{"step": 1, "loss": 10.539107322692871, "lr": 4.0000000000000003e-07, "elapsed_sec": 9.938241720199585, "step_time_sec": 9.882532201008871, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "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": 10.535860061645508, "lr": 8.000000000000001e-07, "elapsed_sec": 18.17241406440735, "step_time_sec": 8.233983158977935, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 10.503301620483398, "lr": 1.2000000000000002e-06, "elapsed_sec": 26.40058469772339, "step_time_sec": 8.227977244998328, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 10.496511459350586, "lr": 1.6000000000000001e-06, "elapsed_sec": 34.628124475479126, "step_time_sec": 8.227348979009548, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 10.501688003540039, "lr": 2e-06, "elapsed_sec": 42.85646605491638, "step_time_sec": 8.228160597995156, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 10.476656913757324, "lr": 2.4000000000000003e-06, "elapsed_sec": 51.084527015686035, "step_time_sec": 8.227923277008813, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 10.437271118164062, "lr": 2.8e-06, "elapsed_sec": 59.313827991485596, "step_time_sec": 8.229128197999671, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 10.430242538452148, "lr": 3.2000000000000003e-06, "elapsed_sec": 67.54275918006897, "step_time_sec": 8.228769823996117, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 10.397146224975586, "lr": 3.6000000000000003e-06, "elapsed_sec": 75.77271389961243, "step_time_sec": 8.229788208001992, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 10.373180389404297, "lr": 4e-06, "elapsed_sec": 84.00137257575989, "step_time_sec": 8.228522552992217, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 10.332941055297852, "lr": 4.4e-06, "elapsed_sec": 92.22867488861084, "step_time_sec": 8.227108271006728, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 10.282548904418945, "lr": 4.800000000000001e-06, "elapsed_sec": 100.45580172538757, "step_time_sec": 8.226931954995962, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 10.279932975769043, "lr": 5.200000000000001e-06, "elapsed_sec": 108.68355441093445, "step_time_sec": 8.227645845006919, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 10.248591423034668, "lr": 5.6e-06, "elapsed_sec": 116.91243982315063, "step_time_sec": 8.228710426017642, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 10.221306800842285, "lr": 6e-06, "elapsed_sec": 125.14092326164246, "step_time_sec": 8.228307215002133, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 10.148557662963867, "lr": 6.4000000000000006e-06, "elapsed_sec": 133.36742687225342, "step_time_sec": 8.226354526996147, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 10.154084205627441, "lr": 6.8e-06, "elapsed_sec": 141.59670686721802, "step_time_sec": 8.22904602799099, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 10.096874237060547, "lr": 7.2000000000000005e-06, "elapsed_sec": 149.82503247261047, "step_time_sec": 8.2281768550165, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 10.075173377990723, "lr": 7.6e-06, "elapsed_sec": 158.0532307624817, "step_time_sec": 8.228041018999647, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 10.071382522583008, "lr": 8e-06, "elapsed_sec": 166.2811644077301, "step_time_sec": 8.227787943003932, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 10.061782836914062, "lr": 8.400000000000001e-06, "elapsed_sec": 174.5090811252594, "step_time_sec": 8.22778239301988, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.982057571411133, "lr": 8.8e-06, "elapsed_sec": 182.73605155944824, "step_time_sec": 8.226779684016947, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.947029113769531, "lr": 9.2e-06, "elapsed_sec": 190.96308636665344, "step_time_sec": 8.226863919000607, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.949165344238281, "lr": 9.600000000000001e-06, "elapsed_sec": 199.19053840637207, "step_time_sec": 8.227236259001074, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.93150806427002, "lr": 1e-05, "elapsed_sec": 207.41911435127258, "step_time_sec": 8.228456440992886, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.870080947875977, "lr": 1.0400000000000002e-05, "elapsed_sec": 215.64768886566162, "step_time_sec": 8.22832029001438, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.855535507202148, "lr": 1.08e-05, "elapsed_sec": 223.8754551410675, "step_time_sec": 8.22765696098213, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.793105125427246, "lr": 1.12e-05, "elapsed_sec": 232.10305213928223, "step_time_sec": 8.227410405001137, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.782771110534668, "lr": 1.16e-05, "elapsed_sec": 240.33261513710022, "step_time_sec": 8.229448399011744, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.751171112060547, "lr": 1.2e-05, "elapsed_sec": 248.56165385246277, "step_time_sec": 8.228880984999705, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.738306045532227, "lr": 1.2400000000000002e-05, "elapsed_sec": 256.78888726234436, "step_time_sec": 8.227032819995657, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.699894905090332, "lr": 1.2800000000000001e-05, "elapsed_sec": 265.01672530174255, "step_time_sec": 8.227728010009741, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.663061141967773, "lr": 1.32e-05, "elapsed_sec": 273.2444624900818, "step_time_sec": 8.227573499985738, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.71033000946045, "lr": 1.36e-05, "elapsed_sec": 281.4715209007263, "step_time_sec": 8.226911770994775, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.625850677490234, "lr": 1.4e-05, "elapsed_sec": 289.6979866027832, "step_time_sec": 8.226277524023317, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.67159366607666, "lr": 1.4400000000000001e-05, "elapsed_sec": 297.92771315574646, "step_time_sec": 8.229591938987141, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.53085708618164, "lr": 1.48e-05, "elapsed_sec": 306.1561007499695, "step_time_sec": 8.228164239000762, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.591432571411133, "lr": 1.52e-05, "elapsed_sec": 314.3843524456024, "step_time_sec": 8.22814195099636, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.477296829223633, "lr": 1.56e-05, "elapsed_sec": 322.6120538711548, "step_time_sec": 8.227544766006758, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.507052421569824, "lr": 1.6e-05, "elapsed_sec": 330.8406009674072, "step_time_sec": 8.228426565008704, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.47337818145752, "lr": 1.6400000000000002e-05, "elapsed_sec": 339.0694143772125, "step_time_sec": 8.228588525002124, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.465594291687012, "lr": 1.6800000000000002e-05, "elapsed_sec": 347.2983136177063, "step_time_sec": 8.228745789005188, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.565919876098633, "lr": 1.72e-05, "elapsed_sec": 355.52542757987976, "step_time_sec": 8.226957421022234, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.463726997375488, "lr": 1.76e-05, "elapsed_sec": 363.75494146347046, "step_time_sec": 8.229367069026921, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.451949119567871, "lr": 1.8e-05, "elapsed_sec": 371.98210406303406, "step_time_sec": 8.226967513997806, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.364481925964355, "lr": 1.84e-05, "elapsed_sec": 380.2104730606079, "step_time_sec": 8.22824935198878, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.370523452758789, "lr": 1.88e-05, "elapsed_sec": 388.43832659721375, "step_time_sec": 8.227726924000308, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.407183647155762, "lr": 1.9200000000000003e-05, "elapsed_sec": 396.6657862663269, "step_time_sec": 8.227269993978553, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.341446876525879, "lr": 1.96e-05, "elapsed_sec": 404.89419174194336, "step_time_sec": 8.228235190006671, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.311203956604004, "lr": 2e-05, "elapsed_sec": 413.12312841415405, "step_time_sec": 8.228746195003623, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.307658195495605, "lr": 2.04e-05, "elapsed_sec": 421.35253405570984, "step_time_sec": 8.229271764983423, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.234817504882812, "lr": 2.0800000000000004e-05, "elapsed_sec": 429.5819480419159, "step_time_sec": 8.229221543006133, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.281026840209961, "lr": 2.12e-05, "elapsed_sec": 437.81028151512146, "step_time_sec": 8.228210138971917, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.354969024658203, "lr": 2.16e-05, "elapsed_sec": 446.0398464202881, "step_time_sec": 8.22938400998828, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.473270416259766, "lr": 2.2000000000000003e-05, "elapsed_sec": 454.26811170578003, "step_time_sec": 8.228103075001854, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.259379386901855, "lr": 2.24e-05, "elapsed_sec": 462.4979655742645, "step_time_sec": 8.229663873004029, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.250909805297852, "lr": 2.2800000000000002e-05, "elapsed_sec": 470.72585821151733, "step_time_sec": 8.227704051008914, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.15527057647705, "lr": 2.32e-05, "elapsed_sec": 478.95394134521484, "step_time_sec": 8.227909973997157, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.223548889160156, "lr": 2.36e-05, "elapsed_sec": 487.1814510822296, "step_time_sec": 8.227357939991634, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.254575729370117, "lr": 2.4e-05, "elapsed_sec": 495.4103629589081, "step_time_sec": 8.228755346994149, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.146458625793457, "lr": 2.44e-05, "elapsed_sec": 503.63934874534607, "step_time_sec": 8.228833354020026, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.255533218383789, "lr": 2.4800000000000003e-05, "elapsed_sec": 511.86804604530334, "step_time_sec": 8.228528347011888, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.136784553527832, "lr": 2.52e-05, "elapsed_sec": 520.0970077514648, "step_time_sec": 8.228772069996921, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.154542922973633, "lr": 2.5600000000000002e-05, "elapsed_sec": 528.3252789974213, "step_time_sec": 8.228154264972545, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.043829917907715, "lr": 2.6000000000000002e-05, "elapsed_sec": 536.5518960952759, "step_time_sec": 8.226433992007514, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.071345329284668, "lr": 2.64e-05, "elapsed_sec": 544.7799417972565, "step_time_sec": 8.227872820978519, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.113154411315918, "lr": 2.68e-05, "elapsed_sec": 553.0068988800049, "step_time_sec": 8.226764396007638, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.123502731323242, "lr": 2.72e-05, "elapsed_sec": 561.2351982593536, "step_time_sec": 8.228149786009453, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.100492477416992, "lr": 2.7600000000000003e-05, "elapsed_sec": 569.4623696804047, "step_time_sec": 8.227031368005555, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.050661087036133, "lr": 2.8e-05, "elapsed_sec": 577.6904747486115, "step_time_sec": 8.22794578500907, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.965362548828125, "lr": 2.8400000000000003e-05, "elapsed_sec": 585.9175674915314, "step_time_sec": 8.226940724998713, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.9718656539917, "lr": 2.8800000000000002e-05, "elapsed_sec": 594.1461131572723, "step_time_sec": 8.228431298979558, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.027297019958496, "lr": 2.92e-05, "elapsed_sec": 602.3761920928955, "step_time_sec": 8.229842600994743, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.99364948272705, "lr": 2.96e-05, "elapsed_sec": 610.6036643981934, "step_time_sec": 8.227347633015597, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.917802810668945, "lr": 3e-05, "elapsed_sec": 618.8315300941467, "step_time_sec": 8.227690527011873, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.879055976867676, "lr": 3.04e-05, "elapsed_sec": 627.0591688156128, "step_time_sec": 8.227440636022948, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.914271354675293, "lr": 3.08e-05, "elapsed_sec": 635.2876262664795, "step_time_sec": 8.228326386015397, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.928272247314453, "lr": 3.12e-05, "elapsed_sec": 643.5164823532104, "step_time_sec": 8.228696694015525, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.876237869262695, "lr": 3.16e-05, "elapsed_sec": 651.7439455986023, "step_time_sec": 8.227294993004762, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.801451683044434, "lr": 3.2e-05, "elapsed_sec": 659.971197605133, "step_time_sec": 8.227098361996468, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.835479736328125, "lr": 3.24e-05, "elapsed_sec": 668.1999478340149, "step_time_sec": 8.228560830000788, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.833176612854004, "lr": 3.2800000000000004e-05, "elapsed_sec": 676.428254365921, "step_time_sec": 8.228149367001606, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.783199310302734, "lr": 3.32e-05, "elapsed_sec": 684.6568856239319, "step_time_sec": 8.228498785989359, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.785578727722168, "lr": 3.3600000000000004e-05, "elapsed_sec": 692.8837337493896, "step_time_sec": 8.226656553015346, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.788888931274414, "lr": 3.4e-05, "elapsed_sec": 701.1104884147644, "step_time_sec": 8.226610653015086, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.792500495910645, "lr": 3.44e-05, "elapsed_sec": 709.3372163772583, "step_time_sec": 8.226559139991878, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.783136367797852, "lr": 3.4800000000000006e-05, "elapsed_sec": 717.5641860961914, "step_time_sec": 8.226852063991828, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.684346199035645, "lr": 3.52e-05, "elapsed_sec": 725.7902190685272, "step_time_sec": 8.225831567018759, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.713064193725586, "lr": 3.56e-05, "elapsed_sec": 734.0175297260284, "step_time_sec": 8.227171245991485, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.684280395507812, "lr": 3.6e-05, "elapsed_sec": 742.2439622879028, "step_time_sec": 8.226198827993358, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.659466743469238, "lr": 3.6400000000000004e-05, "elapsed_sec": 750.4728696346283, "step_time_sec": 8.228781616984634, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.60403823852539, "lr": 3.68e-05, "elapsed_sec": 758.7013294696808, "step_time_sec": 8.228293921012664, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.6467924118042, "lr": 3.72e-05, "elapsed_sec": 766.9293041229248, "step_time_sec": 8.22779516299488, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.564494132995605, "lr": 3.76e-05, "elapsed_sec": 775.1584181785583, "step_time_sec": 8.228965256974334, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.58819580078125, "lr": 3.8e-05, "elapsed_sec": 783.3874189853668, "step_time_sec": 8.22883819101844, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.488486289978027, "lr": 3.8400000000000005e-05, "elapsed_sec": 791.615097284317, "step_time_sec": 8.227495420986088, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.601044654846191, "lr": 3.88e-05, "elapsed_sec": 799.8422968387604, "step_time_sec": 8.2270620620111, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.521957397460938, "lr": 3.92e-05, "elapsed_sec": 808.0706567764282, "step_time_sec": 8.228199170989683, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.550530433654785, "lr": 3.96e-05, "elapsed_sec": 816.2998299598694, "step_time_sec": 8.229021489998559, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.490187644958496, "lr": 4e-05, "elapsed_sec": 824.5280969142914, "step_time_sec": 8.228132624004502, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.43508529663086, "lr": 4.0400000000000006e-05, "elapsed_sec": 832.7553880214691, "step_time_sec": 8.227064855018398, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.484140396118164, "lr": 4.08e-05, "elapsed_sec": 840.9818501472473, "step_time_sec": 8.226254109991714, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.457494735717773, "lr": 4.12e-05, "elapsed_sec": 849.2091951370239, "step_time_sec": 8.227195129991742, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.409165382385254, "lr": 4.160000000000001e-05, "elapsed_sec": 857.4378645420074, "step_time_sec": 8.228468367015012, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.418153762817383, "lr": 4.2000000000000004e-05, "elapsed_sec": 865.6655433177948, "step_time_sec": 8.227526147005847, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.382179260253906, "lr": 4.24e-05, "elapsed_sec": 873.8940553665161, "step_time_sec": 8.228398088976974, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.491363525390625, "lr": 4.2800000000000004e-05, "elapsed_sec": 882.1231973171234, "step_time_sec": 8.22893366299104, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.335529327392578, "lr": 4.32e-05, "elapsed_sec": 890.3526582717896, "step_time_sec": 8.2292755600065, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.357977867126465, "lr": 4.36e-05, "elapsed_sec": 898.5827786922455, "step_time_sec": 8.229969392996281, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.285054206848145, "lr": 4.4000000000000006e-05, "elapsed_sec": 906.8123557567596, "step_time_sec": 8.22940325899981, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.328099250793457, "lr": 4.44e-05, "elapsed_sec": 915.0414357185364, "step_time_sec": 8.228924249997362, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.326068878173828, "lr": 4.48e-05, "elapsed_sec": 923.2703275680542, "step_time_sec": 8.228729492984712, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.277432441711426, "lr": 4.52e-05, "elapsed_sec": 931.4992551803589, "step_time_sec": 8.228763824008638, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.364901542663574, "lr": 4.5600000000000004e-05, "elapsed_sec": 939.7280297279358, "step_time_sec": 8.228609603014775, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.313380241394043, "lr": 4.6e-05, "elapsed_sec": 947.9571285247803, "step_time_sec": 8.22893638900132, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.34488582611084, "lr": 4.64e-05, "elapsed_sec": 956.1879088878632, "step_time_sec": 8.230629024998052, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.173175811767578, "lr": 4.68e-05, "elapsed_sec": 964.4178082942963, "step_time_sec": 8.229724757984513, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.26225757598877, "lr": 4.72e-05, "elapsed_sec": 972.6471076011658, "step_time_sec": 8.229135430010501, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.261497497558594, "lr": 4.7600000000000005e-05, "elapsed_sec": 980.8761739730835, "step_time_sec": 8.228907837998122, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.281538009643555, "lr": 4.8e-05, "elapsed_sec": 989.1080169677734, "step_time_sec": 8.231664199993247, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.19478702545166, "lr": 4.8400000000000004e-05, "elapsed_sec": 997.3369867801666, "step_time_sec": 8.228818012983538, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.307666778564453, "lr": 4.88e-05, "elapsed_sec": 1005.5736405849457, "step_time_sec": 8.23647436601459, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.231106758117676, "lr": 4.92e-05, "elapsed_sec": 1013.8042948246002, "step_time_sec": 8.23047598800622, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.104755401611328, "lr": 4.9600000000000006e-05, "elapsed_sec": 1022.0356676578522, "step_time_sec": 8.231239520013332, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.123236656188965, "lr": 5e-05, "elapsed_sec": 1030.2651808261871, "step_time_sec": 8.229332029004581, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.140379905700684, "lr": 5.04e-05, "elapsed_sec": 1038.4955718517303, "step_time_sec": 8.230252462992212, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.187718391418457, "lr": 5.08e-05, "elapsed_sec": 1046.7246901988983, "step_time_sec": 8.228868484002305, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.131364822387695, "lr": 5.1200000000000004e-05, "elapsed_sec": 1054.954338312149, "step_time_sec": 8.229502484988188, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.154012680053711, "lr": 5.16e-05, "elapsed_sec": 1063.1842272281647, "step_time_sec": 8.229721631010761, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.031145095825195, "lr": 5.2000000000000004e-05, "elapsed_sec": 1071.4145362377167, "step_time_sec": 8.23018237602082, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.079499244689941, "lr": 5.24e-05, "elapsed_sec": 1079.6442458629608, "step_time_sec": 8.22950695900363, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.283516883850098, "lr": 5.28e-05, "elapsed_sec": 1087.8779292106628, "step_time_sec": 8.233509209996555, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.101082801818848, "lr": 5.3200000000000006e-05, "elapsed_sec": 1096.1122934818268, "step_time_sec": 8.234181595995324, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.105387687683105, "lr": 5.36e-05, "elapsed_sec": 1104.3449573516846, "step_time_sec": 8.232475406985031, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.023136138916016, "lr": 5.4e-05, "elapsed_sec": 1112.5771515369415, "step_time_sec": 8.232058813999174, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.049627304077148, "lr": 5.44e-05, "elapsed_sec": 1120.8073709011078, "step_time_sec": 8.230034354986856, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.074769020080566, "lr": 5.4800000000000004e-05, "elapsed_sec": 1129.0374047756195, "step_time_sec": 8.229864883003756, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.065865516662598, "lr": 5.520000000000001e-05, "elapsed_sec": 1137.273466348648, "step_time_sec": 8.235890194016974, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.00903034210205, "lr": 5.56e-05, "elapsed_sec": 1145.5027823448181, "step_time_sec": 8.229154850007035, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.964427471160889, "lr": 5.6e-05, "elapsed_sec": 1153.7364761829376, "step_time_sec": 8.23348826399888, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.00632381439209, "lr": 5.640000000000001e-05, "elapsed_sec": 1161.9655003547668, "step_time_sec": 8.228825380996568, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.98394250869751, "lr": 5.6800000000000005e-05, "elapsed_sec": 1170.199434041977, "step_time_sec": 8.233790492988192, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.992773056030273, "lr": 5.72e-05, "elapsed_sec": 1178.4332225322723, "step_time_sec": 8.233652638999047, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.960933208465576, "lr": 5.7600000000000004e-05, "elapsed_sec": 1186.6650087833405, "step_time_sec": 8.231577175989514, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.957793712615967, "lr": 5.8e-05, "elapsed_sec": 1194.8946063518524, "step_time_sec": 8.22945287800394, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.868411064147949, "lr": 5.84e-05, "elapsed_sec": 1203.1260092258453, "step_time_sec": 8.231233244994655, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.928884029388428, "lr": 5.8800000000000006e-05, "elapsed_sec": 1211.3590741157532, "step_time_sec": 8.232917674002238, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.978039741516113, "lr": 5.92e-05, "elapsed_sec": 1219.5884399414062, "step_time_sec": 8.229205300012836, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 1227.8172104358673, "step_time_sec": 8.228611401980743, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.006260871887207, "lr": 6e-05, "elapsed_sec": 1236.0483763217926, "step_time_sec": 8.23104102400248, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.969548225402832, "lr": 6.0400000000000004e-05, "elapsed_sec": 1244.2790279388428, "step_time_sec": 8.230479691992514, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.917298316955566, "lr": 6.08e-05, "elapsed_sec": 1252.5075364112854, "step_time_sec": 8.228359782980988, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.944379806518555, "lr": 6.120000000000001e-05, "elapsed_sec": 1260.7372722625732, "step_time_sec": 8.229529545991682, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.9007792472839355, "lr": 6.16e-05, "elapsed_sec": 1268.9661660194397, "step_time_sec": 8.228757285018219, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.9255242347717285, "lr": 6.2e-05, "elapsed_sec": 1277.194829940796, "step_time_sec": 8.228496363997692, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.946313858032227, "lr": 6.24e-05, "elapsed_sec": 1285.423888683319, "step_time_sec": 8.228912488004426, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.921523094177246, "lr": 6.280000000000001e-05, "elapsed_sec": 1293.6547253131866, "step_time_sec": 8.230637778993696, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.970492362976074, "lr": 6.32e-05, "elapsed_sec": 1301.8888931274414, "step_time_sec": 8.234006358019542, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.870519638061523, "lr": 6.36e-05, "elapsed_sec": 1310.1214053630829, "step_time_sec": 8.232375253981445, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.875805377960205, "lr": 6.4e-05, "elapsed_sec": 1318.3521447181702, "step_time_sec": 8.230564184981631, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.878680229187012, "lr": 6.44e-05, "elapsed_sec": 1326.581868171692, "step_time_sec": 8.229590684000868, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.951796531677246, "lr": 6.48e-05, "elapsed_sec": 1334.8121209144592, "step_time_sec": 8.230040494003333, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.85944128036499, "lr": 6.520000000000001e-05, "elapsed_sec": 1343.0412514209747, "step_time_sec": 8.228970341006061, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.844451427459717, "lr": 6.560000000000001e-05, "elapsed_sec": 1351.2708950042725, "step_time_sec": 8.229510534001747, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.8650665283203125, "lr": 6.6e-05, "elapsed_sec": 1359.5024027824402, "step_time_sec": 8.23131842300063, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.788283348083496, "lr": 6.64e-05, "elapsed_sec": 1367.7322971820831, "step_time_sec": 8.229755506006768, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.7577667236328125, "lr": 6.68e-05, "elapsed_sec": 1375.960970878601, "step_time_sec": 8.228507138992427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.774542808532715, "lr": 6.720000000000001e-05, "elapsed_sec": 1384.1903176307678, "step_time_sec": 8.229185105010401, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.743876934051514, "lr": 6.76e-05, "elapsed_sec": 1392.4198820590973, "step_time_sec": 8.229376774019329, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.801985263824463, "lr": 6.8e-05, "elapsed_sec": 1400.6513178348541, "step_time_sec": 8.231246325012762, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.757818698883057, "lr": 6.840000000000001e-05, "elapsed_sec": 1408.8804624080658, "step_time_sec": 8.229035241005477, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.917051315307617, "lr": 6.88e-05, "elapsed_sec": 1417.1136560440063, "step_time_sec": 8.23302217599121, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.766968250274658, "lr": 6.92e-05, "elapsed_sec": 1425.3426938056946, "step_time_sec": 8.228834779991303, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.7254838943481445, "lr": 6.960000000000001e-05, "elapsed_sec": 1433.571929216385, "step_time_sec": 8.22907407200546, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.79286003112793, "lr": 7.000000000000001e-05, "elapsed_sec": 1441.8052995204926, "step_time_sec": 8.233195267006522, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.748465538024902, "lr": 7.04e-05, "elapsed_sec": 1450.033995628357, "step_time_sec": 8.228546166996239, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.788219451904297, "lr": 7.08e-05, "elapsed_sec": 1458.2625212669373, "step_time_sec": 8.228351538011339, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.8216657638549805, "lr": 7.12e-05, "elapsed_sec": 1466.4914186000824, "step_time_sec": 8.228806908999104, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.770859241485596, "lr": 7.159999999999999e-05, "elapsed_sec": 1474.7205834388733, "step_time_sec": 8.228954810008872, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.69477653503418, "lr": 7.2e-05, "elapsed_sec": 1482.9533214569092, "step_time_sec": 8.232582344004186, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.7531232833862305, "lr": 7.240000000000001e-05, "elapsed_sec": 1491.182154417038, "step_time_sec": 8.228700360981748, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.7076735496521, "lr": 7.280000000000001e-05, "elapsed_sec": 1499.4118824005127, "step_time_sec": 8.229533081990667, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.655876636505127, "lr": 7.32e-05, "elapsed_sec": 1507.6414070129395, "step_time_sec": 8.229375519003952, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.69143009185791, "lr": 7.36e-05, "elapsed_sec": 1515.8718965053558, "step_time_sec": 8.230318195011932, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.6777238845825195, "lr": 7.400000000000001e-05, "elapsed_sec": 1524.1044373512268, "step_time_sec": 8.232365680974908, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.641297340393066, "lr": 7.44e-05, "elapsed_sec": 1532.3334505558014, "step_time_sec": 8.228849111997988, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.703188896179199, "lr": 7.48e-05, "elapsed_sec": 1540.564547777176, "step_time_sec": 8.230919466004707, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.728152751922607, "lr": 7.52e-05, "elapsed_sec": 1548.7964460849762, "step_time_sec": 8.231789244979154, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.5579962730407715, "lr": 7.56e-05, "elapsed_sec": 1557.0260827541351, "step_time_sec": 8.229460091009969, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.628966331481934, "lr": 7.6e-05, "elapsed_sec": 1565.2558405399323, "step_time_sec": 8.22953892001533, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.7179083824157715, "lr": 7.640000000000001e-05, "elapsed_sec": 1573.4843156337738, "step_time_sec": 8.228355842991732, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.630806922912598, "lr": 7.680000000000001e-05, "elapsed_sec": 1581.71466588974, "step_time_sec": 8.230156176985474, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.637725353240967, "lr": 7.72e-05, "elapsed_sec": 1589.943813085556, "step_time_sec": 8.228964565001661, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.670409202575684, "lr": 7.76e-05, "elapsed_sec": 1598.1735072135925, "step_time_sec": 8.22951621899847, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.5541839599609375, "lr": 7.8e-05, "elapsed_sec": 1606.4029259681702, "step_time_sec": 8.229309802001808, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.639224529266357, "lr": 7.84e-05, "elapsed_sec": 1614.6358211040497, "step_time_sec": 8.23270467601833, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.56196403503418, "lr": 7.88e-05, "elapsed_sec": 1622.8710792064667, "step_time_sec": 8.235087887995178, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.61405611038208, "lr": 7.92e-05, "elapsed_sec": 1631.1032180786133, "step_time_sec": 8.23201977001736, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.60678768157959, "lr": 7.960000000000001e-05, "elapsed_sec": 1639.3328313827515, "step_time_sec": 8.229439553979319, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.529764175415039, "lr": 8e-05, "elapsed_sec": 1647.5613594055176, "step_time_sec": 8.228344421018846, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.666224002838135, "lr": 8.04e-05, "elapsed_sec": 1655.794646024704, "step_time_sec": 8.233115431008628, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.556634426116943, "lr": 8.080000000000001e-05, "elapsed_sec": 1664.0240590572357, "step_time_sec": 8.229255698999623, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.530052185058594, "lr": 8.120000000000001e-05, "elapsed_sec": 1672.254073381424, "step_time_sec": 8.229842066997662, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.537210941314697, "lr": 8.16e-05, "elapsed_sec": 1680.4835617542267, "step_time_sec": 8.22934197500581, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.4870829582214355, "lr": 8.2e-05, "elapsed_sec": 1688.7125492095947, "step_time_sec": 8.228860082977917, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.483022212982178, "lr": 8.24e-05, "elapsed_sec": 1696.945986032486, "step_time_sec": 8.233233008999377, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.466499328613281, "lr": 8.28e-05, "elapsed_sec": 1705.17489695549, "step_time_sec": 8.228772283007856, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.517866611480713, "lr": 8.320000000000002e-05, "elapsed_sec": 1713.4087476730347, "step_time_sec": 8.233678877004422, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.5045166015625, "lr": 8.360000000000001e-05, "elapsed_sec": 1721.6374416351318, "step_time_sec": 8.2285283189849, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.443332672119141, "lr": 8.400000000000001e-05, "elapsed_sec": 1729.8665370941162, "step_time_sec": 8.228944417001912, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.628190994262695, "lr": 8.44e-05, "elapsed_sec": 1738.0993392467499, "step_time_sec": 8.23262363098911, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.365503787994385, "lr": 8.48e-05, "elapsed_sec": 1746.3286805152893, "step_time_sec": 8.229239792010048, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.424105167388916, "lr": 8.52e-05, "elapsed_sec": 1754.5622289180756, "step_time_sec": 8.233337838988518, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.493538856506348, "lr": 8.560000000000001e-05, "elapsed_sec": 1762.7921695709229, "step_time_sec": 8.229772619990399, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.373263359069824, "lr": 8.6e-05, "elapsed_sec": 1771.021583557129, "step_time_sec": 8.229229375981959, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.358706951141357, "lr": 8.64e-05, "elapsed_sec": 1779.2509460449219, "step_time_sec": 8.229222755995579, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.43594217300415, "lr": 8.68e-05, "elapsed_sec": 1787.4808633327484, "step_time_sec": 8.22972766897874, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.5597076416015625, "lr": 8.72e-05, "elapsed_sec": 1795.7107408046722, "step_time_sec": 8.22971267599496, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.629430770874023, "lr": 8.76e-05, "elapsed_sec": 1803.9440433979034, "step_time_sec": 8.233134781010449, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.497067451477051, "lr": 8.800000000000001e-05, "elapsed_sec": 1812.1738831996918, "step_time_sec": 8.229676918999758, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.414957523345947, "lr": 8.840000000000001e-05, "elapsed_sec": 1820.4033527374268, "step_time_sec": 8.229296532023, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.376187801361084, "lr": 8.88e-05, "elapsed_sec": 1828.6339437961578, "step_time_sec": 8.230490493995603, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.359637260437012, "lr": 8.92e-05, "elapsed_sec": 1836.86541223526, "step_time_sec": 8.23125738199451, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.444538593292236, "lr": 8.96e-05, "elapsed_sec": 1845.095998287201, "step_time_sec": 8.230436996993376, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.477090835571289, "lr": 9e-05, "elapsed_sec": 1853.3279721736908, "step_time_sec": 8.231850152020343, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.3540449142456055, "lr": 9.04e-05, "elapsed_sec": 1861.5603206157684, "step_time_sec": 8.232161177002126, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.3545379638671875, "lr": 9.080000000000001e-05, "elapsed_sec": 1869.789499282837, "step_time_sec": 8.228996067016851, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.333803176879883, "lr": 9.120000000000001e-05, "elapsed_sec": 1878.0190880298615, "step_time_sec": 8.229418895003619, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.390617847442627, "lr": 9.16e-05, "elapsed_sec": 1886.2485752105713, "step_time_sec": 8.22934726101812, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.367952823638916, "lr": 9.2e-05, "elapsed_sec": 1894.4786441326141, "step_time_sec": 8.229851381009212, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.371184825897217, "lr": 9.240000000000001e-05, "elapsed_sec": 1902.7080132961273, "step_time_sec": 8.229225652001332, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.323879718780518, "lr": 9.28e-05, "elapsed_sec": 1910.9384095668793, "step_time_sec": 8.230230019980809, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.308291912078857, "lr": 9.32e-05, "elapsed_sec": 1919.1709971427917, "step_time_sec": 8.232444518012926, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.387063980102539, "lr": 9.36e-05, "elapsed_sec": 1927.4049050807953, "step_time_sec": 8.2337525979965, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.341152191162109, "lr": 9.4e-05, "elapsed_sec": 1935.6353752613068, "step_time_sec": 8.230238350020954, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.317520618438721, "lr": 9.44e-05, "elapsed_sec": 1943.8658516407013, "step_time_sec": 8.230320461996598, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.236241817474365, "lr": 9.480000000000001e-05, "elapsed_sec": 1952.098025560379, "step_time_sec": 8.232010864012409, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.2427473068237305, "lr": 9.520000000000001e-05, "elapsed_sec": 1960.3270890712738, "step_time_sec": 8.228923881979426, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.327444553375244, "lr": 9.56e-05, "elapsed_sec": 1968.5573551654816, "step_time_sec": 8.23016286100028, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.2435622215271, "lr": 9.6e-05, "elapsed_sec": 1976.7883696556091, "step_time_sec": 8.230861227988498, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.230710983276367, "lr": 9.64e-05, "elapsed_sec": 1985.0188975334167, "step_time_sec": 8.230323487019632, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.412811279296875, "lr": 9.680000000000001e-05, "elapsed_sec": 1993.252816438675, "step_time_sec": 8.233726466016378, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.270981311798096, "lr": 9.72e-05, "elapsed_sec": 2001.4870154857635, "step_time_sec": 8.23407924597268, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.291118621826172, "lr": 9.76e-05, "elapsed_sec": 2009.7166945934296, "step_time_sec": 8.22946567699546, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.368896961212158, "lr": 9.800000000000001e-05, "elapsed_sec": 2017.946724653244, "step_time_sec": 8.229893187002745, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.210421562194824, "lr": 9.84e-05, "elapsed_sec": 2026.1772530078888, "step_time_sec": 8.23032892800984, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.238697528839111, "lr": 9.88e-05, "elapsed_sec": 2034.406763792038, "step_time_sec": 8.229375059978338, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.263420104980469, "lr": 9.920000000000001e-05, "elapsed_sec": 2042.6373007297516, "step_time_sec": 8.230322135001188, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.154346942901611, "lr": 9.960000000000001e-05, "elapsed_sec": 2050.8704924583435, "step_time_sec": 8.23302095499821, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.1789727210998535, "lr": 0.0001, "elapsed_sec": 2059.1040897369385, "step_time_sec": 8.233447512990097, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.254812240600586, "lr": 0.0001004, "elapsed_sec": 2067.335854768753, "step_time_sec": 8.231592437979998, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.235644340515137, "lr": 0.0001008, "elapsed_sec": 2075.567863225937, "step_time_sec": 8.231826108996756, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.2438578605651855, "lr": 0.0001012, "elapsed_sec": 2083.7996554374695, "step_time_sec": 8.231647920998512, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.1992998123168945, "lr": 0.0001016, "elapsed_sec": 2092.0299167633057, "step_time_sec": 8.230095587001415, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.234037399291992, "lr": 0.00010200000000000001, "elapsed_sec": 2100.2609214782715, "step_time_sec": 8.23083679197589, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.191378116607666, "lr": 0.00010240000000000001, "elapsed_sec": 2108.4918069839478, "step_time_sec": 8.2307746099832, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.230913162231445, "lr": 0.0001028, "elapsed_sec": 2116.7239289283752, "step_time_sec": 8.23191296501318, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.203978538513184, "lr": 0.0001032, "elapsed_sec": 2124.958427667618, "step_time_sec": 8.234374298976036, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.157383918762207, "lr": 0.00010360000000000001, "elapsed_sec": 2133.188116312027, "step_time_sec": 8.229492680024123, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.105503559112549, "lr": 0.00010400000000000001, "elapsed_sec": 2141.420086145401, "step_time_sec": 8.231811206002021, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.242863655090332, "lr": 0.0001044, "elapsed_sec": 2149.649574279785, "step_time_sec": 8.22930120502133, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.2115678787231445, "lr": 0.0001048, "elapsed_sec": 2157.8817896842957, "step_time_sec": 8.232087784010218, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.2273125648498535, "lr": 0.0001052, "elapsed_sec": 2166.1161737442017, "step_time_sec": 8.234226585016586, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.256625652313232, "lr": 0.0001056, "elapsed_sec": 2174.351213693619, "step_time_sec": 8.234840506018372, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.2274088859558105, "lr": 0.00010600000000000002, "elapsed_sec": 2182.5853440761566, "step_time_sec": 8.23397951599327, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.260874271392822, "lr": 0.00010640000000000001, "elapsed_sec": 2190.8184204101562, "step_time_sec": 8.232900929026073, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.1111321449279785, "lr": 0.00010680000000000001, "elapsed_sec": 2199.053811073303, "step_time_sec": 8.23526825098088, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.204519271850586, "lr": 0.0001072, "elapsed_sec": 2207.288024663925, "step_time_sec": 8.234051830979297, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.156832218170166, "lr": 0.0001076, "elapsed_sec": 2215.522755384445, "step_time_sec": 8.234561029996257, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.285793781280518, "lr": 0.000108, "elapsed_sec": 2223.753278493881, "step_time_sec": 8.230340223002713, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.29644250869751, "lr": 0.0001084, "elapsed_sec": 2231.9824702739716, "step_time_sec": 8.229009091999615, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.144976615905762, "lr": 0.0001088, "elapsed_sec": 2240.21258687973, "step_time_sec": 8.229985953017604, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.206099510192871, "lr": 0.00010920000000000001, "elapsed_sec": 2248.4426712989807, "step_time_sec": 8.229887683002744, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.077547550201416, "lr": 0.00010960000000000001, "elapsed_sec": 2256.677523612976, "step_time_sec": 8.23473604398896, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.120008945465088, "lr": 0.00011, "elapsed_sec": 2264.9118840694427, "step_time_sec": 8.23416838599951, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.145639419555664, "lr": 0.00011040000000000001, "elapsed_sec": 2273.141885995865, "step_time_sec": 8.2298214820039, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.116590976715088, "lr": 0.00011080000000000001, "elapsed_sec": 2281.3772506713867, "step_time_sec": 8.23523360901163, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.149325370788574, "lr": 0.0001112, "elapsed_sec": 2289.613400697708, "step_time_sec": 8.235942425002577, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.08897066116333, "lr": 0.0001116, "elapsed_sec": 2297.8473179340363, "step_time_sec": 8.233782822993817, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.057685375213623, "lr": 0.000112, "elapsed_sec": 2306.077873468399, "step_time_sec": 8.230391723976936, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.0908308029174805, "lr": 0.0001124, "elapsed_sec": 2314.3074510097504, "step_time_sec": 8.229416688991478, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.123631000518799, "lr": 0.00011280000000000002, "elapsed_sec": 2322.5431096553802, "step_time_sec": 8.235491738014389, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.009051322937012, "lr": 0.00011320000000000001, "elapsed_sec": 2330.778353691101, "step_time_sec": 8.235129882988986, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.0414509773254395, "lr": 0.00011360000000000001, "elapsed_sec": 2339.013827562332, "step_time_sec": 8.235342664993368, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.2224578857421875, "lr": 0.000114, "elapsed_sec": 2347.247203350067, "step_time_sec": 8.233163642988075, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.051135063171387, "lr": 0.0001144, "elapsed_sec": 2355.483179330826, "step_time_sec": 8.235806005017366, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.1838860511779785, "lr": 0.0001148, "elapsed_sec": 2363.7148101329803, "step_time_sec": 8.231462713010842, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.095859050750732, "lr": 0.00011520000000000001, "elapsed_sec": 2371.9507763385773, "step_time_sec": 8.235836708015995, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.079031467437744, "lr": 0.0001156, "elapsed_sec": 2380.1858270168304, "step_time_sec": 8.23487108500558, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.072887420654297, "lr": 0.000116, "elapsed_sec": 2388.4231429100037, "step_time_sec": 8.23715633500251, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.0125932693481445, "lr": 0.0001164, "elapsed_sec": 2396.6530752182007, "step_time_sec": 8.229770366015146, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.981656074523926, "lr": 0.0001168, "elapsed_sec": 2404.8882608413696, "step_time_sec": 8.23498450597981, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.062839031219482, "lr": 0.00011720000000000002, "elapsed_sec": 2413.1183993816376, "step_time_sec": 8.230030259001069, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.042774200439453, "lr": 0.00011760000000000001, "elapsed_sec": 2421.3489496707916, "step_time_sec": 8.230385583010502, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.9931864738464355, "lr": 0.00011800000000000001, "elapsed_sec": 2429.583897590637, "step_time_sec": 8.234745355002815, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.183349609375, "lr": 0.0001184, "elapsed_sec": 2437.819346189499, "step_time_sec": 8.235339154984104, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.002267360687256, "lr": 0.0001188, "elapsed_sec": 2446.0542147159576, "step_time_sec": 8.234652440994978, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.967225551605225, "lr": 0.0001192, "elapsed_sec": 2454.291349887848, "step_time_sec": 8.237006581010064, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.029119968414307, "lr": 0.00011960000000000001, "elapsed_sec": 2462.528090953827, "step_time_sec": 8.236522233986761, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.063477516174316, "lr": 0.00012, "elapsed_sec": 2470.76295876503, "step_time_sec": 8.234694569022395, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.036959171295166, "lr": 0.00012040000000000001, "elapsed_sec": 2478.998921394348, "step_time_sec": 8.235867104987847, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.1645379066467285, "lr": 0.00012080000000000001, "elapsed_sec": 2487.2354958057404, "step_time_sec": 8.236406505980995, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.925196170806885, "lr": 0.0001212, "elapsed_sec": 2495.4719083309174, "step_time_sec": 8.236254467017716, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.976163864135742, "lr": 0.0001216, "elapsed_sec": 2503.7025752067566, "step_time_sec": 8.23048723800457, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.0362348556518555, "lr": 0.00012200000000000001, "elapsed_sec": 2511.9329941272736, "step_time_sec": 8.230288040009327, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.8864054679870605, "lr": 0.00012240000000000002, "elapsed_sec": 2520.169285774231, "step_time_sec": 8.236092064995319, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.940955638885498, "lr": 0.0001228, "elapsed_sec": 2528.4039521217346, "step_time_sec": 8.2344927630038, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.994656085968018, "lr": 0.0001232, "elapsed_sec": 2536.639855861664, "step_time_sec": 8.235740675008856, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.060554504394531, "lr": 0.0001236, "elapsed_sec": 2544.875918865204, "step_time_sec": 8.235913885000627, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.962956428527832, "lr": 0.000124, "elapsed_sec": 2553.111802816391, "step_time_sec": 8.235684418992605, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.000129222869873, "lr": 0.00012440000000000002, "elapsed_sec": 2561.347995042801, "step_time_sec": 8.236119395005517, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.034297466278076, "lr": 0.0001248, "elapsed_sec": 2569.583331346512, "step_time_sec": 8.235132778994739, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.15023136138916, "lr": 0.0001252, "elapsed_sec": 2577.8203823566437, "step_time_sec": 8.236897174007026, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.931808948516846, "lr": 0.00012560000000000002, "elapsed_sec": 2586.0553135871887, "step_time_sec": 8.234771096991608, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.083316802978516, "lr": 0.000126, "elapsed_sec": 2594.2917473316193, "step_time_sec": 8.236245819018222, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.918760299682617, "lr": 0.0001264, "elapsed_sec": 2602.5212137699127, "step_time_sec": 8.229348895023577, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.99388313293457, "lr": 0.0001268, "elapsed_sec": 2610.7577497959137, "step_time_sec": 8.236323874996742, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.9091010093688965, "lr": 0.0001272, "elapsed_sec": 2618.9937529563904, "step_time_sec": 8.235912188014481, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.008285045623779, "lr": 0.0001276, "elapsed_sec": 2627.2244911193848, "step_time_sec": 8.230560291995062, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.917326927185059, "lr": 0.000128, "elapsed_sec": 2635.4547290802, "step_time_sec": 8.230058431014186, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.923355579376221, "lr": 0.0001284, "elapsed_sec": 2643.686798095703, "step_time_sec": 8.231933987000957, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.9515700340271, "lr": 0.0001288, "elapsed_sec": 2651.9173123836517, "step_time_sec": 8.230346058990108, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.048811912536621, "lr": 0.0001292, "elapsed_sec": 2660.1470518112183, "step_time_sec": 8.22955454501789, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.993483543395996, "lr": 0.0001296, "elapsed_sec": 2668.3766615390778, "step_time_sec": 8.229475731001003, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.13192081451416, "lr": 0.00013000000000000002, "elapsed_sec": 2676.611800432205, "step_time_sec": 8.234950519981794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.86976957321167, "lr": 0.00013040000000000003, "elapsed_sec": 2684.8477873802185, "step_time_sec": 8.23585307598114, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.952634334564209, "lr": 0.0001308, "elapsed_sec": 2693.0780959129333, "step_time_sec": 8.230123525019735, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.815874099731445, "lr": 0.00013120000000000002, "elapsed_sec": 2701.315598487854, "step_time_sec": 8.237381211016327, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.8278632164001465, "lr": 0.0001316, "elapsed_sec": 2709.551155567169, "step_time_sec": 8.235358261008514, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.049564838409424, "lr": 0.000132, "elapsed_sec": 2717.786757707596, "step_time_sec": 8.235419977019774, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.039964199066162, "lr": 0.00013240000000000002, "elapsed_sec": 2726.0222413539886, "step_time_sec": 8.235326594003709, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.898155212402344, "lr": 0.0001328, "elapsed_sec": 2734.2580728530884, "step_time_sec": 8.23567483999068, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.913607120513916, "lr": 0.0001332, "elapsed_sec": 2742.4947583675385, "step_time_sec": 8.236530275986297, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.866842746734619, "lr": 0.0001336, "elapsed_sec": 2750.7311265468597, "step_time_sec": 8.23619186002179, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.874545574188232, "lr": 0.000134, "elapsed_sec": 2758.9676320552826, "step_time_sec": 8.236386865988607, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.875837802886963, "lr": 0.00013440000000000001, "elapsed_sec": 2767.203085422516, "step_time_sec": 8.23526787100127, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.787038803100586, "lr": 0.0001348, "elapsed_sec": 2775.439469575882, "step_time_sec": 8.236214011005359, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.166839599609375, "lr": 0.0001352, "elapsed_sec": 2783.6699497699738, "step_time_sec": 8.230307705001906, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.906242847442627, "lr": 0.0001356, "elapsed_sec": 2791.899886608124, "step_time_sec": 8.229749514983268, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.796223163604736, "lr": 0.000136, "elapsed_sec": 2800.135416507721, "step_time_sec": 8.235382087994367, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.847805976867676, "lr": 0.00013639999999999998, "elapsed_sec": 2808.3709931373596, "step_time_sec": 8.23540058301296, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.206633567810059, "lr": 0.00013680000000000002, "elapsed_sec": 2816.6070504188538, "step_time_sec": 8.235926469991682, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.968772888183594, "lr": 0.00013720000000000003, "elapsed_sec": 2824.8376133441925, "step_time_sec": 8.230371115001617, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.969465732574463, "lr": 0.0001376, "elapsed_sec": 2833.067932844162, "step_time_sec": 8.23020900899428, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.925909042358398, "lr": 0.00013800000000000002, "elapsed_sec": 2841.297975063324, "step_time_sec": 8.229873263015179, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.7676100730896, "lr": 0.0001384, "elapsed_sec": 2849.534722805023, "step_time_sec": 8.236555990006309, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.886618137359619, "lr": 0.0001388, "elapsed_sec": 2857.770028591156, "step_time_sec": 8.235134804999689, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.835052013397217, "lr": 0.00013920000000000002, "elapsed_sec": 2866.0059702396393, "step_time_sec": 8.235785572003806, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.819122314453125, "lr": 0.0001396, "elapsed_sec": 2874.235713481903, "step_time_sec": 8.229621462000068, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.926530838012695, "lr": 0.00014000000000000001, "elapsed_sec": 2882.4666018486023, "step_time_sec": 8.230675504979445, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.882236003875732, "lr": 