e-06, "elapsed_sec": 90729.09304904938, "step_time_sec": 8.233280475018546, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10911, "loss": 3.7610716819763184, "lr": 5.2382180372176114e-06, "elapsed_sec": 90737.32549977303, "step_time_sec": 8.232358476961963, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10912, "loss": 3.652019739151001, "lr": 5.232897017380138e-06, "elapsed_sec": 90745.55956363678, "step_time_sec": 8.233817687956616, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10913, "loss": 3.8728365898132324, "lr": 5.227636026789678e-06, "elapsed_sec": 90753.79240727425, "step_time_sec": 8.232668732991442, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10914, "loss": 3.8219704627990723, "lr": 5.222435068693105e-06, "elapsed_sec": 90762.0257434845, "step_time_sec": 8.233195742941462, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10915, "loss": 3.9409918785095215, "lr": 5.217294146300236e-06, "elapsed_sec": 90770.25840854645, "step_time_sec": 8.232437173021026, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10916, "loss": 4.227794647216797, "lr": 5.212213262783817e-06, "elapsed_sec": 90778.4916934967, "step_time_sec": 8.233102426980622, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10917, "loss": 3.7716195583343506, "lr": 5.207192421279583e-06, "elapsed_sec": 90786.72902989388, "step_time_sec": 8.237165931030177, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10918, "loss": 3.7886340618133545, "lr": 5.202231624886182e-06, "elapsed_sec": 90794.96264839172, "step_time_sec": 8.233447529026307, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10919, "loss": 3.849848508834839, "lr": 5.197330876665214e-06, "elapsed_sec": 90803.19559335709, "step_time_sec": 8.232784127932973, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10920, "loss": 3.9618008136749268, "lr": 5.192490179641209e-06, "elapsed_sec": 90811.43412303925, "step_time_sec": 8.238366081030108, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10921, "loss": 3.726231098175049, "lr": 5.18770953680167e-06, "elapsed_sec": 90819.6658642292, "step_time_sec": 8.231595061952248, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10922, "loss": 3.77856707572937, "lr": 5.182988951096982e-06, "elapsed_sec": 90827.89905047417, "step_time_sec": 8.233049987000413, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10923, "loss": 3.7431066036224365, "lr": 5.178328425440529e-06, "elapsed_sec": 90836.130828619, "step_time_sec": 8.231596246943809, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10924, "loss": 3.8509459495544434, "lr": 5.173727962708569e-06, "elapsed_sec": 90844.36438059807, "step_time_sec": 8.23336365702562, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10925, "loss": 3.8714308738708496, "lr": 5.169187565740315e-06, "elapsed_sec": 90852.59741282463, "step_time_sec": 8.232836610986851, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10926, "loss": 3.7661523818969727, "lr": 5.164707237337941e-06, "elapsed_sec": 90860.83117604256, "step_time_sec": 8.233585939044133, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10927, "loss": 3.6881794929504395, "lr": 5.160286980266492e-06, "elapsed_sec": 90869.06486964226, "step_time_sec": 8.233556270017289, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10928, "loss": 3.899198293685913, "lr": 5.155926797253989e-06, "elapsed_sec": 90877.3002049923, "step_time_sec": 8.235165152931586, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10929, "loss": 3.861643075942993, "lr": 5.15162669099135e-06, "elapsed_sec": 90885.53610253334, "step_time_sec": 8.235694203991443, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10930, "loss": 3.8956291675567627, "lr": 5.147386664132418e-06, "elapsed_sec": 90893.7692284584, "step_time_sec": 8.232948949094862, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10931, "loss": 3.7685914039611816, "lr": 5.143206719293971e-06, "elapsed_sec": 90902.00233530998, "step_time_sec": 8.232928047073074, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10932, "loss": 3.7426435947418213, "lr": 5.1390868590557e-06, "elapsed_sec": 90910.2398929596, "step_time_sec": 8.237427185988054, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10933, "loss": 3.5854368209838867, "lr": 