0.0001404, "elapsed_sec": 2890.6968138217926, "step_time_sec": 8.23003471197444, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.895683288574219, "lr": 0.0001408, "elapsed_sec": 2898.933813095093, "step_time_sec": 8.236901783006033, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.856359958648682, "lr": 0.00014120000000000002, "elapsed_sec": 2907.1703073978424, "step_time_sec": 8.236314863985172, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.841971397399902, "lr": 0.0001416, "elapsed_sec": 2915.4055964946747, "step_time_sec": 8.235087903012754, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.815659046173096, "lr": 0.000142, "elapsed_sec": 2923.6414861679077, "step_time_sec": 8.235770163009875, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.847985744476318, "lr": 0.0001424, "elapsed_sec": 2931.876526594162, "step_time_sec": 8.234860863973154, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.8154072761535645, "lr": 0.0001428, "elapsed_sec": 2940.1120052337646, "step_time_sec": 8.235332871001447, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.938900947570801, "lr": 0.00014319999999999998, "elapsed_sec": 2948.3474564552307, "step_time_sec": 8.235279168991838, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.864852428436279, "lr": 0.0001436, "elapsed_sec": 2956.583218574524, "step_time_sec": 8.235605244990438, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.8585004806518555, "lr": 0.000144, "elapsed_sec": 2964.8189463615417, "step_time_sec": 8.235551439021947, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.912375450134277, "lr": 0.0001444, "elapsed_sec": 2973.054263114929, "step_time_sec": 8.235121528006857, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.758143424987793, "lr": 0.00014480000000000002, "elapsed_sec": 2981.2907390594482, "step_time_sec": 8.236370200989768, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.745064735412598, "lr": 0.0001452, "elapsed_sec": 2989.5270459651947, "step_time_sec": 8.236125063995132, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.937960624694824, "lr": 0.00014560000000000002, "elapsed_sec": 2997.7588431835175, "step_time_sec": 8.231671778019518, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.894201278686523, "lr": 0.00014600000000000003, "elapsed_sec": 3005.9894235134125, "step_time_sec": 8.230403923022095, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.887033939361572, "lr": 0.0001464, "elapsed_sec": 3014.2196865081787, "step_time_sec": 8.230067674012389, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.784755706787109, "lr": 0.00014680000000000002, "elapsed_sec": 3022.4552960395813, "step_time_sec": 8.235504359006882, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.806985378265381, "lr": 0.0001472, "elapsed_sec": 3030.6909551620483, "step_time_sec": 8.235462590004317, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.899438858032227, "lr": 0.0001476, "elapsed_sec": 3038.927004098892, "step_time_sec": 8.235888673982117, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.817554950714111, "lr": 0.00014800000000000002, "elapsed_sec": 3047.1571369171143, "step_time_sec": 8.229949734988622, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.799916744232178, "lr": 0.0001484, "elapsed_sec": 3055.387670516968, "step_time_sec": 8.230361714988248, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.754536151885986, "lr": 0.0001488, "elapsed_sec": 3063.618476629257, "step_time_sec": 8.230673292011488, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.737117767333984, "lr": 0.0001492, "elapsed_sec": 3071.8484921455383, "step_time_sec": 8.229840527987108, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.775265216827393, "lr": 0.0001496, "elapsed_sec": 3080.0784418582916, "step_time_sec": 8.229784991010092, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.837540149688721, "lr": 0.00015, "elapsed_sec": 3088.3084955215454, "step_time_sec": 8.229884611995658, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.803699493408203, "lr": 0.0001504, "elapsed_sec": 3096.5444231033325, "step_time_sec": 8.235790984996129, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.839330673217773, "lr": 0.0001508, "elapsed_sec": 3104.780449151993, "step_time_sec": 8.23583064097329, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.762569904327393, "lr": 0.0001512, "elapsed_sec": 3113.0168056488037, "step_time_sec": 8.23620449998998, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.77072286605835, "lr": 0.00015160000000000003, "elapsed_sec": 3121.24804520607, "step_time_sec": 8.23106278400519, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.852710723876953, "lr": 0.000152, "elapsed_sec": 3129.479416847229, "step_time_sec": 8.231254353013355, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.724658012390137, "lr": 0.00015240000000000002, "elapsed_sec": 3137.714441537857, "step_time_sec": 8.234908523008926, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.733143329620361, "lr": 0.00015280000000000003, "elapsed_sec": 3145.951315641403, "step_time_sec": 8.236659749003593, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.730434894561768, "lr": 0.0001532, "elapsed_sec": 3154.1870963573456, "step_time_sec": 8.235590156022226, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.754579067230225, "lr": 0.00015360000000000002, "elapsed_sec": 3162.423598766327, "step_time_sec": 8.236370036000153, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.774380683898926, "lr": 0.000154, "elapsed_sec": 3170.6586990356445, "step_time_sec": 8.234910614002729, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.7683000564575195, "lr": 0.0001544, "elapsed_sec": 3178.8938915729523, "step_time_sec": 8.235091130976798, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.714135646820068, "lr": 0.00015480000000000002, "elapsed_sec": 3187.129333257675, "step_time_sec": 8.235229334997712, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.857284069061279, "lr": 0.0001552, "elapsed_sec": 3195.3644020557404, "step_time_sec": 8.234895651985426, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.8647966384887695, "lr": 0.00015560000000000001, "elapsed_sec": 3203.6005046367645, "step_time_sec": 8.23596340502263, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.842672824859619, "lr": 0.000156, "elapsed_sec": 3211.8368542194366, "step_time_sec": 8.236161438981071, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.660247802734375, "lr": 0.0001564, "elapsed_sec": 3220.0678765773773, "step_time_sec": 8.23087396597839, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.776867389678955, "lr": 0.0001568, "elapsed_sec": 3228.2979855537415, "step_time_sec": 8.230005569988862, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.7697858810424805, "lr": 0.0001572, "elapsed_sec": 3236.5292150974274, "step_time_sec": 8.231051290989853, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.823585033416748, "lr": 0.0001576, "elapsed_sec": 3244.7661666870117, "step_time_sec": 8.236735859012697, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.787378787994385, "lr": 0.000158, "elapsed_sec": 3253.0033042430878, "step_time_sec": 8.236994164006319, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.582910537719727, "lr": 0.0001584, "elapsed_sec": 3261.2395358085632, "step_time_sec": 8.236082135001197, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.684423923492432, "lr": 0.00015879999999999998, "elapsed_sec": 3269.476384162903, "step_time_sec": 8.236712076002732, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.785495281219482, "lr": 0.00015920000000000002, "elapsed_sec": 3277.7124269008636, "step_time_sec": 8.235894479992567, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.700091361999512, "lr": 0.00015960000000000003, "elapsed_sec": 3285.9488377571106, "step_time_sec": 8.236218442994868, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.805973529815674, "lr": 0.00016, "elapsed_sec": 3294.179003715515, "step_time_sec": 8.230036632012343, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.826656818389893, "lr": 0.00016040000000000002, "elapsed_sec": 3302.4103791713715, "step_time_sec": 8.231234817998484, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.722320079803467, "lr": 0.0001608, "elapsed_sec": 3310.6404745578766, "step_time_sec": 8.229851434996817, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.608851909637451, "lr": 0.00016120000000000002, "elapsed_sec": 3318.8708760738373, "step_time_sec": 8.230284345016116, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.704578399658203, "lr": 0.00016160000000000002, "elapsed_sec": 3327.1015927791595, "step_time_sec": 8.230502580001485, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.818565368652344, "lr": 0.000162, "elapsed_sec": 3335.331165075302, "step_time_sec": 8.229388231993653, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.649557590484619, "lr": 0.00016240000000000002, "elapsed_sec": 3343.5683114528656, "step_time_sec": 8.23702445300296, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.757771015167236, "lr": 0.0001628, "elapsed_sec": 3351.8032665252686, "step_time_sec": 8.234760673018172, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.810235500335693, "lr": 0.0001632, "elapsed_sec": 3360.0404012203217, "step_time_sec": 8.237037218001205, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.766888618469238, "lr": 0.0001636, "elapsed_sec": 3368.2753698825836, "step_time_sec": 8.234823042002972, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.727482795715332, "lr": 0.000164, "elapsed_sec": 3376.512988805771, "step_time_sec": 8.237421472003916, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.716240882873535, "lr": 0.0001644, "elapsed_sec": 3384.7488980293274, "step_time_sec": 8.235732864006422, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.75543737411499, "lr": 0.0001648, "elapsed_sec": 3392.986186027527, "step_time_sec": 8.237127085012617, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.623505592346191, "lr": 0.0001652, "elapsed_sec": 3401.22354722023, "step_time_sec": 8.237203710014, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.712040424346924, "lr": 0.0001656, "elapsed_sec": 3409.4600999355316, "step_time_sec": 8.236341943993466, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.734735012054443, "lr": 0.000166, "elapsed_sec": 3417.69140791893, "step_time_sec": 8.231223723007133, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.690248489379883, "lr": 0.00016640000000000003, "elapsed_sec": 3425.927550792694, "step_time_sec": 8.23594365498866, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.684763431549072, "lr": 0.00016680000000000002, "elapsed_sec": 3434.163593530655, "step_time_sec": 8.235874114994658, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.619856357574463, "lr": 0.00016720000000000003, "elapsed_sec": 3442.3936443328857, "step_time_sec": 8.229902461025631, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.809211254119873, "lr": 0.0001676, "elapsed_sec": 3450.6301443576813, "step_time_sec": 8.236320841999259, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.6645917892456055, "lr": 0.00016800000000000002, "elapsed_sec": 3458.8664667606354, "step_time_sec": 8.236201539984904, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.710045337677002, "lr": 0.00016840000000000003, "elapsed_sec": 3467.102949142456, "step_time_sec": 8.23626848301501, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.534599304199219, "lr": 0.0001688, "elapsed_sec": 3475.339229106903, "step_time_sec": 8.236148202006007, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.779881954193115, "lr": 0.00016920000000000002, "elapsed_sec": 3483.5774483680725, "step_time_sec": 8.238086394994752, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.655633449554443, "lr": 0.0001696, "elapsed_sec": 3491.8131878376007, "step_time_sec": 8.235561378998682, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.654770851135254, "lr": 0.00017, "elapsed_sec": 3500.0439777374268, "step_time_sec": 8.230587266007205, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.7683587074279785, "lr": 0.0001704, "elapsed_sec": 3508.27401638031, "step_time_sec": 8.229842650995124, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.599987983703613, "lr": 0.0001708, "elapsed_sec": 3516.5053040981293, "step_time_sec": 8.231154634995619, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.623669147491455, "lr": 0.00017120000000000001, "elapsed_sec": 3524.7414333820343, "step_time_sec": 8.23594667899306, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.61849308013916, "lr": 0.0001716, "elapsed_sec": 3532.9782478809357, "step_time_sec": 8.2366769349901, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.670934677124023, "lr": 0.000172, "elapsed_sec": 3541.2145459651947, "step_time_sec": 8.236130088975187, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.607108116149902, "lr": 0.0001724, "elapsed_sec": 3549.451231241226, "step_time_sec": 8.236483574000886, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.677918910980225, "lr": 0.0001728, "elapsed_sec": 3557.68750500679, "step_time_sec": 8.236156211001799, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.48624849319458, "lr": 0.0001732, "elapsed_sec": 3565.922139406204, "step_time_sec": 8.234419093991164, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.716592311859131, "lr": 0.0001736, "elapsed_sec": 3574.1590988636017, "step_time_sec": 8.236824143008562, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.631209373474121, "lr": 0.00017400000000000003, "elapsed_sec": 3582.3950839042664, "step_time_sec": 8.235875451005995, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.730271816253662, "lr": 0.0001744, "elapsed_sec": 3590.6323471069336, "step_time_sec": 8.237046683992958, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.628267765045166, "lr": 0.00017480000000000002, "elapsed_sec": 3598.8627548217773, "step_time_sec": 8.23029078199761, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.562857151031494, "lr": 0.0001752, "elapsed_sec": 3607.0995078086853, "step_time_sec": 8.236566792998929, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.679497718811035, "lr": 0.0001756, "elapsed_sec": 3615.335775613785, "step_time_sec": 8.236070815997664, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.7218804359436035, "lr": 0.00017600000000000002, "elapsed_sec": 3623.5732157230377, "step_time_sec": 8.237279337015934, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.549403667449951, "lr": 0.0001764, "elapsed_sec": 3631.808824777603, "step_time_sec": 8.235430236003594, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.628031253814697, "lr": 0.00017680000000000001, "elapsed_sec": 3640.0458946228027, "step_time_sec": 8.236929617996793, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.603500843048096, "lr": 0.0001772, "elapsed_sec": 3648.283147096634, "step_time_sec": 8.237100110010942, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.672660827636719, "lr": 0.0001776, "elapsed_sec": 3656.514111995697, "step_time_sec": 8.230849683022825, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.5961174964904785, "lr": 0.00017800000000000002, "elapsed_sec": 3664.7445652484894, "step_time_sec": 8.230203402985353, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.653379440307617, "lr": 0.0001784, "elapsed_sec": 3672.9753334522247, "step_time_sec": 8.230667622992769, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.690316200256348, "lr": 0.0001788, "elapsed_sec": 3681.2108573913574, "step_time_sec": 8.235336315003224, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.596421241760254, "lr": 0.0001792, "elapsed_sec": 3689.4479093551636, "step_time_sec": 8.236837744014338, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.605165004730225, "lr": 0.0001796, "elapsed_sec": 3697.683347463608, "step_time_sec": 8.235358428995823, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.49411678314209, "lr": 0.00018, "elapsed_sec": 3705.9197268486023, "step_time_sec": 8.236152832978405, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.510256290435791, "lr": 0.0001804, "elapsed_sec": 3714.1567149162292, "step_time_sec": 8.236904997000238, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.546986103057861, "lr": 0.0001808, "elapsed_sec": 3722.393622636795, "step_time_sec": 8.236697604006622, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.699304103851318, "lr": 0.0001812, "elapsed_sec": 3730.6301653385162, "step_time_sec": 8.23640128300758, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.571630001068115, "lr": 0.00018160000000000002, "elapsed_sec": 3738.8658154010773, "step_time_sec": 8.235499625996454, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.739142417907715, "lr": 0.000182, "elapsed_sec": 3747.102233171463, "step_time_sec": 8.236202531028539, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.64375114440918, "lr": 0.00018240000000000002, "elapsed_sec": 3755.3385558128357, "step_time_sec": 8.236149115022272, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.621083736419678, "lr": 0.00018280000000000003, "elapsed_sec": 3763.575896501541, "step_time_sec": 8.237213654007064, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.603884220123291, "lr": 0.0001832, "elapsed_sec": 3771.811760663986, "step_time_sec": 8.235653523996007, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.617647171020508, "lr": 0.00018360000000000002, "elapsed_sec": 3780.0486447811127, "step_time_sec": 8.236754711018875, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.663127899169922, "lr": 0.000184, "elapsed_sec": 3788.284011363983, "step_time_sec": 8.235212465020595, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.6113152503967285, "lr": 0.0001844, "elapsed_sec": 3796.5203330516815, "step_time_sec": 8.236150719982106, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.590612888336182, "lr": 0.00018480000000000002, "elapsed_sec": 3804.7574088573456, "step_time_sec": 8.236883077013772, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.619027614593506, "lr": 0.0001852, "elapsed_sec": 3812.9920206069946, "step_time_sec": 8.23447928999667, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.803319454193115, "lr": 0.0001856, "elapsed_sec": 3821.228347301483, "step_time_sec": 8.236153787001967, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.753872394561768, "lr": 0.000186, "elapsed_sec": 3829.465179204941, "step_time_sec": 8.236640076996991, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.710853099822998, "lr": 0.0001864, "elapsed_sec": 3837.7018332481384, "step_time_sec": 8.236521069018636, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.558346748352051, "lr": 0.00018680000000000001, "elapsed_sec": 3845.9367101192474, "step_time_sec": 8.234674306004308, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.464254856109619, "lr": 0.0001872, "elapsed_sec": 3854.173039674759, "step_time_sec": 8.2361956499808, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.733436584472656, "lr": 0.0001876, "elapsed_sec": 3862.410138130188, "step_time_sec": 8.236964015988633, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.612011432647705, "lr": 0.000188, "elapsed_sec": 3870.646611213684, "step_time_sec": 8.236299855983816, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.5804643630981445, "lr": 0.0001884, "elapsed_sec": 3878.8823742866516, "step_time_sec": 8.235638473008294, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.487613677978516, "lr": 0.0001888, "elapsed_sec": 3887.118618249893, "step_time_sec": 8.236046279984294, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.621426105499268, "lr": 0.00018920000000000002, "elapsed_sec": 3895.3552556037903, "step_time_sec": 8.236474401986925, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.536467552185059, "lr": 0.00018960000000000003, "elapsed_sec": 3903.5918850898743, "step_time_sec": 8.236461826018058, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.572568416595459, "lr": 0.00019, "elapsed_sec": 3911.8295958042145, "step_time_sec": 8.237548227014486, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.53804874420166, "lr": 0.00019040000000000002, "elapsed_sec": 3920.0662183761597, "step_time_sec": 8.236464813991915, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.5229291915893555, "lr": 0.0001908, "elapsed_sec": 3928.3033657073975, "step_time_sec": 8.236969290999696, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.589864730834961, "lr": 0.0001912, "elapsed_sec": 3936.539173603058, "step_time_sec": 8.235648733010748, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.092962265014648, "lr": 0.00019160000000000002, "elapsed_sec": 3944.776082277298, "step_time_sec": 8.236715444014408, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.4925537109375, "lr": 0.000192, "elapsed_sec": 3953.012115716934, "step_time_sec": 8.235900353989564, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.684003829956055, "lr": 0.00019240000000000001, "elapsed_sec": 3961.2491755485535, "step_time_sec": 8.23687105398858, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.48390531539917, "lr": 0.0001928, "elapsed_sec": 3969.48677110672, "step_time_sec": 8.237406244006706, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.416462421417236, "lr": 0.0001932, "elapsed_sec": 3977.723520040512, "step_time_sec": 8.236641219002195, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.5337300300598145, "lr": 0.00019360000000000002, "elapsed_sec": 3985.9601757526398, "step_time_sec": 8.236453657009406, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.5260467529296875, "lr": 0.000194, "elapsed_sec": 3994.196264743805, "step_time_sec": 8.23591251700418, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.535672664642334, "lr": 0.0001944, "elapsed_sec": 4002.433490037918, "step_time_sec": 8.23709363801754, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.394049167633057, "lr": 0.0001948, "elapsed_sec": 4010.6697833538055, "step_time_sec": 8.236102922004648, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.490416526794434, "lr": 0.0001952, "elapsed_sec": 4018.909416437149, "step_time_sec": 8.239531641011126, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.51165246963501, "lr": 0.00019559999999999998, "elapsed_sec": 4027.1452519893646, "step_time_sec": 8.235590830998262, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.490628242492676, "lr": 0.00019600000000000002, "elapsed_sec": 4035.3818924427032, "step_time_sec": 8.236483913002303, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.50009298324585, "lr": 0.00019640000000000003, "elapsed_sec": 4043.6183700561523, "step_time_sec": 8.236363529023947, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.6248321533203125, "lr": 0.0001968, "elapsed_sec": 4051.8541989326477, "step_time_sec": 8.235625497007277, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.464243412017822, "lr": 0.00019720000000000002, "elapsed_sec": 4060.0909967422485, "step_time_sec": 8.236618067981908, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.488579750061035, "lr": 0.0001976, "elapsed_sec": 4068.326994419098, "step_time_sec": 8.235859837004682, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.496064186096191, "lr": 0.00019800000000000002, "elapsed_sec": 4076.5645713806152, "step_time_sec": 8.237443521997193, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.431778907775879, "lr": 0.00019840000000000002, "elapsed_sec": 4084.80161857605, "step_time_sec": 8.23684618101106, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.537414073944092, "lr": 0.0001988, "elapsed_sec": 4093.038901090622, "step_time_sec": 8.237163308018353, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.441727638244629, "lr": 0.00019920000000000002, "elapsed_sec": 4101.276389598846, "step_time_sec": 8.237319159001345, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.401037693023682, "lr": 0.0001996, "elapsed_sec": 4109.513262748718, "step_time_sec": 8.236667937017046, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.511814594268799, "lr": 0.0002, "elapsed_sec": 4117.750268697739, "step_time_sec": 29.771960740996292, "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": 6.465513229370117, "lr": 0.0002, "elapsed_sec": 4147.5311760902405, "step_time_sec": 8.245299313013675, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.437492370605469, "lr": 0.0002, "elapsed_sec": 4155.76044511795, "step_time_sec": 8.229121818993008, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.535143852233887, "lr": 0.0002, "elapsed_sec": 4163.990844964981, "step_time_sec": 8.230231403984362, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.5528764724731445, "lr": 0.0002, "elapsed_sec": 4172.221040964127, "step_time_sec": 8.229993605025811, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.492119789123535, "lr": 0.0002, "elapsed_sec": 4180.452234506607, "step_time_sec": 8.231015486002434, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.368930816650391, "lr": 0.0002, "elapsed_sec": 4188.682370424271, "step_time_sec": 8.230040135997115, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.457215309143066, "lr": 0.0002, "elapsed_sec": 4196.913388967514, "step_time_sec": 8.230848239007173, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.62441873550415, "lr": 0.0002, "elapsed_sec": 4205.148941755295, "step_time_sec": 8.235349612979917, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.560901641845703, "lr": 0.0002, "elapsed_sec": 4213.3857662677765, "step_time_sec": 8.236691457015695, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.5873894691467285, "lr": 0.0002, "elapsed_sec": 4221.622380018234, "step_time_sec": 8.23644629798946, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.403398036956787, "lr": 0.0002, "elapsed_sec": 4229.859136104584, "step_time_sec": 8.236565329018049, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.463426113128662, "lr": 0.0002, "elapsed_sec": 4238.094377279282, "step_time_sec": 8.235042020998662, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.457305431365967, "lr": 0.0002, "elapsed_sec": 4246.330416917801, "step_time_sec": 8.235887178016128, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.457043647766113, "lr": 0.0002, "elapsed_sec": 4254.566330432892, "step_time_sec": 8.235736564995022, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.6872687339782715, "lr": 0.0002, "elapsed_sec": 4262.80281662941, "step_time_sec": 8.236347794998437, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.502204895019531, "lr": 0.0002, "elapsed_sec": 4271.037917613983, "step_time_sec": 8.234954181010835, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.558790683746338, "lr": 0.0002, "elapsed_sec": 4279.275587320328, "step_time_sec": 8.237485749006737, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.605945587158203, "lr": 0.0002, "elapsed_sec": 4287.5136070251465, "step_time_sec": 8.237865019007586, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.51085090637207, "lr": 0.0002, "elapsed_sec": 4295.749089479446, "step_time_sec": 8.235344495013123, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.5013251304626465, "lr": 0.0002, "elapsed_sec": 4303.985656738281, "step_time_sec": 8.236412981990725, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.397944927215576, "lr": 0.0002, "elapsed_sec": 4312.221871376038, "step_time_sec": 8.236061815987341, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.516694068908691, "lr": 0.0002, "elapsed_sec": 4320.46017241478, "step_time_sec": 8.238097312976606, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.579593181610107, "lr": 0.0002, "elapsed_sec": 4328.697982788086, "step_time_sec": 8.237711291003507, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.522622585296631, "lr": 0.0002, "elapsed_sec": 4336.935685157776, "step_time_sec": 8.23751747602364, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.398120880126953, "lr": 0.0002, "elapsed_sec": 4345.171627283096, "step_time_sec": 8.235758305992931, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.560542583465576, "lr": 0.0002, "elapsed_sec": 4353.408048152924, "step_time_sec": 8.23628813799587, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.496139049530029, "lr": 0.0002, "elapsed_sec": 4361.642703294754, "step_time_sec": 8.234434506011894, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.384778022766113, "lr": 0.0002, "elapsed_sec": 4369.87811088562, "step_time_sec": 8.23529421002604, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.553006649017334, "lr": 0.0002, "elapsed_sec": 4378.11644244194, "step_time_sec": 8.238162886991631, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.527829647064209, "lr": 0.0002, "elapsed_sec": 4386.350885391235, "step_time_sec": 8.234315029985737, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.5486249923706055, "lr": 0.0002, "elapsed_sec": 4394.5871431827545, "step_time_sec": 8.236063131014816, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.526352405548096, "lr": 0.0002, "elapsed_sec": 4402.823452949524, "step_time_sec": 8.236083596973913, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.387411117553711, "lr": 0.0002, "elapsed_sec": 4411.0611481666565, "step_time_sec": 8.237550471996656, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.402510166168213, "lr": 0.0002, "elapsed_sec": 4419.298322916031, "step_time_sec": 8.23698654200416, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.4588847160339355, "lr": 0.0002, "elapsed_sec": 4427.535271406174, "step_time_sec": 8.236797719990136, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.3950629234313965, "lr": 0.0002, "elapsed_sec": 4435.7729551792145, "step_time_sec": 8.237493477994576, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.362270832061768, "lr": 0.0002, "elapsed_sec": 4444.010215044022, "step_time_sec": 8.237153644004138, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.644595146179199, "lr": 0.0002, "elapsed_sec": 4452.247295618057, "step_time_sec": 8.236894223984564, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.520115852355957, "lr": 0.0002, "elapsed_sec": 4460.482952833176, "step_time_sec": 8.235459677001927, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.506774425506592, "lr": 0.0002, "elapsed_sec": 4468.719064474106, "step_time_sec": 8.236001351993764, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.7903923988342285, "lr": 0.0002, "elapsed_sec": 4476.953986644745, "step_time_sec": 8.234710952994647, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.4173054695129395, "lr": 0.0002, "elapsed_sec": 4485.189999103546, "step_time_sec": 8.23587521200534, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.573141098022461, "lr": 0.0002, "elapsed_sec": 4493.426500082016, "step_time_sec": 8.23632653101231, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.351966857910156, "lr": 0.0002, "elapsed_sec": 4501.66330242157, "step_time_sec": 8.236606000980828, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.370062828063965, "lr": 0.0002, "elapsed_sec": 4509.900730133057, "step_time_sec": 8.237276610016124, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.506958484649658, "lr": 0.0002, "elapsed_sec": 4518.136046409607, "step_time_sec": 8.235135200986406, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.3949103355407715, "lr": 0.0002, "elapsed_sec": 4526.373791217804, "step_time_sec": 8.237639533006586, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.342252731323242, "lr": 0.0002, "elapsed_sec": 4534.610001087189, "step_time_sec": 8.235984215018107, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.47490119934082, "lr": 0.0002, "elapsed_sec": 4542.847495317459, "step_time_sec": 8.2373506409931, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.402872562408447, "lr": 0.0002, "elapsed_sec": 4551.084161758423, "step_time_sec": 8.236536865995731, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.4583539962768555, "lr": 0.0002, "elapsed_sec": 4559.32098698616, "step_time_sec": 8.236639137991006, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.433820724487305, "lr": 0.0002, "elapsed_sec": 4567.557877540588, "step_time_sec": 8.23674210801255, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.387409687042236, "lr": 0.0002, "elapsed_sec": 4575.795112609863, "step_time_sec": 8.237064524000743, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.449325084686279, "lr": 0.0002, "elapsed_sec": 4584.032459020615, "step_time_sec": 8.23718057401129, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.394170761108398, "lr": 0.0002, "elapsed_sec": 4592.268309593201, "step_time_sec": 8.235692424990702, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.4415974617004395, "lr": 0.0002, "elapsed_sec": 4600.505441427231, "step_time_sec": 8.236943666997831, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.410675525665283, "lr": 0.0002, "elapsed_sec": 4608.741965770721, "step_time_sec": 8.236420947010629, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.414782524108887, "lr": 0.0002, "elapsed_sec": 4616.9796686172485, "step_time_sec": 8.237525525997626, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.450790882110596, "lr": 0.0002, "elapsed_sec": 4625.216801166534, "step_time_sec": 8.23698956798762, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.375936508178711, "lr": 0.0002, "elapsed_sec": 4633.45414018631, "step_time_sec": 8.237184984027408, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.412473678588867, "lr": 0.0002, "elapsed_sec": 4641.691466093063, "step_time_sec": 8.237108612986049, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.449220180511475, "lr": 0.0002, "elapsed_sec": 4649.92747092247, "step_time_sec": 8.235867036011769, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.334491729736328, "lr": 0.0002, "elapsed_sec": 4658.163857221603, "step_time_sec": 8.236215322976932, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.339288711547852, "lr": 0.0002, "elapsed_sec": 4666.398833990097, "step_time_sec": 8.234837108000647, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.336171627044678, "lr": 0.0002, "elapsed_sec": 4674.6354949474335, "step_time_sec": 8.236462711996865, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.443721294403076, "lr": 0.0002, "elapsed_sec": 4682.8711795806885, "step_time_sec": 8.235564782022266, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.349232196807861, "lr": 0.0002, "elapsed_sec": 4691.108012676239, "step_time_sec": 8.236660212976858, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.345146179199219, "lr": 0.0002, "elapsed_sec": 4699.34406208992, "step_time_sec": 8.235886611015303, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.319936752319336, "lr": 0.0002, "elapsed_sec": 4707.579988002777, "step_time_sec": 8.235745500016492, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.415694236755371, "lr": 0.0002, "elapsed_sec": 4715.815859794617, "step_time_sec": 8.235719525982859, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.353046417236328, "lr": 0.0002, "elapsed_sec": 4724.052354812622, "step_time_sec": 8.236376198008657, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.47601842880249, "lr": 0.0002, "elapsed_sec": 4732.289174556732, "step_time_sec": 8.236671278020367, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.3366217613220215, "lr": 0.0002, "elapsed_sec": 4740.525851249695, "step_time_sec": 8.236465199996019, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.3584208488464355, "lr": 0.0002, "elapsed_sec": 4748.763422250748, "step_time_sec": 8.237444252998102, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.293683052062988, "lr": 0.0002, "elapsed_sec": 4757.00030207634, "step_time_sec": 8.236717496009078, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.452631950378418, "lr": 0.0002, "elapsed_sec": 4765.238220214844, "step_time_sec": 8.237725786981173, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.3093976974487305, "lr": 0.0002, "elapsed_sec": 4773.475086450577, "step_time_sec": 8.236722464003833, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.27727746963501, "lr": 0.0002, "elapsed_sec": 4781.711883544922, "step_time_sec": 8.236573066009441, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.396608829498291, "lr": 0.0002, "elapsed_sec": 4789.948237657547, "step_time_sec": 8.2362108459929, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.334529876708984, "lr": 0.0002, "elapsed_sec": 4798.183865785599, "step_time_sec": 8.235477467009332, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.322073459625244, "lr": 0.0002, "elapsed_sec": 4806.420702219009, "step_time_sec": 8.236724381014938, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.47326135635376, "lr": 0.0002, "elapsed_sec": 4814.656122922897, "step_time_sec": 8.235206746991025, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.381850719451904, "lr": 0.0002, "elapsed_sec": 4822.893434762955, "step_time_sec": 8.23714796200511, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.458949089050293, "lr": 0.0002, "elapsed_sec": 4831.1291699409485, "step_time_sec": 8.235593293007696, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.3697190284729, "lr": 0.0002, "elapsed_sec": 4839.366970539093, "step_time_sec": 8.23762798999087, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.347571849822998, "lr": 0.0002, "elapsed_sec": 4847.604754447937, "step_time_sec": 8.237627249996876, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.283129692077637, "lr": 0.0002, "elapsed_sec": 4855.842969655991, "step_time_sec": 8.238043225981528, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.391366004943848, "lr": 0.0002, "elapsed_sec": 4864.08121919632, "step_time_sec": 8.23813524201978, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.3963518142700195, "lr": 0.0002, "elapsed_sec": 4872.3170030117035, "step_time_sec": 8.235618153994437, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.4681196212768555, "lr": 0.0002, "elapsed_sec": 4880.554876327515, "step_time_sec": 8.237689360015793, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.4159979820251465, "lr": 0.0002, "elapsed_sec": 4888.791065692902, "step_time_sec": 8.236100424983306, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.3472795486450195, "lr": 0.0002, "elapsed_sec": 4897.026581287384, "step_time_sec": 8.2352732209838, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.3901448249816895, "lr": 0.0002, "elapsed_sec": 4905.263068675995, "step_time_sec": 8.236333687003935, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.3082780838012695, "lr": 0.0002, "elapsed_sec": 4913.500918149948, "step_time_sec": 8.237704190018121, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.394173622131348, "lr": 0.0002, "elapsed_sec": 4921.73851108551, "step_time_sec": 8.237384556996403, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.35171365737915, "lr": 0.0002, "elapsed_sec": 4929.974933385849, "step_time_sec": 8.236279698990984, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.289848327636719, "lr": 0.0002, "elapsed_sec": 4938.2125561237335, "step_time_sec": 8.237503907992505, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.276395320892334, "lr": 0.0002, "elapsed_sec": 4946.449476003647, "step_time_sec": 8.23677639599191, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.335993766784668, "lr": 0.0002, "elapsed_sec": 4954.685615301132, "step_time_sec": 8.23591222599498, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.257296562194824, "lr": 0.0002, "elapsed_sec": 4962.92113161087, "step_time_sec": 8.235339847014984, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.269532203674316, "lr": 0.0002, "elapsed_sec": 4971.1581382751465, "step_time_sec": 8.236853762005921, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.1217122077941895, "lr": 0.0002, "elapsed_sec": 4979.39456820488, "step_time_sec": 8.236260552017484, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.384222507476807, "lr": 0.0002, "elapsed_sec": 4987.631149768829, "step_time_sec": 8.236460498010274, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.256448268890381, "lr": 0.0002, "elapsed_sec": 4995.867919206619, "step_time_sec": 8.236575684015406, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.283169746398926, "lr": 0.0002, "elapsed_sec": 5004.105050563812, "step_time_sec": 8.237005072005559, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.406441688537598, "lr": 0.0002, "elapsed_sec": 5012.342574119568, "step_time_sec": 8.237357868987601, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.334880828857422, "lr": 0.0002, "elapsed_sec": 5020.578492164612, "step_time_sec": 8.23573036698508, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.350874423980713, "lr": 0.0002, "elapsed_sec": 5028.815098285675, "step_time_sec": 8.236489183997037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.324249267578125, "lr": 0.0002, "elapsed_sec": 5037.052372932434, "step_time_sec": 8.237070986011531, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.317347049713135, "lr": 0.0002, "elapsed_sec": 5045.28963303566, "step_time_sec": 8.237099599005887, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.204692363739014, "lr": 0.0002, "elapsed_sec": 5053.526706933975, "step_time_sec": 8.236951230006525, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.2868971824646, "lr": 0.0002, "elapsed_sec": 5061.763581037521, "step_time_sec": 8.236677514010807, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.235531806945801, "lr": 0.0002, "elapsed_sec": 5070.000533103943, "step_time_sec": 8.236780867009657, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.286681175231934, "lr": 0.0002, "elapsed_sec": 5078.235500335693, "step_time_sec": 8.234807891014498, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.30150032043457, "lr": 0.0002, "elapsed_sec": 5086.471473455429, "step_time_sec": 8.235815102991182, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.294943332672119, "lr": 0.0002, "elapsed_sec": 5094.707632780075, "step_time_sec": 8.236010941007407, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.326326847076416, "lr": 0.0002, "elapsed_sec": 5102.944575309753, "step_time_sec": 8.236779249011306, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.229767322540283, "lr": 0.0002, "elapsed_sec": 5111.180728435516, "step_time_sec": 8.235988878994249, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.160036563873291, "lr": 0.0002, "elapsed_sec": 5119.418201446533, "step_time_sec": 8.237298321007984, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.284679412841797, "lr": 0.0002, "elapsed_sec": 5127.65433549881, "step_time_sec": 8.236026027007028, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.396829605102539, "lr": 0.0002, "elapsed_sec": 5135.892135858536, "step_time_sec": 8.237559583998518, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.214874267578125, "lr": 0.0002, "elapsed_sec": 5144.129953622818, "step_time_sec": 8.237712178990478, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.294672966003418, "lr": 0.0002, "elapsed_sec": 5152.364424705505, "step_time_sec": 8.234310447995085, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.297032356262207, "lr": 0.0002, "elapsed_sec": 5160.601746082306, "step_time_sec": 8.237127075000899, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.19273042678833, "lr": 0.0002, "elapsed_sec": 5168.837215900421, "step_time_sec": 8.235359562007943, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.256440162658691, "lr": 0.0002, "elapsed_sec": 5177.104424238205, "step_time_sec": 8.243719384016003, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.31049108505249, "lr": 0.0002, "elapsed_sec": 5185.34193778038, "step_time_sec": 8.237287705007475, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.492241859436035, "lr": 0.0002, "elapsed_sec": 5193.578457355499, "step_time_sec": 8.236392335995333, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.433403968811035, "lr": 0.0002, "elapsed_sec": 5201.814961671829, "step_time_sec": 8.236333081003977, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.333721160888672, "lr": 0.0002, "elapsed_sec": 5210.050879001617, "step_time_sec": 8.235748966981191, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.358126640319824, "lr": 0.0002, "elapsed_sec": 5218.288207530975, "step_time_sec": 8.237168009974994, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.214243412017822, "lr": 0.0002, "elapsed_sec": 5226.52387547493, "step_time_sec": 8.235558570013382, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.342261791229248, "lr": 0.0002, "elapsed_sec": 5234.7598485946655, "step_time_sec": 8.235771120002028, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.275350570678711, "lr": 0.0002, "elapsed_sec": 5242.995468854904, "step_time_sec": 8.235450349980965, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.337666988372803, "lr": 0.0002, "elapsed_sec": 5251.231266498566, "step_time_sec": 8.235637028003111, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.171314716339111, "lr": 0.0002, "elapsed_sec": 5259.468883514404, "step_time_sec": 8.237453405017732, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.33211612701416, "lr": 0.0002, "elapsed_sec": 5267.703947544098, "step_time_sec": 8.234941792994505, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.444884300231934, "lr": 0.0002, "elapsed_sec": 5275.940528869629, "step_time_sec": 8.236363292002352, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.24905252456665, "lr": 0.0002, "elapsed_sec": 5284.178077459335, "step_time_sec": 8.23743859998649, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.267798900604248, "lr": 0.0002, "elapsed_sec": 5292.41460442543, "step_time_sec": 8.236373805004405, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.274415969848633, "lr": 0.0002, "elapsed_sec": 5300.652025938034, "step_time_sec": 8.23725606201333, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.227463722229004, "lr": 0.0002, "elapsed_sec": 5308.887385129929, "step_time_sec": 8.235227236989886, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.259109973907471, "lr": 0.0002, "elapsed_sec": 5317.1232714653015, "step_time_sec": 8.235684644023422, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.171639442443848, "lr": 0.0002, "elapsed_sec": 5325.360435009003, "step_time_sec": 8.237039069994353, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.214748859405518, "lr": 0.0002, "elapsed_sec": 5333.597056627274, "step_time_sec": 8.23648487799801, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.241093158721924, "lr": 0.0002, "elapsed_sec": 5341.833230018616, "step_time_sec": 8.23596819100203, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.3119120597839355, "lr": 0.0002, "elapsed_sec": 5350.069887638092, "step_time_sec": 8.23646254101186, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.325658321380615, "lr": 0.0002, "elapsed_sec": 5358.306433200836, "step_time_sec": 8.236438849009573, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.223738670349121, "lr": 0.0002, "elapsed_sec": 5366.541900873184, "step_time_sec": 8.235293442005059, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.297028064727783, "lr": 0.0002, "elapsed_sec": 5374.778052806854, "step_time_sec": 8.235938726022141, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.267711162567139, "lr": 0.0002, "elapsed_sec": 5383.014051914215, "step_time_sec": 8.235853293997934, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.352022171020508, "lr": 0.0002, "elapsed_sec": 5391.2517375946045, "step_time_sec": 8.237504774006084, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.160573482513428, "lr": 0.0002, "elapsed_sec": 5399.487666606903, "step_time_sec": 8.235759789007716, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.335890769958496, "lr": 0.0002, "elapsed_sec": 5407.724361896515, "step_time_sec": 8.236582608020399, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.3115105628967285, "lr": 0.0002, "elapsed_sec": 5415.959363698959, "step_time_sec": 8.234828106011264, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.207906723022461, "lr": 0.0002, "elapsed_sec": 5424.197378396988, "step_time_sec": 8.237845087016467, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.263377666473389, "lr": 0.0002, "elapsed_sec": 5432.434098243713, "step_time_sec": 8.236522588995285, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.155126571655273, "lr": 0.0002, "elapsed_sec": 5440.670782089233, "step_time_sec": 8.236533477000194, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.170982360839844, "lr": 0.0002, "elapsed_sec": 5448.908370018005, "step_time_sec": 8.237461754004471, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.257833480834961, "lr": 0.0002, "elapsed_sec": 5457.145871639252, "step_time_sec": 8.237298603984527, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.263462543487549, "lr": 0.0002, "elapsed_sec": 5465.383325338364, "step_time_sec": 8.23734410200268, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.197177410125732, "lr": 0.0002, "elapsed_sec": 5473.6201293468475, "step_time_sec": 8.236618915980216, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.199919700622559, "lr": 0.0002, "elapsed_sec": 5481.858516454697, "step_time_sec": 8.238215246004984, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.14612340927124, "lr": 0.0002, "elapsed_sec": 5490.095381736755, "step_time_sec": 8.236717816995224, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.197697639465332, "lr": 0.0002, "elapsed_sec": 5498.332006454468, "step_time_sec": 8.236462222004775, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.299767971038818, "lr": 0.0002, "elapsed_sec": 5506.569582462311, "step_time_sec": 8.237428616004763, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.301109790802002, "lr": 0.0002, "elapsed_sec": 5514.806369543076, "step_time_sec": 8.236597893002909, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.1219892501831055, "lr": 0.0002, "elapsed_sec": 5523.044779777527, "step_time_sec": 8.238267165026627, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.294464588165283, "lr": 0.0002, "elapsed_sec": 5531.28061580658, "step_time_sec": 8.2357014779991, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.158390045166016, "lr": 0.0002, "elapsed_sec": 5539.51833152771, "step_time_sec": 8.23755260699545, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.218480110168457, "lr": 0.0002, "elapsed_sec": 5547.7531208992, "step_time_sec": 8.234617100009928, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.384010314941406, "lr": 0.0002, "elapsed_sec": 5555.990326166153, "step_time_sec": 8.237055325997062, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.261749267578125, "lr": 0.0002, "elapsed_sec": 5564.2267370224, "step_time_sec": 8.23623309400864, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.241793155670166, "lr": 0.0002, "elapsed_sec": 5572.463107585907, "step_time_sec": 8.236200040002586, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.109353065490723, "lr": 0.0002, "elapsed_sec": 5580.700439929962, "step_time_sec": 8.23715403600363, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.206398010253906, "lr": 0.0002, "elapsed_sec": 5588.935828924179, "step_time_sec": 8.235247407021234, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.1792893409729, "lr": 0.0002, "elapsed_sec": 5597.171872854233, "step_time_sec": 8.23586864999379, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.043578147888184, "lr": 0.0002, "elapsed_sec": 5605.408653974533, "step_time_sec": 8.236602085991763, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.269617080688477, "lr": 0.0002, "elapsed_sec": 5613.645712614059, "step_time_sec": 8.236917451984482, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.152854919433594, "lr": 0.0002, "elapsed_sec": 5621.88160443306, "step_time_sec": 8.235732250992442, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.220953464508057, "lr": 0.0002, "elapsed_sec": 5630.118459701538, "step_time_sec": 8.236695083003724, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.129759788513184, "lr": 0.0002, "elapsed_sec": 5638.355703115463, "step_time_sec": 8.237100088008447, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.244933605194092, "lr": 0.0002, "elapsed_sec": 5646.592414140701, "step_time_sec": 8.236603083001683, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.156457424163818, "lr": 0.0002, "elapsed_sec": 5654.829400062561, "step_time_sec": 8.236806046013953, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.191587924957275, "lr": 0.0002, "elapsed_sec": 5663.065641403198, "step_time_sec": 8.236110539000947, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.226241588592529, "lr": 0.0002, "elapsed_sec": 5671.303357601166, "step_time_sec": 8.237488996004686, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.213083267211914, "lr": 0.0002, "elapsed_sec": 5679.538651704788, "step_time_sec": 8.235173369990662, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.188515663146973, "lr": 0.0002, "elapsed_sec": 5687.774405479431, "step_time_sec": 8.235591818025569, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.19035005569458, "lr": 0.0002, "elapsed_sec": 5696.010578155518, "step_time_sec": 8.235982814978343, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.301524639129639, "lr": 0.0002, "elapsed_sec": 5704.24923324585, "step_time_sec": 8.238515033997828, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.0804948806762695, "lr": 0.0002, "elapsed_sec": 5712.486977815628, "step_time_sec": 8.23753656601184, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.285374641418457, "lr": 0.0002, "elapsed_sec": 5720.721661329269, "step_time_sec": 8.234534123999765, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.174095153808594, "lr": 0.0002, "elapsed_sec": 5728.958292961121, "step_time_sec": 8.236538703000406, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.1554670333862305, "lr": 0.0002, "elapsed_sec": 5737.194600343704, "step_time_sec": 8.236106113996357, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.19868803024292, "lr": 0.0002, "elapsed_sec": 5745.43123459816, "step_time_sec": 8.236469515017234, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.190197944641113, "lr": 0.0002, "elapsed_sec": 5753.666462659836, "step_time_sec": 8.235089728987077, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.348995685577393, "lr": 0.0002, "elapsed_sec": 5761.90444111824, "step_time_sec": 8.237800834991504, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.313975811004639, "lr": 0.0002, "elapsed_sec": 5770.140592813492, "step_time_sec": 8.236003861995414, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.20204496383667, "lr": 0.0002, "elapsed_sec": 5778.376801967621, "step_time_sec": 8.236082294984953, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.075397491455078, "lr": 0.0002, "elapsed_sec": 5786.612948894501, "step_time_sec": 8.235958433011547, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.113461017608643, "lr": 0.0002, "elapsed_sec": 5794.849977731705, "step_time_sec": 8.236857534007868, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.024404048919678, "lr": 0.0002, "elapsed_sec": 5803.086932659149, "step_time_sec": 8.236815078009386, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.222400188446045, "lr": 0.0002, "elapsed_sec": 5811.322628974915, "step_time_sec": 8.235545912000816, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.042140007019043, "lr": 0.0002, "elapsed_sec": 5819.559648036957, "step_time_sec": 8.236859962024027, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.136736869812012, "lr": 0.0002, "elapsed_sec": 5827.7953317165375, "step_time_sec": 8.23552828002721, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.262478828430176, "lr": 0.0002, "elapsed_sec": 5836.033611774445, "step_time_sec": 8.23810941100237, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.169619560241699, "lr": 0.0002, "elapsed_sec": 5844.270262956619, "step_time_sec": 8.236497016012436, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.189877033233643, "lr": 0.0002, "elapsed_sec": 5852.5072548389435, "step_time_sec": 8.23683918901952, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.149435997009277, "lr": 0.0002, "elapsed_sec": 5860.745611667633, "step_time_sec": 8.238168082985794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.220247268676758, "lr": 