5.135027085960195e-06, "elapsed_sec": 90918.4721519947, "step_time_sec": 8.232051448896527, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10934, "loss": 3.7251079082489014, "lr": 5.131027402513007e-06, "elapsed_sec": 90926.7061150074, "step_time_sec": 8.233810249948874, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10935, "loss": 3.808336019515991, "lr": 5.127087811182555e-06, "elapsed_sec": 90934.9413049221, "step_time_sec": 8.2350175549509, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10936, "loss": 3.7708499431610107, "lr": 5.123208314400198e-06, "elapsed_sec": 90943.17458081245, "step_time_sec": 8.233052895986475, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10937, "loss": 3.8612682819366455, "lr": 5.1193889145602054e-06, "elapsed_sec": 90951.40695095062, "step_time_sec": 8.232220072997734, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10938, "loss": 3.998798370361328, "lr": 5.115629614019757e-06, "elapsed_sec": 90959.64357829094, "step_time_sec": 8.23649896401912, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10939, "loss": 3.770989179611206, "lr": 5.1119304150989276e-06, "elapsed_sec": 90967.87989211082, "step_time_sec": 8.23614059202373, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10940, "loss": 3.7710347175598145, "lr": 5.108291320080716e-06, "elapsed_sec": 90976.11858797073, "step_time_sec": 8.238476095022634, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10941, "loss": 3.793602228164673, "lr": 5.104712331211016e-06, "elapsed_sec": 90984.35537719727, "step_time_sec": 8.236617178074084, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10942, "loss": 3.7930638790130615, "lr": 5.101193450698644e-06, "elapsed_sec": 90992.5951397419, "step_time_sec": 8.239575264044106, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10943, "loss": 3.7565338611602783, "lr": 5.097734680715291e-06, "elapsed_sec": 91000.83288431168, "step_time_sec": 8.237570745055564, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10944, "loss": 3.832892417907715, "lr": 5.094336023395594e-06, "elapsed_sec": 91009.07024288177, "step_time_sec": 8.237214277032763, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10945, "loss": 3.7450485229492188, "lr": 5.090997480837037e-06, "elapsed_sec": 91017.3031680584, "step_time_sec": 8.2328681119252, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10946, "loss": 3.7198374271392822, "lr": 5.0877190551000445e-06, "elapsed_sec": 91025.53595280647, "step_time_sec": 8.23243304505013, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10947, "loss": 3.7687628269195557, "lr": 5.084500748207922e-06, "elapsed_sec": 91033.76861953735, "step_time_sec": 8.232557689887471, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10948, "loss": 3.939650297164917, "lr": 5.081342562146865e-06, "elapsed_sec": 91042.00652766228, "step_time_sec": 8.237683319021016, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10949, "loss": 3.9149022102355957, "lr": 5.078244498865999e-06, "elapsed_sec": 91050.24004554749, "step_time_sec": 8.233342295978218, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10950, "loss": 3.871202230453491, "lr": 5.075206560277307e-06, "elapsed_sec": 91058.47267127037, "step_time_sec": 8.23248746804893, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10951, "loss": 3.9149861335754395, "lr": 5.072228748255677e-06, "elapsed_sec": 91066.71008610725, "step_time_sec": 8.23721679893788, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10952, "loss": 4.038631916046143, "lr": 5.069311064638911e-06, "elapsed_sec": 91074.94220685959, "step_time_sec": 8.231964672915637, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10953, "loss": 4.041018962860107, "lr": 5.066453511227651e-06, "elapsed_sec": 91083.1751961708, "step_time_sec": 8.232793650007807, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10954, "loss": 3.847428560256958, "lr": 5.063656089785486e-06, "elapsed_sec": 91091.4072008133, "step_time_sec": 8.231848457013257, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10955, "loss": 3.7365400791168213, "lr": 5.060918802038862e-06, "elapsed_sec": 91099.64243173599, "step_time_sec": 8.235114297014661, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10956, "loss": 