0.0002, "elapsed_sec": 5868.983278274536, "step_time_sec": 8.237587868992705, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.226924419403076, "lr": 0.0002, "elapsed_sec": 5877.221040248871, "step_time_sec": 8.237552025006153, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.086084365844727, "lr": 0.0002, "elapsed_sec": 5885.456640005112, "step_time_sec": 8.235437352996087, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.198526859283447, "lr": 0.0002, "elapsed_sec": 5893.693571329117, "step_time_sec": 8.236811834998662, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.097102165222168, "lr": 0.0002, "elapsed_sec": 5901.930420398712, "step_time_sec": 8.236641914001666, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.212132930755615, "lr": 0.0002, "elapsed_sec": 5910.168699979782, "step_time_sec": 8.238133389007999, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.240853786468506, "lr": 0.0002, "elapsed_sec": 5918.405205011368, "step_time_sec": 8.2363717519911, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.225794315338135, "lr": 0.0002, "elapsed_sec": 5926.642574310303, "step_time_sec": 8.237159554002574, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.156499862670898, "lr": 0.0002, "elapsed_sec": 5934.879477024078, "step_time_sec": 8.236748291994445, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.135499000549316, "lr": 0.0002, "elapsed_sec": 5943.115799188614, "step_time_sec": 8.236163288005628, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.147253513336182, "lr": 0.0002, "elapsed_sec": 5951.35292506218, "step_time_sec": 8.236948796984507, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.198662757873535, "lr": 0.0002, "elapsed_sec": 5959.589326620102, "step_time_sec": 8.236285588005558, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.224717140197754, "lr": 0.0002, "elapsed_sec": 5967.825714826584, "step_time_sec": 8.236230341979535, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.069607734680176, "lr": 0.0002, "elapsed_sec": 5976.062163352966, "step_time_sec": 8.236254660994746, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.118734836578369, "lr": 0.0002, "elapsed_sec": 5984.299378871918, "step_time_sec": 8.237046224996448, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.099090576171875, "lr": 0.0002, "elapsed_sec": 5992.53746676445, "step_time_sec": 8.237932138988981, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.176412105560303, "lr": 0.0002, "elapsed_sec": 6000.773762226105, "step_time_sec": 8.236115175008308, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.212978839874268, "lr": 0.0002, "elapsed_sec": 6009.011416435242, "step_time_sec": 8.23748141300166, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.129892349243164, "lr": 0.0002, "elapsed_sec": 6017.248081684113, "step_time_sec": 8.236550439993152, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.1232733726501465, "lr": 0.0002, "elapsed_sec": 6025.4851195812225, "step_time_sec": 8.236858101998223, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.166321754455566, "lr": 0.0002, "elapsed_sec": 6033.721282958984, "step_time_sec": 8.23598822200438, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.131737232208252, "lr": 0.0002, "elapsed_sec": 6041.959109783173, "step_time_sec": 8.23764623698662, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.051253795623779, "lr": 0.0002, "elapsed_sec": 6050.195812225342, "step_time_sec": 8.236546811007429, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.983297348022461, "lr": 0.0002, "elapsed_sec": 6058.433792591095, "step_time_sec": 8.237864132999675, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.02085018157959, "lr": 0.0002, "elapsed_sec": 6066.672013282776, "step_time_sec": 8.238079567003297, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.090119361877441, "lr": 0.0002, "elapsed_sec": 6074.9085252285, "step_time_sec": 8.236316093010828, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.11627721786499, "lr": 0.0002, "elapsed_sec": 6083.145663022995, "step_time_sec": 8.236967909004306, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.078520774841309, "lr": 0.0002, "elapsed_sec": 6091.382200479507, "step_time_sec": 8.236382378003327, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.182251930236816, "lr": 0.0002, "elapsed_sec": 6099.620315790176, "step_time_sec": 8.237970353977289, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.115259647369385, "lr": 0.0002, "elapsed_sec": 6107.857985019684, "step_time_sec": 8.23747186199762, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.182945251464844, "lr": 0.0002, "elapsed_sec": 6116.094750881195, "step_time_sec": 8.236617140995804, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.992626667022705, "lr": 0.0002, "elapsed_sec": 6124.331707239151, "step_time_sec": 8.236791032977635, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.102729797363281, "lr": 0.0002, "elapsed_sec": 6132.5683534145355, "step_time_sec": 8.236510941002052, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.087621688842773, "lr": 0.0002, "elapsed_sec": 6140.804975986481, "step_time_sec": 8.236408097000094, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.078227519989014, "lr": 0.0002, "elapsed_sec": 6149.040550470352, "step_time_sec": 8.235476652975194, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.079198360443115, "lr": 0.0002, "elapsed_sec": 6157.278214216232, "step_time_sec": 8.237472313980106, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.091529369354248, "lr": 0.0002, "elapsed_sec": 6165.515021800995, "step_time_sec": 8.236598354997113, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.100127220153809, "lr": 0.0002, "elapsed_sec": 6173.752846479416, "step_time_sec": 8.2377257150074, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.114923000335693, "lr": 0.0002, "elapsed_sec": 6181.990217924118, "step_time_sec": 8.237153442983981, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.311781883239746, "lr": 0.0002, "elapsed_sec": 6190.2265911102295, "step_time_sec": 8.236170914984541, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.119918346405029, "lr": 0.0002, "elapsed_sec": 6198.463321447372, "step_time_sec": 8.23664996799198, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.127298831939697, "lr": 0.0002, "elapsed_sec": 6206.698607444763, "step_time_sec": 8.235055180004565, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.1080827713012695, "lr": 0.0002, "elapsed_sec": 6214.936411380768, "step_time_sec": 8.23767231500824, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.0897016525268555, "lr": 0.0002, "elapsed_sec": 6223.172985076904, "step_time_sec": 8.236369838006794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.030884265899658, "lr": 0.0002, "elapsed_sec": 6231.410654783249, "step_time_sec": 8.237540419999277, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.085892677307129, "lr": 0.0002, "elapsed_sec": 6239.647317409515, "step_time_sec": 8.236487015994499, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.120818614959717, "lr": 0.0002, "elapsed_sec": 6247.883566379547, "step_time_sec": 8.23609028098872, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.976187705993652, "lr": 0.0002, "elapsed_sec": 6256.121572256088, "step_time_sec": 8.23785576299997, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.1992974281311035, "lr": 0.0002, "elapsed_sec": 6264.357157230377, "step_time_sec": 8.235428356012562, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.0289692878723145, "lr": 0.0002, "elapsed_sec": 6272.594284296036, "step_time_sec": 8.236962127004517, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.110548496246338, "lr": 0.0002, "elapsed_sec": 6280.830607414246, "step_time_sec": 8.23615425801836, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.930358409881592, "lr": 0.0002, "elapsed_sec": 6289.067999839783, "step_time_sec": 8.23721414298052, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.058958053588867, "lr": 0.0002, "elapsed_sec": 6297.304251432419, "step_time_sec": 8.236093462997815, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.094589710235596, "lr": 0.0002, "elapsed_sec": 6305.540968418121, "step_time_sec": 8.236564833001466, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.0531158447265625, "lr": 0.0002, "elapsed_sec": 6313.777856826782, "step_time_sec": 8.236798311991151, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.1230034828186035, "lr": 0.0002, "elapsed_sec": 6322.015950202942, "step_time_sec": 8.237869200005662, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.130890369415283, "lr": 0.0002, "elapsed_sec": 6330.253108024597, "step_time_sec": 8.236960234993603, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.062107563018799, "lr": 0.0002, "elapsed_sec": 6338.487840890884, "step_time_sec": 8.234555678995093, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.051578998565674, "lr": 0.0002, "elapsed_sec": 6346.725741624832, "step_time_sec": 8.237730849010404, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.163684368133545, "lr": 0.0002, "elapsed_sec": 6354.962342739105, "step_time_sec": 8.236483419983415, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.057170867919922, "lr": 0.0002, "elapsed_sec": 6363.201654672623, "step_time_sec": 8.239096269011497, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.924957275390625, "lr": 0.0002, "elapsed_sec": 6371.438304424286, "step_time_sec": 8.236498423997546, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.1559672355651855, "lr": 0.0002, "elapsed_sec": 6379.675360441208, "step_time_sec": 8.236897405993659, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.076247692108154, "lr": 0.0002, "elapsed_sec": 6387.913080215454, "step_time_sec": 8.237593385973014, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.033394813537598, "lr": 0.0002, "elapsed_sec": 6396.149262428284, "step_time_sec": 8.236005143000511, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.292604446411133, "lr": 0.0002, "elapsed_sec": 6404.385922431946, "step_time_sec": 8.236432617006358, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.063112258911133, "lr": 0.0002, "elapsed_sec": 6412.622247695923, "step_time_sec": 8.23623716601287, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.050293922424316, "lr": 0.0002, "elapsed_sec": 6420.859649419785, "step_time_sec": 8.237167443992803, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.052756309509277, "lr": 0.0002, "elapsed_sec": 6429.097723722458, "step_time_sec": 8.23791648200131, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.056283473968506, "lr": 0.0002, "elapsed_sec": 6437.3341999053955, "step_time_sec": 8.236323960998561, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.050161838531494, "lr": 0.0002, "elapsed_sec": 6445.57182598114, "step_time_sec": 8.237489873019513, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.010027885437012, "lr": 0.0002, "elapsed_sec": 6453.808660507202, "step_time_sec": 8.236663110001246, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.927030086517334, "lr": 0.0002, "elapsed_sec": 6462.04515004158, "step_time_sec": 8.236282917001517, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.0868024826049805, "lr": 0.0002, "elapsed_sec": 6470.281522274017, "step_time_sec": 8.23621333099436, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.048264503479004, "lr": 0.0002, "elapsed_sec": 6478.5195055007935, "step_time_sec": 8.237841627007583, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.0599894523620605, "lr": 0.0002, "elapsed_sec": 6486.755946397781, "step_time_sec": 8.236308880004799, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.985924243927002, "lr": 0.0002, "elapsed_sec": 6494.9939522743225, "step_time_sec": 8.237834972009296, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.1278767585754395, "lr": 0.0002, "elapsed_sec": 6503.232009410858, "step_time_sec": 8.237871600023936, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.073634624481201, "lr": 0.0002, "elapsed_sec": 6511.469335079193, "step_time_sec": 8.237151492998237, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.978202819824219, "lr": 0.0002, "elapsed_sec": 6519.707597732544, "step_time_sec": 8.23809744999744, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.984084129333496, "lr": 0.0002, "elapsed_sec": 6527.943871021271, "step_time_sec": 8.236148085998138, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.022728443145752, "lr": 0.0002, "elapsed_sec": 6536.18253493309, "step_time_sec": 8.238534460979281, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.990434169769287, "lr": 0.0002, "elapsed_sec": 6544.419070482254, "step_time_sec": 8.236310502019478, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.146414756774902, "lr": 0.0002, "elapsed_sec": 6552.656081438065, "step_time_sec": 8.236864628997864, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.990077495574951, "lr": 0.0002, "elapsed_sec": 6560.892368793488, "step_time_sec": 8.236150337004801, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.999990463256836, "lr": 0.0002, "elapsed_sec": 6569.130359649658, "step_time_sec": 8.2378425639763, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.031276702880859, "lr": 0.0002, "elapsed_sec": 6577.368027687073, "step_time_sec": 8.237481465999736, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.01285982131958, "lr": 0.0002, "elapsed_sec": 6585.604742050171, "step_time_sec": 8.236540859012166, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.046972751617432, "lr": 0.0002, "elapsed_sec": 6593.841906070709, "step_time_sec": 8.237080966995563, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.11688756942749, "lr": 0.0002, "elapsed_sec": 6602.079381227493, "step_time_sec": 8.23723090198473, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.036430358886719, "lr": 0.0002, "elapsed_sec": 6610.316689014435, "step_time_sec": 8.237161810015095, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.0859832763671875, "lr": 0.0002, "elapsed_sec": 6618.552015304565, "step_time_sec": 8.23517378201359, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.041625022888184, "lr": 0.0002, "elapsed_sec": 6626.787858724594, "step_time_sec": 8.235629222996067, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.029399394989014, "lr": 0.0002, "elapsed_sec": 6635.024663925171, "step_time_sec": 8.236673606996192, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.9881062507629395, "lr": 0.0002, "elapsed_sec": 6643.261088609695, "step_time_sec": 8.236198384023737, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.037793159484863, "lr": 0.0002, "elapsed_sec": 6651.4987099170685, "step_time_sec": 8.237472276989138, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.122070789337158, "lr": 0.0002, "elapsed_sec": 6659.733437776566, "step_time_sec": 8.234575337992283, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.04747200012207, "lr": 0.0002, "elapsed_sec": 6667.9710965156555, "step_time_sec": 8.237519228016026, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.060126304626465, "lr": 0.0002, "elapsed_sec": 6676.20808339119, "step_time_sec": 8.236805602005916, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.9853339195251465, "lr": 0.0002, "elapsed_sec": 6684.446483373642, "step_time_sec": 8.23824142900412, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.002078056335449, "lr": 0.0002, "elapsed_sec": 6692.682155370712, "step_time_sec": 8.235497019981267, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.052155017852783, "lr": 0.0002, "elapsed_sec": 6700.918377399445, "step_time_sec": 8.236094795021927, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.117384433746338, "lr": 0.0002, "elapsed_sec": 6722.212227344513, "step_time_sec": 21.293671216990333, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.0155720710754395, "lr": 0.0002, "elapsed_sec": 6730.442498445511, "step_time_sec": 8.230119076994015, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.97226619720459, "lr": 0.0002, "elapsed_sec": 6738.678616285324, "step_time_sec": 8.235972518014023, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.943271160125732, "lr": 0.0002, "elapsed_sec": 6746.916189432144, "step_time_sec": 8.2373287990049, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.864638328552246, "lr": 0.0002, "elapsed_sec": 6755.153941631317, "step_time_sec": 8.237637543003075, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.003595352172852, "lr": 0.0002, "elapsed_sec": 6763.389990091324, "step_time_sec": 8.235833623009967, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.0582356452941895, "lr": 0.0002, "elapsed_sec": 6771.6280953884125, "step_time_sec": 8.237956720986404, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.096101760864258, "lr": 0.0002, "elapsed_sec": 6779.865149259567, "step_time_sec": 8.236940165021224, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.974308490753174, "lr": 0.0002, "elapsed_sec": 6788.103606700897, "step_time_sec": 8.238252626004396, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.047638893127441, "lr": 0.0002, "elapsed_sec": 6796.340350151062, "step_time_sec": 8.236556560004828, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.99217414855957, "lr": 0.0002, "elapsed_sec": 6804.577848434448, "step_time_sec": 8.237348944996484, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.129955291748047, "lr": 0.0002, "elapsed_sec": 6812.815900802612, "step_time_sec": 8.237867633986752, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.046621322631836, "lr": 0.0002, "elapsed_sec": 6821.051472902298, "step_time_sec": 8.235438924020855, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.969282627105713, "lr": 0.0002, "elapsed_sec": 6829.28880572319, "step_time_sec": 8.237165793980239, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.937124729156494, "lr": 0.0002, "elapsed_sec": 6837.524507522583, "step_time_sec": 8.235518728994066, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.023293495178223, "lr": 0.0002, "elapsed_sec": 6845.762650728226, "step_time_sec": 8.23798035699292, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.021817207336426, "lr": 0.0002, "elapsed_sec": 6853.997086286545, "step_time_sec": 8.23430852999445, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.1503424644470215, "lr": 0.0002, "elapsed_sec": 6862.235659360886, "step_time_sec": 8.238378407026175, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.966644287109375, "lr": 0.0002, "elapsed_sec": 6870.473114490509, "step_time_sec": 8.237334129982628, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.0780487060546875, "lr": 0.0002, "elapsed_sec": 6878.710217952728, "step_time_sec": 8.236924700991949, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.046118259429932, "lr": 0.0002, "elapsed_sec": 6886.94860458374, "step_time_sec": 8.238161222019698, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.0031418800354, "lr": 0.0002, "elapsed_sec": 6895.1858167648315, "step_time_sec": 8.237066247995244, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.0674967765808105, "lr": 0.0002, "elapsed_sec": 6903.423884630203, "step_time_sec": 8.237886635994073, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.005364418029785, "lr": 0.0002, "elapsed_sec": 6911.660587310791, "step_time_sec": 8.236572783003794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.014676570892334, "lr": 0.0002, "elapsed_sec": 6919.899700164795, "step_time_sec": 8.23891950401594, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.971290588378906, "lr": 0.0002, "elapsed_sec": 6928.13791847229, "step_time_sec": 8.2380840620026, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.002715587615967, "lr": 0.0002, "elapsed_sec": 6936.373075008392, "step_time_sec": 8.234964433009736, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.080801010131836, "lr": 0.0002, "elapsed_sec": 6944.610759973526, "step_time_sec": 8.237524302006932, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.00225305557251, "lr": 0.0002, "elapsed_sec": 6952.846914052963, "step_time_sec": 8.235964835010236, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.976263523101807, "lr": 0.0002, "elapsed_sec": 6961.084822654724, "step_time_sec": 8.237797929992666, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.944404125213623, "lr": 0.0002, "elapsed_sec": 6969.320554733276, "step_time_sec": 8.235571021999931, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.948369026184082, "lr": 0.0002, "elapsed_sec": 6977.557336807251, "step_time_sec": 8.236592678993475, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.995006561279297, "lr": 0.0002, "elapsed_sec": 6985.795157909393, "step_time_sec": 8.237621620995924, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.920032501220703, "lr": 0.0002, "elapsed_sec": 6994.032251834869, "step_time_sec": 8.236966718017356, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.8711724281311035, "lr": 0.0002, "elapsed_sec": 7002.269958496094, "step_time_sec": 8.23750306700822, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.956996917724609, "lr": 0.0002, "elapsed_sec": 7010.507392406464, "step_time_sec": 8.23732850298984, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.033047676086426, "lr": 0.0002, "elapsed_sec": 7018.746652603149, "step_time_sec": 8.239105268992716, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.997239112854004, "lr": 0.0002, "elapsed_sec": 7026.983635425568, "step_time_sec": 8.236763142980635, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.9298834800720215, "lr": 0.0002, "elapsed_sec": 7035.220988035202, "step_time_sec": 8.237208578997524, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.895323753356934, "lr": 0.0002, "elapsed_sec": 7043.458063364029, "step_time_sec": 8.23690109400195, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.957430839538574, "lr": 0.0002, "elapsed_sec": 7051.695625305176, "step_time_sec": 8.237407214008272, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.019507884979248, "lr": 0.0002, "elapsed_sec": 7059.933775186539, "step_time_sec": 8.237968219997128, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.928638458251953, "lr": 0.0002, "elapsed_sec": 7068.170227050781, "step_time_sec": 8.236293683003169, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.9171576499938965, "lr": 0.0002, "elapsed_sec": 7076.408440113068, "step_time_sec": 8.238096953980858, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.010960578918457, "lr": 0.0002, "elapsed_sec": 7084.64436173439, "step_time_sec": 8.2357066340046, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.972098350524902, "lr": 0.0002, "elapsed_sec": 7092.881776094437, "step_time_sec": 8.23723501097993, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.084504127502441, "lr": 0.0002, "elapsed_sec": 7101.1186945438385, "step_time_sec": 8.236768600996584, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.054569721221924, "lr": 0.0002, "elapsed_sec": 7109.355339288712, "step_time_sec": 8.23651121702278, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.971547603607178, "lr": 0.0002, "elapsed_sec": 7117.593157052994, "step_time_sec": 8.237618793005822, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.970885276794434, "lr": 0.0002, "elapsed_sec": 7125.830463886261, "step_time_sec": 8.237134757015156, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.925214767456055, "lr": 0.0002, "elapsed_sec": 7134.067965745926, "step_time_sec": 8.237370893999469, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.932915687561035, "lr": 0.0002, "elapsed_sec": 7142.3048684597015, "step_time_sec": 8.236693405982805, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.753531455993652, "lr": 0.0002, "elapsed_sec": 7150.542473077774, "step_time_sec": 8.237477517017396, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.868962287902832, "lr": 0.0002, "elapsed_sec": 7158.779146671295, "step_time_sec": 8.236499250982888, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.009196758270264, "lr": 0.0002, "elapsed_sec": 7167.016592264175, "step_time_sec": 8.237244403979275, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.01922607421875, "lr": 0.0002, "elapsed_sec": 7175.254040718079, "step_time_sec": 8.237297085986938, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.949853420257568, "lr": 0.0002, "elapsed_sec": 7183.490799188614, "step_time_sec": 8.236588850006228, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.976675987243652, "lr": 0.0002, "elapsed_sec": 7191.727389574051, "step_time_sec": 8.236426352988929, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.974020957946777, "lr": 0.0002, "elapsed_sec": 7199.963918685913, "step_time_sec": 8.236411493009655, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.00684118270874, "lr": 0.0002, "elapsed_sec": 7208.201222658157, "step_time_sec": 8.237095475982642, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.915319919586182, "lr": 0.0002, "elapsed_sec": 7216.4374170303345, "step_time_sec": 8.236011992004933, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.939164638519287, "lr": 0.0002, "elapsed_sec": 7224.675018310547, "step_time_sec": 8.237434829992708, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.889339923858643, "lr": 0.0002, "elapsed_sec": 7232.911639213562, "step_time_sec": 8.236435917002382, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.914118766784668, "lr": 0.0002, "elapsed_sec": 7241.148112773895, "step_time_sec": 8.23636338498909, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.920680046081543, "lr": 0.0002, "elapsed_sec": 7249.385782718658, "step_time_sec": 8.237474958994426, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.899847507476807, "lr": 0.0002, "elapsed_sec": 7257.621677398682, "step_time_sec": 8.235735731985187, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.920300483703613, "lr": 0.0002, "elapsed_sec": 7265.859512090683, "step_time_sec": 8.237670312984847, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.997809886932373, "lr": 0.0002, "elapsed_sec": 7274.095411777496, "step_time_sec": 8.235690498986514, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.779327392578125, "lr": 0.0002, "elapsed_sec": 7282.332247018814, "step_time_sec": 8.236684973002411, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.842092990875244, "lr": 0.0002, "elapsed_sec": 7290.569040298462, "step_time_sec": 8.236611630010884, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.910144329071045, "lr": 0.0002, "elapsed_sec": 7298.804836988449, "step_time_sec": 8.235661550977966, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.969423294067383, "lr": 0.0002, "elapsed_sec": 7307.042111873627, "step_time_sec": 8.237064241984626, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.975088119506836, "lr": 0.0002, "elapsed_sec": 7315.27822637558, "step_time_sec": 8.23598746798234, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.977208137512207, "lr": 0.0002, "elapsed_sec": 7323.515495061874, "step_time_sec": 8.237067744979868, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.9344072341918945, "lr": 0.0002, "elapsed_sec": 7331.752177476883, "step_time_sec": 8.236514835007256, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.937851905822754, "lr": 0.0002, "elapsed_sec": 7339.990102529526, "step_time_sec": 8.237746535014594, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.985236644744873, "lr": 0.0002, "elapsed_sec": 7348.227266788483, "step_time_sec": 8.237000969005749, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.871114730834961, "lr": 0.0002, "elapsed_sec": 7356.464807987213, "step_time_sec": 8.23738905999926, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.889721393585205, "lr": 0.0002, "elapsed_sec": 7364.701236963272, "step_time_sec": 8.236302691017045, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.935098171234131, "lr": 0.0002, "elapsed_sec": 7372.93839597702, "step_time_sec": 8.23696413898142, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.939993858337402, "lr": 0.0002, "elapsed_sec": 7381.177031517029, "step_time_sec": 8.238468392984942, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.902981281280518, "lr": 0.0002, "elapsed_sec": 7389.413003206253, "step_time_sec": 8.235814665007638, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.878725528717041, "lr": 0.0002, "elapsed_sec": 7397.649966478348, "step_time_sec": 8.236813796014758, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.784796237945557, "lr": 0.0002, "elapsed_sec": 7405.885729789734, "step_time_sec": 8.235617336002178, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.856338024139404, "lr": 0.0002, "elapsed_sec": 7414.123877048492, "step_time_sec": 8.23794388800161, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.866729736328125, "lr": 0.0002, "elapsed_sec": 7422.359893083572, "step_time_sec": 8.235886339010904, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.897148132324219, "lr": 0.0002, "elapsed_sec": 7430.597174167633, "step_time_sec": 8.237156004994176, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.8050689697265625, "lr": 0.0002, "elapsed_sec": 7438.834415912628, "step_time_sec": 8.237067154987017, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.048725605010986, "lr": 0.0002, "elapsed_sec": 7447.070693016052, "step_time_sec": 8.236061808012892, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.847606658935547, "lr": 0.0002, "elapsed_sec": 7455.307766914368, "step_time_sec": 8.236918084992794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.928175449371338, "lr": 0.0002, "elapsed_sec": 7463.543378353119, "step_time_sec": 8.235472325992305, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.912534713745117, "lr": 0.0002, "elapsed_sec": 7471.78101849556, "step_time_sec": 8.237470245017903, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.860657691955566, "lr": 0.0002, "elapsed_sec": 7480.016602039337, "step_time_sec": 8.235392375005176, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.185755729675293, "lr": 0.0002, "elapsed_sec": 7488.253737211227, "step_time_sec": 8.237005126982694, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.982276439666748, "lr": 0.0002, "elapsed_sec": 7496.4902131557465, "step_time_sec": 8.236356488021556, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.927794456481934, "lr": 0.0002, "elapsed_sec": 7504.727532148361, "step_time_sec": 8.237115827010712, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.88494348526001, "lr": 0.0002, "elapsed_sec": 7512.964071035385, "step_time_sec": 8.236340696981642, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.101069927215576, "lr": 0.0002, "elapsed_sec": 7521.200751543045, "step_time_sec": 8.236550334986532, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.804977893829346, "lr": 0.0002, "elapsed_sec": 7529.436897039413, "step_time_sec": 8.235985464998521, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.878833293914795, "lr": 0.0002, "elapsed_sec": 7537.6714062690735, "step_time_sec": 8.234368100995198, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.889571666717529, "lr": 0.0002, "elapsed_sec": 7545.908013820648, "step_time_sec": 8.236421901994618, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.90641975402832, "lr": 0.0002, "elapsed_sec": 7554.144714832306, "step_time_sec": 8.236554265022278, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.951208114624023, "lr": 0.0002, "elapsed_sec": 7562.383562803268, "step_time_sec": 8.238634039997123, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.891989231109619, "lr": 0.0002, "elapsed_sec": 7570.620109796524, "step_time_sec": 8.236425946000963, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.920272350311279, "lr": 0.0002, "elapsed_sec": 7578.8575003147125, "step_time_sec": 8.237246117991162, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.767282962799072, "lr": 0.0002, "elapsed_sec": 7587.095595836639, "step_time_sec": 8.237884917994961, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.842907428741455, "lr": 0.0002, "elapsed_sec": 7595.331633806229, "step_time_sec": 8.235846618015785, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 6.044867992401123, "lr": 0.0002, "elapsed_sec": 7603.568266630173, "step_time_sec": 8.236477546015522, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.845921039581299, "lr": 0.0002, "elapsed_sec": 7611.805160284042, "step_time_sec": 8.236727300012717, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.8097357749938965, "lr": 0.0002, "elapsed_sec": 7620.042300224304, "step_time_sec": 8.236977842985652, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.923910140991211, "lr": 0.0002, "elapsed_sec": 7628.278832435608, "step_time_sec": 8.236344545002794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.914360523223877, "lr": 0.0002, "elapsed_sec": 7636.5155329704285, "step_time_sec": 8.236529393005185, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.86039400100708, "lr": 0.0002, "elapsed_sec": 7644.75387597084, "step_time_sec": 8.23822744100471, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.928979873657227, "lr": 0.0002, "elapsed_sec": 7652.991692781448, "step_time_sec": 8.237623178982176, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.796677112579346, "lr": 0.0002, "elapsed_sec": 7661.227825164795, "step_time_sec": 8.235993128997507, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.850037097930908, "lr": 0.0002, "elapsed_sec": 7669.4640600681305, "step_time_sec": 8.236057492002146, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.784773349761963, "lr": 0.0002, "elapsed_sec": 7677.70262336731, "step_time_sec": 8.238438992993906, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.8460869789123535, "lr": 0.0002, "elapsed_sec": 7685.938832521439, "step_time_sec": 8.236015353992116, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.837003707885742, "lr": 0.0002, "elapsed_sec": 7694.176451683044, "step_time_sec": 8.237408274988411, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.869614601135254, "lr": 0.0002, "elapsed_sec": 7702.413859844208, "step_time_sec": 8.237266306998208, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.860983848571777, "lr": 0.0002, "elapsed_sec": 7710.651837110519, "step_time_sec": 8.237837195018074, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.890157699584961, "lr": 0.0002, "elapsed_sec": 7718.889470100403, "step_time_sec": 8.237502136005787, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.718137264251709, "lr": 0.0002, "elapsed_sec": 7727.125346899033, "step_time_sec": 8.235652431001654, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.79058313369751, "lr": 0.0002, "elapsed_sec": 7735.362640142441, "step_time_sec": 8.23715882797842, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.964085102081299, "lr": 0.0002, "elapsed_sec": 7743.600488901138, "step_time_sec": 8.237691230984638, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.820768356323242, "lr": 0.0002, "elapsed_sec": 7751.838768482208, "step_time_sec": 8.238141375011764, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.8342084884643555, "lr": 0.0002, "elapsed_sec": 7760.0765380859375, "step_time_sec": 8.23759772797348, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.820967674255371, "lr": 0.0002, "elapsed_sec": 7768.313724994659, "step_time_sec": 8.237013017991558, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.803562641143799, "lr": 0.0002, "elapsed_sec": 7776.551270484924, "step_time_sec": 8.237367091001943, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.8424506187438965, "lr": 0.0002, "elapsed_sec": 7784.788303613663, "step_time_sec": 8.236883793986635, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.934298515319824, "lr": 0.0002, "elapsed_sec": 7793.026845932007, "step_time_sec": 8.238331564003602, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.834733486175537, "lr": 0.0002, "elapsed_sec": 7801.264855861664, "step_time_sec": 8.237903934001224, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.74855375289917, "lr": 0.0002, "elapsed_sec": 7809.5033259391785, "step_time_sec": 8.238259064004524, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.817145824432373, "lr": 0.0002, "elapsed_sec": 7817.741291046143, "step_time_sec": 8.237871261982946, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.926942825317383, "lr": 0.0002, "elapsed_sec": 7825.978618860245, "step_time_sec": 8.237159825017443, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.808021068572998, "lr": 0.0002, "elapsed_sec": 7834.214910030365, "step_time_sec": 8.236069702979876, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.9782891273498535, "lr": 0.0002, "elapsed_sec": 7842.451607227325, "step_time_sec": 8.23655954998685, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.880135536193848, "lr": 0.0002, "elapsed_sec": 7850.690924167633, "step_time_sec": 8.239093896001577, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.815313339233398, "lr": 0.0002, "elapsed_sec": 7858.9286296367645, "step_time_sec": 8.237593033991288, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.875834941864014, "lr": 0.0002, "elapsed_sec": 7867.1661467552185, "step_time_sec": 8.237284798990004, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.818961143493652, "lr": 0.0002, "elapsed_sec": 7875.40373301506, "step_time_sec": 8.237405438005226, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.732355117797852, "lr": 0.0002, "elapsed_sec": 7883.640203714371, "step_time_sec": 8.236300324992044, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.846202850341797, "lr": 0.0002, "elapsed_sec": 7891.87934088707, "step_time_sec": 8.238979045010637, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.7709832191467285, "lr": 0.0002, "elapsed_sec": 7900.116387605667, "step_time_sec": 8.236918956012232, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.797693252563477, "lr": 0.0002, "elapsed_sec": 7908.353396892548, "step_time_sec": 8.236829418019624, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.799264430999756, "lr": 0.0002, "elapsed_sec": 7916.590812921524, "step_time_sec": 8.237271527003031, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.831547260284424, "lr": 0.0002, "elapsed_sec": 7924.828589439392, "step_time_sec": 8.23756740099634, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.8811259269714355, "lr": 0.0002, "elapsed_sec": 7933.06637430191, "step_time_sec": 8.237622149987146, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.812170028686523, "lr": 0.0002, "elapsed_sec": 7941.302293300629, "step_time_sec": 8.23579429500387, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.853306770324707, "lr": 0.0002, "elapsed_sec": 7949.539560079575, "step_time_sec": 8.23706812900491, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.85029935836792, "lr": 0.0002, "elapsed_sec": 7957.775779724121, "step_time_sec": 8.236097993009025, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.8588128089904785, "lr": 0.0002, "elapsed_sec": 7966.012952804565, "step_time_sec": 8.23701723999693, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.880611419677734, "lr": 0.0002, "elapsed_sec": 7974.249973058701, "step_time_sec": 8.236877527990146, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.7804765701293945, "lr": 0.0002, "elapsed_sec": 7982.487449884415, "step_time_sec": 8.237250749982195, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.850569725036621, "lr": 0.0002, "elapsed_sec": 7990.724542140961, "step_time_sec": 8.236920657975134, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.851251125335693, "lr": 0.0002, "elapsed_sec": 7998.961110830307, "step_time_sec": 8.236458022001898, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.789341449737549, "lr": 0.0002, "elapsed_sec": 8007.198881864548, "step_time_sec": 8.237589915981516, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.912563323974609, "lr": 0.0002, "elapsed_sec": 8015.434274196625, "step_time_sec": 8.235190081992187, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.797858238220215, "lr": 0.0002, "elapsed_sec": 8023.671329975128, "step_time_sec": 8.236889961990528, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.698955535888672, "lr": 0.0002, "elapsed_sec": 8031.9080901145935, "step_time_sec": 8.236638998991111, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.708131313323975, "lr": 0.0002, "elapsed_sec": 8040.146644830704, "step_time_sec": 8.238363688986283, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.753489971160889, "lr": 0.0002, "elapsed_sec": 8048.38297867775, "step_time_sec": 8.236166009010049, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.829041957855225, "lr": 0.0002, "elapsed_sec": 8056.620685815811, "step_time_sec": 8.237547202006681, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.901814937591553, "lr": 0.0002, "elapsed_sec": 8064.858589172363, "step_time_sec": 8.237730313994689, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.820355415344238, "lr": 0.0002, "elapsed_sec": 8073.095122337341, "step_time_sec": 8.23639448301401, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.850785255432129, "lr": 0.0002, "elapsed_sec": 8081.332142353058, "step_time_sec": 8.236864495003829, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.8989691734313965, "lr": 0.0002, "elapsed_sec": 8089.56755566597, "step_time_sec": 8.235187248006696, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.774062633514404, "lr": 0.0002, "elapsed_sec": 8097.806143522263, "step_time_sec": 8.238495997997234, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.970680236816406, "lr": 0.0002, "elapsed_sec": 8106.043536424637, "step_time_sec": 8.237177334987791, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.854933261871338, "lr": 0.0002, "elapsed_sec": 8114.280451774597, "step_time_sec": 8.236753540986683, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.755605220794678, "lr": 0.0002, "elapsed_sec": 8122.51718711853, "step_time_sec": 8.236567215994, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.831167697906494, "lr": 0.0002, "elapsed_sec": 8130.753693342209, "step_time_sec": 8.236331218009582, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.914340972900391, "lr": 0.0002, "elapsed_sec": 8138.990778207779, "step_time_sec": 8.236973448016215, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.690707683563232, "lr": 0.0002, "elapsed_sec": 8147.226870536804, "step_time_sec": 8.235880129010184, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.936673164367676, "lr": 0.0002, "elapsed_sec": 8155.463293552399, "step_time_sec": 8.23622796998825, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.809830188751221, "lr": 0.0002, "elapsed_sec": 8163.699496984482, "step_time_sec": 8.236054790002527, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.79762601852417, "lr": 0.0002, "elapsed_sec": 8171.936800956726, "step_time_sec": 8.23713386899908, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.759929180145264, "lr": 0.0002, "elapsed_sec": 8180.173734903336, "step_time_sec": 8.236772666976321, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.888732433319092, "lr": 0.0002, "elapsed_sec": 8188.411318778992, "step_time_sec": 8.237423843995202, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.743844509124756, "lr": 0.0002, "elapsed_sec": 8196.648805856705, "step_time_sec": 8.237290791992564, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.742275238037109, "lr": 0.0002, "elapsed_sec": 8204.885270118713, "step_time_sec": 8.23631948599359, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.809820652008057, "lr": 0.0002, "elapsed_sec": 8213.12358880043, "step_time_sec": 8.238153951999266, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.766637802124023, "lr": 0.0002, "elapsed_sec": 8221.361166715622, "step_time_sec": 8.237473919987679, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.830974102020264, "lr": 0.0002, "elapsed_sec": 8229.598257780075, "step_time_sec": 8.236892624001484, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.797764301300049, "lr": 0.0002, "elapsed_sec": 8237.835694789886, "step_time_sec": 8.237287262018071, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.803894996643066, "lr": 0.0002, "elapsed_sec": 8246.0726583004, "step_time_sec": 8.236774772987701, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.799333095550537, "lr": 0.0002, "elapsed_sec": 8254.31106877327, "step_time_sec": 8.238240272999974, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.78566312789917, "lr": 0.0002, "elapsed_sec": 8262.548253774643, "step_time_sec": 8.237087874993449, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.8581743240356445, "lr": 0.0002, "elapsed_sec": 8270.78662109375, "step_time_sec": 53.22661468500155, "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": 1.0771121399884578, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 1001, "loss": 5.759693622589111, "lr": 0.0002, "elapsed_sec": 8324.018045663834, "step_time_sec": 8.242908666987205, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.7391676902771, "lr": 0.0002, "elapsed_sec": 8332.248308897018, "step_time_sec": 8.230095120990882, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.84536075592041, "lr": 0.0002, "elapsed_sec": 8340.485296964645, "step_time_sec": 8.236837261996698, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.822324752807617, "lr": 0.0002, "elapsed_sec": 8348.722545862198, "step_time_sec": 8.237061675987206, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.695547580718994, "lr": 0.0002, "elapsed_sec": 8356.96180653572, "step_time_sec": 8.23907546600094, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.788199424743652, "lr": 0.0002, "elapsed_sec": 8365.200627326965, "step_time_sec": 8.238661243987735, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.855216026306152, "lr": 0.0002, "elapsed_sec": 8373.437537431717, "step_time_sec": 8.236758330982411, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.785156726837158, "lr": 0.0002, "elapsed_sec": 8381.67481803894, "step_time_sec": 8.237110426009167, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.810042858123779, "lr": 0.0002, "elapsed_sec": 8389.910770654678, "step_time_sec": 8.235725900012767, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.761291027069092, "lr": 0.0002, "elapsed_sec": 8398.149317264557, "step_time_sec": 8.23842420699657, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.720808506011963, "lr": 0.0002, "elapsed_sec": 8406.387645959854, "step_time_sec": 8.238202304986771, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.7300214767456055, "lr": 0.0002, "elapsed_sec": 8414.625737667084, "step_time_sec": 8.237878807005472, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.7202653884887695, "lr": 0.0002, "elapsed_sec": 8422.863728523254, "step_time_sec": 8.237870640004985, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.763113498687744, "lr": 0.0002, "elapsed_sec": 8431.100826978683, "step_time_sec": 8.236901733005652, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.728540420532227, "lr": 0.0002, "elapsed_sec": 8439.340238332748, "step_time_sec": 8.23920160101261, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.716302871704102, "lr": 0.0002, "elapsed_sec": 8447.578392744064, "step_time_sec": 8.238030559994513, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.744083404541016, "lr": 0.0002, "elapsed_sec": 8455.81620645523, "step_time_sec": 8.237640252016718, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.9691314697265625, "lr": 0.0002, "elapsed_sec": 8464.053763151169, "step_time_sec": 8.23735559702618, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.618669033050537, "lr": 0.0002, "elapsed_sec": 8472.290515899658, "step_time_sec": 8.236585213977378, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.773624897003174, "lr": 0.0002, "elapsed_sec": 8480.527341842651, "step_time_sec": 8.236672842002008, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.845386028289795, "lr": 0.0002, "elapsed_sec": 8488.763563156128, "step_time_sec": 8.236071937019005, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.764163017272949, "lr": 0.0002, "elapsed_sec": 8497.00185918808, "step_time_sec": 8.23809604998678, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.663121700286865, "lr": 0.0002, "elapsed_sec": 8505.239139080048, "step_time_sec": 8.237146825995296, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.775842666625977, "lr": 0.0002, "elapsed_sec": 8513.477318286896, "step_time_sec": 8.238016664981842, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.813530445098877, "lr": 0.0002, "elapsed_sec": 8521.71540927887, "step_time_sec": 8.237936709017958, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.767189025878906, "lr": 0.0002, "elapsed_sec": 8529.952260255814, "step_time_sec": 8.236623370001325, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.626843452453613, "lr": 0.0002, "elapsed_sec": 8538.189921855927, "step_time_sec": 8.237525214004563, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.606561183929443, "lr": 0.0002, "elapsed_sec": 8546.425962686539, "step_time_sec": 8.235911900002975, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.65390682220459, "lr": 0.0002, "elapsed_sec": 8554.66528582573, "step_time_sec": 8.239167735009687, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.7157158851623535, "lr": 0.0002, "elapsed_sec": 8562.904156684875, "step_time_sec": 8.238696187007008, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.814980983734131, "lr": 0.0002, "elapsed_sec": 8571.14169549942, "step_time_sec": 8.237358991988003, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.8004608154296875, "lr": 0.0002, "elapsed_sec": 8579.380250692368, "step_time_sec": 8.2384493870195, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.789820194244385, "lr": 0.0002, "elapsed_sec": 8587.616949558258, "step_time_sec": 8.236500939005055, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.729423522949219, "lr": 0.0002, "elapsed_sec": 8595.855839967728, "step_time_sec": 8.23871818400221, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.766083717346191, "lr": 0.0002, "elapsed_sec": 8604.093057394028, "step_time_sec": 8.23700109499623, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.889943599700928, "lr": 0.0002, "elapsed_sec": 8612.33068728447, "step_time_sec": 8.237464453006396, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.714629650115967, "lr": 0.0002, "elapsed_sec": 8620.568038463593, "step_time_sec": 8.237204964010743, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.674260139465332, "lr": 0.0002, "elapsed_sec": 8628.8057513237, "step_time_sec": 8.237575167004252, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.81787633895874, "lr": 0.0002, "elapsed_sec": 8637.043917655945, "step_time_sec": 8.237960743019357, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.707958698272705, "lr": 0.0002, "elapsed_sec": 8645.280309438705, "step_time_sec": 8.236217094992753, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.681210994720459, "lr": 0.0002, "elapsed_sec": 8653.517660617828, "step_time_sec": 8.237218606984243, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.719813823699951, "lr": 0.0002, "elapsed_sec": 8661.754129886627, "step_time_sec": 8.236295062000863, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.722426414489746, "lr": 0.0002, "elapsed_sec": 8669.990890264511, "step_time_sec": 8.236535965988878, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.737937927246094, "lr": 0.0002, "elapsed_sec": 8678.227781295776, "step_time_sec": 8.23673798199161, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.692834854125977, "lr": 0.0002, "elapsed_sec": 8686.46479678154, "step_time_sec": 8.236866004008334, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.794095993041992, "lr": 0.0002, "elapsed_sec": 8694.70149898529, "step_time_sec": 8.236541081016185, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.766648769378662, "lr": 0.0002, "elapsed_sec": 8702.937763690948, "step_time_sec": 8.236091865983326, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.769759178161621, "lr": 0.0002, "elapsed_sec": 8711.173951387405, "step_time_sec": 8.236006839986658, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.812398910522461, "lr": 0.0002, "elapsed_sec": 8719.41043806076, "step_time_sec": 8.236371067003347, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.815721035003662, "lr": 0.0002, "elapsed_sec": 8727.65019249916, "step_time_sec": 8.239604155998677, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.6471099853515625, "lr": 0.0002, "elapsed_sec": 8735.887768745422, "step_time_sec": 8.237417018972337, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.581630706787109, "lr": 0.0002, "elapsed_sec": 8744.125544071198, "step_time_sec": 8.237611060991185, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.8644633293151855, "lr": 0.0002, "elapsed_sec": 8752.363335371017, "step_time_sec": 8.237641651998274, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.7782487869262695, "lr": 0.0002, "elapsed_sec": 8760.60097360611, "step_time_sec": 8.237472423003055, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.69982385635376, "lr": 0.0002, "elapsed_sec": 8768.837849140167, "step_time_sec": 8.236740476015257, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.728814601898193, "lr": 0.0002, "elapsed_sec": 8777.075101852417, "step_time_sec": 8.23704969600658, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.388880729675293, "lr": 0.0002, "elapsed_sec": 8785.312616109848, "step_time_sec": 8.237339606013848, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.763959884643555, "lr": 0.0002, "elapsed_sec": 8793.55000114441, "step_time_sec": 8.237248587975046, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.6665873527526855, "lr": 0.0002, "elapsed_sec": 8801.787603378296, "step_time_sec": 8.237458249001065, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.659310817718506, "lr": 0.0002, "elapsed_sec": 8810.025331020355, "step_time_sec": 8.23754337299033, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.876332759857178, "lr": 0.0002, "elapsed_sec": 8818.261570930481, "step_time_sec": 8.236066387005849, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.737668991088867, "lr": 0.0002, "elapsed_sec": 8826.49835562706, "step_time_sec": 8.23669969200273, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.7659502029418945, "lr": 0.0002, "elapsed_sec": 8834.735675811768, "step_time_sec": 8.237116987002082, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.718911647796631, "lr": 0.0002, "elapsed_sec": 8842.972807645798, "step_time_sec": 8.237015580991283, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.680848598480225, "lr": 0.0002, "elapsed_sec": 8851.210770368576, "step_time_sec": 8.23780706099933, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.757038593292236, "lr": 0.0002, "elapsed_sec": 8859.44830083847, "step_time_sec": 8.23732685201685, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.911016941070557, "lr": 0.0002, "elapsed_sec": 8867.685842752457, "step_time_sec": 8.23741494701244, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.645969867706299, "lr": 0.0002, "elapsed_sec": 8875.922528266907, "step_time_sec": 8.236486959998729, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.574141979217529, "lr": 0.0002, "elapsed_sec": 8884.160932779312, "step_time_sec": 8.238273182010744, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.7064642906188965, "lr": 0.0002, "elapsed_sec": 8892.39843082428, "step_time_sec": 8.237343740009237, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.58833646774292, "lr": 0.0002, "elapsed_sec": 8900.635911464691, "step_time_sec": 8.237325085006887, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.7549872398376465, "lr": 0.0002, "elapsed_sec": 8908.873208761215, "step_time_sec": 8.237153661990305, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.6417388916015625, "lr": 0.0002, "elapsed_sec": 8917.109629154205, "step_time_sec": 8.236240202008048, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.682008266448975, "lr": 0.0002, "elapsed_sec": 8925.347229719162, "step_time_sec": 8.237405338993995, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.705203056335449, "lr": 0.0002, "elapsed_sec": 8933.583564043045, "step_time_sec": 8.236171193013433, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.727677345275879, "lr": 0.0002, "elapsed_sec": 8941.821832895279, "step_time_sec": 