3.7981865406036377, "lr": 5.058241649677124e-06, "elapsed_sec": 91107.8745675087, "step_time_sec": 8.231973970890976, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10957, "loss": 3.860525369644165, "lr": 5.055624634352495e-06, "elapsed_sec": 91116.1072678566, "step_time_sec": 8.232467586058192, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10958, "loss": 3.8220558166503906, "lr": 5.0530677576800825e-06, "elapsed_sec": 91124.33992242813, "step_time_sec": 8.232512070913799, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10959, "loss": 3.817570686340332, "lr": 5.0505710212378945e-06, "elapsed_sec": 91132.57364153862, "step_time_sec": 8.233585531939752, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10960, "loss": 3.8341519832611084, "lr": 5.048134426566825e-06, "elapsed_sec": 91140.80661654472, "step_time_sec": 8.23278116399888, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10961, "loss": 3.879969596862793, "lr": 5.045757975170617e-06, "elapsed_sec": 91149.0391318798, "step_time_sec": 8.23231426498387, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10962, "loss": 3.7376580238342285, "lr": 5.043441668515925e-06, "elapsed_sec": 91157.2722415924, "step_time_sec": 8.232989785959944, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10963, "loss": 3.8471741676330566, "lr": 5.041185508032289e-06, "elapsed_sec": 91165.5109398365, "step_time_sec": 8.238471488002688, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10964, "loss": 3.8777146339416504, "lr": 5.03898949511211e-06, "elapsed_sec": 91173.74599790573, "step_time_sec": 8.23488688794896, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10965, "loss": 3.7988028526306152, "lr": 5.0368536311106785e-06, "elapsed_sec": 91181.9827773571, "step_time_sec": 8.236606399063021, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10966, "loss": 3.868990898132324, "lr": 5.0347779173461585e-06, "elapsed_sec": 91190.21663045883, "step_time_sec": 8.23373473691754, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10967, "loss": 3.755276679992676, "lr": 5.032762355099598e-06, "elapsed_sec": 91198.45585656166, "step_time_sec": 8.238997527048923, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10968, "loss": 3.9171392917633057, "lr": 5.030806945614912e-06, "elapsed_sec": 91206.69158601761, "step_time_sec": 8.235574366990477, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10969, "loss": 3.6689722537994385, "lr": 5.028911690098913e-06, "elapsed_sec": 91214.92973470688, "step_time_sec": 8.238059201044962, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10970, "loss": 3.662024736404419, "lr": 5.02707658972126e-06, "elapsed_sec": 91223.16276836395, "step_time_sec": 8.23279288795311, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10971, "loss": 3.8332793712615967, "lr": 5.025301645614507e-06, "elapsed_sec": 91231.39522218704, "step_time_sec": 8.232280146097764, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10972, "loss": 3.8910739421844482, "lr": 5.023586858874089e-06, "elapsed_sec": 91239.62787723541, "step_time_sec": 8.232490033027716, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10973, "loss": 3.7566123008728027, "lr": 5.02193223055828e-06, "elapsed_sec": 91247.86049866676, "step_time_sec": 8.232491613947786, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10974, "loss": 3.849137306213379, "lr": 5.020337761688258e-06, "elapsed_sec": 91256.09291934967, "step_time_sec": 8.232217259937897, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10975, "loss": 3.818800210952759, "lr": 5.018803453248063e-06, "elapsed_sec": 91264.32626271248, "step_time_sec": 8.233166448073462, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10976, "loss": 4.030967712402344, "lr": 5.017329306184615e-06, "elapsed_sec": 91272.55859327316, "step_time_sec": 8.232182139065117, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10977, "loss": 3.7455551624298096, "lr": 5.015915321407684e-06, "elapsed_sec": 91280.79246807098, "step_time_sec": 8.233666334068403, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10978, "loss": 3.866455078125, "lr": 5.014561499789924e-06, "elapsed_sec": 91289.02427244186, "step_time_sec": 8.231589495087974, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10979, "loss": 