8.238158762978856, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.711902618408203, "lr": 0.0002, "elapsed_sec": 8950.05736041069, "step_time_sec": 8.235344671993516, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.645553112030029, "lr": 0.0002, "elapsed_sec": 8958.295033931732, "step_time_sec": 8.23748287500348, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.632348537445068, "lr": 0.0002, "elapsed_sec": 8966.53280878067, "step_time_sec": 8.237644084991189, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.685232639312744, "lr": 0.0002, "elapsed_sec": 8974.769669532776, "step_time_sec": 8.236667290999321, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.587974548339844, "lr": 0.0002, "elapsed_sec": 8983.006075382233, "step_time_sec": 8.236278844997287, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.650007724761963, "lr": 0.0002, "elapsed_sec": 8991.242751836777, "step_time_sec": 8.236518757999875, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.710019588470459, "lr": 0.0002, "elapsed_sec": 8999.480228185654, "step_time_sec": 8.237370238028234, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.697711944580078, "lr": 0.0002, "elapsed_sec": 9007.715534448624, "step_time_sec": 8.235113384987926, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.688304424285889, "lr": 0.0002, "elapsed_sec": 9015.95272898674, "step_time_sec": 8.237015054997755, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.651940822601318, "lr": 0.0002, "elapsed_sec": 9024.188536643982, "step_time_sec": 8.235665348998737, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.8284687995910645, "lr": 0.0002, "elapsed_sec": 9032.425976753235, "step_time_sec": 8.237228809011867, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.7142014503479, "lr": 0.0002, "elapsed_sec": 9040.66400027275, "step_time_sec": 8.23788997301017, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.665709018707275, "lr": 0.0002, "elapsed_sec": 9048.902587890625, "step_time_sec": 8.238477461010916, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.894207000732422, "lr": 0.0002, "elapsed_sec": 9057.141539812088, "step_time_sec": 8.23877814799198, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.732616424560547, "lr": 0.0002, "elapsed_sec": 9065.378744363785, "step_time_sec": 8.23698775700177, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.6060051918029785, "lr": 0.0002, "elapsed_sec": 9073.617750883102, "step_time_sec": 8.238861336983973, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.600897789001465, "lr": 0.0002, "elapsed_sec": 9081.855552434921, "step_time_sec": 8.237658239988377, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.737110614776611, "lr": 0.0002, "elapsed_sec": 9090.093654632568, "step_time_sec": 8.237901732994942, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.622531414031982, "lr": 0.0002, "elapsed_sec": 9098.33140873909, "step_time_sec": 8.237623645021813, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.682069778442383, "lr": 0.0002, "elapsed_sec": 9106.568172931671, "step_time_sec": 8.236601951008197, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.584131717681885, "lr": 0.0002, "elapsed_sec": 9114.80539393425, "step_time_sec": 8.237058397004148, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.652855396270752, "lr": 0.0002, "elapsed_sec": 9123.04236960411, "step_time_sec": 8.236845857987646, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.686306476593018, "lr": 0.0002, "elapsed_sec": 9131.279727697372, "step_time_sec": 8.237182770011714, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.602626323699951, "lr": 0.0002, "elapsed_sec": 9139.516967058182, "step_time_sec": 8.237055022997083, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.7803192138671875, "lr": 0.0002, "elapsed_sec": 9147.754417181015, "step_time_sec": 8.23730573899229, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.69775915145874, "lr": 0.0002, "elapsed_sec": 9155.992880821228, "step_time_sec": 8.238318414980313, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.706762790679932, "lr": 0.0002, "elapsed_sec": 9164.229305744171, "step_time_sec": 8.236240357015049, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.730007171630859, "lr": 0.0002, "elapsed_sec": 9172.465523481369, "step_time_sec": 8.236042748991167, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.682034492492676, "lr": 0.0002, "elapsed_sec": 9180.702156066895, "step_time_sec": 8.236491843010299, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.573151111602783, "lr": 0.0002, "elapsed_sec": 9188.939190864563, "step_time_sec": 8.236916555004427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.625747203826904, "lr": 0.0002, "elapsed_sec": 9197.175845861435, "step_time_sec": 8.236456856015138, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.7104291915893555, "lr": 0.0002, "elapsed_sec": 9205.41393995285, "step_time_sec": 8.237891917990055, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.662698745727539, "lr": 0.0002, "elapsed_sec": 9213.651715993881, "step_time_sec": 8.23760884697549, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.637563228607178, "lr": 0.0002, "elapsed_sec": 9221.889132738113, "step_time_sec": 8.237264407012844, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.581355571746826, "lr": 0.0002, "elapsed_sec": 9230.126948356628, "step_time_sec": 8.23764353900333, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.691904067993164, "lr": 0.0002, "elapsed_sec": 9238.363339662552, "step_time_sec": 8.236260586010758, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.608743190765381, "lr": 0.0002, "elapsed_sec": 9246.601191997528, "step_time_sec": 8.237725754996063, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.529472351074219, "lr": 0.0002, "elapsed_sec": 9254.838333845139, "step_time_sec": 8.236948126985226, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.762320041656494, "lr": 0.0002, "elapsed_sec": 9263.076517820358, "step_time_sec": 8.238004765007645, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.688467025756836, "lr": 0.0002, "elapsed_sec": 9271.31407737732, "step_time_sec": 8.23742593100178, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.6957783699035645, "lr": 0.0002, "elapsed_sec": 9279.550727844238, "step_time_sec": 8.2364550929924, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.5359697341918945, "lr": 0.0002, "elapsed_sec": 9287.787028074265, "step_time_sec": 8.236157853010809, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.586021900177002, "lr": 0.0002, "elapsed_sec": 9296.023540496826, "step_time_sec": 8.23635598900728, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.84481143951416, "lr": 0.0002, "elapsed_sec": 9304.26298713684, "step_time_sec": 8.239266098011285, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.618277549743652, "lr": 0.0002, "elapsed_sec": 9312.50039768219, "step_time_sec": 8.237249787984183, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.612917900085449, "lr": 0.0002, "elapsed_sec": 9320.737974405289, "step_time_sec": 8.237406628002645, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.578474998474121, "lr": 0.0002, "elapsed_sec": 9328.974528551102, "step_time_sec": 8.23644918599166, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.691777229309082, "lr": 0.0002, "elapsed_sec": 9337.212006568909, "step_time_sec": 8.237320423999336, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.593930244445801, "lr": 0.0002, "elapsed_sec": 9345.45068860054, "step_time_sec": 8.238471654010937, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.540892601013184, "lr": 0.0002, "elapsed_sec": 9353.687102079391, "step_time_sec": 8.236253596027382, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.652067184448242, "lr": 0.0002, "elapsed_sec": 9361.92575621605, "step_time_sec": 8.238522785017267, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.620809078216553, "lr": 0.0002, "elapsed_sec": 9370.162146806717, "step_time_sec": 8.23622937098844, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.6335320472717285, "lr": 0.0002, "elapsed_sec": 9378.399954557419, "step_time_sec": 8.237649611983215, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.579326629638672, "lr": 0.0002, "elapsed_sec": 9386.63692164421, "step_time_sec": 8.236822050006595, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.717384338378906, "lr": 0.0002, "elapsed_sec": 9394.87392282486, "step_time_sec": 8.23680452699773, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.52669620513916, "lr": 0.0002, "elapsed_sec": 9403.11017537117, "step_time_sec": 8.236091670987662, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.572129249572754, "lr": 0.0002, "elapsed_sec": 9411.345790624619, "step_time_sec": 8.23548466898501, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.717071533203125, "lr": 0.0002, "elapsed_sec": 9419.585496902466, "step_time_sec": 8.239543736010091, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.607456684112549, "lr": 0.0002, "elapsed_sec": 9427.822964191437, "step_time_sec": 8.237316660990473, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.66383695602417, "lr": 0.0002, "elapsed_sec": 9436.06075143814, "step_time_sec": 8.237603640998714, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.6607842445373535, "lr": 0.0002, "elapsed_sec": 9444.2973985672, "step_time_sec": 8.23650528700091, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.576639175415039, "lr": 0.0002, "elapsed_sec": 9452.535326480865, "step_time_sec": 8.23780927999178, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.607294082641602, "lr": 0.0002, "elapsed_sec": 9460.772146463394, "step_time_sec": 8.236609172978206, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.59021520614624, "lr": 0.0002, "elapsed_sec": 9469.008030891418, "step_time_sec": 8.235755388013786, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.723301887512207, "lr": 0.0002, "elapsed_sec": 9477.245524406433, "step_time_sec": 8.237332764983876, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.597795486450195, "lr": 0.0002, "elapsed_sec": 9485.481351852417, "step_time_sec": 8.235664751002332, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.573676109313965, "lr": 0.0002, "elapsed_sec": 9493.719899177551, "step_time_sec": 8.238360187009675, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.53679895401001, "lr": 0.0002, "elapsed_sec": 9501.956386327744, "step_time_sec": 8.236313072004123, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.692109107971191, "lr": 0.0002, "elapsed_sec": 9510.194447994232, "step_time_sec": 8.237903594999807, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.6470627784729, "lr": 0.0002, "elapsed_sec": 9518.43199801445, "step_time_sec": 8.237399209989235, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.772937297821045, "lr": 0.0002, "elapsed_sec": 9526.668841600418, "step_time_sec": 8.236672230006661, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.63196325302124, "lr": 0.0002, "elapsed_sec": 9534.906803131104, "step_time_sec": 8.237790834973566, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.446396350860596, "lr": 0.0002, "elapsed_sec": 9543.145068883896, "step_time_sec": 8.238163191010244, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.556735038757324, "lr": 0.0002, "elapsed_sec": 9551.38368177414, "step_time_sec": 8.238430828001583, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.655851364135742, "lr": 0.0002, "elapsed_sec": 9559.621138572693, "step_time_sec": 8.237276050000219, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.602168083190918, "lr": 0.0002, "elapsed_sec": 9567.857901096344, "step_time_sec": 8.236609404004412, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.5677618980407715, "lr": 0.0002, "elapsed_sec": 9576.095810890198, "step_time_sec": 8.237725814018631, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.590300559997559, "lr": 0.0002, "elapsed_sec": 9584.33119559288, "step_time_sec": 8.235231395985465, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.59955358505249, "lr": 0.0002, "elapsed_sec": 9592.568769693375, "step_time_sec": 8.23743127999478, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.591731548309326, "lr": 0.0002, "elapsed_sec": 9600.806294679642, "step_time_sec": 8.237339895014884, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.538600921630859, "lr": 0.0002, "elapsed_sec": 9609.044194459915, "step_time_sec": 8.237678667006548, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.600287914276123, "lr": 0.0002, "elapsed_sec": 9617.281606197357, "step_time_sec": 8.237314205995062, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.578134536743164, "lr": 0.0002, "elapsed_sec": 9625.519073963165, "step_time_sec": 8.237237916997401, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.530975818634033, "lr": 0.0002, "elapsed_sec": 9633.756128072739, "step_time_sec": 8.236870034976164, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.590819835662842, "lr": 0.0002, "elapsed_sec": 9641.992664337158, "step_time_sec": 8.236404520022916, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.623713970184326, "lr": 0.0002, "elapsed_sec": 9650.230005025864, "step_time_sec": 8.237161907978589, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.635095596313477, "lr": 0.0002, "elapsed_sec": 9658.467552661896, "step_time_sec": 8.237421034980798, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.555048942565918, "lr": 0.0002, "elapsed_sec": 9666.706048965454, "step_time_sec": 8.238355551002314, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.468771934509277, "lr": 0.0002, "elapsed_sec": 9674.94355916977, "step_time_sec": 8.237295099010225, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.643409252166748, "lr": 0.0002, "elapsed_sec": 9683.181780338287, "step_time_sec": 8.238048553990666, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.685905456542969, "lr": 0.0002, "elapsed_sec": 9691.420350551605, "step_time_sec": 8.23840924698743, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.518862724304199, "lr": 0.0002, "elapsed_sec": 9699.65751838684, "step_time_sec": 8.237011020013597, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.699189186096191, "lr": 0.0002, "elapsed_sec": 9707.894608974457, "step_time_sec": 8.236933815991506, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.659247875213623, "lr": 0.0002, "elapsed_sec": 9716.13118672371, "step_time_sec": 8.236442070017802, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.483818531036377, "lr": 0.0002, "elapsed_sec": 9724.367653369904, "step_time_sec": 8.23625779800932, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.720613956451416, "lr": 0.0002, "elapsed_sec": 9732.605289697647, "step_time_sec": 8.237523521995172, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.4807047843933105, "lr": 0.0002, "elapsed_sec": 9740.843019485474, "step_time_sec": 8.237540517991874, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.533736228942871, "lr": 0.0002, "elapsed_sec": 9749.081064224243, "step_time_sec": 8.237902327993652, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.5642571449279785, "lr": 0.0002, "elapsed_sec": 9757.318075418472, "step_time_sec": 8.236816460994305, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.6956868171691895, "lr": 0.0002, "elapsed_sec": 9765.556066036224, "step_time_sec": 8.237824890995398, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.520901203155518, "lr": 0.0002, "elapsed_sec": 9773.792737960815, "step_time_sec": 8.236539782985346, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.430897235870361, "lr": 0.0002, "elapsed_sec": 9782.03203034401, "step_time_sec": 8.239159049000591, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.560309410095215, "lr": 0.0002, "elapsed_sec": 9790.269681453705, "step_time_sec": 8.237438844022108, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.665749549865723, "lr": 0.0002, "elapsed_sec": 9798.507179498672, "step_time_sec": 8.237322991975816, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.6471710205078125, "lr": 0.0002, "elapsed_sec": 9806.74446773529, "step_time_sec": 8.237143179983832, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.567424774169922, "lr": 0.0002, "elapsed_sec": 9814.97964978218, "step_time_sec": 8.235013463010546, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.62756872177124, "lr": 0.0002, "elapsed_sec": 9823.217738866806, "step_time_sec": 8.237922900007106, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.633543014526367, "lr": 0.0002, "elapsed_sec": 9831.454811573029, "step_time_sec": 8.236904263001634, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.779295444488525, "lr": 0.0002, "elapsed_sec": 9839.6921377182, "step_time_sec": 8.237176223017741, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.524999618530273, "lr": 0.0002, "elapsed_sec": 9847.929363489151, "step_time_sec": 8.237077347992454, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.591286659240723, "lr": 0.0002, "elapsed_sec": 9856.16629076004, "step_time_sec": 8.236733772006119, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.547974586486816, "lr": 0.0002, "elapsed_sec": 9864.403460741043, "step_time_sec": 8.237018448999152, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.666354179382324, "lr": 0.0002, "elapsed_sec": 9872.639180660248, "step_time_sec": 8.235592073993757, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.400753498077393, "lr": 0.0002, "elapsed_sec": 9880.876899003983, "step_time_sec": 8.237509714992484, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.539851188659668, "lr": 0.0002, "elapsed_sec": 9889.113666534424, "step_time_sec": 8.236595463997219, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.666132926940918, "lr": 0.0002, "elapsed_sec": 9897.352584600449, "step_time_sec": 8.238722743000835, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.7139458656311035, "lr": 0.0002, "elapsed_sec": 9905.59084224701, "step_time_sec": 8.238097762980033, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.506638050079346, "lr": 0.0002, "elapsed_sec": 9913.828690767288, "step_time_sec": 8.237703760009026, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.4933013916015625, "lr": 0.0002, "elapsed_sec": 9922.066054821014, "step_time_sec": 8.237224971991964, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.517407417297363, "lr": 0.0002, "elapsed_sec": 9930.302808046341, "step_time_sec": 8.236542708997149, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.613697528839111, "lr": 0.0002, "elapsed_sec": 9938.540607690811, "step_time_sec": 8.237631707976107, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.557631015777588, "lr": 0.0002, "elapsed_sec": 9946.77735877037, "step_time_sec": 8.236649248981848, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.5191168785095215, "lr": 0.0002, "elapsed_sec": 9955.015645980835, "step_time_sec": 8.238078624999616, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.723733425140381, "lr": 0.0002, "elapsed_sec": 9963.254178762436, "step_time_sec": 8.23839633600437, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.492008209228516, "lr": 0.0002, "elapsed_sec": 9971.490766525269, "step_time_sec": 8.236385027994402, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.589022636413574, "lr": 0.0002, "elapsed_sec": 9979.72833776474, "step_time_sec": 8.237467576982453, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.619746685028076, "lr": 0.0002, "elapsed_sec": 9987.965198755264, "step_time_sec": 8.236699725996004, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.523195743560791, "lr": 0.0002, "elapsed_sec": 9996.204363107681, "step_time_sec": 8.238993351988029, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.575541973114014, "lr": 0.0002, "elapsed_sec": 10004.441660165787, "step_time_sec": 8.237129884015303, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.559513568878174, "lr": 0.0002, "elapsed_sec": 10012.680143356323, "step_time_sec": 8.238361816009274, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.504007816314697, "lr": 0.0002, "elapsed_sec": 10020.918865442276, "step_time_sec": 8.238527296023676, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.627912521362305, "lr": 0.0002, "elapsed_sec": 10029.155102968216, "step_time_sec": 8.236084056989057, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.53504753112793, "lr": 0.0002, "elapsed_sec": 10037.392228126526, "step_time_sec": 8.23694499200792, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.463786602020264, "lr": 0.0002, "elapsed_sec": 10045.629029989243, "step_time_sec": 8.236630787985632, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.521615982055664, "lr": 0.0002, "elapsed_sec": 10053.86774611473, "step_time_sec": 8.238595849019475, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.646785259246826, "lr": 0.0002, "elapsed_sec": 10062.106160402298, "step_time_sec": 8.238264844985679, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.480492115020752, "lr": 0.0002, "elapsed_sec": 10070.344297647476, "step_time_sec": 8.23796000200673, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.567540645599365, "lr": 0.0002, "elapsed_sec": 10078.582874536514, "step_time_sec": 8.238424181006849, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.736468315124512, "lr": 0.0002, "elapsed_sec": 10086.819206476212, "step_time_sec": 8.23614410799928, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.599093914031982, "lr": 0.0002, "elapsed_sec": 10095.057350635529, "step_time_sec": 8.237952196999686, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.550646781921387, "lr": 0.0002, "elapsed_sec": 10103.294134140015, "step_time_sec": 8.236579308984801, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.596011161804199, "lr": 0.0002, "elapsed_sec": 10111.531909942627, "step_time_sec": 8.237594077014364, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.555633544921875, "lr": 0.0002, "elapsed_sec": 10119.769517660141, "step_time_sec": 8.237486936995992, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.620336055755615, "lr": 0.0002, "elapsed_sec": 10128.006809949875, "step_time_sec": 8.237161558005027, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.534176349639893, "lr": 0.0002, "elapsed_sec": 10136.246803045273, "step_time_sec": 8.239822080999147, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.591699123382568, "lr": 0.0002, "elapsed_sec": 10144.483105897903, "step_time_sec": 8.23618630997953, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.518958568572998, "lr": 0.0002, "elapsed_sec": 10152.721675634384, "step_time_sec": 8.238329348008847, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.562099933624268, "lr": 0.0002, "elapsed_sec": 10160.958615064621, "step_time_sec": 8.236784946988337, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.45017671585083, "lr": 0.0002, "elapsed_sec": 10169.1960978508, "step_time_sec": 8.23728953601676, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.601998329162598, "lr": 0.0002, "elapsed_sec": 10177.433616876602, "step_time_sec": 8.237324387999251, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.583358287811279, "lr": 0.0002, "elapsed_sec": 10185.669536590576, "step_time_sec": 8.235752298001898, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.520330905914307, "lr": 0.0002, "elapsed_sec": 10193.906022548676, "step_time_sec": 8.236362815019675, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.509858131408691, "lr": 0.0002, "elapsed_sec": 10202.142318487167, "step_time_sec": 8.236116612009937, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.514407634735107, "lr": 0.0002, "elapsed_sec": 10210.380222558975, "step_time_sec": 8.237699061981402, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.542941093444824, "lr": 0.0002, "elapsed_sec": 10218.617103815079, "step_time_sec": 8.236783016996924, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.553707122802734, "lr": 0.0002, "elapsed_sec": 10226.853824138641, "step_time_sec": 8.236498832004145, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.546749591827393, "lr": 0.0002, "elapsed_sec": 10235.090663909912, "step_time_sec": 8.236719540000195, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.535150527954102, "lr": 0.0002, "elapsed_sec": 10243.32842040062, "step_time_sec": 8.237584571004845, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.604800224304199, "lr": 0.0002, "elapsed_sec": 10251.565884590149, "step_time_sec": 8.237285370996688, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.575399875640869, "lr": 0.0002, "elapsed_sec": 10259.802268505096, "step_time_sec": 8.236227553978097, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.609633922576904, "lr": 0.0002, "elapsed_sec": 10268.038575649261, "step_time_sec": 8.236194661993068, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.487806797027588, "lr": 0.0002, "elapsed_sec": 10276.276098251343, "step_time_sec": 8.237329567986308, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.4989495277404785, "lr": 0.0002, "elapsed_sec": 10284.514199733734, "step_time_sec": 8.237924221000867, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.566027641296387, "lr": 0.0002, "elapsed_sec": 10292.750728845596, "step_time_sec": 8.23636764701223, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.494907379150391, "lr": 0.0002, "elapsed_sec": 10300.988848209381, "step_time_sec": 8.237970828980906, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.564068794250488, "lr": 0.0002, "elapsed_sec": 10309.227615594864, "step_time_sec": 8.238582730002236, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.429793834686279, "lr": 0.0002, "elapsed_sec": 10317.463307857513, "step_time_sec": 8.235576766019221, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.562089443206787, "lr": 0.0002, "elapsed_sec": 10325.701562643051, "step_time_sec": 8.238096022017999, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.514034748077393, "lr": 0.0002, "elapsed_sec": 10333.938295125961, "step_time_sec": 8.236602564022178, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.6043853759765625, "lr": 0.0002, "elapsed_sec": 10342.22907781601, "step_time_sec": 8.246452457999112, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.503715991973877, "lr": 0.0002, "elapsed_sec": 10350.466955661774, "step_time_sec": 8.23772107597324, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.54874849319458, "lr": 0.0002, "elapsed_sec": 10358.703696012497, "step_time_sec": 8.236605832993519, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.513980388641357, "lr": 0.0002, "elapsed_sec": 10366.94197010994, "step_time_sec": 8.238074665016029, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.623108386993408, "lr": 0.0002, "elapsed_sec": 10375.179199934006, "step_time_sec": 8.23707396700047, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.605390548706055, "lr": 0.0002, "elapsed_sec": 10383.416594982147, "step_time_sec": 8.237223349016858, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.551279067993164, "lr": 0.0002, "elapsed_sec": 10391.653798341751, "step_time_sec": 8.237028231000295, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.589983940124512, "lr": 0.0002, "elapsed_sec": 10399.891669511795, "step_time_sec": 8.237714198010508, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.586594104766846, "lr": 0.0002, "elapsed_sec": 10408.130490779877, "step_time_sec": 8.238706831994932, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.571986198425293, "lr": 0.0002, "elapsed_sec": 10416.366468667984, "step_time_sec": 8.235788799007423, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.489558219909668, "lr": 0.0002, "elapsed_sec": 10424.604707479477, "step_time_sec": 8.238078485010192, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.6913347244262695, "lr": 0.0002, "elapsed_sec": 10432.841365337372, "step_time_sec": 8.236508345988113, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.510307312011719, "lr": 0.0002, "elapsed_sec": 10441.080401420593, "step_time_sec": 8.238878850999754, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.55867338180542, "lr": 0.0002, "elapsed_sec": 10449.317981243134, "step_time_sec": 8.237443905993132, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.601202964782715, "lr": 0.0002, "elapsed_sec": 10457.555433750153, "step_time_sec": 8.237308646988822, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.518582344055176, "lr": 0.0002, "elapsed_sec": 10465.792423009872, "step_time_sec": 8.236791628005449, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.475037097930908, "lr": 0.0002, "elapsed_sec": 10474.028847455978, "step_time_sec": 8.236324124009116, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.545795917510986, "lr": 0.0002, "elapsed_sec": 10482.266616106033, "step_time_sec": 8.237605584989069, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.478725433349609, "lr": 0.0002, "elapsed_sec": 10490.503005743027, "step_time_sec": 8.2362050130032, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.553271293640137, "lr": 0.0002, "elapsed_sec": 10498.741462469101, "step_time_sec": 8.238339055009419, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.421362400054932, "lr": 0.0002, "elapsed_sec": 10506.979064941406, "step_time_sec": 8.237418858974706, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.494930267333984, "lr": 0.0002, "elapsed_sec": 10515.215543031693, "step_time_sec": 8.236272192007164, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.536724090576172, "lr": 0.0002, "elapsed_sec": 10523.453561782837, "step_time_sec": 8.237885948998155, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.68066930770874, "lr": 0.0002, "elapsed_sec": 10531.69014430046, "step_time_sec": 8.23640147500555, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.42241907119751, "lr": 0.0002, "elapsed_sec": 10539.928735017776, "step_time_sec": 8.238412448001327, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.548980236053467, "lr": 0.0002, "elapsed_sec": 10548.16544508934, "step_time_sec": 8.236614312016172, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.491379737854004, "lr": 0.0002, "elapsed_sec": 10556.404071331024, "step_time_sec": 8.23844823098625, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.427245140075684, "lr": 0.0002, "elapsed_sec": 10564.641414403915, "step_time_sec": 8.237150426022708, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.469512462615967, "lr": 0.0002, "elapsed_sec": 10572.87861585617, "step_time_sec": 8.237025470007211, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.620713233947754, "lr": 0.0002, "elapsed_sec": 10581.115975856781, "step_time_sec": 8.237209434009856, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.418733596801758, "lr": 0.0002, "elapsed_sec": 10589.352176427841, "step_time_sec": 8.236059679009486, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.566924571990967, "lr": 0.0002, "elapsed_sec": 10597.592483997345, "step_time_sec": 8.240136918000644, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.419778347015381, "lr": 0.0002, "elapsed_sec": 10605.830227851868, "step_time_sec": 8.23755387897836, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.464039325714111, "lr": 0.0002, "elapsed_sec": 10614.068066120148, "step_time_sec": 8.237728458014317, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.484614849090576, "lr": 0.0002, "elapsed_sec": 10622.307778835297, "step_time_sec": 8.239508943981491, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.485362529754639, "lr": 0.0002, "elapsed_sec": 10630.544555425644, "step_time_sec": 8.236627431993838, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.451778411865234, "lr": 0.0002, "elapsed_sec": 10638.780860185623, "step_time_sec": 8.236174097983167, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.547266006469727, "lr": 0.0002, "elapsed_sec": 10647.01731967926, "step_time_sec": 8.236256413016235, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.464949131011963, "lr": 0.0002, "elapsed_sec": 10655.254950523376, "step_time_sec": 8.237478513998212, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.615945339202881, "lr": 0.0002, "elapsed_sec": 10663.491909503937, "step_time_sec": 8.236844846018357, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.437317848205566, "lr": 0.0002, "elapsed_sec": 10671.7287068367, "step_time_sec": 8.236587235995103, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.621802806854248, "lr": 0.0002, "elapsed_sec": 10679.96833372116, "step_time_sec": 8.239475236012368, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.520289897918701, "lr": 0.0002, "elapsed_sec": 10688.205577611923, "step_time_sec": 8.237106267013587, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.571896076202393, "lr": 0.0002, "elapsed_sec": 10696.443105459213, "step_time_sec": 8.23731494799722, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.483381748199463, "lr": 0.0002, "elapsed_sec": 10704.679417848587, "step_time_sec": 8.236210292001488, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.444550514221191, "lr": 0.0002, "elapsed_sec": 10712.917072057724, "step_time_sec": 8.237433342001168, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.4926438331604, "lr": 0.0002, "elapsed_sec": 10721.15385890007, "step_time_sec": 8.236649408994708, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.5511956214904785, "lr": 0.0002, "elapsed_sec": 10729.391453742981, "step_time_sec": 8.237412254005903, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.421509265899658, "lr": 0.0002, "elapsed_sec": 10737.629971981049, "step_time_sec": 8.238392417988507, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.443878650665283, "lr": 0.0002, "elapsed_sec": 10745.866881608963, "step_time_sec": 8.236759744002484, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.432158470153809, "lr": 0.0002, "elapsed_sec": 10754.105034351349, "step_time_sec": 8.23796834499808, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.49329948425293, "lr": 0.0002, "elapsed_sec": 10762.341604948044, "step_time_sec": 8.236407054006122, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.582098484039307, "lr": 0.0002, "elapsed_sec": 10770.579727172852, "step_time_sec": 8.237962058978155, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.476711750030518, "lr": 0.0002, "elapsed_sec": 10778.817477703094, "step_time_sec": 8.237621328007663, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.4536356925964355, "lr": 0.0002, "elapsed_sec": 10787.055055856705, "step_time_sec": 8.237364639004227, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.495048999786377, "lr": 0.0002, "elapsed_sec": 10795.29366016388, "step_time_sec": 8.238423008995596, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.475602626800537, "lr": 0.0002, "elapsed_sec": 10803.53091430664, "step_time_sec": 8.237141612014966, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.589604377746582, "lr": 0.0002, "elapsed_sec": 10811.769154548645, "step_time_sec": 8.238053612993099, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.483760356903076, "lr": 0.0002, "elapsed_sec": 10820.006915330887, "step_time_sec": 8.237579667998943, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.400275707244873, "lr": 0.0002, "elapsed_sec": 10828.24526810646, "step_time_sec": 8.238227611000184, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.512385368347168, "lr": 0.0002, "elapsed_sec": 10836.483089208603, "step_time_sec": 8.237628161994508, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.49968147277832, "lr": 0.0002, "elapsed_sec": 10844.72045302391, "step_time_sec": 8.237205918005202, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.494216442108154, "lr": 0.0002, "elapsed_sec": 10852.958406209946, "step_time_sec": 8.237778734008316, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.406598091125488, "lr": 0.0002, "elapsed_sec": 10861.196066856384, "step_time_sec": 8.237486081023235, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.46937370300293, "lr": 0.0002, "elapsed_sec": 10869.433903694153, "step_time_sec": 8.237700437020976, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.403533458709717, "lr": 0.0002, "elapsed_sec": 10877.671909570694, "step_time_sec": 8.237849250988802, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.429104328155518, "lr": 0.0002, "elapsed_sec": 10885.908703327179, "step_time_sec": 8.236605343001429, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.383822441101074, "lr": 0.0002, "elapsed_sec": 10894.146266698837, "step_time_sec": 8.237405041989405, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.523291110992432, "lr": 0.0002, "elapsed_sec": 10902.383134126663, "step_time_sec": 8.236693344981177, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.593332767486572, "lr": 0.0002, "elapsed_sec": 10910.619873523712, "step_time_sec": 8.236600803007605, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.469661235809326, "lr": 0.0002, "elapsed_sec": 10918.858013868332, "step_time_sec": 8.23800387600204, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.618001461029053, "lr": 0.0002, "elapsed_sec": 10927.096114873886, "step_time_sec": 8.23789653199492, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.4762678146362305, "lr": 0.0002, "elapsed_sec": 10935.335161447525, "step_time_sec": 8.238881839992246, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.375906944274902, "lr": 0.0002, "elapsed_sec": 10943.571992635727, "step_time_sec": 8.236699608009076, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.387034893035889, "lr": 0.0002, "elapsed_sec": 10951.810276985168, "step_time_sec": 8.238062137010274, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.492986679077148, "lr": 0.0002, "elapsed_sec": 10960.0470662117, "step_time_sec": 8.23662674400839, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.485050201416016, "lr": 0.0002, "elapsed_sec": 10968.284798145294, "step_time_sec": 8.237583048990928, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.4506988525390625, "lr": 0.0002, "elapsed_sec": 10976.522115468979, "step_time_sec": 8.237199836992659, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.379250526428223, "lr": 0.0002, "elapsed_sec": 10984.760746240616, "step_time_sec": 8.23842900499585, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.507585525512695, "lr": 0.0002, "elapsed_sec": 10992.999064922333, "step_time_sec": 8.238136366009712, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.499029636383057, "lr": 0.0002, "elapsed_sec": 11001.235566139221, "step_time_sec": 8.236325913021574, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.496823310852051, "lr": 0.0002, "elapsed_sec": 11009.473481416702, "step_time_sec": 8.237756671005627, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.418405055999756, "lr": 0.0002, "elapsed_sec": 11017.710312128067, "step_time_sec": 8.23668354502297, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.47814416885376, "lr": 0.0002, "elapsed_sec": 11025.948434352875, "step_time_sec": 8.237941714003682, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.4898200035095215, "lr": 0.0002, "elapsed_sec": 11034.18592596054, "step_time_sec": 8.237383892992511, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.452550411224365, "lr": 0.0002, "elapsed_sec": 11042.424487113953, "step_time_sec": 8.238359806011431, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.478292465209961, "lr": 0.0002, "elapsed_sec": 11050.662581443787, "step_time_sec": 8.237926382978912, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.515748500823975, "lr": 0.0002, "elapsed_sec": 11058.898928880692, "step_time_sec": 8.236212400021031, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.498211860656738, "lr": 0.0002, "elapsed_sec": 11067.135851860046, "step_time_sec": 8.236738749023061, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.4730658531188965, "lr": 0.0002, "elapsed_sec": 11075.371441602707, "step_time_sec": 8.235459247021936, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.507211208343506, "lr": 0.0002, "elapsed_sec": 11083.610543966293, "step_time_sec": 8.238895628019236, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.486413478851318, "lr": 0.0002, "elapsed_sec": 11091.847753047943, "step_time_sec": 8.237010579992784, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.455544948577881, "lr": 0.0002, "elapsed_sec": 11100.086498737335, "step_time_sec": 8.238603250007145, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.316229820251465, "lr": 0.0002, "elapsed_sec": 11108.323343515396, "step_time_sec": 8.23667618300533, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.375499248504639, "lr": 0.0002, "elapsed_sec": 11116.559561014175, "step_time_sec": 8.236056468012976, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.513635635375977, "lr": 0.0002, "elapsed_sec": 11124.797756671906, "step_time_sec": 8.238040197989903, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.368907928466797, "lr": 0.0002, "elapsed_sec": 11133.0342066288, "step_time_sec": 8.236298402014654, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.482186794281006, "lr": 0.0002, "elapsed_sec": 11141.272255420685, "step_time_sec": 8.237916974001564, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.451324462890625, "lr": 0.0002, "elapsed_sec": 11149.510309696198, "step_time_sec": 8.237884717993438, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.448064804077148, "lr": 0.0002, "elapsed_sec": 11157.749209403992, "step_time_sec": 8.238710141013144, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.437567710876465, "lr": 0.0002, "elapsed_sec": 11165.986810445786, "step_time_sec": 8.237422487000003, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.4195637702941895, "lr": 0.0002, "elapsed_sec": 11174.222684144974, "step_time_sec": 8.235714147012914, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.419911861419678, "lr": 0.0002, "elapsed_sec": 11182.46175289154, "step_time_sec": 8.238929122977424, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.424424171447754, "lr": 0.0002, "elapsed_sec": 11190.69917345047, "step_time_sec": 8.23727131501073, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.430663585662842, "lr": 0.0002, "elapsed_sec": 11198.93698143959, "step_time_sec": 8.237650905008195, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.485802173614502, "lr": 0.0002, "elapsed_sec": 11207.174860954285, "step_time_sec": 8.237687361019198, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.5225396156311035, "lr": 0.0002, "elapsed_sec": 11215.412765026093, "step_time_sec": 8.237740214011865, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.452729225158691, "lr": 0.0002, "elapsed_sec": 11223.650746107101, "step_time_sec": 8.237855397979729, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.414124488830566, "lr": 0.0002, "elapsed_sec": 11231.887902975082, "step_time_sec": 8.236999894987093, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.459560871124268, "lr": 0.0002, "elapsed_sec": 11240.126920223236, "step_time_sec": 8.238832847011508, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.314943790435791, "lr": 0.0002, "elapsed_sec": 11248.364812374115, "step_time_sec": 8.237771732005058, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.461194038391113, "lr": 0.0002, "elapsed_sec": 11256.602988004684, "step_time_sec": 8.237969117006287, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.440371036529541, "lr": 0.0002, "elapsed_sec": 11264.841548919678, "step_time_sec": 8.238406247022795, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.465603828430176, "lr": 0.0002, "elapsed_sec": 11273.076820135117, "step_time_sec": 8.235129273001803, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.358201026916504, "lr": 0.0002, "elapsed_sec": 11281.314926624298, "step_time_sec": 8.237910879019182, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.312955379486084, "lr": 0.0002, "elapsed_sec": 11289.552915096283, "step_time_sec": 8.237848165998003, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.391745090484619, "lr": 0.0002, "elapsed_sec": 11297.790077924728, "step_time_sec": 8.237013729987666, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.262738227844238, "lr": 0.0002, "elapsed_sec": 11306.027200698853, "step_time_sec": 8.2369369789958, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.361663341522217, "lr": 0.0002, "elapsed_sec": 11314.264331817627, "step_time_sec": 8.236933300999226, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.545799732208252, "lr": 0.0002, "elapsed_sec": 11322.501845121384, "step_time_sec": 8.237344767985633, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.465174674987793, "lr": 0.0002, "elapsed_sec": 11330.73782491684, "step_time_sec": 8.235812242986867, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.246612548828125, "lr": 0.0002, "elapsed_sec": 11338.97488117218, "step_time_sec": 8.23693050100701, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.395740509033203, "lr": 0.0002, "elapsed_sec": 11347.212531328201, "step_time_sec": 8.237470407009823, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.414810657501221, "lr": 0.0002, "elapsed_sec": 11355.450818300247, "step_time_sec": 8.23807876399951, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.491816520690918, "lr": 0.0002, "elapsed_sec": 11363.687577009201, "step_time_sec": 8.236643117008498, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.3616790771484375, "lr": 0.0002, "elapsed_sec": 11371.92492556572, "step_time_sec": 8.237151735986117, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.358441352844238, "lr": 0.0002, "elapsed_sec": 11380.162410974503, "step_time_sec": 8.237325272988528, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.439538478851318, "lr": 0.0002, "elapsed_sec": 11388.399301052094, "step_time_sec": 8.236773364013061, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.497256278991699, "lr": 0.0002, "elapsed_sec": 11396.63794016838, "step_time_sec": 8.238468481984455, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.398837566375732, "lr": 0.0002, "elapsed_sec": 11404.874941587448, "step_time_sec": 8.236806752014672, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.438497066497803, "lr": 0.0002, "elapsed_sec": 11413.113213777542, "step_time_sec": 8.238128396013053, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.40796422958374, "lr": 0.0002, "elapsed_sec": 11421.351360082626, "step_time_sec": 8.237959451013012, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.516676425933838, "lr": 0.0002, "elapsed_sec": 11429.58902144432, "step_time_sec": 8.237490850005997, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.4044647216796875, "lr": 0.0002, "elapsed_sec": 11437.826952695847, "step_time_sec": 8.237779440998565, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.460461139678955, "lr": 0.0002, "elapsed_sec": 11446.062672138214, "step_time_sec": 8.235514945990872, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.512850761413574, "lr": 0.0002, "elapsed_sec": 11454.300029039383, "step_time_sec": 8.237247575016227, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.429794788360596, "lr": 0.0002, "elapsed_sec": 11462.538739204407, "step_time_sec": 8.238513326999964, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.3431715965271, "lr": 0.0002, "elapsed_sec": 11470.776595830917, "step_time_sec": 8.237700834986754, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.437534332275391, "lr": 0.0002, "elapsed_sec": 11479.01417708397, "step_time_sec": 8.237473849003436, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.4145965576171875, "lr": 0.0002, "elapsed_sec": 11487.251671791077, "step_time_sec": 8.237281930982135, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.265428066253662, "lr": 0.0002, "elapsed_sec": 11495.49037694931, "step_time_sec": 8.238567106978735, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.3579535484313965, "lr": 0.0002, "elapsed_sec": 11503.72708272934, "step_time_sec": 8.236491780990036, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.2533769607543945, "lr": 0.0002, "elapsed_sec": 11511.96473455429, "step_time_sec": 8.237535386986565, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.444495677947998, "lr": 0.0002, "elapsed_sec": 11520.2007625103, "step_time_sec": 8.235859703971073, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.412936210632324, "lr": 0.0002, "elapsed_sec": 11528.439306020737, "step_time_sec": 8.238387038989458, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.597115516662598, "lr": 0.0002, "elapsed_sec": 11536.677062749863, "step_time_sec": 8.237566022988176, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.390893459320068, "lr": 0.0002, "elapsed_sec": 11544.913356542587, "step_time_sec": 8.236188121984014, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.377254009246826, "lr": 0.0002, "elapsed_sec": 11553.149770021439, "step_time_sec": 8.236229776986875, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.400972366333008, "lr": 0.0002, "elapsed_sec": 11561.385953903198, "step_time_sec": 8.236018890020205, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.3586859703063965, "lr": 0.0002, "elapsed_sec": 11569.62355542183, "step_time_sec": 8.237451602006331, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.3942108154296875, "lr": 0.0002, "elapsed_sec": 11577.861196994781, "step_time_sec": 8.23748970401357, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.422717094421387, "lr": 0.0002, "elapsed_sec": 11586.100795507431, "step_time_sec": 8.23940638397471, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.2935709953308105, "lr": 0.0002, "elapsed_sec": 11594.337039709091, "step_time_sec": 8.23610828400706, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.198644638061523, "lr": 0.0002, "elapsed_sec": 11602.573557376862, "step_time_sec": 8.236363895004615, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.389023780822754, "lr": 0.0002, "elapsed_sec": 11610.809430599213, "step_time_sec": 8.235693794005783, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.332301139831543, "lr": 0.0002, "elapsed_sec": 11619.04507946968, "step_time_sec": 8.235523629991803, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.339681148529053, "lr": 0.0002, "elapsed_sec": 11627.282008647919, "step_time_sec": 8.236729244992603, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.377445220947266, "lr": 0.0002, "elapsed_sec": 11635.517427682877, "step_time_sec": 8.235265883995453, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.481765270233154, "lr": 0.0002, "elapsed_sec": 11643.75610756874, "step_time_sec": 8.238540884980466, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.45772123336792, "lr": 0.0002, "elapsed_sec": 11651.993944644928, "step_time_sec": 8.237664257001597, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.3139119148254395, "lr": 0.0002, "elapsed_sec": 11660.232187271118, "step_time_sec": 8.238077326008352, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.2540788650512695, "lr": 0.0002, "elapsed_sec": 11668.469451904297, "step_time_sec": 8.237117662007222, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.467703819274902, "lr": 0.0002, "elapsed_sec": 11676.706456184387, "step_time_sec": 8.236865707993275, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.348790645599365, "lr": 0.0002, "elapsed_sec": 11684.944114923477, "step_time_sec": 8.237489366001682, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.372278213500977, "lr": 0.0002, "elapsed_sec": 11693.181372404099, "step_time_sec": 8.237047570990399, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.331405162811279, "lr": 0.0002, "elapsed_sec": 11701.419665575027, "step_time_sec": 8.238152616017032, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.343088150024414, "lr": 0.0002, "elapsed_sec": 11709.656763792038, "step_time_sec": 8.236946274992079, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.392487049102783, "lr": 0.0002, "elapsed_sec": 11717.894641399384, "step_time_sec": 8.237712475995068, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.354831695556641, "lr": 0.0002, "elapsed_sec": 11726.13169002533, "step_time_sec": 8.236873436981114, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.467643737792969, "lr": 0.0002, "elapsed_sec": 11734.367995977402, "step_time_sec": 8.236164544010535, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.387803077697754, "lr": 0.0002, "elapsed_sec": 11742.604892730713, "step_time_sec": 8.236740624008235, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.386620044708252, "lr": 0.0002, "elapsed_sec": 11750.842271566391, "step_time_sec": 8.23725390600157, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.3493971824646, "lr": 0.0002, "elapsed_sec": 11759.080666542053, "step_time_sec": 8.238237533019856, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.419776439666748, "lr": 0.0002, "elapsed_sec": 11767.317638158798, "step_time_sec": 8.236815696989652, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.453867435455322, "lr": 0.0002, "elapsed_sec": 11775.554925441742, "step_time_sec": 8.23709156300174, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.413870811462402, "lr": 0.0002, "elapsed_sec": 11783.793310642242, "step_time_sec": 8.23821261501871, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.363373279571533, "lr": 0.0002, "elapsed_sec": 11792.029611825943, "step_time_sec": 8.236151949997293, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.285003662109375, "lr": 0.0002, "elapsed_sec": 11800.26698589325, "step_time_sec": 8.23719124298077, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.257654190063477, "lr": 0.0002, "elapsed_sec": 11808.50284075737, "step_time_sec": 8.235732431989163, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.305545806884766, "lr": 0.0002, "elapsed_sec": 11816.74134516716, "step_time_sec": 8.238329112005886, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.392931938171387, "lr": 0.0002, "elapsed_sec": 11824.97747373581, "step_time_sec": 8.235970740002813, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.432465553283691, "lr": 0.0002, "elapsed_sec": 11833.215530633926, "step_time_sec": 8.237876090977807, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.432668209075928, "lr": 0.0002, "elapsed_sec": 11841.45328116417, "step_time_sec": 8.237581974011846, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.426959991455078, "lr": 0.0002, "elapsed_sec": 11849.690190076828, "step_time_sec": 8.236757695995038, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.343288898468018, "lr": 0.0002, "elapsed_sec": 11857.927723407745, "step_time_sec": 8.23737061501015, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.430453777313232, "lr": 0.0002, "elapsed_sec": 11866.164977550507, "step_time_sec": 8.237116618984146, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.250524520874023, "lr": 0.0002, "elapsed_sec": 11874.40292263031, "step_time_sec": 8.23775911598932, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.419616222381592, "lr": 0.0002, "elapsed_sec": 11882.640466928482, "step_time_sec": 8.237353381002322, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.299338340759277, "lr": 0.0002, "elapsed_sec": 11890.877878904343, "step_time_sec": 8.237262856011512, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.443199157714844, "lr": 0.0002, "elapsed_sec": 11899.11591553688, "step_time_sec": 8.237910708994605, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.35878324508667, "lr": 0.0002, "elapsed_sec": 11907.352014303207, "step_time_sec": 8.235915091994684, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.4341325759887695, "lr": 0.0002, "elapsed_sec": 11915.589759111404, "step_time_sec": 8.237573537015123, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.337796688079834, "lr": 0.0002, "elapsed_sec": 11923.826950073242, "step_time_sec": 8.237028119998286, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.312045097351074, "lr": 0.0002, "elapsed_sec": 11932.06602525711, "step_time_sec": 8.238908444007393, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.396066665649414, "lr": 0.0002, "elapsed_sec": 11940.303566932678, "step_time_sec": 8.237390286987647, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.328559398651123, "lr": 0.0002, "elapsed_sec": 11948.541532278061, "step_time_sec": 8.237764853984118, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.491306781768799, "lr": 0.0002, "elapsed_sec": 11956.779579162598, "step_time_sec": 8.237881498993374, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.3295183181762695, "lr": 0.0002, "elapsed_sec": 11965.016033411026, "step_time_sec": 8.23630493998644, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.285393714904785, "lr": 0.0002, "elapsed_sec": 11973.253456830978, "step_time_sec": 8.23725336100324, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.447789669036865, "lr": 0.0002, "elapsed_sec": 11981.489703416824, "step_time_sec": 8.236129225988407, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.463622093200684, "lr": 0.0002, "elapsed_sec": 11989.72826719284, "step_time_sec": 8.238363415002823, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.271623134613037, "lr": 0.0002, "elapsed_sec": 11997.96604847908, "step_time_sec": 8.237621402018704, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.238972187042236, "lr": 0.0002, "elapsed_sec": 12006.202870607376, "step_time_sec": 8.236659240006702, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.369658470153809, "lr": 0.0002, "elapsed_sec": 12014.442026138306, "step_time_sec": 8.238983337010723, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.291617393493652, "lr": 0.0002, "elapsed_sec": 12022.677595615387, "step_time_sec": 8.235396687989123, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.284217834472656, "lr": 0.0002, "elapsed_sec": 12030.915225744247, "step_time_sec": 8.23749761498766, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.545286655426025, "lr": 0.0002, "elapsed_sec": 12039.151518583298, "step_time_sec": 8.236112457991112, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.302093982696533, "lr": 0.0002, "elapsed_sec": 12047.390124559402, "step_time_sec": 8.23842637098278, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.355355739593506, "lr": 0.0002, "elapsed_sec": 12055.62744307518, "step_time_sec": 8.23714672101778, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.354175567626953, "lr": 0.0002, "elapsed_sec": 12063.865554332733, "step_time_sec": 8.237942100997316, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.337337017059326, "lr": 0.0002, "elapsed_sec": 12072.103874921799, "step_time_sec": 8.238191823009402, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.302898406982422, "lr": 0.0002, "elapsed_sec": 12080.340564966202, "step_time_sec": 8.236489064001944, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.496868133544922, "lr": 0.0002, "elapsed_sec": 12088.577482938766, "step_time_sec": 8.236761747975834, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.120859146118164, "lr": 0.0002, "elapsed_sec": 12096.816566467285, "step_time_sec": 8.238924286997644, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.369993686676025, "lr": 0.0002, "elapsed_sec": 12105.054952383041, "step_time_sec": 8.238232634001179, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.334239482879639, "lr": 0.0002, "elapsed_sec": 12113.291673660278, "step_time_sec": 8.236569383007009, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.488234996795654, "lr": 0.0002, "elapsed_sec": 12121.530885219574, "step_time_sec": 8.23907879399485, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.390545845031738, "lr": 0.0002, "elapsed_sec": 12129.768650054932, "step_time_sec": 8.237591229990358, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.476963996887207, "lr": 0.0002, "elapsed_sec": 12138.004061698914, "step_time_sec": 8.235233742016135, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.406778335571289, "lr": 0.0002, "elapsed_sec": 12146.242013216019, "step_time_sec": 8.23783824700513, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.320772171020508, "lr": 0.0002, "elapsed_sec": 12154.47880935669, "step_time_sec": 8.236642175004818, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.232151031494141, "lr": 0.0002, "elapsed_sec": 12162.716054916382, "step_time_sec": 8.237047676986549, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.189246654510498, "lr": 0.0002, "elapsed_sec": 12170.953912258148, "step_time_sec": 8.23772218299564, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.489042282104492, "lr": 0.0002, "elapsed_sec": 12179.190469264984, "step_time_sec": 8.236426641989965, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.316493988037109, "lr": 0.0002, "elapsed_sec": 12187.428327083588, "step_time_sec": 8.237672771007055, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.446714401245117, "lr": 0.0002, "elapsed_sec": 12195.66359448433, "step_time_sec": 8.235110375011573, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.364014148712158, "lr": 0.0002, "elapsed_sec": 12203.900694608688, "step_time_sec": 8.23692526598461, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.450648784637451, "lr": 0.0002, "elapsed_sec": 12212.138958454132, "step_time_sec": 8.238094972999534, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.355210781097412, "lr": 0.0002, "elapsed_sec": 12220.377476453781, "step_time_sec": 8.238344540004618, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.483749866485596, "lr": 0.0002, "elapsed_sec": 12228.615676641464, "step_time_sec": 8.238065420009661, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.3102593421936035, "lr": 0.0002, "elapsed_sec": 12236.853586435318, "step_time_sec": 8.237699408986373, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.295679569244385, "lr": 0.0002, "elapsed_sec": 12245.092378377914, "step_time_sec": 8.238631580985384, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.260632514953613, "lr": 0.0002, "elapsed_sec": 12253.328825473785, "step_time_sec": 8.236295081005665, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.266820907592773, "lr": 0.0002, "elapsed_sec": 12261.567570924759, "step_time_sec": 8.238601099001244, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.423098087310791, "lr": 0.0002, "elapsed_sec": 12269.805493116379, "step_time_sec": 8.237722340010805, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.262634754180908, "lr": 0.0002, "elapsed_sec": 12278.04229927063, "step_time_sec": 8.236697289976291, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.350804805755615, "lr": 0.0002, "elapsed_sec": 12286.280554771423, "step_time_sec": 8.238024963997304, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.249330043792725, "lr": 0.0002, "elapsed_sec": 12294.517078399658, "step_time_sec": 8.23641405699891, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.374478816986084, "lr": 0.0002, "elapsed_sec": 12302.756260871887, "step_time_sec": 8.238990012003342, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.323276042938232, "lr": 0.0002, "elapsed_sec": 12310.992384433746, "step_time_sec": 8.23597510001855, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.234821319580078, "lr": 0.0002, "elapsed_sec": 12319.23026585579, "step_time_sec": 8.237742479977896, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.272463321685791, "lr": 0.0002, "elapsed_sec": 12327.46721625328, "step_time_sec": 8.236754712008405, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.301645755767822, "lr": 0.0002, "elapsed_sec": 12335.705474615097, "step_time_sec": 8.238081620016601, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.219340801239014, "lr": 0.0002, "elapsed_sec": 12343.942480802536, "step_time_sec": 8.236838911019731, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.432836532592773, "lr": 0.0002, "elapsed_sec": 12352.178969860077, "step_time_sec": 8.23637080899789, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.275291442871094, "lr": 0.0002, "elapsed_sec": 12360.417343854904, "step_time_sec": 8.238167458010139, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.260319232940674, "lr": 0.0002, "elapsed_sec": 12368.65396952629, "step_time_sec": 8.236515319993487, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.374512195587158, "lr": 0.0002, "elapsed_sec": 12376.893702745438, "step_time_sec": 8.239551047008717, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.342883110046387, "lr": 0.0002, "elapsed_sec": 12385.132806062698, "step_time_sec": 8.238922366988845, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.320346355438232, "lr": 0.0002, "elapsed_sec": 12393.369216918945, "step_time_sec": 8.236266980995424, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.118985652923584, "lr": 0.0002, "elapsed_sec": 12401.607464551926, "step_time_sec": 8.23805163500947, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.2758331298828125, "lr": 0.0002, "elapsed_sec": 12409.844250679016, "step_time_sec": 8.23666451900499, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.191242694854736, "lr": 0.0002, "elapsed_sec": 12418.083068609238, "step_time_sec": 8.238668853009585, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.407506942749023, "lr": 0.0002, "elapsed_sec": 12426.320022583008, "step_time_sec": 8.236821884987876, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.312568664550781, "lr": 0.0002, "elapsed_sec": 12434.55818772316, "step_time_sec": 30.35610932702548, "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": 5.394394397735596, "lr": 0.0002, "elapsed_sec": 12464.926676750183, "step_time_sec": 8.249853096989682, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.269169330596924, "lr": 0.0002, "elapsed_sec": 12473.1578707695, "step_time_sec": 8.23102104102145, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.455951690673828, "lr": 0.0002, "elapsed_sec": 12481.393612861633, "step_time_sec": 8.235578036023071, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.352729797363281, "lr": 0.0002, "elapsed_sec": 12489.630088806152, "step_time_sec": 8.236365577002289, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.366023540496826, "lr": 0.0002, "elapsed_sec": 12497.866926908493, "step_time_sec": 8.236631480016513, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.3593525886535645, "lr": 0.0002, "elapsed_sec": 12506.104448795319, "step_time_sec": 8.237362612999277, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.307405471801758, "lr": 0.0002, "elapsed_sec": 12514.340198755264, "step_time_sec": 8.235580194974318, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.251858234405518, "lr": 0.0002, "elapsed_sec": 12522.579126358032, "step_time_sec": 8.238770408002893, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.438937664031982, "lr": 0.0002, "elapsed_sec": 12530.815981149673, "step_time_sec": 8.236664532014402, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.289389610290527, "lr": 0.0002, "elapsed_sec": 12539.054655790329, "step_time_sec": 8.238557365984889, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.327630043029785, "lr": 0.0002, "elapsed_sec": 12547.29246211052, "step_time_sec": 8.237609019008232, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.426877975463867, "lr": 0.0002, "elapsed_sec": 12555.529816150665, "step_time_sec": 8.237175855989335, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.1854095458984375, "lr": 0.0002, "elapsed_sec": 12563.76786661148, "step_time_sec": 8.237890238000546, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.392857551574707, "lr": 0.0002, "elapsed_sec": 12572.004140138626, "step_time_sec": 8.236123011010932, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.184781074523926, "lr": 0.0002, "elapsed_sec": 12580.243203401566, "step_time_sec": 8.238905986014288, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.338922023773193, "lr": 0.0002, "elapsed_sec": 12588.479677677155, "step_time_sec": 8.236319925985299, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.125316619873047, "lr": 0.0002, "elapsed_sec": 12596.718012571335, "step_time_sec": 8.238148367003305, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.382548809051514, "lr": 0.0002, "elapsed_sec": 12604.956246376038, "step_time_sec": 8.238102920004167, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.204871654510498, "lr": 0.0002, "elapsed_sec": 12613.193873643875, "step_time_sec": 8.237448853004025, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.28168249130249, "lr": 0.0002, "elapsed_sec": 12621.433054685593, "step_time_sec": 8.239016393985366, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.298373222351074, "lr": 0.0002, "elapsed_sec": 12629.670642375946, "step_time_sec": 8.23746246198425, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.254115104675293, "lr": 0.0002, "elapsed_sec": 12637.909447193146, "step_time_sec": 8.238606298982631, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.137169361114502, "lr": 0.0002, "elapsed_sec": 12646.14787888527, "step_time_sec": 8.238286964013241, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.379271507263184, "lr": 0.0002, "elapsed_sec": 12654.384641885757, "step_time_sec": 8.23662812000839, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.259393215179443, "lr": 0.0002, "elapsed_sec": 12662.62399482727, "step_time_sec": 8.23915632700664, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.130316257476807, "lr": 0.0002, "elapsed_sec": 12670.861046552658, "step_time_sec": 8.23689122198266, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.318426132202148, "lr": 0.0002, "elapsed_sec": 12679.099407434464, "step_time_sec": 8.238217167003313, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.288117408752441, "lr": 0.0002, "elapsed_sec": 12687.337101221085, "step_time_sec": 8.237531837017741, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.288695812225342, "lr": 0.0002, "elapsed_sec": 12695.57581782341, "step_time_sec": 8.238541763013927, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.134374141693115, "lr": 0.0002, "elapsed_sec": 12703.813883543015, "step_time_sec": 8.237918005994288, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.318931579589844, "lr": 0.0002, "elapsed_sec": 12712.050076246262, "step_time_sec": 8.236047217011219, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.233508586883545, "lr": 0.0002, "elapsed_sec": 12720.28984117508, "step_time_sec": 8.239617434999673, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.347756862640381, "lr": 0.0002, "elapsed_sec": 12728.526134729385, "step_time_sec": 8.236133511993103, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.335978984832764, "lr": 0.0002, "elapsed_sec": 12736.76482462883, "step_time_sec": 8.23852223099675, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.145904541015625, "lr": 0.0002, "elapsed_sec": 12745.001697540283, "step_time_sec": 8.236743225017563, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.412615776062012, "lr": 0.0002, "elapsed_sec": 12753.238991498947, "step_time_sec": 8.237151231005555, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.162635803222656, "lr": 0.0002, "elapsed_sec": 12761.477368593216, "step_time_sec": 8.238173257996095, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.3744330406188965, "lr": 0.0002, "elapsed_sec": 12769.713960886002, "step_time_sec": 8.236452648998238, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.166582107543945, "lr": 0.0002, "elapsed_sec": 12777.954357385635, "step_time_sec": 8.240211356984219, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.279036521911621, "lr": 0.0002, "elapsed_sec": 12786.190733909607, "step_time_sec": 8.236207986978116, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.123805999755859, "lr": 0.0002, "elapsed_sec": 12794.428117752075, "step_time_sec": 8.237252644001273, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.266399383544922, "lr": 0.0002, "elapsed_sec": 12802.667307853699, "step_time_sec": 8.238996783009497, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.193399906158447, "lr": 0.0002, "elapsed_sec": 12810.903911113739, "step_time_sec": 8.236424839997198, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.359682083129883, "lr": 0.0002, "elapsed_sec": 12819.142162561417, "step_time_sec": 8.238093299005413, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.332095146179199, "lr": 0.0002, "elapsed_sec": 12827.379632472992, "step_time_sec": 8.237336093996419, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.270133018493652, "lr": 0.0002, "elapsed_sec": 12835.618460893631, "step_time_sec": 8.238669411017327, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.465456008911133, "lr": 0.0002, "elapsed_sec": 12843.855829238892, "step_time_sec": 8.237201735988492, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.204163074493408, "lr": 0.0002, "elapsed_sec": 12852.09422492981, "step_time_sec": 8.238244022999424, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.285242080688477, "lr": 0.0002, "elapsed_sec": 12860.331523418427, "step_time_sec": 8.23711492298753, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.3123040199279785, "lr": 0.0002, "elapsed_sec": 12868.568416833878, "step_time_sec": 8.236760625004536, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.225358009338379, "lr": 0.0002, "elapsed_sec": 12876.805990219116, "step_time_sec": 8.237385685002664, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.3108649253845215, "lr": 0.0002, "elapsed_sec": 12885.044031143188, "step_time_sec": 8.237918692990206, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.206407070159912, "lr": 0.0002, "elapsed_sec": 12893.282402276993, "step_time_sec": 8.23824590400909, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.2322163581848145, "lr": 0.0002, "elapsed_sec": 12901.521137714386, "step_time_sec": 8.238526747009018, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.204904556274414, "lr": 0.0002, "elapsed_sec": 12909.758891582489, "step_time_sec": 8.237589349009795, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.197480201721191, "lr": 0.0002, "elapsed_sec": 12917.997209072113, "step_time_sec": 8.23814656198374, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.263538837432861, "lr": 0.0002, "elapsed_sec": 12926.234414100647, "step_time_sec": 8.237062200991204, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.1203813552856445, "lr": 0.0002, "elapsed_sec": 12934.473025560379, "step_time_sec": 8.238399570022011, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.226943016052246, "lr": 0.0002, "elapsed_sec": 12942.710772037506, "step_time_sec": 8.237597642000765, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.241118907928467, "lr": 0.0002, "elapsed_sec": 12950.946724414825, "step_time_sec": 8.235775684006512, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.275466442108154, "lr": 0.0002, "elapsed_sec": 12959.18447470665, "step_time_sec": 8.237655696022557, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.233409404754639, "lr": 0.0002, "elapsed_sec": 12967.42108631134, "step_time_sec": 8.236434460995952, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.324085712432861, "lr": 0.0002, "elapsed_sec": 12975.65903544426, "step_time_sec": 8.237743528996361, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.235313892364502, "lr": 0.0002, "elapsed_sec": 12983.897604227066, "step_time_sec": 8.238437302003149, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.253385066986084, "lr": 0.0002, "elapsed_sec": 12992.135941028595, "step_time_sec": 8.238147772994125, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.34339714050293, "lr": 0.0002, "elapsed_sec": 13000.373527526855, "step_time_sec": 8.237467364000622, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.206543922424316, "lr": 0.0002, "elapsed_sec": 13008.60961985588, "step_time_sec": 8.2359246959968, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.157937049865723, "lr": 0.0002, "elapsed_sec": 13016.848121404648, "step_time_sec": 8.238282089994755, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.2351579666137695, "lr": 0.0002, "elapsed_sec": 13025.085296869278, "step_time_sec": 8.237046866008313, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.177433967590332, "lr": 0.0002, "elapsed_sec": 13033.325163602829, "step_time_sec": 8.23969655498513, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.3239545822143555, "lr": 0.0002, "elapsed_sec": 13041.562946796417, "step_time_sec": 8.237573135003913, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.210295677185059, "lr": 0.0002, "elapsed_sec": 13049.800489664078, "step_time_sec": 8.237410934001673, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.437309741973877, "lr": 0.0002, "elapsed_sec": 13058.037414550781, "step_time_sec": 8.236778705002507, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.262451648712158, "lr": 0.0002, "elapsed_sec": 13066.273421764374, "step_time_sec": 8.235826705989894, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.204892158508301, "lr": 0.0002, "elapsed_sec": 13074.510843753815, "step_time_sec": 8.237272861995734, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.320728302001953, "lr": 0.0002, "elapsed_sec": 13082.747681379318, "step_time_sec": 8.236666315002367, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.190659999847412, "lr": 0.0002, "elapsed_sec": 13090.986117601395, "step_time_sec": 8.23832919698907, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.19606876373291, "lr": 0.0002, "elapsed_sec": 13099.224900245667, "step_time_sec": 8.238565953011857, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.2912092208862305, "lr": 0.0002, "elapsed_sec": 13107.463836431503, "step_time_sec": 8.23877771300613, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.370738506317139, "lr": 0.0002, "elapsed_sec": 13115.701488494873, "step_time_sec": 8.237503569980618, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.140071868896484, "lr": 0.0002, "elapsed_sec": 13123.938254117966, "step_time_sec": 8.236645312979817, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.230533123016357, "lr": 0.0002, "elapsed_sec": 13132.177317142487, "step_time_sec": 8.238877374998992, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.237411975860596, "lr": 0.0002, "elapsed_sec": 13140.415095567703, "step_time_sec": 8.237642330001108, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.236165523529053, "lr": 0.0002, "elapsed_sec": 13148.65261888504, "step_time_sec": 8.23736874901806, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.246597766876221, "lr": 0.0002, "elapsed_sec": 13156.889627456665, "step_time_sec": 8.236815773998387, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.241607189178467, "lr": 0.0002, "elapsed_sec": 13165.127453327179, "step_time_sec": 8.237719744007336, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.132216453552246, "lr": 0.0002, "elapsed_sec": 13173.36494898796, "step_time_sec": 8.237275163002778, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.244050025939941, "lr": 0.0002, "elapsed_sec": 13181.601527929306, "step_time_sec": 8.236416230007308, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.199542999267578, "lr": 0.0002, "elapsed_sec": 13189.84031701088, "step_time_sec": 8.238662699994165, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.222890377044678, "lr": 0.0002, "elapsed_sec": 13198.07935333252, "step_time_sec": 8.238853688002564, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.1852803230285645, "lr": 0.0002, "elapsed_sec": 13206.317237615585, "step_time_sec": 8.237759581010323, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.292542934417725, "lr": 0.0002, "elapsed_sec": 13214.55604839325, "step_time_sec": 8.238584850012558, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.322769641876221, "lr": 0.0002, "elapsed_sec": 13222.7918009758, "step_time_sec": 8.235637829988264, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.159582614898682, "lr": 0.0002, "elapsed_sec": 13231.029373884201, "step_time_sec": 8.237381151993759, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.4032816886901855, "lr": 0.0002, "elapsed_sec": 13239.26656293869, "step_time_sec": 8.237061672989512, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.263344764709473, "lr": 0.0002, "elapsed_sec": 13247.505887269974, "step_time_sec": 8.239125039021019, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.244779109954834, "lr": 0.0002, "elapsed_sec": 13255.743936777115, "step_time_sec": 8.23789397100336, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.219179153442383, "lr": 0.0002, "elapsed_sec": 13263.981977939606, "step_time_sec": 8.23788270301884, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.211555480957031, "lr": 0.0002, "elapsed_sec": 13272.219312906265, "step_time_sec": 8.23720193799818, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.14410400390625, "lr": 0.0002, "elapsed_sec": 13280.455798149109, "step_time_sec": 8.236333704990102, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.295172691345215, "lr": 0.0002, "elapsed_sec": 13288.695014715195, "step_time_sec": 8.239097246987512, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.32474946975708, "lr": 0.0002, "elapsed_sec": 13296.93205165863, "step_time_sec": 8.236814228992444, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.224703311920166, "lr": 0.0002, "elapsed_sec": 13305.169189691544, "step_time_sec": 8.237005882983794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.1223626136779785, "lr": 0.0002, "elapsed_sec": 13313.407238006592, "step_time_sec": 8.237854503997369, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.147593975067139, "lr": 0.0002, "elapsed_sec": 13321.644572257996, "step_time_sec": 8.23717709199991, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.129763126373291, "lr": 0.0002, "elapsed_sec": 13329.882376670837, "step_time_sec": 8.237689961999422, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.268677711486816, "lr": 0.0002, "elapsed_sec": 13338.11880493164, "step_time_sec": 8.236243480991106, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.122448921203613, "lr": 0.0002, "elapsed_sec": 13346.356082439423, "step_time_sec": 8.23714774599648, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.28801965713501, "lr": 0.0002, "elapsed_sec": 13354.594340801239, "step_time_sec": 8.238104071991984, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.145168781280518, "lr": 0.0002, "elapsed_sec": 13362.832605600357, "step_time_sec": 8.23807917398517, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.424874305725098, "lr": 0.0002, "elapsed_sec": 13371.070899486542, "step_time_sec": 8.23815943201771, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.181631088256836, "lr": 0.0002, "elapsed_sec": 13379.309145450592, "step_time_sec": 8.238119621004444, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.405335426330566, "lr": 0.0002, "elapsed_sec": 13387.548770427704, "step_time_sec": 8.239432038011728, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.253519058227539, "lr": 0.0002, "elapsed_sec": 13395.786249876022, "step_time_sec": 8.237292448990047, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.1916069984436035, "lr": 0.0002, "elapsed_sec": 13404.02372598648, "step_time_sec": 8.237310511991382, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.288763046264648, "lr": 0.0002, "elapsed_sec": 13412.26200914383, "step_time_sec": 8.238179640989983, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.15554666519165, "lr": 0.0002, "elapsed_sec": 13420.498741149902, "step_time_sec": 8.236551290989155, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.247806072235107, "lr": 0.0002, "elapsed_sec": 13428.738269805908, "step_time_sec": 8.239388059999328, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.236217498779297, "lr": 0.0002, "elapsed_sec": 13436.976699590683, "step_time_sec": 8.238227952999296, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.159672260284424, "lr": 0.0002, "elapsed_sec": 13445.216186523438, "step_time_sec": 8.239332174998708, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.13564920425415, "lr": 0.0002, "elapsed_sec": 13453.453997373581, "step_time_sec": 8.23765473801177, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.389355659484863, "lr": 0.0002, "elapsed_sec": 13461.691744089127, "step_time_sec": 8.237572358979378, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.1834588050842285, "lr": 0.0002, "elapsed_sec": 13483.13504576683, "step_time_sec": 21.44321722499444, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.249416351318359, "lr": 0.0002, "elapsed_sec": 13491.36627316475, "step_time_sec": 8.231012395001017, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.296346664428711, "lr": 0.0002, "elapsed_sec": 13499.599054336548, "step_time_sec": 8.23261261699372, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.25272274017334, "lr": 0.0002, "elapsed_sec": 13507.829690933228, "step_time_sec": 8.230469626985723, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.200801849365234, "lr": 0.0002, "elapsed_sec": 13516.062238693237, "step_time_sec": 8.232377838983666, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.191614627838135, "lr": 0.0002, "elapsed_sec": 13524.294739484787, "step_time_sec": 8.232362970011309, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.149508953094482, "lr": 0.0002, "elapsed_sec": 13532.5251891613, "step_time_sec": 8.2302889470011, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.189505100250244, "lr": 0.0002, "elapsed_sec": 13540.75770497322, "step_time_sec": 8.232379502005642, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.312041282653809, "lr": 0.0002, "elapsed_sec": 13548.989944696426, "step_time_sec": 8.23211393400561, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.213010787963867, "lr": 0.0002, "elapsed_sec": 13557.22572684288, "step_time_sec": 8.235612426971784, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.229742527008057, "lr": 0.0002, "elapsed_sec": 13565.463589191437, "step_time_sec": 8.237703413993586, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.243343830108643, "lr": 0.0002, "elapsed_sec": 13573.701701402664, "step_time_sec": 8.237933308002539, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.25315523147583, "lr": 0.0002, "elapsed_sec": 13581.939574241638, "step_time_sec": 8.237692283000797, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.298964977264404, "lr": 0.0002, "elapsed_sec": 13590.176717996597, "step_time_sec": 8.236987481999677, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.258061408996582, "lr": 0.0002, "elapsed_sec": 13598.416242837906, "step_time_sec": 8.239382097002817, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.080819129943848, "lr": 0.0002, "elapsed_sec": 13606.653576135635, "step_time_sec": 8.237150508008199, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.213656902313232, "lr": 0.0002, "elapsed_sec": 13614.890910625458, "step_time_sec": 8.237229588994524, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.219189643859863, "lr": 0.0002, "elapsed_sec": 13623.128061532974, "step_time_sec": 8.236975715990411, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.204822540283203, "lr": 0.0002, "elapsed_sec": 13631.36495065689, "step_time_sec": 8.23672121600248, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.351865768432617, "lr": 0.0002, "elapsed_sec": 13639.603792905807, "step_time_sec": 8.238663084019208, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.275698184967041, "lr": 0.0002, "elapsed_sec": 13647.840425491333, "step_time_sec": 8.236448229989037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.1153364181518555, "lr": 0.0002, "elapsed_sec": 13656.078261137009, "step_time_sec": 8.237701264995849, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.165585994720459, "lr": 0.0002, "elapsed_sec": 13664.31565284729, "step_time_sec": 8.237225711985957, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.958221435546875, "lr": 0.0002, "elapsed_sec": 13672.553053379059, "step_time_sec": 8.237221458985005, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.204303741455078, "lr": 0.0002, "elapsed_sec": 13680.791491270065, "step_time_sec": 8.238305755017791, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.124622821807861, "lr": 0.0002, "elapsed_sec": 13689.028663158417, "step_time_sec": 8.237034769990714, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.202895641326904, "lr": 0.0002, "elapsed_sec": 13697.26683807373, "step_time_sec": 8.237964783998905, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.2242937088012695, "lr": 0.0002, "elapsed_sec": 13705.504127979279, "step_time_sec": 8.237162417004583, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.178075790405273, "lr": 0.0002, "elapsed_sec": 13713.74300646782, "step_time_sec": 8.238662440999178, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.319624900817871, "lr": 0.0002, "elapsed_sec": 13721.981632947922, "step_time_sec": 8.238495916011743, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.2720208168029785, "lr": 0.0002, "elapsed_sec": 13730.218653917313, "step_time_sec": 8.23683012498077, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.340462684631348, "lr": 0.0002, "elapsed_sec": 13738.458318710327, "step_time_sec": 8.239500992989633, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.180654525756836, "lr": 0.0002, "elapsed_sec": 13746.693904161453, "step_time_sec": 8.235478682006942, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.263378143310547, "lr": 0.0002, "elapsed_sec": 13754.931158781052, "step_time_sec": 8.237029324984178, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.186205863952637, "lr": 0.0002, "elapsed_sec": 13763.167999744415, "step_time_sec": 8.236678816989297, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.219064712524414, "lr": 0.0002, "elapsed_sec": 13771.405677556992, "step_time_sec": 8.237517995003145, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.037460803985596, "lr": 0.0002, "elapsed_sec": 13779.644345521927, "step_time_sec": 8.238531884009717, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.207411289215088, "lr": 0.0002, "elapsed_sec": 13787.87995004654, "step_time_sec": 8.23546583999996, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.20258092880249, "lr": 0.0002, "elapsed_sec": 13796.118708372116, "step_time_sec": 8.238567669992335, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.150393486022949, "lr": 0.0002, "elapsed_sec": 13804.356909036636, "step_time_sec": 8.238064559991471, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.2528791427612305, "lr": 0.0002, "elapsed_sec": 13812.596286773682, "step_time_sec": 8.23919080500491, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.238315582275391, "lr": 0.0002, "elapsed_sec": 13820.832505464554, "step_time_sec": 8.2360750779917, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.287368297576904, "lr": 0.0002, "elapsed_sec": 13829.070392608643, "step_time_sec": 8.237713403010275, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.147230625152588, "lr": 0.0002, "elapsed_sec": 13837.308336257935, "step_time_sec": 8.237772357999347, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.149064540863037, "lr": 0.0002, "elapsed_sec": 13845.545016288757, "step_time_sec": 8.236522478982806, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.122503280639648, "lr": 0.0002, "elapsed_sec": 13853.784132003784, "step_time_sec": 8.238925980986096, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.294088840484619, "lr": 0.0002, "elapsed_sec": 13862.02141880989, "step_time_sec": 8.237149469001452, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.121148586273193, "lr": 0.0002, "elapsed_sec": 13870.260088205338, "step_time_sec": 8.238492286007386, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.155667304992676, "lr": 0.0002, "elapsed_sec": 13878.497885227203, "step_time_sec": 8.237667791981949, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.277134418487549, "lr": 0.0002, "elapsed_sec": 13886.736211061478, "step_time_sec": 8.23814596500597, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.229856967926025, "lr": 0.0002, "elapsed_sec": 13894.973202943802, "step_time_sec": 8.236843084014254, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.187621116638184, "lr": 0.0002, "elapsed_sec": 13903.210252523422, "step_time_sec": 8.236847011983627, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.21621561050415, "lr": 0.0002, "elapsed_sec": 13911.44923901558, "step_time_sec": 8.238820525002666, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.391763687133789, "lr": 0.0002, "elapsed_sec": 13919.687973737717, "step_time_sec": 8.238578820019029, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.145547389984131, "lr": 0.0002, "elapsed_sec": 13927.926803350449, "step_time_sec": 8.238647543999832, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.228546619415283, "lr": 0.0002, "elapsed_sec": 13936.164400815964, "step_time_sec": 8.237456240982283, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.20768928527832, "lr": 0.0002, "elapsed_sec": 13944.40023636818, "step_time_sec": 8.235659519996261, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.185385704040527, "lr": 0.0002, "elapsed_sec": 13952.637954235077, "step_time_sec": 8.237569352000719, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.2006144523620605, "lr": 0.0002, "elapsed_sec": 13960.874725818634, "step_time_sec": 8.236623789009172, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.151601314544678, "lr": 0.0002, "elapsed_sec": 13969.11409497261, "step_time_sec": 8.239207265985897, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.255289554595947, "lr": 0.0002, "elapsed_sec": 13977.352010011673, "step_time_sec": 8.23773602899746, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.244237899780273, "lr": 0.0002, "elapsed_sec": 13985.589442491531, "step_time_sec": 8.23728467800538, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.265559196472168, "lr": 0.0002, "elapsed_sec": 13993.827278614044, "step_time_sec": 8.237680802994873, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.368333339691162, "lr": 0.0002, "elapsed_sec": 14002.064344406128, "step_time_sec": 8.236875712987967, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.046143054962158, "lr": 0.0002, "elapsed_sec": 14010.303629875183, "step_time_sec": 8.239126567001222, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.268409252166748, "lr": 0.0002, "elapsed_sec": 14018.539751768112, "step_time_sec": 8.2359810390044, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.213136196136475, "lr": 0.0002, "elapsed_sec": 14026.777091741562, "step_time_sec": 8.237165558995912, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.226988315582275, "lr": 0.0002, "elapsed_sec": 14035.014121055603, "step_time_sec": 8.236889853025787, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.223604679107666, "lr": 0.0002, "elapsed_sec": 14043.252193450928, "step_time_sec": 8.237897487008013, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.452968120574951, "lr": 0.0002, "elapsed_sec": 14051.48848414421, "step_time_sec": 8.23611956401146, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.251842498779297, "lr": 0.0002, "elapsed_sec": 14059.725481033325, "step_time_sec": 8.2368482099846, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.172296047210693, "lr": 0.0002, "elapsed_sec": 14067.962767601013, "step_time_sec": 8.237146934989141, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.213274002075195, "lr": 0.0002, "elapsed_sec": 14076.199142694473, "step_time_sec": 8.236226793989772, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.157473087310791, "lr": 0.0002, "elapsed_sec": 14084.437985181808, "step_time_sec": 8.2386303319945, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.114118576049805, "lr": 0.0002, "elapsed_sec": 14092.67545413971, "step_time_sec": 8.237320568005089, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.224738121032715, "lr": 0.0002, "elapsed_sec": 14100.912285327911, "step_time_sec": 8.236720194021473, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.125312805175781, "lr": 0.0002, "elapsed_sec": 14109.15121126175, "step_time_sec": 8.238686369993957, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.234676361083984, "lr": 0.0002, "elapsed_sec": 14117.38691830635, "step_time_sec": 8.235551757010398, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.283975124359131, "lr": 0.0002, "elapsed_sec": 14125.624715805054, "step_time_sec": 8.23764896998182, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.066939830780029, "lr": 0.0002, "elapsed_sec": 14133.86228442192, "step_time_sec": 8.23739794100402, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.307112216949463, "lr": 0.0002, "elapsed_sec": 14142.100722789764, "step_time_sec": 8.238259163015755, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.265880107879639, "lr": 0.0002, "elapsed_sec": 14150.33938741684, "step_time_sec": 8.23849253699882, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.185434818267822, "lr": 0.0002, "elapsed_sec": 14158.57698059082, "step_time_sec": 8.237437734002015, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.126810073852539, "lr": 0.0002, "elapsed_sec": 14166.815521001816, "step_time_sec": 8.23836907799705, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.190891265869141, "lr": 0.0002, "elapsed_sec": 14175.05181312561, "step_time_sec": 8.236180507024983, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.267226696014404, "lr": 0.0002, "elapsed_sec": 14183.289039850235, "step_time_sec": 8.237002907000715, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.090719699859619, "lr": 0.0002, "elapsed_sec": 14191.525908470154, "step_time_sec": 8.236710728990147, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.133795261383057, "lr": 0.0002, "elapsed_sec": 14199.764369010925, "step_time_sec": 8.238328040984925, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.165402412414551, "lr": 0.0002, "elapsed_sec": 14208.00129365921, "step_time_sec": 8.23680633798358, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.129800796508789, "lr": 0.0002, "elapsed_sec": 14216.238565444946, "step_time_sec": 8.237083339015953, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.9824676513671875, "lr": 0.0002, "elapsed_sec": 14224.477471113205, "step_time_sec": 8.23872257801122, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.24062442779541, "lr": 0.0002, "elapsed_sec": 14232.71405172348, "step_time_sec": 8.236394710984314, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.224229335784912, "lr": 0.0002, "elapsed_sec": 14240.952909946442, "step_time_sec": 8.238743655994767, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.174773216247559, "lr": 0.0002, "elapsed_sec": 14249.19173836708, "step_time_sec": 8.238652074011043, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.162736415863037, "lr": 0.0002, "elapsed_sec": 14257.429275035858, "step_time_sec": 8.237409697001567, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.2048797607421875, "lr": 0.0002, "elapsed_sec": 14265.666094779968, "step_time_sec": 8.23662796401186, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.0071797370910645, "lr": 0.0002, "elapsed_sec": 14273.903432130814, "step_time_sec": 8.237181932025123, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.075562477111816, "lr": 0.0002, "elapsed_sec": 14282.14088511467, "step_time_sec": 8.237256222986616, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.149485111236572, "lr": 0.0002, "elapsed_sec": 14290.378581285477, "step_time_sec": 8.237496679998003, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.171418190002441, "lr": 0.0002, "elapsed_sec": 14298.617161989212, "step_time_sec": 8.238499188999413, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.210690498352051, "lr": 0.0002, "elapsed_sec": 14306.854060411453, "step_time_sec": 8.236709100980079, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.0979485511779785, "lr": 0.0002, "elapsed_sec": 14315.09092617035, "step_time_sec": 8.236656224995386, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.144603729248047, "lr": 0.0002, "elapsed_sec": 14323.327944755554, "step_time_sec": 8.236866793013178, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.212184906005859, "lr": 0.0002, "elapsed_sec": 14331.565513849258, "step_time_sec": 8.237420270976145, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.10283899307251, "lr": 0.0002, "elapsed_sec": 14339.804443359375, "step_time_sec": 8.238777043006849, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.0332746505737305, "lr": 0.0002, "elapsed_sec": 14348.040704250336, "step_time_sec": 8.23611778797931, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.1595563888549805, "lr": 0.0002, "elapsed_sec": 14356.278762578964, "step_time_sec": 8.237895722006215, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.216962814331055, "lr": 0.0002, "elapsed_sec": 14364.517139911652, "step_time_sec": 8.238182119996054, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.027144908905029, "lr": 0.0002, "elapsed_sec": 14372.754296302795, "step_time_sec": 8.237008329975652, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.093922138214111, "lr": 0.0002, "elapsed_sec": 14380.992068529129, "step_time_sec": 8.237587401992641, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.218013286590576, "lr": 0.0002, "elapsed_sec": 14389.229194402695, "step_time_sec": 8.236961499991594, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.113111972808838, "lr": 0.0002, "elapsed_sec": 14397.468341350555, "step_time_sec": 8.238965138996718, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.091163635253906, "lr": 0.0002, "elapsed_sec": 14405.706612586975, "step_time_sec": 8.238116850989172, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.216073036193848, "lr": 0.0002, "elapsed_sec": 14413.945291519165, "step_time_sec": 8.23854218100314, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.093718528747559, "lr": 0.0002, "elapsed_sec": 14422.183214902878, "step_time_sec": 8.237725353013957, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.127114295959473, "lr": 0.0002, "elapsed_sec": 14430.419098377228, "step_time_sec": 8.235699746001046, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.136131763458252, "lr": 0.0002, "elapsed_sec": 14438.657607078552, "step_time_sec": 8.238391110993689, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.105370998382568, "lr": 0.0002, "elapsed_sec": 14446.893563747406, "step_time_sec": 8.23573744497844, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.178045272827148, "lr": 0.0002, "elapsed_sec": 14455.131609201431, "step_time_sec": 8.237884534988552, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.989068984985352, "lr": 0.0002, "elapsed_sec": 14463.36960363388, "step_time_sec": 8.237828726007137, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.143304347991943, "lr": 0.0002, "elapsed_sec": 14471.60656094551, "step_time_sec": 8.236784662003629, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.1806960105896, "lr": 0.0002, "elapsed_sec": 14479.84317445755, "step_time_sec": 8.23645890498301, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.192770004272461, "lr": 0.0002, "elapsed_sec": 14488.080564498901, "step_time_sec": 8.237244907999411, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.22744083404541, "lr": 0.0002, "elapsed_sec": 14496.319235563278, "step_time_sec": 8.238555002986686, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.150350570678711, "lr": 0.0002, "elapsed_sec": 14504.554940223694, "step_time_sec": 8.235457523987861, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.1283369064331055, "lr": 0.0002, "elapsed_sec": 14512.794147968292, "step_time_sec": 8.239063059008913, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.212568283081055, "lr": 0.0002, "elapsed_sec": 14521.032104730606, "step_time_sec": 8.23782137298258, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.134167194366455, "lr": 0.0002, "elapsed_sec": 14529.270028352737, "step_time_sec": 8.237733957997989, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.136016368865967, "lr": 0.0002, "elapsed_sec": 14537.507636070251, "step_time_sec": 8.237483067001449, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.175662517547607, "lr": 0.0002, "elapsed_sec": 14545.744525909424, "step_time_sec": 8.236685558978934, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.133482456207275, "lr": 0.0002, "elapsed_sec": 14553.982294559479, "step_time_sec": 8.237610095005948, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.198638916015625, "lr": 0.0002, "elapsed_sec": 14562.219358921051, "step_time_sec": 8.2368976160069, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.034807205200195, "lr": 0.0002, "elapsed_sec": 14570.457350969315, "step_time_sec": 8.237823589006439, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.158194541931152, "lr": 0.0002, "elapsed_sec": 14578.695078372955, "step_time_sec": 8.237641684012488, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.257108211517334, "lr": 0.0002, "elapsed_sec": 14586.932797193527, "step_time_sec": 8.237487400008831, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.014054775238037, "lr": 0.0002, "elapsed_sec": 14595.170847654343, "step_time_sec": 8.237879940978019, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.134880542755127, "lr": 0.0002, "elapsed_sec": 14603.408092021942, "step_time_sec": 8.237087467015954, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.114322662353516, "lr": 0.0002, "elapsed_sec": 14611.647542715073, "step_time_sec": 8.239276177977445, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.235638618469238, "lr": 0.0002, "elapsed_sec": 14619.885395288467, "step_time_sec": 8.237694099982036, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.034665107727051, "lr": 0.0002, "elapsed_sec": 14628.121761798859, "step_time_sec": 8.236227926012361, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.018316745758057, "lr": 0.0002, "elapsed_sec": 14636.360243558884, "step_time_sec": 8.238329865998821, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.1994476318359375, "lr": 0.0002, "elapsed_sec": 14644.596554756165, "step_time_sec": 8.236104643001454, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.177206993103027, "lr": 0.0002, "elapsed_sec": 14652.833552837372, "step_time_sec": 8.236832332011545, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.185940742492676, "lr": 0.0002, "elapsed_sec": 14661.070544242859, "step_time_sec": 8.236857551994035, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.196025371551514, "lr": 0.0002, "elapsed_sec": 14669.309431314468, "step_time_sec": 8.238697083026636, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.120103359222412, "lr": 0.0002, "elapsed_sec": 14677.546719789505, "step_time_sec": 8.237181686010445, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.176280498504639, "lr": 0.0002, "elapsed_sec": 14685.78440451622, "step_time_sec": 8.237489784980426, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.095302581787109, "lr": 0.0002, "elapsed_sec": 14694.02341222763, "step_time_sec": 8.238894532987615, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.1927714347839355, "lr": 0.0002, "elapsed_sec": 14702.26062130928, "step_time_sec": 8.23700082799769, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.1467132568359375, "lr": 0.0002, "elapsed_sec": 14710.499871015549, "step_time_sec": 8.239059204002842, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.058391094207764, "lr": 0.0002, "elapsed_sec": 14718.737454891205, "step_time_sec": 8.237499306997051, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.102205276489258, "lr": 0.0002, "elapsed_sec": 14726.976194143295, "step_time_sec": 8.23852239298867, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.09459924697876, "lr": 0.0002, "elapsed_sec": 14735.21462726593, "step_time_sec": 8.238251489005052, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.029178142547607, "lr": 0.0002, "elapsed_sec": 14743.451038599014, "step_time_sec": 8.23626505600987, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.11535120010376, "lr": 0.0002, "elapsed_sec": 14751.689384937286, "step_time_sec": 8.238213546021143, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.060381889343262, "lr": 0.0002, "elapsed_sec": 14759.92540550232, "step_time_sec": 8.235816478001652, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.067785263061523, "lr": 0.0002, "elapsed_sec": 14768.163443565369, "step_time_sec": 8.23789909100742, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.249032974243164, "lr": 0.0002, "elapsed_sec": 14776.400251150131, "step_time_sec": 8.236685235984623, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.084648609161377, "lr": 0.0002, "elapsed_sec": 14784.63893532753, "step_time_sec": 8.238483516004635, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.218665599822998, "lr": 0.0002, "elapsed_sec": 14792.87816619873, "step_time_sec": 8.239050545991631, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.046334266662598, "lr": 0.0002, "elapsed_sec": 14801.11497926712, "step_time_sec": 8.236675535998074, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.178956985473633, "lr": 0.0002, "elapsed_sec": 14809.35454082489, "step_time_sec": 8.239417333999882, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.115974426269531, "lr": 0.0002, "elapsed_sec": 14817.592063903809, "step_time_sec": 8.237368845002493, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.131571292877197, "lr": 0.0002, "elapsed_sec": 14825.830707788467, "step_time_sec": 8.238460360997124, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.993197441101074, "lr": 0.0002, "elapsed_sec": 14834.068606615067, "step_time_sec": 8.23773851597798, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.106697082519531, "lr": 0.0002, "elapsed_sec": 14842.305231332779, "step_time_sec": 8.236444558977382, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.971419811248779, "lr": 0.0002, "elapsed_sec": 14850.543233633041, "step_time_sec": 8.237823432020377, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.255711555480957, "lr": 0.0002, "elapsed_sec": 14858.779852628708, "step_time_sec": 8.236505093984306, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.110722541809082, "lr": 0.0002, "elapsed_sec": 14867.019029140472, "step_time_sec": 8.239016817999072, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.113140106201172, "lr": 0.0002, "elapsed_sec": 14875.256271362305, "step_time_sec": 8.237025028007338, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.0351457595825195, "lr": 0.0002, "elapsed_sec": 14883.495348453522, "step_time_sec": 8.238930522988085, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.013983726501465, "lr": 0.0002, "elapsed_sec": 14891.733051538467, "step_time_sec": 8.237564664013917, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.178585052490234, "lr": 0.0002, "elapsed_sec": 14899.969257354736, "step_time_sec": 8.236015772010433, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.020552158355713, "lr": 0.0002, "elapsed_sec": 14908.207235336304, "step_time_sec": 8.237841117981588, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.110262870788574, "lr": 0.0002, "elapsed_sec": 14916.443980693817, "step_time_sec": 8.236585420003394, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.118047714233398, "lr": 0.0002, "elapsed_sec": 14924.683067560196, "step_time_sec": 8.238895529997535, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.158466815948486, "lr": 0.0002, "elapsed_sec": 14932.921412467957, "step_time_sec": 8.238177021994488, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.094090461730957, "lr": 0.0002, "elapsed_sec": 14941.159208536148, "step_time_sec": 8.237625887006288, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.1352057456970215, "lr": 0.0002, "elapsed_sec": 14949.396546840668, "step_time_sec": 8.237178352981573, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.044894218444824, "lr": 0.0002, "elapsed_sec": 14957.632691383362, "step_time_sec": 8.236030828993535, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.147581100463867, "lr": 0.0002, "elapsed_sec": 14965.87035036087, "step_time_sec": 8.237460098986048, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.02544641494751, "lr": 0.0002, "elapsed_sec": 14974.108598709106, "step_time_sec": 8.238093919993844, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.051618576049805, "lr": 0.0002, "elapsed_sec": 14982.34754371643, "step_time_sec": 8.238786434027134, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.073084831237793, "lr": 0.0002, "elapsed_sec": 14990.585876703262, "step_time_sec": 8.238155501021538, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.955908298492432, "lr": 