3.796623945236206, "lr": 5.013267842166869e-06, "elapsed_sec": 91297.25699543953, "step_time_sec": 8.232586508034728, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10980, "loss": 3.7433409690856934, "lr": 5.0120343493369036e-06, "elapsed_sec": 91305.48957753181, "step_time_sec": 8.232390081975609, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10981, "loss": 3.8089704513549805, "lr": 5.010861022061296e-06, "elapsed_sec": 91313.72334265709, "step_time_sec": 8.233601888059638, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10982, "loss": 4.044589519500732, "lr": 5.009747861064161e-06, "elapsed_sec": 91321.95563435555, "step_time_sec": 8.232093075988814, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10983, "loss": 3.7483160495758057, "lr": 5.0086948670325e-06, "elapsed_sec": 91330.18850779533, "step_time_sec": 8.232714268029667, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10984, "loss": 4.021202087402344, "lr": 5.007702040616191e-06, "elapsed_sec": 91338.42600464821, "step_time_sec": 8.237300042062998, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10985, "loss": 3.8699474334716797, "lr": 5.006769382427945e-06, "elapsed_sec": 91346.66379928589, "step_time_sec": 8.237667776062153, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10986, "loss": 3.883211612701416, "lr": 5.0058968930433845e-06, "elapsed_sec": 91354.8990790844, "step_time_sec": 8.235051744035445, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10987, "loss": 3.9114863872528076, "lr": 5.005084573000952e-06, "elapsed_sec": 91363.13233971596, "step_time_sec": 8.23308651894331, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10988, "loss": 3.759514331817627, "lr": 5.00433242280199e-06, "elapsed_sec": 91371.36710739136, "step_time_sec": 8.234604569966905, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10989, "loss": 3.725457191467285, "lr": 5.003640442910685e-06, "elapsed_sec": 91379.60617160797, "step_time_sec": 8.238895561080426, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10990, "loss": 3.8843047618865967, "lr": 5.0030086337541105e-06, "elapsed_sec": 91387.84277296066, "step_time_sec": 8.236488798051141, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10991, "loss": 3.797151565551758, "lr": 5.002436995722186e-06, "elapsed_sec": 91396.0777130127, "step_time_sec": 8.234730947995558, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10992, "loss": 3.612673282623291, "lr": 5.001925529167705e-06, "elapsed_sec": 91404.30958938599, "step_time_sec": 8.231717237038538, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10993, "loss": 3.7283008098602295, "lr": 5.001474234406318e-06, "elapsed_sec": 91412.548620224, "step_time_sec": 8.238887661020271, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10994, "loss": 3.6881356239318848, "lr": 5.001083111716554e-06, "elapsed_sec": 91420.78062677383, "step_time_sec": 8.231837844010442, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10995, "loss": 3.8190412521362305, "lr": 5.000752161339792e-06, "elapsed_sec": 91429.01412415504, "step_time_sec": 8.23332073702477, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10996, "loss": 3.8264005184173584, "lr": 5.0004813834802835e-06, "elapsed_sec": 91437.24604058266, "step_time_sec": 8.231748637976125, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10997, "loss": 3.7417545318603516, "lr": 5.000270778305139e-06, "elapsed_sec": 91445.4804649353, "step_time_sec": 8.23420994693879, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10998, "loss": 3.799510955810547, "lr": 5.000120345944342e-06, "elapsed_sec": 91453.71212100983, "step_time_sec": 8.231523412978277, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10999, "loss": 3.7595748901367188, "lr": 5.000030086490724e-06, "elapsed_sec": 91461.94906616211, "step_time_sec": 8.236733599100262, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 11000, "loss": 3.664522171020508, "lr": 5e-06, "elapsed_sec": 91470.18182826042, "step_time_sec": 51.38879422505852, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": true, "validation_ran": true, "probe_ran": true, "probe_elapsed_sec": 0.9798749899491668, "warmup_excluded": false, "peak_vram_bytes": 13269960192}