0.0002, "elapsed_sec": 14998.823055505753, "step_time_sec": 8.237068001995794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.95569372177124, "lr": 0.0002, "elapsed_sec": 15007.060560464859, "step_time_sec": 8.237292541016359, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.034017086029053, "lr": 0.0002, "elapsed_sec": 15015.298516988754, "step_time_sec": 8.237807051016716, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.056347370147705, "lr": 0.0002, "elapsed_sec": 15023.537538528442, "step_time_sec": 8.238864012993872, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.139965534210205, "lr": 0.0002, "elapsed_sec": 15031.774193525314, "step_time_sec": 8.236539502977394, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.155852794647217, "lr": 0.0002, "elapsed_sec": 15040.011081457138, "step_time_sec": 8.236689568992006, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.9895758628845215, "lr": 0.0002, "elapsed_sec": 15048.247928857803, "step_time_sec": 8.23673141899053, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.127640724182129, "lr": 0.0002, "elapsed_sec": 15056.484254598618, "step_time_sec": 8.236208110000007, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.038715362548828, "lr": 0.0002, "elapsed_sec": 15064.722319364548, "step_time_sec": 8.237858397013042, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.056773662567139, "lr": 0.0002, "elapsed_sec": 15072.961179733276, "step_time_sec": 8.238652065978386, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.230171203613281, "lr": 0.0002, "elapsed_sec": 15081.199803113937, "step_time_sec": 8.238445741997566, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.065550804138184, "lr": 0.0002, "elapsed_sec": 15089.436945438385, "step_time_sec": 8.236988154996652, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.995993137359619, "lr": 0.0002, "elapsed_sec": 15097.673963785172, "step_time_sec": 8.236882528988644, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.036993503570557, "lr": 0.0002, "elapsed_sec": 15105.912484645844, "step_time_sec": 8.238337793009123, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.105478763580322, "lr": 0.0002, "elapsed_sec": 15114.149500846863, "step_time_sec": 8.236912865017075, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.179108142852783, "lr": 0.0002, "elapsed_sec": 15122.388890028, "step_time_sec": 8.239181978016859, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.043497562408447, "lr": 0.0002, "elapsed_sec": 15130.6261947155, "step_time_sec": 8.237122096004896, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.95767068862915, "lr": 0.0002, "elapsed_sec": 15138.86342215538, "step_time_sec": 8.237113310024142, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.167346000671387, "lr": 0.0002, "elapsed_sec": 15147.100448608398, "step_time_sec": 8.23682588301017, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.145210266113281, "lr": 0.0002, "elapsed_sec": 15155.337764263153, "step_time_sec": 8.237124613980995, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.050331115722656, "lr": 0.0002, "elapsed_sec": 15163.576081752777, "step_time_sec": 8.238215092016617, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.227225303649902, "lr": 0.0002, "elapsed_sec": 15171.812197208405, "step_time_sec": 8.235937310004374, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.013734340667725, "lr": 0.0002, "elapsed_sec": 15180.05286192894, "step_time_sec": 8.24046582498704, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.045483589172363, "lr": 0.0002, "elapsed_sec": 15188.29001235962, "step_time_sec": 8.237010240001837, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.127965927124023, "lr": 0.0002, "elapsed_sec": 15196.528108358383, "step_time_sec": 8.237948180001695, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.20252799987793, "lr": 0.0002, "elapsed_sec": 15204.766129493713, "step_time_sec": 8.237888547999319, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.103487014770508, "lr": 0.0002, "elapsed_sec": 15213.00344491005, "step_time_sec": 8.237108916015131, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.10581636428833, "lr": 0.0002, "elapsed_sec": 15221.242102622986, "step_time_sec": 8.238491864991374, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.026205062866211, "lr": 0.0002, "elapsed_sec": 15229.479651927948, "step_time_sec": 8.237391871021828, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.968350887298584, "lr": 0.0002, "elapsed_sec": 15237.718456029892, "step_time_sec": 8.238670948019717, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.01998233795166, "lr": 0.0002, "elapsed_sec": 15245.956857204437, "step_time_sec": 8.238232731004246, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.22914981842041, "lr": 0.0002, "elapsed_sec": 15254.193422794342, "step_time_sec": 8.236402519018156, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.941863059997559, "lr": 0.0002, "elapsed_sec": 15262.430976390839, "step_time_sec": 8.237372800998855, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.069939136505127, "lr": 0.0002, "elapsed_sec": 15270.66787481308, "step_time_sec": 8.236744824011112, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.17672872543335, "lr": 0.0002, "elapsed_sec": 15278.906702756882, "step_time_sec": 8.238692293991335, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.158458232879639, "lr": 0.0002, "elapsed_sec": 15287.14513373375, "step_time_sec": 8.23824000597233, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.012870788574219, "lr": 0.0002, "elapsed_sec": 15295.383377552032, "step_time_sec": 8.23804698800086, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.174057960510254, "lr": 0.0002, "elapsed_sec": 15303.621902942657, "step_time_sec": 8.238355304987635, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.080540657043457, "lr": 0.0002, "elapsed_sec": 15311.859373092651, "step_time_sec": 8.237366947025293, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.1891279220581055, "lr": 0.0002, "elapsed_sec": 15320.09725689888, "step_time_sec": 8.237667548004538, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.9455084800720215, "lr": 0.0002, "elapsed_sec": 15328.335303068161, "step_time_sec": 8.23790679697413, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.1084089279174805, "lr": 0.0002, "elapsed_sec": 15336.573600292206, "step_time_sec": 8.238118048000615, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.081230640411377, "lr": 0.0002, "elapsed_sec": 15344.81239438057, "step_time_sec": 8.238605972990626, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.05029296875, "lr": 0.0002, "elapsed_sec": 15353.048415660858, "step_time_sec": 8.23586276001879, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.065269947052002, "lr": 0.0002, "elapsed_sec": 15361.288093328476, "step_time_sec": 8.23949514402193, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.104930400848389, "lr": 0.0002, "elapsed_sec": 15369.524589061737, "step_time_sec": 8.236374389991397, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.053078651428223, "lr": 0.0002, "elapsed_sec": 15377.763488769531, "step_time_sec": 8.238762738008518, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.069220542907715, "lr": 0.0002, "elapsed_sec": 15385.999674081802, "step_time_sec": 8.235977596021257, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.1088128089904785, "lr": 0.0002, "elapsed_sec": 15394.238312005997, "step_time_sec": 8.238465625996469, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.149667263031006, "lr": 0.0002, "elapsed_sec": 15402.476842403412, "step_time_sec": 8.238362565025454, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.123683452606201, "lr": 0.0002, "elapsed_sec": 15410.713245630264, "step_time_sec": 8.23624513600953, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.95670747756958, "lr": 0.0002, "elapsed_sec": 15418.94985818863, "step_time_sec": 8.236475751007674, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.012176513671875, "lr": 0.0002, "elapsed_sec": 15427.186363220215, "step_time_sec": 8.236321169999428, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.1126251220703125, "lr": 0.0002, "elapsed_sec": 15435.423585176468, "step_time_sec": 8.237007999006892, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.085558891296387, "lr": 0.0002, "elapsed_sec": 15443.660153627396, "step_time_sec": 8.236406781012192, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.04786491394043, "lr": 0.0002, "elapsed_sec": 15451.897983789444, "step_time_sec": 8.237720647011884, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.09515905380249, "lr": 0.0002, "elapsed_sec": 15460.135288000107, "step_time_sec": 8.237124927021796, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.237826347351074, "lr": 0.0002, "elapsed_sec": 15468.371919631958, "step_time_sec": 8.23647275401163, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.06374979019165, "lr": 0.0002, "elapsed_sec": 15476.610100746155, "step_time_sec": 8.23806062398944, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.099347114562988, "lr": 0.0002, "elapsed_sec": 15484.8479616642, "step_time_sec": 8.237655539996922, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.137527942657471, "lr": 0.0002, "elapsed_sec": 15493.087054729462, "step_time_sec": 8.238956943008816, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.084342002868652, "lr": 0.0002, "elapsed_sec": 15501.325273990631, "step_time_sec": 8.238035190006485, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.1208343505859375, "lr": 0.0002, "elapsed_sec": 15509.562268733978, "step_time_sec": 8.236843191989465, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.0058794021606445, "lr": 0.0002, "elapsed_sec": 15517.80021572113, "step_time_sec": 8.23777077300474, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.879157066345215, "lr": 0.0002, "elapsed_sec": 15526.100558280945, "step_time_sec": 8.242443952010944, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.112834930419922, "lr": 0.0002, "elapsed_sec": 15534.337787628174, "step_time_sec": 8.237079371989239, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.018529415130615, "lr": 0.0002, "elapsed_sec": 15542.577363967896, "step_time_sec": 8.239382294006646, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.993391990661621, "lr": 0.0002, "elapsed_sec": 15550.814177274704, "step_time_sec": 8.236673574981978, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.973626136779785, "lr": 0.0002, "elapsed_sec": 15559.05281496048, "step_time_sec": 8.238520288985455, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.143945217132568, "lr": 0.0002, "elapsed_sec": 15567.29010128975, "step_time_sec": 8.23707577001187, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.264761447906494, "lr": 0.0002, "elapsed_sec": 15575.526881694794, "step_time_sec": 8.23656681299326, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.9676618576049805, "lr": 0.0002, "elapsed_sec": 15583.763361692429, "step_time_sec": 8.2363166710129, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.125551223754883, "lr": 0.0002, "elapsed_sec": 15592.000412225723, "step_time_sec": 8.23693617299432, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.105132579803467, "lr": 0.0002, "elapsed_sec": 15600.239140510559, "step_time_sec": 8.238551976013696, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.089249134063721, "lr": 0.0002, "elapsed_sec": 15608.476792573929, "step_time_sec": 8.237433706992306, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.104232311248779, "lr": 0.0002, "elapsed_sec": 15616.71646952629, "step_time_sec": 8.23950630199397, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.0355353355407715, "lr": 0.0002, "elapsed_sec": 15624.954807281494, "step_time_sec": 8.238175058009801, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.163853168487549, "lr": 0.0002, "elapsed_sec": 15633.19179058075, "step_time_sec": 8.236812481016386, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.953299045562744, "lr": 0.0002, "elapsed_sec": 15641.430419445038, "step_time_sec": 8.238496977020986, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.017730236053467, "lr": 0.0002, "elapsed_sec": 15649.667315244675, "step_time_sec": 8.23669567098841, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.067231178283691, "lr": 0.0002, "elapsed_sec": 15657.90595984459, "step_time_sec": 8.23848972999258, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.072965145111084, "lr": 0.0002, "elapsed_sec": 15666.143072128296, "step_time_sec": 8.236940959992353, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.032601356506348, "lr": 0.0002, "elapsed_sec": 15674.381091833115, "step_time_sec": 8.237881078995997, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.062229156494141, "lr": 0.0002, "elapsed_sec": 15682.618793725967, "step_time_sec": 8.23754237499088, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.087886333465576, "lr": 0.0002, "elapsed_sec": 15690.855512142181, "step_time_sec": 8.236547367996536, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.186297416687012, "lr": 0.0002, "elapsed_sec": 15699.09385061264, "step_time_sec": 8.238150491990382, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.05950403213501, "lr": 0.0002, "elapsed_sec": 15707.33097410202, "step_time_sec": 8.236993420985527, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.086407661437988, "lr": 0.0002, "elapsed_sec": 15715.569983005524, "step_time_sec": 8.238828664994799, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.062439441680908, "lr": 0.0002, "elapsed_sec": 15723.808690786362, "step_time_sec": 8.238488128001336, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.990935802459717, "lr": 0.0002, "elapsed_sec": 15732.04559135437, "step_time_sec": 8.236788107984466, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.899148941040039, "lr": 0.0002, "elapsed_sec": 15740.281628131866, "step_time_sec": 8.235901119012851, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.133002281188965, "lr": 0.0002, "elapsed_sec": 15748.519449710846, "step_time_sec": 8.237614843004849, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.949366092681885, "lr": 0.0002, "elapsed_sec": 15756.757710933685, "step_time_sec": 8.238072620006278, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.04243278503418, "lr": 0.0002, "elapsed_sec": 15764.995311021805, "step_time_sec": 8.237469398998655, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.076997756958008, "lr": 0.0002, "elapsed_sec": 15773.232865810394, "step_time_sec": 8.237392638984602, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.0691986083984375, "lr": 0.0002, "elapsed_sec": 15781.472410678864, "step_time_sec": 8.239370427996619, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.018735408782959, "lr": 0.0002, "elapsed_sec": 15789.708668231964, "step_time_sec": 8.236094952997519, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.135227203369141, "lr": 0.0002, "elapsed_sec": 15797.945002555847, "step_time_sec": 8.23618430798524, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.961730480194092, "lr": 0.0002, "elapsed_sec": 15806.181407928467, "step_time_sec": 8.236233555013314, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.953001976013184, "lr": 0.0002, "elapsed_sec": 15814.419211149216, "step_time_sec": 8.23768803299754, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.06337308883667, "lr": 0.0002, "elapsed_sec": 15822.655210494995, "step_time_sec": 8.235811727994587, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.075684547424316, "lr": 0.0002, "elapsed_sec": 15830.893066167831, "step_time_sec": 8.237708353000926, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.8890461921691895, "lr": 0.0002, "elapsed_sec": 15839.133872509003, "step_time_sec": 8.240658960014116, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.042243003845215, "lr": 0.0002, "elapsed_sec": 15847.370755195618, "step_time_sec": 8.23672474900377, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.057312488555908, "lr": 0.0002, "elapsed_sec": 15855.609194993973, "step_time_sec": 8.238235280994559, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.0274128913879395, "lr": 0.0002, "elapsed_sec": 15863.846256017685, "step_time_sec": 8.236888966988772, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.053271293640137, "lr": 0.0002, "elapsed_sec": 15872.085191726685, "step_time_sec": 8.23879890498938, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.006120204925537, "lr": 0.0002, "elapsed_sec": 15880.323424100876, "step_time_sec": 8.238056133995997, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.050276279449463, "lr": 0.0002, "elapsed_sec": 15888.561531305313, "step_time_sec": 8.237953870004276, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.070277214050293, "lr": 0.0002, "elapsed_sec": 15896.799627065659, "step_time_sec": 8.237903911998728, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.0407257080078125, "lr": 0.0002, "elapsed_sec": 15905.03563117981, "step_time_sec": 8.23585736899986, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.015465259552002, "lr": 0.0002, "elapsed_sec": 15913.274706363678, "step_time_sec": 8.238929023995297, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.09940242767334, "lr": 0.0002, "elapsed_sec": 15921.51230430603, "step_time_sec": 8.237417630996788, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.058826446533203, "lr": 0.0002, "elapsed_sec": 15929.75030207634, "step_time_sec": 8.237828953017015, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.981389045715332, "lr": 0.0002, "elapsed_sec": 15937.987411022186, "step_time_sec": 8.236979344976135, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.019990921020508, "lr": 0.0002, "elapsed_sec": 15946.223909854889, "step_time_sec": 8.236278172000311, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.013602256774902, "lr": 0.0002, "elapsed_sec": 15954.462800264359, "step_time_sec": 8.238791902986122, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.919231414794922, "lr": 0.0002, "elapsed_sec": 15962.69969010353, "step_time_sec": 8.236666287004482, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.0683794021606445, "lr": 0.0002, "elapsed_sec": 15970.937905073166, "step_time_sec": 8.238109454017831, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.0773162841796875, "lr": 0.0002, "elapsed_sec": 15979.175874710083, "step_time_sec": 8.237767700018594, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.059515953063965, "lr": 0.0002, "elapsed_sec": 15987.413507461548, "step_time_sec": 8.237507949990686, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.136651515960693, "lr": 0.0002, "elapsed_sec": 15995.651739120483, "step_time_sec": 8.238031607004814, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.97283935546875, "lr": 0.0002, "elapsed_sec": 16003.888986110687, "step_time_sec": 8.237100416008616, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.934141159057617, "lr": 0.0002, "elapsed_sec": 16012.127723455429, "step_time_sec": 8.23855657398235, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.075517654418945, "lr": 0.0002, "elapsed_sec": 16020.366078138351, "step_time_sec": 8.238224338012515, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.035972595214844, "lr": 0.0002, "elapsed_sec": 16028.60397696495, "step_time_sec": 8.237725246988703, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.000176429748535, "lr": 0.0002, "elapsed_sec": 16036.842668294907, "step_time_sec": 8.23852539001382, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.907454490661621, "lr": 0.0002, "elapsed_sec": 16045.080116987228, "step_time_sec": 8.237299964996055, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.974312782287598, "lr": 0.0002, "elapsed_sec": 16053.318464279175, "step_time_sec": 8.238163953996263, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.9681782722473145, "lr": 0.0002, "elapsed_sec": 16061.555189371109, "step_time_sec": 8.236543456005165, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.969987869262695, "lr": 0.0002, "elapsed_sec": 16069.79306769371, "step_time_sec": 8.237684480001917, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.980356693267822, "lr": 0.0002, "elapsed_sec": 16078.030745506287, "step_time_sec": 8.237528666999424, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.170294761657715, "lr": 0.0002, "elapsed_sec": 16086.26915693283, "step_time_sec": 8.238245633983752, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.11378812789917, "lr": 0.0002, "elapsed_sec": 16094.507339954376, "step_time_sec": 8.238045764999697, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.975762367248535, "lr": 0.0002, "elapsed_sec": 16102.743337869644, "step_time_sec": 8.235853894992033, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.990817070007324, "lr": 0.0002, "elapsed_sec": 16110.981678009033, "step_time_sec": 8.238139833993046, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.0208611488342285, "lr": 0.0002, "elapsed_sec": 16119.219089746475, "step_time_sec": 8.237243337003747, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.038038730621338, "lr": 0.0002, "elapsed_sec": 16127.456776618958, "step_time_sec": 8.237568655022187, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.043548107147217, "lr": 0.0002, "elapsed_sec": 16135.695348024368, "step_time_sec": 8.238373489992227, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.049440383911133, "lr": 0.0002, "elapsed_sec": 16143.933880329132, "step_time_sec": 8.238377735979157, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.886075973510742, "lr": 0.0002, "elapsed_sec": 16152.172827482224, "step_time_sec": 8.238766767986817, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.8983540534973145, "lr": 0.0002, "elapsed_sec": 16160.409427165985, "step_time_sec": 8.236442812980385, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.051055431365967, "lr": 0.0002, "elapsed_sec": 16168.648599624634, "step_time_sec": 8.239056001009885, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.057409286499023, "lr": 0.0002, "elapsed_sec": 16176.88583278656, "step_time_sec": 8.237018091022037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.999445915222168, "lr": 0.0002, "elapsed_sec": 16185.12441444397, "step_time_sec": 8.23841357199126, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.040074348449707, "lr": 0.0002, "elapsed_sec": 16193.362425327301, "step_time_sec": 8.237892150995322, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.853452682495117, "lr": 0.0002, "elapsed_sec": 16201.599839687347, "step_time_sec": 8.237246354983654, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.985223770141602, "lr": 0.0002, "elapsed_sec": 16209.838273763657, "step_time_sec": 8.238221540988889, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.1272664070129395, "lr": 0.0002, "elapsed_sec": 16218.075913906097, "step_time_sec": 8.23750438800198, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.962246417999268, "lr": 0.0002, "elapsed_sec": 16226.313703775406, "step_time_sec": 8.237642425985541, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.844414710998535, "lr": 0.0002, "elapsed_sec": 16234.551881074905, "step_time_sec": 8.237987119005993, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.880814075469971, "lr": 0.0002, "elapsed_sec": 16242.789041280746, "step_time_sec": 8.237022678978974, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.966894149780273, "lr": 0.0002, "elapsed_sec": 16251.027637958527, "step_time_sec": 8.238394461019197, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.956801891326904, "lr": 0.0002, "elapsed_sec": 16259.264830350876, "step_time_sec": 8.23707010401995, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.913849353790283, "lr": 0.0002, "elapsed_sec": 16267.502967357635, "step_time_sec": 8.238010264991317, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.918597221374512, "lr": 0.0002, "elapsed_sec": 16275.74072766304, "step_time_sec": 8.23753206297988, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.059228897094727, "lr": 0.0002, "elapsed_sec": 16283.978014945984, "step_time_sec": 8.237136890995316, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.085511207580566, "lr": 0.0002, "elapsed_sec": 16292.215780735016, "step_time_sec": 8.237616224010708, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.010023593902588, "lr": 0.0002, "elapsed_sec": 16300.453861951828, "step_time_sec": 8.237916329002474, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.979783535003662, "lr": 0.0002, "elapsed_sec": 16308.69209742546, "step_time_sec": 8.238122759998078, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.932241439819336, "lr": 0.0002, "elapsed_sec": 16316.92914390564, "step_time_sec": 8.236823250015732, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.97825813293457, "lr": 0.0002, "elapsed_sec": 16325.167875051498, "step_time_sec": 8.238562104990706, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.091817855834961, "lr": 0.0002, "elapsed_sec": 16333.407056093216, "step_time_sec": 8.239039163017878, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.950571537017822, "lr": 0.0002, "elapsed_sec": 16341.644044160843, "step_time_sec": 8.236802507017273, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.976985454559326, "lr": 0.0002, "elapsed_sec": 16349.882344007492, "step_time_sec": 8.238128101016628, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.056337356567383, "lr": 0.0002, "elapsed_sec": 16358.118882417679, "step_time_sec": 8.236402330978308, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.023752212524414, "lr": 0.0002, "elapsed_sec": 16366.357507705688, "step_time_sec": 8.238410284015117, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.154882431030273, "lr": 0.0002, "elapsed_sec": 16374.594771623611, "step_time_sec": 8.237121940997895, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.1052117347717285, "lr": 0.0002, "elapsed_sec": 16382.83253288269, "step_time_sec": 8.23762124098721, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.025475978851318, "lr": 0.0002, "elapsed_sec": 16391.06947827339, "step_time_sec": 8.236733039986575, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.0454816818237305, "lr": 0.0002, "elapsed_sec": 16399.306536197662, "step_time_sec": 8.236880117998226, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.952042102813721, "lr": 0.0002, "elapsed_sec": 16407.545730113983, "step_time_sec": 8.239040782005759, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.8971357345581055, "lr": 0.0002, "elapsed_sec": 16415.782079219818, "step_time_sec": 8.236186726018786, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.993683338165283, "lr": 0.0002, "elapsed_sec": 16424.02046394348, "step_time_sec": 8.238255133997882, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.99247407913208, "lr": 0.0002, "elapsed_sec": 16432.25858283043, "step_time_sec": 8.237962070008507, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.030678749084473, "lr": 0.0002, "elapsed_sec": 16440.49645423889, "step_time_sec": 8.23764234400005, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.098254203796387, "lr": 0.0002, "elapsed_sec": 16448.73336458206, "step_time_sec": 8.236809654015815, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.953746318817139, "lr": 0.0002, "elapsed_sec": 16456.969892263412, "step_time_sec": 8.236318447015947, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.960512638092041, "lr": 0.0002, "elapsed_sec": 16465.208740472794, "step_time_sec": 8.23867876100121, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.028336524963379, "lr": 0.0002, "elapsed_sec": 16473.445604085922, "step_time_sec": 8.23671783899772, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.001224517822266, "lr": 0.0002, "elapsed_sec": 16481.683672189713, "step_time_sec": 8.23790579198976, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.886001110076904, "lr": 0.0002, "elapsed_sec": 16489.920983552933, "step_time_sec": 8.23713702801615, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.454108715057373, "lr": 0.0002, "elapsed_sec": 16498.159115552902, "step_time_sec": 8.237991522008087, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.970856189727783, "lr": 0.0002, "elapsed_sec": 16506.39703822136, "step_time_sec": 8.237813557992922, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.056257247924805, "lr": 0.0002, "elapsed_sec": 16514.63407588005, "step_time_sec": 8.236849434993928, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.9278435707092285, "lr": 0.0002, "elapsed_sec": 16522.873131275177, "step_time_sec": 8.238913310982753, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.146968841552734, "lr": 0.0002, "elapsed_sec": 16531.11089205742, "step_time_sec": 8.237534189014696, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.042311191558838, "lr": 0.0002, "elapsed_sec": 16539.34849715233, "step_time_sec": 8.237447472987697, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.802256107330322, "lr": 0.0002, "elapsed_sec": 16547.58679819107, "step_time_sec": 8.238144538976485, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.953668117523193, "lr": 0.0002, "elapsed_sec": 16555.823704719543, "step_time_sec": 8.236744207999436, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.021764755249023, "lr": 0.0002, "elapsed_sec": 16564.062953710556, "step_time_sec": 8.23911987899919, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.018695831298828, "lr": 0.0002, "elapsed_sec": 16572.29958629608, "step_time_sec": 8.236465997004416, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.929755210876465, "lr": 0.0002, "elapsed_sec": 16580.536997318268, "step_time_sec": 8.23722138197627, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.959891319274902, "lr": 0.0002, "elapsed_sec": 16588.774585962296, "step_time_sec": 51.557499764021486, "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": 0.9995068030257244, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 2001, "loss": 4.922317981719971, "lr": 0.0002, "elapsed_sec": 16640.338764429092, "step_time_sec": 8.243986916990252, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.018273830413818, "lr": 0.0002, "elapsed_sec": 16648.57446026802, "step_time_sec": 8.235567927011289, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.879498481750488, "lr": 0.0002, "elapsed_sec": 16656.812227249146, "step_time_sec": 8.23756127699744, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.941476821899414, "lr": 0.0002, "elapsed_sec": 16665.050041913986, "step_time_sec": 8.237657616002252, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.9486002922058105, "lr": 0.0002, "elapsed_sec": 16673.28837442398, "step_time_sec": 8.238203951972537, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.052792549133301, "lr": 0.0002, "elapsed_sec": 16681.527391910553, "step_time_sec": 8.238819075020729, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.977291584014893, "lr": 0.0002, "elapsed_sec": 16689.762707471848, "step_time_sec": 8.235152187990025, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.148684024810791, "lr": 0.0002, "elapsed_sec": 16697.999825954437, "step_time_sec": 8.236996169027407, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.929917812347412, "lr": 0.0002, "elapsed_sec": 16706.23602938652, "step_time_sec": 8.236017888004426, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.933377742767334, "lr": 0.0002, "elapsed_sec": 16714.473121643066, "step_time_sec": 8.236910539999371, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.955269813537598, "lr": 0.0002, "elapsed_sec": 16722.71112537384, "step_time_sec": 8.237850643985439, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.937071323394775, "lr": 0.0002, "elapsed_sec": 16730.948351860046, "step_time_sec": 8.237034032994416, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.031503200531006, "lr": 0.0002, "elapsed_sec": 16739.185732126236, "step_time_sec": 8.237259323999751, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.977535724639893, "lr": 0.0002, "elapsed_sec": 16747.422523021698, "step_time_sec": 8.236593960988102, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.9416913986206055, "lr": 0.0002, "elapsed_sec": 16755.659690856934, "step_time_sec": 8.237020250991918, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.029823303222656, "lr": 0.0002, "elapsed_sec": 16763.895843744278, "step_time_sec": 8.235975893010618, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.981520175933838, "lr": 0.0002, "elapsed_sec": 16772.13254904747, "step_time_sec": 8.236571597022703, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.008968353271484, "lr": 0.0002, "elapsed_sec": 16780.370329618454, "step_time_sec": 8.237597994011594, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.920145034790039, "lr": 0.0002, "elapsed_sec": 16788.607828855515, "step_time_sec": 8.237386359018274, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.958712100982666, "lr": 0.0002, "elapsed_sec": 16796.84540629387, "step_time_sec": 8.237367093999637, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.896470069885254, "lr": 0.0002, "elapsed_sec": 16805.082494735718, "step_time_sec": 8.236918285983847, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.9255547523498535, "lr": 0.0002, "elapsed_sec": 16813.32045316696, "step_time_sec": 8.237853541009827, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.875556468963623, "lr": 0.0002, "elapsed_sec": 16821.556654930115, "step_time_sec": 8.235982482001418, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.986764430999756, "lr": 0.0002, "elapsed_sec": 16829.79680967331, "step_time_sec": 8.240010798006551, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.933938980102539, "lr": 0.0002, "elapsed_sec": 16838.03373980522, "step_time_sec": 8.236794493015623, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.889519214630127, "lr": 0.0002, "elapsed_sec": 16846.270320892334, "step_time_sec": 8.236381650989642, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.964709758758545, "lr": 0.0002, "elapsed_sec": 16854.507917165756, "step_time_sec": 8.237425468018046, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.931051731109619, "lr": 0.0002, "elapsed_sec": 16862.746431827545, "step_time_sec": 8.238349310006015, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.983593463897705, "lr": 0.0002, "elapsed_sec": 16870.984478235245, "step_time_sec": 8.237888694973662, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.924543857574463, "lr": 0.0002, "elapsed_sec": 16879.220808267593, "step_time_sec": 8.236162746994523, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.867337703704834, "lr": 0.0002, "elapsed_sec": 16887.46020412445, "step_time_sec": 8.239217094000196, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.980381965637207, "lr": 0.0002, "elapsed_sec": 16895.698229312897, "step_time_sec": 8.23787509900285, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.950557231903076, "lr": 0.0002, "elapsed_sec": 16903.93601155281, "step_time_sec": 8.23759525999776, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.005340099334717, "lr": 0.0002, "elapsed_sec": 16912.1722676754, "step_time_sec": 8.23614157299744, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.964684963226318, "lr": 0.0002, "elapsed_sec": 16920.40876340866, "step_time_sec": 8.236306164006237, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.067929267883301, "lr": 0.0002, "elapsed_sec": 16928.646108150482, "step_time_sec": 8.237241066002753, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.14166784286499, "lr": 0.0002, "elapsed_sec": 16936.883333683014, "step_time_sec": 8.237010116019519, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.045099258422852, "lr": 0.0002, "elapsed_sec": 16945.121299743652, "step_time_sec": 8.237810907012317, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.943898677825928, "lr": 0.0002, "elapsed_sec": 16953.358091831207, "step_time_sec": 8.23667285501142, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.969322681427002, "lr": 0.0002, "elapsed_sec": 16961.595056056976, "step_time_sec": 8.236758312996244, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.9469099044799805, "lr": 0.0002, "elapsed_sec": 16969.832603931427, "step_time_sec": 8.237373042997206, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.031070232391357, "lr": 0.0002, "elapsed_sec": 16978.069526195526, "step_time_sec": 8.23679130099481, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.934319019317627, "lr": 0.0002, "elapsed_sec": 16986.307393550873, "step_time_sec": 8.237679924000986, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.955069541931152, "lr": 0.0002, "elapsed_sec": 16994.544864177704, "step_time_sec": 8.237309852003818, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.101129531860352, "lr": 0.0002, "elapsed_sec": 17002.78227329254, "step_time_sec": 8.237249347992474, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.057223320007324, "lr": 0.0002, "elapsed_sec": 17011.01860356331, "step_time_sec": 8.236159621999832, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.86039924621582, "lr": 0.0002, "elapsed_sec": 17019.25584435463, "step_time_sec": 8.237082684005145, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.909441947937012, "lr": 0.0002, "elapsed_sec": 17027.4938519001, "step_time_sec": 8.237846037984127, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.963437557220459, "lr": 0.0002, "elapsed_sec": 17035.73006772995, "step_time_sec": 8.236063087999355, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.065816402435303, "lr": 0.0002, "elapsed_sec": 17043.96851015091, "step_time_sec": 8.23830168301356, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.934907913208008, "lr": 0.0002, "elapsed_sec": 17052.20628428459, "step_time_sec": 8.237640984007157, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.994115352630615, "lr": 0.0002, "elapsed_sec": 17060.443489551544, "step_time_sec": 8.237018465006258, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.920947074890137, "lr": 0.0002, "elapsed_sec": 17068.681352853775, "step_time_sec": 8.237750425993, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.928854942321777, "lr": 0.0002, "elapsed_sec": 17076.918492794037, "step_time_sec": 8.236920598021243, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.915828227996826, "lr": 0.0002, "elapsed_sec": 17085.15734887123, "step_time_sec": 8.238702326023486, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.828474998474121, "lr": 0.0002, "elapsed_sec": 17093.394190073013, "step_time_sec": 8.236717710009543, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.778115272521973, "lr": 0.0002, "elapsed_sec": 17101.633816242218, "step_time_sec": 8.23939935999806, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.008076190948486, "lr": 0.0002, "elapsed_sec": 17109.871213674545, "step_time_sec": 8.237256594991777, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.030848979949951, "lr": 0.0002, "elapsed_sec": 17118.10812497139, "step_time_sec": 8.236780478997389, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.984222412109375, "lr": 0.0002, "elapsed_sec": 17126.344144821167, "step_time_sec": 8.23583976898226, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.102366924285889, "lr": 0.0002, "elapsed_sec": 17134.58138179779, "step_time_sec": 8.237082998995902, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.89292573928833, "lr": 0.0002, "elapsed_sec": 17142.8191614151, "step_time_sec": 8.237551928992616, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.916752338409424, "lr": 0.0002, "elapsed_sec": 17151.054930448532, "step_time_sec": 8.235601635009516, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.916094779968262, "lr": 0.0002, "elapsed_sec": 17159.291259288788, "step_time_sec": 8.236185110989027, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.088181972503662, "lr": 0.0002, "elapsed_sec": 17167.528670072556, "step_time_sec": 8.237250120000681, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.90495491027832, "lr": 0.0002, "elapsed_sec": 17175.76707649231, "step_time_sec": 8.238233418989694, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.943191051483154, "lr": 0.0002, "elapsed_sec": 17184.004276752472, "step_time_sec": 8.237088802998187, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.772139549255371, "lr": 0.0002, "elapsed_sec": 17192.241828918457, "step_time_sec": 8.237372592004249, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.97415018081665, "lr": 0.0002, "elapsed_sec": 17200.478677749634, "step_time_sec": 8.236665723001352, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.07147741317749, "lr": 0.0002, "elapsed_sec": 17208.71455025673, "step_time_sec": 8.235775064997142, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.829383373260498, "lr": 0.0002, "elapsed_sec": 17216.953721761703, "step_time_sec": 8.238970553007675, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.931319236755371, "lr": 0.0002, "elapsed_sec": 17225.19121336937, "step_time_sec": 8.237346558016725, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.98878288269043, "lr": 0.0002, "elapsed_sec": 17233.42936563492, "step_time_sec": 8.238010367989773, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.9841790199279785, "lr": 0.0002, "elapsed_sec": 17241.666072368622, "step_time_sec": 8.236492848984199, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.891979217529297, "lr": 0.0002, "elapsed_sec": 17249.9034948349, "step_time_sec": 8.237275065999711, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.058010578155518, "lr": 0.0002, "elapsed_sec": 17258.140533447266, "step_time_sec": 8.236906817008276, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.0397868156433105, "lr": 0.0002, "elapsed_sec": 17266.377341747284, "step_time_sec": 8.236652224004501, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.950645446777344, "lr": 0.0002, "elapsed_sec": 17274.614148378372, "step_time_sec": 8.236643818003358, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.911172866821289, "lr": 0.0002, "elapsed_sec": 17282.851773500443, "step_time_sec": 8.237462663993938, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.771438121795654, "lr": 0.0002, "elapsed_sec": 17291.09065580368, "step_time_sec": 8.238711470999988, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.938629627227783, "lr": 0.0002, "elapsed_sec": 17299.328293800354, "step_time_sec": 8.237498100002995, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.048403739929199, "lr": 0.0002, "elapsed_sec": 17307.56494784355, "step_time_sec": 8.236482190026436, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.0065155029296875, "lr": 0.0002, "elapsed_sec": 17315.803382873535, "step_time_sec": 8.238314546004403, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.945117473602295, "lr": 0.0002, "elapsed_sec": 17324.03972530365, "step_time_sec": 8.236216401011916, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.998676776885986, "lr": 0.0002, "elapsed_sec": 17332.27664375305, "step_time_sec": 8.236703573988052, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.871273994445801, "lr": 0.0002, "elapsed_sec": 17340.513706684113, "step_time_sec": 8.236897644994315, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.899975299835205, "lr": 0.0002, "elapsed_sec": 17348.75117111206, "step_time_sec": 8.237348385999212, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.814776420593262, "lr": 0.0002, "elapsed_sec": 17356.988587379456, "step_time_sec": 8.237194729008479, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.895139217376709, "lr": 0.0002, "elapsed_sec": 17365.22504258156, "step_time_sec": 8.23629993700888, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.876270771026611, "lr": 0.0002, "elapsed_sec": 17373.46253156662, "step_time_sec": 8.237379036989296, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.880076885223389, "lr": 0.0002, "elapsed_sec": 17381.69891524315, "step_time_sec": 8.236185643007047, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.9100728034973145, "lr": 0.0002, "elapsed_sec": 17389.936772584915, "step_time_sec": 8.237698262993945, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.814282417297363, "lr": 0.0002, "elapsed_sec": 17398.17423725128, "step_time_sec": 8.237304862996098, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.9488983154296875, "lr": 0.0002, "elapsed_sec": 17406.41279411316, "step_time_sec": 8.238444852991961, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.860019683837891, "lr": 0.0002, "elapsed_sec": 17414.650740861893, "step_time_sec": 8.237740839016624, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.843703269958496, "lr": 0.0002, "elapsed_sec": 17422.887296438217, "step_time_sec": 8.236386577016674, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.936622619628906, "lr": 0.0002, "elapsed_sec": 17431.12592124939, "step_time_sec": 8.238453110010596, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.935842990875244, "lr": 0.0002, "elapsed_sec": 17439.362448453903, "step_time_sec": 8.236383865005337, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.956331253051758, "lr": 0.0002, "elapsed_sec": 17447.598928451538, "step_time_sec": 8.236323478980921, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.886780738830566, "lr": 0.0002, "elapsed_sec": 17455.83549594879, "step_time_sec": 8.23639988800278, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.870883941650391, "lr": 0.0002, "elapsed_sec": 17464.07375741005, "step_time_sec": 8.238093470979948, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.839310169219971, "lr": 0.0002, "elapsed_sec": 17472.31144618988, "step_time_sec": 8.237557726009982, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.865331649780273, "lr": 0.0002, "elapsed_sec": 17480.54829144478, "step_time_sec": 8.236657506000483, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.023869514465332, "lr": 0.0002, "elapsed_sec": 17488.78643465042, "step_time_sec": 8.238089009013493, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.861941337585449, "lr": 0.0002, "elapsed_sec": 17497.023758411407, "step_time_sec": 8.237095891992794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.901723384857178, "lr": 0.0002, "elapsed_sec": 17505.26115822792, "step_time_sec": 8.23726561901276, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.8742570877075195, "lr": 0.0002, "elapsed_sec": 17513.4976875782, "step_time_sec": 8.236399849003647, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.966376304626465, "lr": 0.0002, "elapsed_sec": 17521.737217903137, "step_time_sec": 8.239291136007523, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.965059280395508, "lr": 0.0002, "elapsed_sec": 17529.9743540287, "step_time_sec": 8.236980626999866, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.898397445678711, "lr": 0.0002, "elapsed_sec": 17538.21203970909, "step_time_sec": 8.237523678981233, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.037445545196533, "lr": 0.0002, "elapsed_sec": 17546.449709177017, "step_time_sec": 8.237528589001158, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.8505682945251465, "lr": 0.0002, "elapsed_sec": 17554.685554742813, "step_time_sec": 8.235662700986722, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.964027404785156, "lr": 0.0002, "elapsed_sec": 17562.92296242714, "step_time_sec": 8.237237241002731, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.991113185882568, "lr": 0.0002, "elapsed_sec": 17571.160011053085, "step_time_sec": 8.236899142997572, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.049830436706543, "lr": 0.0002, "elapsed_sec": 17579.398910284042, "step_time_sec": 8.238748547009891, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.743207931518555, "lr": 0.0002, "elapsed_sec": 17587.63632965088, "step_time_sec": 8.237276226980612, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.928578853607178, "lr": 0.0002, "elapsed_sec": 17595.874156951904, "step_time_sec": 8.237638116988819, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.940062046051025, "lr": 0.0002, "elapsed_sec": 17604.112698554993, "step_time_sec": 8.238386927987449, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.962982177734375, "lr": 0.0002, "elapsed_sec": 17612.34845972061, "step_time_sec": 8.235595188016305, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.885808944702148, "lr": 0.0002, "elapsed_sec": 17620.584813594818, "step_time_sec": 8.236187976988731, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.884152412414551, "lr": 0.0002, "elapsed_sec": 17628.82208275795, "step_time_sec": 8.23716494400287, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.749636173248291, "lr": 0.0002, "elapsed_sec": 17637.059899806976, "step_time_sec": 8.237596589984605, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.891236782073975, "lr": 0.0002, "elapsed_sec": 17645.297009944916, "step_time_sec": 8.236977403983474, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.8675336837768555, "lr": 0.0002, "elapsed_sec": 17653.533594608307, "step_time_sec": 8.236426624003798, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.962010860443115, "lr": 0.0002, "elapsed_sec": 17661.77130794525, "step_time_sec": 8.237565769988578, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.995230197906494, "lr": 0.0002, "elapsed_sec": 17670.006962776184, "step_time_sec": 8.235501716990257, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.0866193771362305, "lr": 0.0002, "elapsed_sec": 17678.244525671005, "step_time_sec": 8.23740061101853, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.954995632171631, "lr": 0.0002, "elapsed_sec": 17686.48163342476, "step_time_sec": 8.236965639982373, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.9726338386535645, "lr": 0.0002, "elapsed_sec": 17694.7199716568, "step_time_sec": 8.238193672004854, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.995445728302002, "lr": 0.0002, "elapsed_sec": 17702.95708179474, "step_time_sec": 8.236944527016021, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.788989067077637, "lr": 0.0002, "elapsed_sec": 17711.19457435608, "step_time_sec": 8.237364866014104, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.857951641082764, "lr": 0.0002, "elapsed_sec": 17719.433766126633, "step_time_sec": 8.23902509600157, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.821610927581787, "lr": 0.0002, "elapsed_sec": 17727.670498371124, "step_time_sec": 8.236572474008426, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.914651393890381, "lr": 0.0002, "elapsed_sec": 17735.907785654068, "step_time_sec": 8.237164622027194, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.8138957023620605, "lr": 0.0002, "elapsed_sec": 17744.144774913788, "step_time_sec": 8.23678725300124, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.975130081176758, "lr": 0.0002, "elapsed_sec": 17752.382718086243, "step_time_sec": 8.237800064991461, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.813164710998535, "lr": 0.0002, "elapsed_sec": 17760.619901180267, "step_time_sec": 8.23702356099966, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.883731842041016, "lr": 0.0002, "elapsed_sec": 17768.856474161148, "step_time_sec": 8.236442358989734, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.979604244232178, "lr": 0.0002, "elapsed_sec": 17777.093991279602, "step_time_sec": 8.237322630011477, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.908714294433594, "lr": 0.0002, "elapsed_sec": 17785.329610824585, "step_time_sec": 8.235445911006536, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.880540370941162, "lr": 0.0002, "elapsed_sec": 17793.568523406982, "step_time_sec": 8.23877120402176, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.8465471267700195, "lr": 0.0002, "elapsed_sec": 17801.806138277054, "step_time_sec": 8.237427704007132, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.968297958374023, "lr": 0.0002, "elapsed_sec": 17810.04456448555, "step_time_sec": 8.238241113984259, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.979217052459717, "lr": 0.0002, "elapsed_sec": 17818.281057834625, "step_time_sec": 8.236343318014406, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.87858247756958, "lr": 0.0002, "elapsed_sec": 17826.51861834526, "step_time_sec": 8.23744358800468, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.764455795288086, "lr": 0.0002, "elapsed_sec": 17834.756377220154, "step_time_sec": 8.237595056998543, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.938248634338379, "lr": 0.0002, "elapsed_sec": 17842.992381811142, "step_time_sec": 8.235867530980613, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.941536903381348, "lr": 0.0002, "elapsed_sec": 17851.22960472107, "step_time_sec": 8.237003347981954, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.957389831542969, "lr": 0.0002, "elapsed_sec": 17859.465997219086, "step_time_sec": 8.236219735990744, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.824416160583496, "lr": 0.0002, "elapsed_sec": 17867.704349279404, "step_time_sec": 8.238277138007106, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.878081321716309, "lr": 0.0002, "elapsed_sec": 17875.941658735275, "step_time_sec": 8.237116759002674, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.869254112243652, "lr": 0.0002, "elapsed_sec": 17884.17972421646, "step_time_sec": 8.23791175999213, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.849368095397949, "lr": 0.0002, "elapsed_sec": 17892.417156934738, "step_time_sec": 8.237287052994361, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.787104606628418, "lr": 0.0002, "elapsed_sec": 17900.653171300888, "step_time_sec": 8.235810140002286, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.7905073165893555, "lr": 0.0002, "elapsed_sec": 17908.88972735405, "step_time_sec": 8.236418087995844, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.98231315612793, "lr": 0.0002, "elapsed_sec": 17917.12757086754, "step_time_sec": 8.237682608014438, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.691681861877441, "lr": 0.0002, "elapsed_sec": 17925.36661839485, "step_time_sec": 8.238846933003515, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.0128493309021, "lr": 0.0002, "elapsed_sec": 17933.60382604599, "step_time_sec": 8.23703874097555, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.783822059631348, "lr": 0.0002, "elapsed_sec": 17941.842436790466, "step_time_sec": 8.238471349002793, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.974362850189209, "lr": 0.0002, "elapsed_sec": 17950.081119537354, "step_time_sec": 8.238521159015363, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.914764881134033, "lr": 0.0002, "elapsed_sec": 17958.31631755829, "step_time_sec": 8.23500847700052, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.929507255554199, "lr": 0.0002, "elapsed_sec": 17966.55336165428, "step_time_sec": 8.236938837973867, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.005839824676514, "lr": 0.0002, "elapsed_sec": 17974.789794445038, "step_time_sec": 8.236241662991233, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.85650634765625, "lr": 0.0002, "elapsed_sec": 17983.026623010635, "step_time_sec": 8.236681629990926, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.77199125289917, "lr": 0.0002, "elapsed_sec": 17991.26473379135, "step_time_sec": 8.237939021986676, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.820580959320068, "lr": 0.0002, "elapsed_sec": 17999.502660512924, "step_time_sec": 8.23775490699336, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.849888801574707, "lr": 0.0002, "elapsed_sec": 18007.73982834816, "step_time_sec": 8.23704950598767, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.87212610244751, "lr": 0.0002, "elapsed_sec": 18015.976546287537, "step_time_sec": 8.236546867992729, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.83692741394043, "lr": 0.0002, "elapsed_sec": 18024.213044166565, "step_time_sec": 8.236384838994127, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.82995080947876, "lr": 0.0002, "elapsed_sec": 18032.4498295784, "step_time_sec": 8.236553882015869, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.822033882141113, "lr": 0.0002, "elapsed_sec": 18040.687429904938, "step_time_sec": 8.237453569017816, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.84463357925415, "lr": 0.0002, "elapsed_sec": 18048.925553798676, "step_time_sec": 8.238014290982392, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.885371208190918, "lr": 0.0002, "elapsed_sec": 18057.163380384445, "step_time_sec": 8.237665382999694, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.019178867340088, "lr": 0.0002, "elapsed_sec": 18065.400576353073, "step_time_sec": 8.236995660990942, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.894569396972656, "lr": 0.0002, "elapsed_sec": 18073.637414216995, "step_time_sec": 8.236690477991942, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.817649841308594, "lr": 0.0002, "elapsed_sec": 18081.875251054764, "step_time_sec": 8.237661184015451, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.939760208129883, "lr": 0.0002, "elapsed_sec": 18090.113322496414, "step_time_sec": 8.237925237015588, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.899044513702393, "lr": 0.0002, "elapsed_sec": 18098.35142493248, "step_time_sec": 8.237934817007044, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.892051696777344, "lr": 0.0002, "elapsed_sec": 18106.58827137947, "step_time_sec": 8.236713204998523, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.800619125366211, "lr": 0.0002, "elapsed_sec": 18114.826059818268, "step_time_sec": 8.237590217002435, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.904012203216553, "lr": 0.0002, "elapsed_sec": 18123.063755750656, "step_time_sec": 8.237537509994581, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.794198036193848, "lr": 0.0002, "elapsed_sec": 18131.300538778305, "step_time_sec": 8.23665501899086, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.809601306915283, "lr": 0.0002, "elapsed_sec": 18139.5381565094, "step_time_sec": 8.237423534010304, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.9251179695129395, "lr": 0.0002, "elapsed_sec": 18147.774596214294, "step_time_sec": 8.236289619002491, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.857892036437988, "lr": 0.0002, "elapsed_sec": 18156.012194395065, "step_time_sec": 8.237440060998779, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.926611423492432, "lr": 0.0002, "elapsed_sec": 18164.24888253212, "step_time_sec": 8.23655015899567, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.927740573883057, "lr": 0.0002, "elapsed_sec": 18172.486387252808, "step_time_sec": 8.237315653997939, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.067748069763184, "lr": 0.0002, "elapsed_sec": 18180.724178791046, "step_time_sec": 8.237638146005338, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.887503147125244, "lr": 0.0002, "elapsed_sec": 18188.95997619629, "step_time_sec": 8.235666986001888, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.804415225982666, "lr": 0.0002, "elapsed_sec": 18197.197169542313, "step_time_sec": 8.237010144017404, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.755861759185791, "lr": 0.0002, "elapsed_sec": 18205.43365073204, "step_time_sec": 8.236286007013405, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.895271301269531, "lr": 0.0002, "elapsed_sec": 18213.672261238098, "step_time_sec": 8.23842438601423, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.8690643310546875, "lr": 0.0002, "elapsed_sec": 18221.91100883484, "step_time_sec": 8.23858616600046, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.812520980834961, "lr": 0.0002, "elapsed_sec": 18230.149382829666, "step_time_sec": 8.238196569000138, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.83732271194458, "lr": 0.0002, "elapsed_sec": 18238.389017105103, "step_time_sec": 8.239479388983455, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.873443126678467, "lr": 0.0002, "elapsed_sec": 18246.625346660614, "step_time_sec": 8.236218281002948, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.844269275665283, "lr": 0.0002, "elapsed_sec": 18254.863380670547, "step_time_sec": 8.237840546004009, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.700755596160889, "lr": 0.0002, "elapsed_sec": 18263.10095667839, "step_time_sec": 8.237396840995643, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.82513952255249, "lr": 0.0002, "elapsed_sec": 18271.3388736248, "step_time_sec": 8.237749161984539, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.887399673461914, "lr": 0.0002, "elapsed_sec": 18279.57635974884, "step_time_sec": 8.23734010400949, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.86674690246582, "lr": 0.0002, "elapsed_sec": 18287.812946796417, "step_time_sec": 8.236417554988293, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.935785293579102, "lr": 0.0002, "elapsed_sec": 18296.05133342743, "step_time_sec": 8.238212398980977, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.92485237121582, "lr": 0.0002, "elapsed_sec": 18304.28826379776, "step_time_sec": 8.23679964299663, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.859181880950928, "lr": 0.0002, "elapsed_sec": 18312.52716445923, "step_time_sec": 8.238716857973486, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.959662914276123, "lr": 0.0002, "elapsed_sec": 18320.764696598053, "step_time_sec": 8.23736243500025, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.805685997009277, "lr": 0.0002, "elapsed_sec": 18329.00308728218, "step_time_sec": 8.238284991995897, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.822354316711426, "lr": 0.0002, "elapsed_sec": 18337.241212129593, "step_time_sec": 8.23795946998871, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.747927188873291, "lr": 0.0002, "elapsed_sec": 18345.47918701172, "step_time_sec": 8.23777563800104, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.798769950866699, "lr": 0.0002, "elapsed_sec": 18353.71909403801, "step_time_sec": 8.239734506991226, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.831726551055908, "lr": 0.0002, "elapsed_sec": 18361.956867933273, "step_time_sec": 8.237624867004342, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.869139194488525, "lr": 0.0002, "elapsed_sec": 18370.19609093666, "step_time_sec": 8.239057844009949, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.83256721496582, "lr": 0.0002, "elapsed_sec": 18378.43381381035, "step_time_sec": 8.237577892985428, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.78084659576416, "lr": 0.0002, "elapsed_sec": 18386.66956925392, "step_time_sec": 8.235591367003508, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.936152458190918, "lr": 0.0002, "elapsed_sec": 18394.907593011856, "step_time_sec": 8.237841769994702, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.762695789337158, "lr": 0.0002, "elapsed_sec": 18403.14358639717, "step_time_sec": 8.23584952901001, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.860875606536865, "lr": 0.0002, "elapsed_sec": 18411.38308405876, "step_time_sec": 8.239348520990461, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.906262397766113, "lr": 0.0002, "elapsed_sec": 18419.619757175446, "step_time_sec": 8.236515869008144, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.772244453430176, "lr": 0.0002, "elapsed_sec": 18427.85739850998, "step_time_sec": 8.237481805990683, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.747926235198975, "lr": 0.0002, "elapsed_sec": 18436.095207452774, "step_time_sec": 8.23762795899529, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.8444504737854, "lr": 0.0002, "elapsed_sec": 18444.33072566986, "step_time_sec": 8.235372299997834, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.843423843383789, "lr": 0.0002, "elapsed_sec": 18452.56813764572, "step_time_sec": 8.23721019999357, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.924961566925049, "lr": 0.0002, "elapsed_sec": 18460.80453658104, "step_time_sec": 8.236250310001196, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.792482852935791, "lr": 0.0002, "elapsed_sec": 18469.043452501297, "step_time_sec": 8.238807910995092, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.7810564041137695, "lr": 0.0002, "elapsed_sec": 18477.281103372574, "step_time_sec": 8.237431996996747, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.865940093994141, "lr": 0.0002, "elapsed_sec": 18485.518626213074, "step_time_sec": 8.237370821996592, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.850314140319824, "lr": 0.0002, "elapsed_sec": 18493.756586313248, "step_time_sec": 8.237824572017416, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.819536209106445, "lr": 0.0002, "elapsed_sec": 18501.99300646782, "step_time_sec": 8.236256408010377, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.875093936920166, "lr": 0.0002, "elapsed_sec": 18510.231241703033, "step_time_sec": 8.238048175000586, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.735835075378418, "lr": 0.0002, "elapsed_sec": 18518.467663526535, "step_time_sec": 8.236322872980963, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.898956298828125, "lr": 0.0002, "elapsed_sec": 18526.70569205284, "step_time_sec": 8.237797490990488, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.768278121948242, "lr": 0.0002, "elapsed_sec": 18534.942603111267, "step_time_sec": 8.236752497003181, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.840126037597656, "lr": 0.0002, "elapsed_sec": 18543.181168794632, "step_time_sec": 8.23842350000632, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.808236598968506, "lr": 0.0002, "elapsed_sec": 18551.41842198372, "step_time_sec": 8.237130989989964, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.811075687408447, "lr": 0.0002, "elapsed_sec": 18559.65476179123, "step_time_sec": 8.236179897008697, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.698203086853027, "lr": 0.0002, "elapsed_sec": 18567.89365887642, "step_time_sec": 8.238719227985712, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.821139812469482, "lr": 0.0002, "elapsed_sec": 18576.130820035934, "step_time_sec": 8.236942599003669, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.796663284301758, "lr": 0.0002, "elapsed_sec": 18584.370609760284, "step_time_sec": 8.239662531996146, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.813948154449463, "lr": 0.0002, "elapsed_sec": 18592.608622074127, "step_time_sec": 8.237827568984358, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.671389102935791, "lr": 0.0002, "elapsed_sec": 18600.847421884537, "step_time_sec": 8.238638564012945, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.837009429931641, "lr": 0.0002, "elapsed_sec": 18609.085086345673, "step_time_sec": 8.23755336998147, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.73391580581665, "lr": 0.0002, "elapsed_sec": 18617.322112321854, "step_time_sec": 8.236823121987982, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.690979957580566, "lr": 0.0002, "elapsed_sec": 18625.562014341354, "step_time_sec": 8.23975885802065, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.859065532684326, "lr": 0.0002, "elapsed_sec": 18633.799990177155, "step_time_sec": 8.237759912008187, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.905369281768799, "lr": 0.0002, "elapsed_sec": 18642.038397550583, "step_time_sec": 8.238263831008226, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.6854023933410645, "lr": 0.0002, "elapsed_sec": 18650.275719165802, "step_time_sec": 8.237164636026137, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.822281837463379, "lr": 0.0002, "elapsed_sec": 18658.512931346893, "step_time_sec": 8.237068318005186, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.944601058959961, "lr": 0.0002, "elapsed_sec": 18666.749747037888, "step_time_sec": 8.236695436004084, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.824206352233887, "lr": 0.0002, "elapsed_sec": 18674.986824274063, "step_time_sec": 8.236864209000487, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.87254524230957, "lr": 0.0002, "elapsed_sec": 18683.224027872086, "step_time_sec": 8.237048593990039, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.761673450469971, "lr": 0.0002, "elapsed_sec": 18691.461257457733, "step_time_sec": 8.237076943012653, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.759700775146484, "lr": 0.0002, "elapsed_sec": 18699.698306322098, "step_time_sec": 8.236916849011322, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.820101261138916, "lr": 0.0002, "elapsed_sec": 18707.935975551605, "step_time_sec": 8.237507501995424, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.837000370025635, "lr": 0.0002, "elapsed_sec": 18716.17267394066, "step_time_sec": 8.236531750997528, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.6594929695129395, "lr": 0.0002, "elapsed_sec": 18724.410669088364, "step_time_sec": 8.237847865995718, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.85169792175293, "lr": 0.0002, "elapsed_sec": 18732.64877986908, "step_time_sec": 8.237874546000967, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.837411880493164, "lr": 0.0002, "elapsed_sec": 18740.886979818344, "step_time_sec": 8.238052328000776, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.692224979400635, "lr": 0.0002, "elapsed_sec": 18749.12300467491, "step_time_sec": 8.235882232023869, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.833120346069336, "lr": 0.0002, "elapsed_sec": 18757.36011338234, "step_time_sec": 8.23694331900333, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.802032947540283, "lr": 0.0002, "elapsed_sec": 18765.598603487015, "step_time_sec": 8.23831475400948, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.803821563720703, "lr": 0.0002, "elapsed_sec": 18773.83501648903, "step_time_sec": 8.236232036986621, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.808932781219482, "lr": 0.0002, "elapsed_sec": 18782.074066877365, "step_time_sec": 8.23891734998324, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.649994373321533, "lr": 0.0002, "elapsed_sec": 18790.310083389282, "step_time_sec": 8.235836125008063, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.800133228302002, "lr": 0.0002, "elapsed_sec": 18798.54663515091, "step_time_sec": 8.236444885027595, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.777222156524658, "lr": 0.0002, "elapsed_sec": 18806.783309698105, "step_time_sec": 8.236480104998918, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.793380260467529, "lr": 0.0002, "elapsed_sec": 18815.02102613449, "step_time_sec": 8.237570159020834, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.7557220458984375, "lr": 0.0002, "elapsed_sec": 18823.259113788605, "step_time_sec": 8.237941108003724, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.762966632843018, "lr": 0.0002, "elapsed_sec": 18831.49543237686, "step_time_sec": 8.236121352994815, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.78695821762085, "lr": 0.0002, "elapsed_sec": 18839.732706785202, "step_time_sec": 8.237151737994282, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.95292329788208, "lr": 0.0002, "elapsed_sec": 18847.97057032585, "step_time_sec": 8.237717798008816, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.826848030090332, "lr": 0.0002, "elapsed_sec": 18856.20862174034, "step_time_sec": 8.237825130025158, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.73650598526001, "lr": 0.0002, "elapsed_sec": 18864.44502568245, "step_time_sec": 8.236256613978185, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.849381923675537, "lr": 0.0002, "elapsed_sec": 18872.68272280693, "step_time_sec": 8.237528347992338, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.82782506942749, "lr": 0.0002, "elapsed_sec": 18880.92159795761, "step_time_sec": 8.238721463014372, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.854355335235596, "lr": 0.0002, "elapsed_sec": 18889.157691001892, "step_time_sec": 8.23593743899255, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.755821228027344, "lr": 0.0002, "elapsed_sec": 18897.39737844467, "step_time_sec": 8.23952584998915, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.714648246765137, "lr": 0.0002, "elapsed_sec": 18905.634851694107, "step_time_sec": 8.237314252997749, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.6456217765808105, "lr": 0.0002, "elapsed_sec": 18913.87211751938, "step_time_sec": 8.237106974993367, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.748382568359375, "lr": 0.0002, "elapsed_sec": 18922.110637426376, "step_time_sec": 8.238375849003205, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.777681827545166, "lr": 0.0002, "elapsed_sec": 18930.347103834152, "step_time_sec": 8.236290852975799, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.782050132751465, "lr": 0.0002, "elapsed_sec": 18938.585783720016, "step_time_sec": 8.238557367003523, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.888965129852295, "lr": 0.0002, "elapsed_sec": 18946.822489261627, "step_time_sec": 8.236530044028768, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.7376227378845215, "lr": 0.0002, "elapsed_sec": 18955.060787200928, "step_time_sec": 8.238144407019718, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.82653284072876, "lr": 0.0002, "elapsed_sec": 18963.297380447388, "step_time_sec": 8.236425743991276, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.7420783042907715, "lr": 0.0002, "elapsed_sec": 18971.535790205002, "step_time_sec": 8.238276635995135, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.788546562194824, "lr": 0.0002, "elapsed_sec": 18979.773529291153, "step_time_sec": 8.23757382002077, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.784666061401367, "lr": 0.0002, "elapsed_sec": 18988.011266231537, "step_time_sec": 8.237552240985679, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.826764106750488, "lr": 0.0002, "elapsed_sec": 18996.249166727066, "step_time_sec": 8.237717707990669, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.753589153289795, "lr": 0.0002, "elapsed_sec": 19004.486067295074, "step_time_sec": 8.236707967997063, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.699190139770508, "lr": 0.0002, "elapsed_sec": 19012.72449374199, "step_time_sec": 8.238279937999323, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.800255298614502, "lr": 0.0002, "elapsed_sec": 19020.96361231804, "step_time_sec": 8.238942516007228, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.6666388511657715, "lr": 0.0002, "elapsed_sec": 19029.200279951096, "step_time_sec": 8.23659662398859, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.7406907081604, "lr": 0.0002, "elapsed_sec": 19037.439126729965, "step_time_sec": 8.238614816014888, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.853107929229736, "lr": 0.0002, "elapsed_sec": 19045.676959514618, "step_time_sec": 8.237705809006002, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.700039386749268, "lr": 0.0002, "elapsed_sec": 19053.916085481644, "step_time_sec": 8.238943415984977, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.874125003814697, "lr": 0.0002, "elapsed_sec": 19062.15374970436, "step_time_sec": 8.237527546007186, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.748518466949463, "lr": 0.0002, "elapsed_sec": 19070.39195561409, "step_time_sec": 8.238020190008683, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.818775653839111, "lr": 0.0002, "elapsed_sec": 19078.6306040287, "step_time_sec": 8.238506021007197, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.810321807861328, "lr": 0.0002, "elapsed_sec": 19086.866559028625, "step_time_sec": 8.235787399986293, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.763361930847168, "lr": 0.0002, "elapsed_sec": 19095.105077266693, "step_time_sec": 8.238377968984423, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.868476390838623, "lr": 0.0002, "elapsed_sec": 19103.34195303917, "step_time_sec": 8.236678921006387, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.745630741119385, "lr": 0.0002, "elapsed_sec": 19111.579442501068, "step_time_sec": 8.237375752010848, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.731749057769775, "lr": 0.0002, "elapsed_sec": 19119.818650722504, "step_time_sec": 8.239045501017245, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.771446704864502, "lr": 0.0002, "elapsed_sec": 19128.056527137756, "step_time_sec": 8.237696112017147, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.732754707336426, "lr": 0.0002, "elapsed_sec": 19136.295098781586, "step_time_sec": 8.238417015003506, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.912272930145264, "lr": 0.0002, "elapsed_sec": 19144.530974149704, "step_time_sec": 8.235710407985607, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.744431018829346, "lr": 0.0002, "elapsed_sec": 19152.76903104782, "step_time_sec": 8.2378372180101, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.765602111816406, "lr": 0.0002, "elapsed_sec": 19161.006860733032, "step_time_sec": 8.237676093995105, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.749931335449219, "lr": 0.0002, "elapsed_sec": 19169.243893623352, "step_time_sec": 8.236927658988861, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.827975749969482, "lr": 0.0002, "elapsed_sec": 19177.48220682144, "step_time_sec": 8.238099737995071, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.679810523986816, "lr": 0.0002, "elapsed_sec": 19185.718473911285, "step_time_sec": 8.236104602983687, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.895175933837891, "lr": 0.0002, "elapsed_sec": 19193.955278396606, "step_time_sec": 8.236661480012117, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.710312366485596, "lr": 0.0002, "elapsed_sec": 19202.19276690483, "step_time_sec": 8.237290085991845, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.670753002166748, "lr": 0.0002, "elapsed_sec": 19210.4320602417, "step_time_sec": 8.23914234698168, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.797358512878418, "lr": 0.0002, "elapsed_sec": 19218.67038655281, "step_time_sec": 8.23820757001522, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.735703468322754, "lr": 0.0002, "elapsed_sec": 19226.90917778015, "step_time_sec": 8.238564186001895, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.729925155639648, "lr": 0.0002, "elapsed_sec": 19235.147212982178, "step_time_sec": 8.237893133016769, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.649356842041016, "lr": 0.0002, "elapsed_sec": 19243.383618593216, "step_time_sec": 8.236196335987188, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.8103251457214355, "lr": 0.0002, "elapsed_sec": 19251.61951971054, "step_time_sec": 8.235751606000122, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.711422920227051, "lr": 0.0002, "elapsed_sec": 19259.858055353165, "step_time_sec": 8.238419237022754, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.834702014923096, "lr": 0.0002, "elapsed_sec": 19268.095902204514, "step_time_sec": 8.237683662009658, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.849392890930176, "lr": 0.0002, "elapsed_sec": 19276.332451820374, "step_time_sec": 8.236384503019508, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.800748348236084, "lr": 0.0002, "elapsed_sec": 19284.570248365402, "step_time_sec": 8.237597809988074, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.85848331451416, "lr": 0.0002, "elapsed_sec": 19292.808809518814, "step_time_sec": 8.23838767700363, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.743308067321777, "lr": 0.0002, "elapsed_sec": 19301.044832229614, "step_time_sec": 8.235910382005386, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.813770771026611, "lr": 0.0002, "elapsed_sec": 19309.282936811447, "step_time_sec": 8.237902084016241, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.73013162612915, "lr": 0.0002, "elapsed_sec": 19317.521248102188, "step_time_sec": 8.238161278015468, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.693339824676514, "lr": 0.0002, "elapsed_sec": 19325.75895857811, "step_time_sec": 8.237528072000714, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.6929168701171875, "lr": 0.0002, "elapsed_sec": 19333.99626994133, "step_time_sec": 8.23713795200456, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.674023628234863, "lr": 0.0002, "elapsed_sec": 19342.234360933304, "step_time_sec": 8.237956094002584, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.781546115875244, "lr": 0.0002, "elapsed_sec": 19350.472784519196, "step_time_sec": 8.238270538015058, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.779456615447998, "lr": 0.0002, "elapsed_sec": 19358.70933032036, "step_time_sec": 8.23640342798899, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.827245712280273, "lr": 0.0002, "elapsed_sec": 19366.94786310196, "step_time_sec": 8.238303642021492, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.611898422241211, "lr": 0.0002, "elapsed_sec": 19375.184468269348, "step_time_sec": 8.236492930998793, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.781529903411865, "lr": 0.0002, "elapsed_sec": 19383.42279458046, "step_time_sec": 8.23811893898528, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.696959972381592, "lr": 0.0002, "elapsed_sec": 19391.660876989365, "step_time_sec": 8.237942863022909, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.8536529541015625, "lr": 0.0002, "elapsed_sec": 19399.897834539413, "step_time_sec": 8.236802244005958, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.556898593902588, "lr": 0.0002, "elapsed_sec": 19408.136251211166, "step_time_sec": 8.238265363994287, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.7660675048828125, "lr": 0.0002, "elapsed_sec": 19416.373700618744, "step_time_sec": 8.237263225018978, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.840282917022705, "lr": 0.0002, "elapsed_sec": 19424.612272024155, "step_time_sec": 8.238412853010232, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.747722625732422, "lr": 0.0002, "elapsed_sec": 19432.851185798645, "step_time_sec": 8.238747954979772, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.674228191375732, "lr": 0.0002, "elapsed_sec": 19441.087645292282, "step_time_sec": 8.236280349985464, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.760984897613525, "lr": 0.0002, "elapsed_sec": 19449.324376821518, "step_time_sec": 8.236566687992308, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.965673446655273, "lr": 0.0002, "elapsed_sec": 19457.56179380417, "step_time_sec": 8.237234169006115, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.720442295074463, "lr": 0.0002, "elapsed_sec": 19465.799832582474, "step_time_sec": 8.23788953901385, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.715677738189697, "lr": 0.0002, "elapsed_sec": 19474.03752374649, "step_time_sec": 8.237534582003718, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.773320198059082, "lr": 0.0002, "elapsed_sec": 19482.275869607925, "step_time_sec": 8.238192011020146, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.851242542266846, "lr": 0.0002, "elapsed_sec": 19490.514234542847, "step_time_sec": 8.23820949200308, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.955886363983154, "lr": 0.0002, "elapsed_sec": 19498.75042462349, "step_time_sec": 8.236104936018819, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.731644630432129, "lr": 0.0002, "elapsed_sec": 19506.98799443245, "step_time_sec": 8.237318065017462, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.6566033363342285, "lr": 0.0002, "elapsed_sec": 19515.224655866623, "step_time_sec": 8.236558800010243, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.639529705047607, "lr": 0.0002, "elapsed_sec": 19523.462250947952, "step_time_sec": 8.23737323799287, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.732475757598877, "lr": 0.0002, "elapsed_sec": 19531.698207855225, "step_time_sec": 8.235847661999287, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.732004165649414, "lr": 0.0002, "elapsed_sec": 19539.936893701553, "step_time_sec": 8.23842039698502, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.725571632385254, "lr": 0.0002, "elapsed_sec": 19548.17420887947, "step_time_sec": 8.237197431008099, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.728036880493164, "lr": 0.0002, "elapsed_sec": 19556.410679340363, "step_time_sec": 8.236337769019883, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.6975555419921875, "lr": 0.0002, "elapsed_sec": 19564.64822745323, "step_time_sec": 8.2373503120034, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.7225565910339355, "lr": 0.0002, "elapsed_sec": 19572.884482860565, "step_time_sec": 8.236035326990532, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.680305480957031, "lr": 0.0002, "elapsed_sec": 19581.122620344162, "step_time_sec": 8.237933812983101, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.776636600494385, "lr": 0.0002, "elapsed_sec": 19589.360647678375, "step_time_sec": 8.237901357992087, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.762897968292236, "lr": 0.0002, "elapsed_sec": 19597.59950041771, "step_time_sec": 8.238720209017629, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.757110595703125, "lr": 0.0002, "elapsed_sec": 19605.838180541992, "step_time_sec": 8.238469711999642, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.715514183044434, "lr": 0.0002, "elapsed_sec": 19614.07459616661, "step_time_sec": 8.236299969023094, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.668313503265381, "lr": 0.0002, "elapsed_sec": 19622.31414580345, "step_time_sec": 8.23934498699964, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.6941118240356445, "lr": 0.0002, "elapsed_sec": 19630.550542116165, "step_time_sec": 8.236270476016216, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.821915149688721, "lr": 0.0002, "elapsed_sec": 19638.789384126663, "step_time_sec": 8.238674903986976, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.645662784576416, "lr": 0.0002, "elapsed_sec": 19647.026454925537, "step_time_sec": 8.236865441023838, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.784804344177246, "lr": 0.0002, "elapsed_sec": 19655.26378583908, "step_time_sec": 8.237184892001096, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.654390811920166, "lr": 0.0002, "elapsed_sec": 19663.500550746918, "step_time_sec": 8.236607348982943, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.930296897888184, "lr": 0.0002, "elapsed_sec": 19671.736666202545, "step_time_sec": 8.235929433983983, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.763442516326904, "lr": 0.0002, "elapsed_sec": 19679.97330236435, "step_time_sec": 8.236507405992597, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.600585460662842, "lr": 0.0002, "elapsed_sec": 19688.20854496956, "step_time_sec": 8.235100491991034, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.68079137802124, "lr": 0.0002, "elapsed_sec": 19696.44771885872, "step_time_sec": 8.238979218003806, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.699245452880859, "lr": 0.0002, "elapsed_sec": 19704.684639930725, "step_time_sec": 8.236760723986663, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.809914588928223, "lr": 0.0002, "elapsed_sec": 19712.92195725441, "step_time_sec": 8.23712717401213, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.628736972808838, "lr": 0.0002, "elapsed_sec": 19721.160076618195, "step_time_sec": 8.237955290998798, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.825097560882568, "lr": 0.0002, "elapsed_sec": 19729.39709353447, "step_time_sec": 8.236852267989889, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.71022891998291, "lr": 0.0002, "elapsed_sec": 19737.636216402054, "step_time_sec": 8.238945842022076, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.589655876159668, "lr": 0.0002, "elapsed_sec": 19745.87360882759, "step_time_sec": 8.237251357000787, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.737056255340576, "lr": 0.0002, "elapsed_sec": 19754.111898183823, "step_time_sec": 8.238117725995835, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.791776180267334, "lr": 0.0002, "elapsed_sec": 19762.34863758087, "step_time_sec": 8.236604347010143, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.680016994476318, "lr": 0.0002, "elapsed_sec": 19770.586584568024, "step_time_sec": 8.237812097009737, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.7286248207092285, "lr": 0.0002, "elapsed_sec": 19778.8245947361, "step_time_sec": 8.237871089018881, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.8078107833862305, "lr": 0.0002, "elapsed_sec": 19787.062297582626, "step_time_sec": 8.237490395986242, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.90829610824585, "lr": 0.0002, "elapsed_sec": 19795.300063848495, "step_time_sec": 8.23762314900523, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.814359188079834, "lr": 0.0002, "elapsed_sec": 19803.53679537773, "step_time_sec": 8.23654883998097, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.81473445892334, "lr": 0.0002, "elapsed_sec": 19811.775212049484, "step_time_sec": 8.238269394991221, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.623017311096191, "lr": 0.0002, "elapsed_sec": 19820.01465988159, "step_time_sec": 8.23928140400676, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.762202739715576, "lr": 0.0002, "elapsed_sec": 19828.252566576004, "step_time_sec": 8.237746711994987, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.726574420928955, "lr": 0.0002, "elapsed_sec": 19836.49088692665, "step_time_sec": 8.238187493989244, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.732025146484375, "lr": 0.0002, "elapsed_sec": 19844.728523492813, "step_time_sec": 8.237459751020651, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.727934837341309, "lr": 0.0002, "elapsed_sec": 19852.966715812683, "step_time_sec": 8.238011667999672, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.699817657470703, "lr": 0.0002, "elapsed_sec": 19861.20466375351, "step_time_sec": 8.237813694024226, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.780637264251709, "lr": 0.0002, "elapsed_sec": 19869.442326784134, "step_time_sec": 8.23746438199305, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.754037380218506, "lr": 0.0002, "elapsed_sec": 19877.680520772934, "step_time_sec": 8.238031692017103, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.598733425140381, "lr": 0.0002, "elapsed_sec": 19885.916445732117, "step_time_sec": 8.235767994017806, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.772981643676758, "lr": 0.0002, "elapsed_sec": 19894.15383529663, "step_time_sec": 8.237188808008796, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.779630184173584, "lr": 0.0002, "elapsed_sec": 19902.39191031456, "step_time_sec": 8.23791686099139, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.7976460456848145, "lr": 0.0002, "elapsed_sec": 19910.630100011826, "step_time_sec": 8.238073802000144, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.769190311431885, "lr": 0.0002, "elapsed_sec": 19918.867841005325, "step_time_sec": 8.237562611990143, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.60880184173584, "lr": 0.0002, "elapsed_sec": 19927.10488653183, "step_time_sec": 8.236828549997881, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.71339225769043, "lr": 0.0002, "elapsed_sec": 19935.343921899796, "step_time_sec": 8.23891839600401, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.726098537445068, "lr": 0.0002, "elapsed_sec": 19943.580721139908, "step_time_sec": 8.236598502990091, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.713596343994141, "lr": 0.0002, "elapsed_sec": 19951.819590568542, "step_time_sec": 8.238754566991702, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.735457897186279, "lr": 0.0002, "elapsed_sec": 19960.057120084763, "step_time_sec": 8.237319884006865, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.79528284072876, "lr": 0.0002, "elapsed_sec": 19968.295023679733, "step_time_sec": 8.237742295023054, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.762667655944824, "lr": 0.0002, "elapsed_sec": 19976.533984422684, "step_time_sec": 8.238833192008315, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.7700581550598145, "lr": 0.0002, "elapsed_sec": 19984.772369861603, "step_time_sec": 8.238240086997394, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.654966831207275, "lr": 0.0002, "elapsed_sec": 19993.012177228928, "step_time_sec": 8.239592683006776, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.704759120941162, "lr": 0.0002, "elapsed_sec": 20001.248528957367, "step_time_sec": 8.23617580300197, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.726579189300537, "lr": 0.0002, "elapsed_sec": 20009.486835718155, "step_time_sec": 8.238155661994824, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.649464130401611, "lr": 0.0002, "elapsed_sec": 20017.724745750427, "step_time_sec": 8.23778737799148, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.715010643005371, "lr": 0.0002, "elapsed_sec": 20025.961777448654, "step_time_sec": 8.23686419898877, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.425398349761963, "lr": 0.0002, "elapsed_sec": 20034.19980120659, "step_time_sec": 8.237883341993438, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.679272651672363, "lr": 0.0002, "elapsed_sec": 20042.43772125244, "step_time_sec": 8.237709356995765, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.831843852996826, "lr": 0.0002, "elapsed_sec": 20050.676508903503, "step_time_sec": 8.238668597012293, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.701044082641602, "lr": 0.0002, "elapsed_sec": 20058.91379594803, "step_time_sec": 8.237134360999335, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.66092586517334, "lr": 0.0002, "elapsed_sec": 20067.15236377716, "step_time_sec": 8.238369864993729, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.707097053527832, "lr": 0.0002, "elapsed_sec": 20075.38943386078, "step_time_sec": 8.23689907998778, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.746358871459961, "lr": 0.0002, "elapsed_sec": 20083.626540899277, "step_time_sec": 8.236964151001303, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.696541786193848, "lr": 0.0002, "elapsed_sec": 20091.865224838257, "step_time_sec": 8.23852234298829, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.727778911590576, "lr": 0.0002, "elapsed_sec": 20100.102915763855, "step_time_sec": 8.237519821996102, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.695029258728027, "lr": 0.0002, "elapsed_sec": 20108.340529203415, "step_time_sec": 8.237414237984922, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.770766735076904, "lr": 0.0002, "elapsed_sec": 20116.57867860794, "step_time_sec": 8.237978529010434, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.729476451873779, "lr": 0.0002, "elapsed_sec": 20124.816304445267, "step_time_sec": 8.237463986006333, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.741345405578613, "lr": 0.0002, "elapsed_sec": 20133.054896831512, "step_time_sec": 8.238450697012013, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.701321125030518, "lr": 0.0002, "elapsed_sec": 20141.29189825058, "step_time_sec": 8.236855458992068, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.5887346267700195, "lr": 0.0002, "elapsed_sec": 20149.531465530396, "step_time_sec": 8.239340954984073, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.7488837242126465, "lr": 0.0002, "elapsed_sec": 20157.769805431366, "step_time_sec": 8.238201642991044, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.754698276519775, "lr": 0.0002, "elapsed_sec": 20166.00693464279, "step_time_sec": 8.236997365980642, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.677996635437012, "lr": 0.0002, "elapsed_sec": 20174.24513053894, "step_time_sec": 8.237972456001444, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.705740928649902, "lr": 0.0002, "elapsed_sec": 20182.481761455536, "step_time_sec": 8.236455839010887, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.53590202331543, "lr": 0.0002, "elapsed_sec": 20190.71914958954, "step_time_sec": 8.237262300011935, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.864216327667236, "lr": 0.0002, "elapsed_sec": 20198.95710349083, "step_time_sec": 8.237790743005462, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.7768025398254395, "lr": 0.0002, "elapsed_sec": 20220.605512857437, "step_time_sec": 21.64821676298743, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.727342128753662, "lr": 0.0002, "elapsed_sec": 20228.83700323105, "step_time_sec": 8.231336248019943, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.739968776702881, "lr": 0.0002, "elapsed_sec": 20237.06845831871, "step_time_sec": 8.231261712993728, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.6468825340271, "lr": 0.0002, "elapsed_sec": 20245.30561041832, "step_time_sec": 8.236980019981274, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.8349432945251465, "lr": 0.0002, "elapsed_sec": 20253.54176735878, "step_time_sec": 8.236009137006477, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.781454086303711, "lr": 0.0002, "elapsed_sec": 20261.780022621155, "step_time_sec": 8.238074869994307, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.855122089385986, "lr": 0.0002, "elapsed_sec": 20270.017567634583, "step_time_sec": 8.237370120012201, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.748664379119873, "lr": 0.0002, "elapsed_sec": 20278.255811214447, "step_time_sec": 8.238096757006133, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.787644386291504, "lr": 0.0002, "elapsed_sec": 20286.494302034378, "step_time_sec": 8.238379431015346, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.688591003417969, "lr": 0.0002, "elapsed_sec": 20294.731332063675, "step_time_sec": 8.236830770008964, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.666113376617432, "lr": 0.0002, "elapsed_sec": 20302.969805002213, "step_time_sec": 8.238317164010368, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.630930423736572, "lr": 0.0002, "elapsed_sec": 20311.208060741425, "step_time_sec": 8.238057640992338, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.696735858917236, "lr": 0.0002, "elapsed_sec": 20319.4454267025, "step_time_sec": 8.237219586997526, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.708733558654785, "lr": 0.0002, "elapsed_sec": 20327.684350013733, "step_time_sec": 8.238742265995825, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.692295074462891, "lr": 0.0002, "elapsed_sec": 20335.92187142372, "step_time_sec": 8.237408258981304, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.732201099395752, "lr": 0.0002, "elapsed_sec": 20344.161147117615, "step_time_sec": 8.239079001010396, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.812173843383789, "lr": 0.0002, "elapsed_sec": 20352.39858031273, "step_time_sec": 8.237245399010135, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.686822414398193, "lr": 0.0002, "elapsed_sec": 20360.636865615845, "step_time_sec": 8.238137291016756, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.8750104904174805, "lr": 0.0002, "elapsed_sec": 20368.875492095947, "step_time_sec": 8.238511033006944, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.694077491760254, "lr": 0.0002, "elapsed_sec": 20377.11102604866, "step_time_sec": 8.235340437997365, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.703069686889648, "lr": 0.0002, "elapsed_sec": 20385.349724531174, "step_time_sec": 8.238545772997895, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.740596294403076, "lr": 0.0002, "elapsed_sec": 20393.586144924164, "step_time_sec": 8.236241428996436, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.706242084503174, "lr": 0.0002, "elapsed_sec": 20401.824833393097, "step_time_sec": 8.238510775001487, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.730598449707031, "lr": 0.0002, "elapsed_sec": 20410.062956809998, "step_time_sec": 8.23797709000064, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.696741104125977, "lr": 0.0002, "elapsed_sec": 20418.300113916397, "step_time_sec": 8.23698225201224, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.805856227874756, "lr": 0.0002, "elapsed_sec": 20426.537680387497, "step_time_sec": 8.23742301799939, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.7278642654418945, "lr": 0.0002, "elapsed_sec": 20434.77523946762, "step_time_sec": 8.23736264501349, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.686431884765625, "lr": 0.0002, "elapsed_sec": 20443.014253616333, "step_time_sec": 8.23890754400054, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.595880031585693, "lr": 0.0002, "elapsed_sec": 20451.25178551674, "step_time_sec": 8.23734234899166, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.790724754333496, "lr": 0.0002, "elapsed_sec": 20459.49010491371, "step_time_sec": 8.238175975013291, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.6887030601501465, "lr": 0.0002, "elapsed_sec": 20467.728267908096, "step_time_sec": 8.23799010599032, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.708544731140137, "lr": 0.0002, "elapsed_sec": 20475.965925216675, "step_time_sec": 8.237504719989374, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.63438081741333, "lr": 0.0002, "elapsed_sec": 20484.204621076584, "step_time_sec": 8.238519942999119, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.777081489562988, "lr": 0.0002, "elapsed_sec": 20492.442600250244, "step_time_sec": 8.237799323018407, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.818986892700195, "lr": 0.0002, "elapsed_sec": 20500.680920362473, "step_time_sec": 8.23815822697361, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.836279392242432, "lr": 0.0002, "elapsed_sec": 20508.919180631638, "step_time_sec": 8.23813776098541, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.6648054122924805, "lr": 0.0002, "elapsed_sec": 20517.1572701931, "step_time_sec": 8.237890919001075, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.813978672027588, "lr": 0.0002, "elapsed_sec": 20525.39591026306, "step_time_sec": 8.23845274199266, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.6755146980285645, "lr": 0.0002, "elapsed_sec": 20533.63234257698, "step_time_sec": 8.236289546999615, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.64857292175293, "lr": 0.0002, "elapsed_sec": 20541.870722055435, "step_time_sec": 8.238200144987786, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.790399074554443, "lr": 0.0002, "elapsed_sec": 20550.10890364647, "step_time_sec": 8.238002490979852, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.749671936035156, "lr": 0.0002, "elapsed_sec": 20558.346683740616, "step_time_sec": 8.237626441987231, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.667248249053955, "lr": 0.0002, "elapsed_sec": 20566.584176540375, "step_time_sec": 8.237302656983957, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.666927337646484, "lr": 0.0002, "elapsed_sec": 20574.821393728256, "step_time_sec": 8.23703686898807, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.672236442565918, "lr": 0.0002, "elapsed_sec": 20583.0605905056, "step_time_sec": 8.239065840025432, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.915714263916016, "lr": 0.0002, "elapsed_sec": 20591.29806160927, "step_time_sec": 8.237292353005614, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.682288646697998, "lr": 0.0002, "elapsed_sec": 20599.53648877144, "step_time_sec": 8.238245473010466, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.680552959442139, "lr": 0.0002, "elapsed_sec": 20607.77492237091, "step_time_sec": 8.238275465002516, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.708984851837158, "lr": 0.0002, "elapsed_sec": 20616.012004375458, "step_time_sec": 8.236952190985903, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.678790092468262, "lr": 0.0002, "elapsed_sec": 20624.25110888481, "step_time_sec": 8.23893084100564, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.712210178375244, "lr": 0.0002, "elapsed_sec": 20632.488131523132, "step_time_sec": 8.236838864977472, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.65994930267334, "lr": 0.0002, "elapsed_sec": 20640.728746414185, "step_time_sec": 8.240441892005038, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.793960094451904, "lr": 0.0002, "elapsed_sec": 20648.966301202774, "step_time_sec": 8.237436625990085, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.705924034118652, "lr": 0.0002, "elapsed_sec": 20657.204268693924, "step_time_sec": 8.237796613015234, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.679427623748779, "lr": 0.0002, "elapsed_sec": 20665.44220638275, "step_time_sec": 8.237765092984773, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.654810905456543, "lr": 0.0002, "elapsed_sec": 20673.67925786972, "step_time_sec": 8.236901491007302, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.686132907867432, "lr": 0.0002, "elapsed_sec": 20681.918623209, "step_time_sec": 8.239158774988027, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.608835220336914, "lr": 0.0002, "elapsed_sec": 20690.221245765686, "step_time_sec": 8.244367175007937, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.783135414123535, "lr": 0.0002, "elapsed_sec": 20698.45873427391, "step_time_sec": 8.23725938599091, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.675750732421875, "lr": 0.0002, "elapsed_sec": 20706.69705605507, "step_time_sec": 8.238199054991128, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.771296977996826, "lr": 0.0002, "elapsed_sec": 20714.935692548752, "step_time_sec": 8.238427419011714, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.6276350021362305, "lr": 0.0002, "elapsed_sec": 20723.172852754593, "step_time_sec": 8.23701581399655, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.719144344329834, "lr": 0.0002, "elapsed_sec": 20731.41159439087, "step_time_sec": 8.238593818998197, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.655930042266846, "lr": 0.0002, "elapsed_sec": 20739.647317171097, "step_time_sec": 8.235541653004475, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.88724946975708, "lr": 0.0002, "elapsed_sec": 20747.885095596313, "step_time_sec": 8.237604183988879, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.715617656707764, "lr": 0.0002, "elapsed_sec": 20756.12232899666, "step_time_sec": 8.23709399800282, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.682052135467529, "lr": 0.0002, "elapsed_sec": 20764.36016511917, "step_time_sec": 30.6936377150123, "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": 4.779757499694824, "lr": 0.0002, "elapsed_sec": 20795.074093341827, "step_time_sec": 8.257377772999462, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.598711967468262, "lr": 0.0002, "elapsed_sec": 20803.305492401123, "step_time_sec": 8.231279122002888, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.808159828186035, "lr": 0.0002, "elapsed_sec": 20811.536378622055, "step_time_sec": 8.230696298007388, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.693366050720215, "lr": 0.0002, "elapsed_sec": 20819.77427339554, "step_time_sec": 8.237706652987981, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.7263078689575195, "lr": 0.0002, "elapsed_sec": 20828.009417533875, "step_time_sec": 8.235019726009341, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.715254783630371, "lr": 0.0002, "elapsed_sec": 20836.247584581375, "step_time_sec": 8.238017208990641, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.6466875076293945, "lr": 0.0002, "elapsed_sec": 20844.485851049423, "step_time_sec": 8.23805684200488, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.70835542678833, "lr": 0.0002, "elapsed_sec": 20852.723741531372, "step_time_sec": 8.237733777001267, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.691956520080566, "lr": 0.0002, "elapsed_sec": 20860.96172475815, "step_time_sec": 8.237824915995589, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.537163734436035, "lr": 0.0002, "elapsed_sec": 20869.19872689247, "step_time_sec": 8.236816944991006, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.694226264953613, "lr": 0.0002, "elapsed_sec": 20877.436913728714, "step_time_sec": 8.238013880007202, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.632923603057861, "lr": 0.0002, "elapsed_sec": 20885.673924684525, "step_time_sec": 8.236874438007362, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.735004425048828, "lr": 0.0002, "elapsed_sec": 20893.91154026985, "step_time_sec": 8.237460748001467, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.843364715576172, "lr": 0.0002, "elapsed_sec": 20902.15108895302, "step_time_sec": 8.239350982999895, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.741669654846191, "lr": 0.0002, "elapsed_sec": 20910.3877825737, "step_time_sec": 8.23655061199679, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.706450462341309, "lr": 0.0002, "elapsed_sec": 20918.625076770782, "step_time_sec": 8.237136567011476, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.806919574737549, "lr": 0.0002, "elapsed_sec": 20926.862495422363, "step_time_sec": 8.237261020985898, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.6774468421936035, "lr": 0.0002, "elapsed_sec": 20935.101647615433, "step_time_sec": 8.238966561009875, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.598939418792725, "lr": 0.0002, "elapsed_sec": 20943.339714050293, "step_time_sec": 8.237921735999407, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.6106181144714355, "lr": 0.0002, "elapsed_sec": 20951.576154708862, "step_time_sec": 8.236290038999869, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.702090740203857, "lr": 0.0002, "elapsed_sec": 20959.81498169899, "step_time_sec": 8.238604169979226, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.653429985046387, "lr": 0.0002, "elapsed_sec": 20968.050827264786, "step_time_sec": 8.23570669000037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.646914005279541, "lr": 0.0002, "elapsed_sec": 20976.289163827896, "step_time_sec": 8.238172874000156, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.684702396392822, "lr": 0.0002, "elapsed_sec": 20984.526340007782, "step_time_sec": 8.23697275199811, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.745511531829834, "lr": 0.0002, "elapsed_sec": 20992.763841867447, "step_time_sec": 8.23733130400069, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.731769561767578, "lr": 0.0002, "elapsed_sec": 21001.00314307213, "step_time_sec": 8.239120999001898, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.677311420440674, "lr": 0.0002, "elapsed_sec": 21009.239270210266, "step_time_sec": 8.235957318014698, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.8200201988220215, "lr": 0.0002, "elapsed_sec": 21017.475320339203, "step_time_sec": 8.235898042010376, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.631782054901123, "lr": 0.0002, "elapsed_sec": 21025.71163702011, "step_time_sec": 8.236167555005522, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.697166919708252, "lr": 0.0002, "elapsed_sec": 21033.949246883392, "step_time_sec": 8.237413737981115, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.787389755249023, "lr": 0.0002, "elapsed_sec": 21042.18637943268, "step_time_sec": 8.236988491989905, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.655288219451904, "lr": 0.0002, "elapsed_sec": 21050.425416707993, "step_time_sec": 8.238910323998425, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.603604793548584, "lr": 0.0002, "elapsed_sec": 21058.663271427155, "step_time_sec": 8.237657443998614, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.767101764678955, "lr": 0.0002, "elapsed_sec": 21066.900780439377, "step_time_sec": 8.237373726995429, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.679776191711426, "lr": 0.0002, "elapsed_sec": 21075.13979101181, "step_time_sec": 8.238784774002852, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.677847862243652, "lr": 0.0002, "elapsed_sec": 21083.376608610153, "step_time_sec": 8.23667029698845, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.518040657043457, "lr": 0.0002, "elapsed_sec": 21091.616087198257, "step_time_sec": 8.239360872976249, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.652716159820557, "lr": 0.0002, "elapsed_sec": 21099.853484392166, "step_time_sec": 8.23720181902172, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.826125144958496, "lr": 0.0002, "elapsed_sec": 21108.091369867325, "step_time_sec": 8.237741695018485, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.698060512542725, "lr": 0.0002, "elapsed_sec": 21116.329649686813, "step_time_sec": 8.238121654023416, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.60862398147583, "lr": 0.0002, "elapsed_sec": 21124.567600488663, "step_time_sec": 8.237779324990697, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.638875484466553, "lr": 0.0002, "elapsed_sec": 21132.806269407272, "step_time_sec": 8.238473435019841, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.716042518615723, "lr": 0.0002, "elapsed_sec": 21141.043728113174, "step_time_sec": 8.237268389988458, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.686909198760986, "lr": 0.0002, "elapsed_sec": 21149.2812128067, "step_time_sec": 8.23732159199426, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.611596584320068, "lr": 0.0002, "elapsed_sec": 21157.518701314926, "step_time_sec": 8.237345984991407, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.545157432556152, "lr": 0.0002, "elapsed_sec": 21165.75647854805, "step_time_sec": 8.237598709994927, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.6973395347595215, "lr": 0.0002, "elapsed_sec": 21173.99462723732, "step_time_sec": 8.238019595999504, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.729614734649658, "lr": 0.0002, "elapsed_sec": 21182.232018232346, "step_time_sec": 8.237202240008628, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.708763122558594, "lr": 0.0002, "elapsed_sec": 21190.47034883499, "step_time_sec": 8.238142246991629, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.650079727172852, "lr": 0.0002, "elapsed_sec": 21198.708709955215, "step_time_sec": 8.238253760995576, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.723671913146973, "lr": 0.0002, "elapsed_sec": 21206.946201086044, "step_time_sec": 8.237306761002401, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.672360420227051, "lr": 0.0002, "elapsed_sec": 21215.184207201004, "step_time_sec": 8.237815194006544, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.669442176818848, "lr": 0.0002, "elapsed_sec": 21223.421373605728, "step_time_sec": 8.23701362599968, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.720096588134766, "lr": 0.0002, "elapsed_sec": 21231.660519599915, "step_time_sec": 8.239018000982469, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.625270843505859, "lr": 0.0002, "elapsed_sec": 21239.89843225479, "step_time_sec": 8.237731319008162, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.762387275695801, "lr": 0.0002, "elapsed_sec": 21248.137089967728, "step_time_sec": 8.23849772498943, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.641890048980713, "lr": 0.0002, "elapsed_sec": 21256.376028060913, "step_time_sec": 8.238787223992404, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.7642107009887695, "lr": 0.0002, "elapsed_sec": 21264.612862825394, "step_time_sec": 8.236646654986544, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.742194652557373, "lr": 0.0002, "elapsed_sec": 21272.849103450775, "step_time_sec": 8.236080242000753, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.639270305633545, "lr": 0.0002, "elapsed_sec": 21281.086357831955, "step_time_sec": 8.237082991981879, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.731553077697754, "lr": 0.0002, "elapsed_sec": 21289.32408285141, "step_time_sec": 8.237571362988092, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.564743995666504, "lr": 0.0002, "elapsed_sec": 21297.56280851364, "step_time_sec": 8.238531483017141, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.665171146392822, "lr": 0.0002, "elapsed_sec": 21305.80086827278, "step_time_sec": 8.237928683985956, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.7391791343688965, "lr": 0.0002, "elapsed_sec": 21314.03915166855, "step_time_sec": 8.238101380993612, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.595208168029785, "lr": 0.0002, "elapsed_sec": 21322.2767329216, "step_time_sec": 8.237428544991417, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.7009711265563965, "lr": 0.0002, "elapsed_sec": 21330.515278577805, "step_time_sec": 8.238366998994024, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.693456649780273, "lr": 0.0002, "elapsed_sec": 21338.7528090477, "step_time_sec": 8.237438531999942, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.619643688201904, "lr": 0.0002, "elapsed_sec": 21346.991223573685, "step_time_sec": 8.238188089977484, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.595999717712402, "lr": 0.0002, "elapsed_sec": 21355.229025363922, "step_time_sec": 8.237656644982053, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.7229390144348145, "lr": 0.0002, "elapsed_sec": 21363.465759277344, "step_time_sec": 8.236533863004297, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.6204681396484375, "lr": 0.0002, "elapsed_sec": 21371.70353770256, "step_time_sec": 8.237652376003098, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.640058994293213, "lr": 0.0002, "elapsed_sec": 21379.939229488373, "step_time_sec": 8.235491538012866, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.724812030792236, "lr": 0.0002, "elapsed_sec": 21388.176650762558, "step_time_sec": 8.23728193200077, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.63579797744751, "lr": 0.0002, "elapsed_sec": 21396.413616657257, "step_time_sec": 8.236776434991043, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.499373912811279, "lr": 0.0002, "elapsed_sec": 21404.651938915253, "step_time_sec": 8.238169549003942, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.629740238189697, "lr": 0.0002, "elapsed_sec": 21412.890234947205, "step_time_sec": 8.238139583991142, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.622286796569824, "lr": 0.0002, "elapsed_sec": 21421.12852883339, "step_time_sec": 8.238102885981789, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.56022834777832, "lr": 0.0002, "elapsed_sec": 21429.366131305695, "step_time_sec": 8.237423002021387, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.6608452796936035, "lr": 0.0002, "elapsed_sec": 21437.60183954239, "step_time_sec": 8.23554931601393, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.607532501220703, "lr": 0.0002, "elapsed_sec": 21445.840022325516, "step_time_sec": 8.23802002199227, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.673922538757324, "lr": 0.0002, "elapsed_sec": 21454.077509641647, "step_time_sec": 8.237338240985991, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.706782341003418, "lr": 0.0002, "elapsed_sec": 21462.316519498825, "step_time_sec": 8.238881330005825, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.6251749992370605, "lr": 0.0002, "elapsed_sec": 21470.554990053177, "step_time_sec": 8.238284585997462, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.761871337890625, "lr": 0.0002, "elapsed_sec": 21478.792207479477, "step_time_sec": 8.2371164909855, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.629623889923096, "lr": 0.0002, "elapsed_sec": 21487.030357837677, "step_time_sec": 8.237948224006686, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.629298210144043, "lr": 0.0002, "elapsed_sec": 21495.266728162766, "step_time_sec": 8.236207677982748, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.62616491317749, "lr": 0.0002, "elapsed_sec": 21503.505480527878, "step_time_sec": 8.238565130974166, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.6426873207092285, "lr": 0.0002, "elapsed_sec": 21511.743426322937, "step_time_sec": 8.237818495021202, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.661685943603516, "lr": 0.0002, "elapsed_sec": 21519.980707883835, "step_time_sec": 8.237070261995541, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.714945316314697, "lr": 0.0002, "elapsed_sec": 21528.21730542183, "step_time_sec": 8.236455415986711, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.678370475769043, "lr": 0.0002, "elapsed_sec": 21536.453620672226, "step_time_sec": 8.236178401013603, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.608642578125, "lr": 0.0002, "elapsed_sec": 21544.690650463104, "step_time_sec": 8.236855273018591, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.599237442016602, "lr": 0.0002, "elapsed_sec": 21552.926691055298, "step_time_sec": 8.235853430989664, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.713299751281738, "lr": 0.0002, "elapsed_sec": 21561.165489673615, "step_time_sec": 8.23861318899435, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.597353458404541, "lr": 0.0002, "elapsed_sec": 21569.401898145676, "step_time_sec": 8.236248594010249, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.6885576248168945, "lr": 0.0002, "elapsed_sec": 21577.639326810837, "step_time_sec": 8.23726594800246, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.653203964233398, "lr": 0.0002, "elapsed_sec": 21585.8754901886, "step_time_sec": 8.235993153008167, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.617085933685303, "lr": 0.0002, "elapsed_sec": 21594.112555742264, "step_time_sec": 8.236933878011769, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.6350531578063965, "lr": 0.0002, "elapsed_sec": 21602.35213494301, "step_time_sec": 8.239383458014345, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.611987590789795, "lr": 0.0002, "elapsed_sec": 21610.589242219925, "step_time_sec": 8.236939931986853, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.573317527770996, "lr": 0.0002, "elapsed_sec": 21618.827582359314, "step_time_sec": 8.238213317003101, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.6474151611328125, "lr": 0.0002, "elapsed_sec": 21627.06640458107, "step_time_sec": 8.23865751302219, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.657855987548828, "lr": 0.0002, "elapsed_sec": 21635.304251670837, "step_time_sec": 8.237665574997663, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.645389556884766, "lr": 0.0002, "elapsed_sec": 21643.542619228363, "step_time_sec": 8.238224860018818, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.639029502868652, "lr": 0.0002, "elapsed_sec": 21651.77902841568, "step_time_sec": 8.236237589997472, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.655653953552246, "lr": 0.0002, "elapsed_sec": 21660.018602848053, "step_time_sec": 8.239416868978878, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.653644561767578, "lr": 0.0002, "elapsed_sec": 21668.254048347473, "step_time_sec": 8.235245827003382, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.671348571777344, "lr": 0.0002, "elapsed_sec": 21676.491526842117, "step_time_sec": 8.237299492990132, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.62208890914917, "lr": 0.0002, "elapsed_sec": 21684.729172468185, "step_time_sec": 8.237474383000517, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.703965663909912, "lr": 0.0002, "elapsed_sec": 21692.96687078476, "step_time_sec": 8.237545281008352, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.57684850692749, "lr": 0.0002, "elapsed_sec": 21701.204731941223, "step_time_sec": 8.237694078998175, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.696823596954346, "lr": 0.0002, "elapsed_sec": 21709.441678762436, "step_time_sec": 8.236832505004713, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.613909721374512, "lr": 0.0002, "elapsed_sec": 21717.680325508118, "step_time_sec": 8.238428861979628, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.615787982940674, "lr": 0.0002, "elapsed_sec": 21725.919176101685, "step_time_sec": 8.238735448016087, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.55595064163208, "lr": 0.0002, "elapsed_sec": 21734.157842874527, "step_time_sec": 8.238484683999559, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.671749591827393, "lr": 0.0002, "elapsed_sec": 21742.396187782288, "step_time_sec": 8.23818544097594, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.521440505981445, "lr": 0.0002, "elapsed_sec": 21750.632363796234, "step_time_sec": 8.236014731985051, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.52898645401001, "lr": 0.0002, "elapsed_sec": 21758.871571540833, "step_time_sec": 8.23907161000534, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.680166721343994, "lr": 0.0002, "elapsed_sec": 21767.107927322388, "step_time_sec": 8.236173823999707, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.573197841644287, "lr": 0.0002, "elapsed_sec": 21775.34770131111, "step_time_sec": 8.239569900004426, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.759476661682129, "lr": 0.0002, "elapsed_sec": 21783.586265325546, "step_time_sec": 8.238393303006887, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.628368854522705, "lr": 0.0002, "elapsed_sec": 21791.823464870453, "step_time_sec": 8.237071881012525, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.644216060638428, "lr": 0.0002, "elapsed_sec": 21800.0612950325, "step_time_sec": 8.237692850001622, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.599142074584961, "lr": 0.0002, "elapsed_sec": 21808.29770040512, "step_time_sec": 8.23621211698628, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.634516716003418, "lr": 0.0002, "elapsed_sec": 21816.536239385605, "step_time_sec": 8.238371703017037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.488820552825928, "lr": 0.0002, "elapsed_sec": 21824.772784471512, "step_time_sec": 8.236371638980927, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.658653259277344, "lr": 0.0002, "elapsed_sec": 21833.008992671967, "step_time_sec": 8.23605482801213, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.754936218261719, "lr": 0.0002, "elapsed_sec": 21841.24797964096, "step_time_sec": 8.238821273000212, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.682060718536377, "lr": 0.0002, "elapsed_sec": 21849.484700918198, "step_time_sec": 8.236561497004004, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.695809364318848, "lr": 0.0002, "elapsed_sec": 21857.722820997238, "step_time_sec": 8.237952698022127, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.667831897735596, "lr": 0.0002, "elapsed_sec": 21865.960328817368, "step_time_sec": 8.237322278000647, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.689744472503662, "lr": 0.0002, "elapsed_sec": 21874.200055360794, "step_time_sec": 8.239585796982283, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.576429843902588, "lr": 0.0002, "elapsed_sec": 21882.43829727173, "step_time_sec": 8.238093681982718, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.6347270011901855, "lr": 0.0002, "elapsed_sec": 21890.677361249924, "step_time_sec": 8.238902578013949, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.643517971038818, "lr": 0.0002, "elapsed_sec": 21898.91485142708, "step_time_sec": 8.237344550987473, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.6871747970581055, "lr": 0.0002, "elapsed_sec": 21907.151532411575, "step_time_sec": 8.23650249899947, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.677406311035156, "lr": 0.0002, "elapsed_sec": 21915.388773679733, "step_time_sec": 8.237064264016226, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.628626346588135, "lr": 0.0002, "elapsed_sec": 21923.62509036064, "step_time_sec": 8.236147657007677, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.544016361236572, "lr": 0.0002, "elapsed_sec": 21931.86396741867, "step_time_sec": 8.23877087899018, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.638637542724609, "lr": 0.0002, "elapsed_sec": 21940.102019309998, "step_time_sec": 8.237838364992058, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.732720375061035, "lr": 0.0002, "elapsed_sec": 21948.338770627975, "step_time_sec": 8.236595361988293, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.621179580688477, "lr": 0.0002, "elapsed_sec": 21956.578909873962, "step_time_sec": 8.240005360014038, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.5726447105407715, "lr": 0.0002, "elapsed_sec": 21964.814655303955, "step_time_sec": 8.235538629989605, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.716270446777344, "lr": 0.0002, "elapsed_sec": 21973.053590536118, "step_time_sec": 8.23876809500507, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.50132417678833, "lr": 0.0002, "elapsed_sec": 21981.29066181183, "step_time_sec": 8.236921598989284, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.7024312019348145, "lr": 0.0002, "elapsed_sec": 21989.529147148132, "step_time_sec": 8.238322674995288, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.598111152648926, "lr": 0.0002, "elapsed_sec": 21997.767470359802, "step_time_sec": 8.238194984005531, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.615213871002197, "lr": 0.0002, "elapsed_sec": 22006.004513263702, "step_time_sec": 8.23684410698479, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.539103984832764, "lr": 0.0002, "elapsed_sec": 22014.243062257767, "step_time_sec": 8.238441331981448, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.6916913986206055, "lr": 0.0002, "elapsed_sec": 22022.48083591461, "step_time_sec": 8.237567142990883, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.661520004272461, "lr": 0.0002, "elapsed_sec": 22030.719670295715, "step_time_sec": 8.238671050989069, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.705374717712402, "lr": 0.0002, "elapsed_sec": 22038.957471370697, "step_time_sec": 8.237690366018796, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.567711353302002, "lr": 0.0002, "elapsed_sec": 22047.195180892944, "step_time_sec": 8.237526353012072, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.558777809143066, "lr": 0.0002, "elapsed_sec": 22055.433612585068, "step_time_sec": 8.238233014999423, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.476033687591553, "lr": 0.0002, "elapsed_sec": 22063.67113995552, "step_time_sec": 8.237366640008986, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.606191635131836, "lr": 0.0002, "elapsed_sec": 22071.910041570663, "step_time_sec": 8.238707832992077, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.623610973358154, "lr": 0.0002, "elapsed_sec": 22080.147892713547, "step_time_sec": 8.237707174994284, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.399093151092529, "lr": 0.0002, "elapsed_sec": 22088.385668992996, "step_time_sec": 8.237675930984551, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.569406986236572, "lr": 0.0002, "elapsed_sec": 22096.623673200607, "step_time_sec": 8.237832841987256, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.725725173950195, "lr": 0.0002, "elapsed_sec": 22104.860066652298, "step_time_sec": 8.236194686993258, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.589970588684082, "lr": 0.0002, "elapsed_sec": 22113.098691225052, "step_time_sec": 8.238471591001144, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.633251190185547, "lr": 0.0002, "elapsed_sec": 22121.336490631104, "step_time_sec": 8.23761245899368, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.615111827850342, "lr": 0.0002, "elapsed_sec": 22129.575626850128, "step_time_sec": 8.239017715997761, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.621021747589111, "lr": 0.0002, "elapsed_sec": 22137.815110206604, "step_time_sec": 8.23931726301089, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.567451000213623, "lr": 0.0002, "elapsed_sec": 22146.053370714188, "step_time_sec": 8.238082367024617, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.670942783355713, "lr": 0.0002, "elapsed_sec": 22154.291409492493, "step_time_sec": 8.237827817996731, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.6053667068481445, "lr": 0.0002, "elapsed_sec": 22162.52877521515, "step_time_sec": 8.237210657011019, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.7099995613098145, "lr": 0.0002, "elapsed_sec": 22170.765521526337, "step_time_sec": 8.2365753390186, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.682890892028809, "lr": 0.0002, "elapsed_sec": 22179.00318288803, "step_time_sec": 8.237554154999088, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.688735008239746, "lr": 0.0002, "elapsed_sec": 22187.241835832596, "step_time_sec": 8.238478813989786, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.606507301330566, "lr": 0.0002, "elapsed_sec": 22195.47835469246, "step_time_sec": 8.23632458798238, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.702685832977295, "lr": 0.0002, "elapsed_sec": 22203.714997053146, "step_time_sec": 8.23647823301144, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.5286455154418945, "lr": 0.0002, "elapsed_sec": 22211.952459812164, "step_time_sec": 8.237321000982774, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.619298934936523, "lr": 0.0002, "elapsed_sec": 22220.190778017044, "step_time_sec": 8.23812833201373, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.570486545562744, "lr": 0.0002, "elapsed_sec": 22228.429926633835, "step_time_sec": 8.238980879017618, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.593095779418945, "lr": 0.0002, "elapsed_sec": 22236.667819023132, "step_time_sec": 8.23774467498879, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.676438331604004, "lr": 0.0002, "elapsed_sec": 22244.90534877777, "step_time_sec": 8.237387153989403, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.584624767303467, "lr": 0.0002, "elapsed_sec": 22253.14325118065, "step_time_sec": 8.237692466005683, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.639968395233154, "lr": 0.0002, "elapsed_sec": 22261.380716323853, "step_time_sec": 8.23731854200014, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.510120868682861, "lr": 0.0002, "elapsed_sec": 22269.61949491501, "step_time_sec": 8.23860894201789, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.775114059448242, "lr": 0.0002, "elapsed_sec": 22277.85690164566, "step_time_sec": 8.237280898989411, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.538266658782959, "lr": 0.0002, "elapsed_sec": 22286.09481525421, "step_time_sec": 8.237720642006025, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.68477201461792, "lr": 0.0002, "elapsed_sec": 22294.331784248352, "step_time_sec": 8.236806711007375, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.5056867599487305, "lr": 0.0002, "elapsed_sec": 22302.568360090256, "step_time_sec": 8.236453379999148, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.53419303894043, "lr": 0.0002, "elapsed_sec": 22310.806485891342, "step_time_sec": 8.2379172439978, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.62399959564209, "lr": 0.0002, "elapsed_sec": 22319.043630361557, "step_time_sec": 8.237030568008777, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.577573776245117, "lr": 0.0002, "elapsed_sec": 22327.28184223175, "step_time_sec": 8.238047679013107, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4011406898498535, "lr": 0.0002, "elapsed_sec": 22335.51875281334, "step_time_sec": 8.236714235012187, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.608364105224609, "lr": 0.0002, "elapsed_sec": 22343.75681900978, "step_time_sec": 8.237913666001987, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.623701572418213, "lr": 0.0002, "elapsed_sec": 22351.99453186989, "step_time_sec": 8.237557103013387, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.495904445648193, "lr": 0.0002, "elapsed_sec": 22360.231424331665, "step_time_sec": 8.23673128397786, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.639565944671631, "lr": 0.0002, "elapsed_sec": 22368.4697599411, "step_time_sec": 8.238195243000519, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.6151509284973145, "lr": 0.0002, "elapsed_sec": 22376.707626342773, "step_time_sec": 8.237704221974127, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.640517711639404, "lr": 0.0002, "elapsed_sec": 22384.945880174637, "step_time_sec": 8.238126531010494, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.600674629211426, "lr": 0.0002, "elapsed_sec": 22393.18425011635, "step_time_sec": 8.238141909998376, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.632373809814453, "lr": 0.0002, "elapsed_sec": 22401.42099428177, "step_time_sec": 8.236588068015408, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.530712127685547, "lr": 0.0002, "elapsed_sec": 22409.65820622444, "step_time_sec": 8.237059831008082, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.594788074493408, "lr": 0.0002, "elapsed_sec": 22417.896188497543, "step_time_sec": 8.237862607988063, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.580802917480469, "lr": 0.0002, "elapsed_sec": 22426.13565802574, "step_time_sec": 8.239236035995418, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.560682773590088, "lr": 0.0002, "elapsed_sec": 22434.37291097641, "step_time_sec": 8.237082612991799, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.643097877502441, "lr": 0.0002, "elapsed_sec": 22442.611610412598, "step_time_sec": 8.238608129002387, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.746896743774414, "lr": 0.0002, "elapsed_sec": 22450.849030971527, "step_time_sec": 8.237217232002877, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.592897415161133, "lr": 0.0002, "elapsed_sec": 22459.08645439148, "step_time_sec": 8.237312103010481, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.627895832061768, "lr": 0.0002, "elapsed_sec": 22467.325328350067, "step_time_sec": 8.238714506995166, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.5332770347595215, "lr": 0.0002, "elapsed_sec": 22475.56183075905, "step_time_sec": 8.236274352995679, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.5456037521362305, "lr": 0.0002, "elapsed_sec": 22483.798813819885, "step_time_sec": 8.236814670002786, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.584880828857422, "lr": 0.0002, "elapsed_sec": 22492.036700725555, "step_time_sec": 8.237781031988561, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.5644426345825195, "lr": 0.0002, "elapsed_sec": 22500.27585220337, "step_time_sec": 8.238938664027955, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.517221450805664, "lr": 0.0002, "elapsed_sec": 22508.512589931488, "step_time_sec": 8.236559721990488, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.653583526611328, "lr": 0.0002, "elapsed_sec": 22516.750220775604, "step_time_sec": 8.23749623200274, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.818925857543945, "lr": 0.0002, "elapsed_sec": 22524.988483190536, "step_time_sec": 8.238102248986252, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.6162943840026855, "lr": 0.0002, "elapsed_sec": 22533.225613594055, "step_time_sec": 8.236956401000498, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.628647804260254, "lr": 0.0002, "elapsed_sec": 22541.463848114014, "step_time_sec": 8.23809404001804, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.562366485595703, "lr": 0.0002, "elapsed_sec": 22549.70294380188, "step_time_sec": 8.238866635016166, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.467998027801514, "lr": 0.0002, "elapsed_sec": 22557.940214395523, "step_time_sec": 8.237166377017274, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.714346885681152, "lr": 0.0002, "elapsed_sec": 22566.17732977867, "step_time_sec": 8.236900759016862, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.797787189483643, "lr": 0.0002, "elapsed_sec": 22574.413238048553, "step_time_sec": 8.235710692999419, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.588383674621582, "lr": 0.0002, "elapsed_sec": 22582.651298999786, "step_time_sec": 8.237972725997679, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.664253234863281, "lr": 0.0002, "elapsed_sec": 22590.888916015625, "step_time_sec": 8.237399709003512, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.588319301605225, "lr": 0.0002, "elapsed_sec": 22599.126230239868, "step_time_sec": 8.237197307986207, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.617103099822998, "lr": 0.0002, "elapsed_sec": 22607.36492562294, "step_time_sec": 8.238481192995096, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.619024276733398, "lr": 0.0002, "elapsed_sec": 22615.601989507675, "step_time_sec": 8.236966277007014, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.615792274475098, "lr": 0.0002, "elapsed_sec": 22623.841543912888, "step_time_sec": 8.239336625003489, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.599827289581299, "lr": 0.0002, "elapsed_sec": 22632.078955888748, "step_time_sec": 8.237265042989748, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.607896327972412, "lr": 0.0002, "elapsed_sec": 22640.31791806221, "step_time_sec": 8.238796897989232, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.583054542541504, "lr": 0.0002, "elapsed_sec": 22648.556270837784, "step_time_sec": 8.2382275799755, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.714932918548584, "lr": 0.0002, "elapsed_sec": 22656.793693065643, "step_time_sec": 8.237279537977884, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.543171405792236, "lr": 0.0002, "elapsed_sec": 22665.032276153564, "step_time_sec": 8.238413126004161, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.599619388580322, "lr": 0.0002, "elapsed_sec": 22673.26879620552, "step_time_sec": 8.23640120899654, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.500417709350586, "lr": 0.0002, "elapsed_sec": 22681.50800204277, "step_time_sec": 8.23896580102155, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.532864570617676, "lr": 0.0002, "elapsed_sec": 22689.74605321884, "step_time_sec": 8.237901754997438, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.543172359466553, "lr": 0.0002, "elapsed_sec": 22697.984713554382, "step_time_sec": 8.238567221997073, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.685383319854736, "lr": 0.0002, "elapsed_sec": 22706.222328186035, "step_time_sec": 8.237438460986596, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.612987041473389, "lr": 0.0002, "elapsed_sec": 22714.460696220398, "step_time_sec": 8.23816197598353, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.767245769500732, "lr": 0.0002, "elapsed_sec": 22722.69920349121, "step_time_sec": 8.238388552010292, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.555084705352783, "lr": 0.0002, "elapsed_sec": 22730.936733722687, "step_time_sec": 8.237326377013233, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.572932720184326, "lr": 0.0002, "elapsed_sec": 22739.175933361053, "step_time_sec": 8.239086377987405, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.529351234436035, "lr": 0.0002, "elapsed_sec": 22747.41399049759, "step_time_sec": 8.237858603009954, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.55593204498291, "lr": 0.0002, "elapsed_sec": 22755.65064740181, "step_time_sec": 8.236492291005561, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.568856716156006, "lr": 0.0002, "elapsed_sec": 22763.889178276062, "step_time_sec": 8.23838685901137, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.497682571411133, "lr": 0.0002, "elapsed_sec": 22772.126359462738, "step_time_sec": 8.237015233025886, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.418493747711182, "lr": 0.0002, "elapsed_sec": 22780.364636659622, "step_time_sec": 8.238121823000256, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.662720680236816, "lr": 0.0002, "elapsed_sec": 22788.601472377777, "step_time_sec": 8.236712790996535, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.589578628540039, "lr": 0.0002, "elapsed_sec": 22796.8391726017, "step_time_sec": 8.237466742022661, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.478949546813965, "lr": 0.0002, "elapsed_sec": 22805.077291488647, "step_time_sec": 8.237984648993006, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.64387845993042, "lr": 0.0002, "elapsed_sec": 22813.314481019974, "step_time_sec": 8.237034559977474, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.537414073944092, "lr": 0.0002, "elapsed_sec": 22821.55267238617, "step_time_sec": 8.237995057977969, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.55338716506958, "lr": 0.0002, "elapsed_sec": 22829.789923906326, "step_time_sec": 8.237088207009947, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.536045551300049, "lr": 0.0002, "elapsed_sec": 22838.027973890305, "step_time_sec": 8.237917098012986, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.540973663330078, "lr": 0.0002, "elapsed_sec": 22846.266943216324, "step_time_sec": 8.238777869992191, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.539345741271973, "lr": 0.0002, "elapsed_sec": 22854.504431009293, "step_time_sec": 8.237335746001918, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.400106430053711, "lr": 0.0002, "elapsed_sec": 22862.742207288742, "step_time_sec": 8.237640358973294, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.5219807624816895, "lr": 0.0002, "elapsed_sec": 22870.978312253952, "step_time_sec": 8.235867657000199, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.520870685577393, "lr": 0.0002, "elapsed_sec": 22879.216804742813, "step_time_sec": 8.238333563000197, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.448083400726318, "lr": 0.0002, "elapsed_sec": 22887.453722715378, "step_time_sec": 8.236755273013841, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.535109996795654, "lr": 0.0002, "elapsed_sec": 22895.69188761711, "step_time_sec": 8.238023632002296, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.712809085845947, "lr": 0.0002, "elapsed_sec": 22903.930287361145, "step_time_sec": 8.23823397399974, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.718941688537598, "lr": 0.0002, "elapsed_sec": 22912.167651176453, "step_time_sec": 8.23723803300527, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.6107869148254395, "lr": 0.0002, "elapsed_sec": 22920.40775704384, "step_time_sec": 8.239905547001399, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.534419536590576, "lr": 0.0002, "elapsed_sec": 22928.646085739136, "step_time_sec": 8.238205565983662, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.497578144073486, "lr": 0.0002, "elapsed_sec": 22936.885115623474, "step_time_sec": 8.238798759994097, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.6310319900512695, "lr": 0.0002, "elapsed_sec": 22945.12254858017, "step_time_sec": 8.237308014999144, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.616804122924805, "lr": 0.0002, "elapsed_sec": 22953.360610961914, "step_time_sec": 8.237892331002513, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.6059184074401855, "lr": 0.0002, "elapsed_sec": 22961.60009121895, "step_time_sec": 8.239303624024615, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.563573360443115, "lr": 0.0002, "elapsed_sec": 22969.8366792202, "step_time_sec": 8.236416634987108, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.541926860809326, "lr": 0.0002, "elapsed_sec": 22978.073039770126, "step_time_sec": 8.23624202699284, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.528733730316162, "lr": 0.0002, "elapsed_sec": 22986.310784578323, "step_time_sec": 8.237584898015484, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.777179718017578, "lr": 0.0002, "elapsed_sec": 22994.548947811127, "step_time_sec": 8.237940341001377, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.584120273590088, "lr": 0.0002, "elapsed_sec": 23002.786887168884, "step_time_sec": 8.23781508297543, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.602491855621338, "lr": 0.0002, "elapsed_sec": 23011.02533340454, "step_time_sec": 8.23826119399746, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.598355293273926, "lr": 0.0002, "elapsed_sec": 23019.26314640045, "step_time_sec": 8.237624640983995, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.698653221130371, "lr": 0.0002, "elapsed_sec": 23027.4988322258, "step_time_sec": 8.235527986020315, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.568430423736572, "lr": 0.0002, "elapsed_sec": 23035.737406492233, "step_time_sec": 8.238431848003529, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.535412311553955, "lr": 0.0002, "elapsed_sec": 23043.976894140244, "step_time_sec": 8.239308090007398, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.6115264892578125, "lr": 0.0002, "elapsed_sec": 23052.214395046234, "step_time_sec": 8.237336482008686, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.54271936416626, "lr": 0.0002, "elapsed_sec": 23060.454016447067, "step_time_sec": 8.239452395995613, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.616098880767822, "lr": 0.0002, "elapsed_sec": 23068.691212415695, "step_time_sec": 8.237098766985582, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.740482330322266, "lr": 0.0002, "elapsed_sec": 23076.928898096085, "step_time_sec": 8.237420144985663, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.5666422843933105, "lr": 0.0002, "elapsed_sec": 23085.16509962082, "step_time_sec": 8.236065701988991, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.592806816101074, "lr": 0.0002, "elapsed_sec": 23093.403688907623, "step_time_sec": 8.238461277011083, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.564150333404541, "lr": 0.0002, "elapsed_sec": 23101.642217874527, "step_time_sec": 8.238305120001314, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.588181018829346, "lr": 0.0002, "elapsed_sec": 23109.879663944244, "step_time_sec": 8.237282045010943, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.619681358337402, "lr": 0.0002, "elapsed_sec": 23118.11964058876, "step_time_sec": 8.239863905007951, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.587174415588379, "lr": 0.0002, "elapsed_sec": 23126.35648870468, "step_time_sec": 8.236705652001547, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.548460483551025, "lr": 0.0002, "elapsed_sec": 23134.594891309738, "step_time_sec": 8.238210820010863, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.424116611480713, "lr": 0.0002, "elapsed_sec": 23142.833471775055, "step_time_sec": 8.238416606007377, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.736181735992432, "lr": 0.0002, "elapsed_sec": 23151.07123684883, "step_time_sec": 8.237611338990973, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.580018520355225, "lr": 0.0002, "elapsed_sec": 23159.309410333633, "step_time_sec": 8.238078132009832, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.537936210632324, "lr": 0.0002, "elapsed_sec": 23167.546750307083, "step_time_sec": 8.23717064599623, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.719261169433594, "lr": 0.0002, "elapsed_sec": 23175.785836219788, "step_time_sec": 8.238875431008637, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.708008289337158, "lr": 0.0002, "elapsed_sec": 23184.024203777313, "step_time_sec": 8.238226205983665, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.562666893005371, "lr": 0.0002, "elapsed_sec": 23192.26138305664, "step_time_sec": 8.236964821000583, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.567478179931641, "lr": 0.0002, "elapsed_sec": 23200.499496936798, "step_time_sec": 8.237973839015467, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.568338394165039, "lr": 0.0002, "elapsed_sec": 23208.735791683197, "step_time_sec": 8.23615954699926, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.629532337188721, "lr": 0.0002, "elapsed_sec": 23216.97476363182, "step_time_sec": 8.238760093983728, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.479454517364502, "lr": 0.0002, "elapsed_sec": 23225.21147608757, "step_time_sec": 8.236570796987507, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.63501501083374, "lr": 0.0002, "elapsed_sec": 23233.449605703354, "step_time_sec": 8.237971947004553, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.617061614990234, "lr": 0.0002, "elapsed_sec": 23241.68770122528, "step_time_sec": 8.237905114976456, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.500502109527588, "lr": 0.0002, "elapsed_sec": 23249.925369501114, "step_time_sec": 8.237507818004815, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.589376926422119, "lr": 0.0002, "elapsed_sec": 23258.163586854935, "step_time_sec": 8.238106236996828, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.5300140380859375, "lr": 0.0002, "elapsed_sec": 23266.400627613068, "step_time_sec": 8.236825176980346, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.528377056121826, "lr": 0.0002, "elapsed_sec": 23274.63911175728, "step_time_sec": 8.238353188993642, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.517440319061279, "lr": 0.0002, "elapsed_sec": 23282.8780169487, "step_time_sec": 8.23868578497786, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.594846248626709, "lr": 0.0002, "elapsed_sec": 23291.115795373917, "step_time_sec": 8.237605964997783, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.616888523101807, "lr": 0.0002, "elapsed_sec": 23299.355672121048, "step_time_sec": 8.23979166400386, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.664924144744873, "lr": 0.0002, "elapsed_sec": 23307.59200334549, "step_time_sec": 8.236106222990202, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.573761940002441, "lr": 0.0002, "elapsed_sec": 23315.829924583435, "step_time_sec": 8.237743958015926, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.673463344573975, "lr": 0.0002, "elapsed_sec": 23324.067855596542, "step_time_sec": 8.237763018987607, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.637468338012695, "lr": 0.0002, "elapsed_sec": 23332.306024074554, "step_time_sec": 8.238033020985313, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.543626308441162, "lr": 0.0002, "elapsed_sec": 23340.545991182327, "step_time_sec": 8.239801590010757, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.580399036407471, "lr": 0.0002, "elapsed_sec": 23348.782340288162, "step_time_sec": 8.236235340998974, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.50419807434082, "lr": 0.0002, "elapsed_sec": 23357.02196073532, "step_time_sec": 8.23939841700485, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4810943603515625, "lr": 0.0002, "elapsed_sec": 23365.259718179703, "step_time_sec": 8.237640306993853, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.534909725189209, "lr": 0.0002, "elapsed_sec": 23373.497526407242, "step_time_sec": 8.237616094993427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.680541038513184, "lr": 0.0002, "elapsed_sec": 23381.734582901, "step_time_sec": 8.236894185014535, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.588202953338623, "lr": 0.0002, "elapsed_sec": 23389.971836328506, "step_time_sec": 8.23709099800908, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.555620193481445, "lr": 0.0002, "elapsed_sec": 23398.210393190384, "step_time_sec": 8.238418041990371, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.754929065704346, "lr": 0.0002, "elapsed_sec": 23406.447560071945, "step_time_sec": 8.237016220024088, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.562585830688477, "lr": 0.0002, "elapsed_sec": 23414.688014507294, "step_time_sec": 8.240262010018341, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.606460094451904, "lr": 0.0002, "elapsed_sec": 23422.926223278046, "step_time_sec": 8.23802695699851, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.592846393585205, "lr": 0.0002, "elapsed_sec": 23431.164032936096, "step_time_sec": 8.237644867011113, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.553095817565918, "lr": 0.0002, "elapsed_sec": 23439.403394460678, "step_time_sec": 8.239204288984183, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.525940418243408, "lr": 0.0002, "elapsed_sec": 23447.640181064606, "step_time_sec": 8.236651977000292, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.572963714599609, "lr": 0.0002, "elapsed_sec": 23455.877775669098, "step_time_sec": 8.237421502009965, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.47057580947876, "lr": 0.0002, "elapsed_sec": 23464.115724802017, "step_time_sec": 8.237773666973226, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.603867053985596, "lr": 0.0002, "elapsed_sec": 23472.354058027267, "step_time_sec": 8.238189547002548, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.605414867401123, "lr": 0.0002, "elapsed_sec": 23480.59268260002, "step_time_sec": 8.238433598016854, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.469859600067139, "lr": 0.0002, "elapsed_sec": 23488.830824375153, "step_time_sec": 8.238041759992484, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.62543249130249, "lr": 0.0002, "elapsed_sec": 23497.069126844406, "step_time_sec": 8.238078830996528, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.548169136047363, "lr": 0.0002, "elapsed_sec": 23505.306234121323, "step_time_sec": 8.23696535199997, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.558518886566162, "lr": 0.0002, "elapsed_sec": 23513.544404506683, "step_time_sec": 8.238053911976749, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.559842586517334, "lr": 0.0002, "elapsed_sec": 23521.78204870224, "step_time_sec": 8.237434246984776, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.527712345123291, "lr": 0.0002, "elapsed_sec": 23530.019842386246, "step_time_sec": 8.237673541996628, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.521951675415039, "lr": 0.0002, "elapsed_sec": 23538.256488084793, "step_time_sec": 8.236495752993505, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.584660053253174, "lr": 0.0002, "elapsed_sec": 23546.493693351746, "step_time_sec": 8.237021497014211, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4588189125061035, "lr": 0.0002, "elapsed_sec": 23554.731731891632, "step_time_sec": 8.23785339598544, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.598663330078125, "lr": 0.0002, "elapsed_sec": 23562.969515562057, "step_time_sec": 8.237608605006244, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.494728088378906, "lr": 0.0002, "elapsed_sec": 23571.208260297775, "step_time_sec": 8.238629480998497, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.594218730926514, "lr": 0.0002, "elapsed_sec": 23579.447311639786, "step_time_sec": 8.238857425982133, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.5703630447387695, "lr": 0.0002, "elapsed_sec": 23587.684243440628, "step_time_sec": 8.23679429999902, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.550136566162109, "lr": 0.0002, "elapsed_sec": 23595.923510313034, "step_time_sec": 8.239059184998041, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.566086769104004, "lr": 0.0002, "elapsed_sec": 23604.161429405212, "step_time_sec": 8.237807099008933, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.432150840759277, "lr": 0.0002, "elapsed_sec": 23612.400232076645, "step_time_sec": 8.238616682996508, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.635239601135254, "lr": 0.0002, "elapsed_sec": 23620.63837003708, "step_time_sec": 8.237944744993001, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.545440673828125, "lr": 0.0002, "elapsed_sec": 23628.875604629517, "step_time_sec": 8.237124395993305, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.606405735015869, "lr": 0.0002, "elapsed_sec": 23637.113258600235, "step_time_sec": 8.23746422000113, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.438167095184326, "lr": 0.0002, "elapsed_sec": 23645.349901676178, "step_time_sec": 8.236478458013153, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.598146915435791, "lr": 0.0002, "elapsed_sec": 23653.587629556656, "step_time_sec": 8.237599658983527, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.631957530975342, "lr": 0.0002, "elapsed_sec": 23661.824746370316, "step_time_sec": 8.236906531004934, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.529594421386719, "lr": 0.0002, "elapsed_sec": 23670.062617778778, "step_time_sec": 8.237758099014172, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.5306620597839355, "lr": 0.0002, "elapsed_sec": 23678.30055809021, "step_time_sec": 8.237775132991374, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.690031051635742, "lr": 0.0002, "elapsed_sec": 23686.537877082825, "step_time_sec": 8.23711135602207, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4786810874938965, "lr": 0.0002, "elapsed_sec": 23694.77539896965, "step_time_sec": 8.237365679990035, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.657088756561279, "lr": 0.0002, "elapsed_sec": 23703.01349592209, "step_time_sec": 8.237943285988877, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.517789363861084, "lr": 0.0002, "elapsed_sec": 23711.252952575684, "step_time_sec": 8.239309208001941, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.568476676940918, "lr": 0.0002, "elapsed_sec": 23719.490670204163, "step_time_sec": 8.237525703007123, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.56658935546875, "lr": 0.0002, "elapsed_sec": 23727.728021144867, "step_time_sec": 8.237185623002006, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4671406745910645, "lr": 0.0002, "elapsed_sec": 23735.966922283173, "step_time_sec": 8.238753294019261, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.649815559387207, "lr": 0.0002, "elapsed_sec": 23744.204269886017, "step_time_sec": 8.237181349017192, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.469109535217285, "lr": 0.0002, "elapsed_sec": 23752.441823482513, "step_time_sec": 8.23737753299065, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.649037837982178, "lr": 0.0002, "elapsed_sec": 23760.679489135742, "step_time_sec": 8.237503087992081, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.474819183349609, "lr": 0.0002, "elapsed_sec": 23768.917614221573, "step_time_sec": 8.238032915978692, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.624330997467041, "lr": 0.0002, "elapsed_sec": 23777.156352758408, "step_time_sec": 8.238515916018514, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.507408618927002, "lr": 0.0002, "elapsed_sec": 23785.392386198044, "step_time_sec": 8.23585968799307, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.611911296844482, "lr": 0.0002, "elapsed_sec": 23793.630118131638, "step_time_sec": 8.237605163012631, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.574161529541016, "lr": 0.0002, "elapsed_sec": 23801.86637353897, "step_time_sec": 8.236036896996666, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.586167335510254, "lr": 0.0002, "elapsed_sec": 23810.104826450348, "step_time_sec": 8.238293350994354, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.5428900718688965, "lr": 0.0002, "elapsed_sec": 23818.342086553574, "step_time_sec": 8.237068818008993, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.636773586273193, "lr": 0.0002, "elapsed_sec": 23826.58028626442, "step_time_sec": 8.2381068719842, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.434237957000732, "lr": 0.0002, "elapsed_sec": 23834.819150924683, "step_time_sec": 8.2386300599901, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.509871482849121, "lr": 0.0002, "elapsed_sec": 23843.05663084984, "step_time_sec": 8.237311526987469, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.575530052185059, "lr": 0.0002, "elapsed_sec": 23851.29611158371, "step_time_sec": 8.239331798977219, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.5204010009765625, "lr": 0.0002, "elapsed_sec": 23859.53431200981, "step_time_sec": 8.238036487979116, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4990153312683105, "lr": 0.0002, "elapsed_sec": 23867.772169828415, "step_time_sec": 8.237695223011542, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.582627773284912, "lr": 0.0002, "elapsed_sec": 23876.011115074158, "step_time_sec": 8.238779155013617, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.5896382331848145, "lr": 0.0002, "elapsed_sec": 23884.24711251259, "step_time_sec": 8.235839427012252, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.644832611083984, "lr": 0.0002, "elapsed_sec": 23892.485525608063, "step_time_sec": 8.238250098016579, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.089484214782715, "lr": 0.0002, "elapsed_sec": 23900.722499370575, "step_time_sec": 8.236803652020171, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.665006637573242, "lr": 0.0002, "elapsed_sec": 23908.961183786392, "step_time_sec": 8.238568174972897, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.7059149742126465, "lr": 0.0002, "elapsed_sec": 23917.198877334595, "step_time_sec": 8.237446318002185, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.5082502365112305, "lr": 0.0002, "elapsed_sec": 23925.43786716461, "step_time_sec": 8.23884015501244, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.588400363922119, "lr": 0.0002, "elapsed_sec": 23933.677872896194, "step_time_sec": 8.239882577006938, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.59529447555542, "lr": 0.0002, "elapsed_sec": 23941.91487264633, "step_time_sec": 8.236793081014184, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.6882243156433105, "lr": 0.0002, "elapsed_sec": 23950.15334868431, "step_time_sec": 8.238344145996962, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.551661968231201, "lr": 0.0002, "elapsed_sec": 23958.39127969742, "step_time_sec": 8.237750013999175, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.544261455535889, "lr": 0.0002, "elapsed_sec": 23966.629776477814, "step_time_sec": 8.238348264014348, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.601990699768066, "lr": 0.0002, "elapsed_sec": 23974.86835360527, "step_time_sec": 8.238411924015963, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.542189121246338, "lr": 0.0002, "elapsed_sec": 23983.10519218445, "step_time_sec": 8.236667115997989, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.585024833679199, "lr": 0.0002, "elapsed_sec": 23991.344123363495, "step_time_sec": 8.238764347013785, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.523182392120361, "lr": 0.0002, "elapsed_sec": 23999.583184719086, "step_time_sec": 8.238847108994378, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.509928226470947, "lr": 0.0002, "elapsed_sec": 24007.82125067711, "step_time_sec": 8.237927479989594, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.602270603179932, "lr": 0.0002, "elapsed_sec": 24016.059548139572, "step_time_sec": 8.23811483301688, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.443925380706787, "lr": 0.0002, "elapsed_sec": 24024.296591997147, "step_time_sec": 8.236887053993996, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.582144260406494, "lr": 0.0002, "elapsed_sec": 24032.535175800323, "step_time_sec": 8.238416987005621, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.497056007385254, "lr": 0.0002, "elapsed_sec": 24040.771721601486, "step_time_sec": 8.236384593008552, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.482672691345215, "lr": 0.0002, "elapsed_sec": 24049.010188102722, "step_time_sec": 8.238350351981353, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.451012134552002, "lr": 0.0002, "elapsed_sec": 24057.248123407364, "step_time_sec": 8.237703077000333, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.55324649810791, "lr": 0.0002, "elapsed_sec": 24065.485276937485, "step_time_sec": 8.237013438018039, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.470335960388184, "lr": 0.0002, "elapsed_sec": 24073.723749160767, "step_time_sec": 8.238308024010621, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.432615756988525, "lr": 0.0002, "elapsed_sec": 24081.96111392975, "step_time_sec": 8.237146120984107, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.59962272644043, "lr": 0.0002, "elapsed_sec": 24090.198668718338, "step_time_sec": 8.237415168987354, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.580652236938477, "lr": 0.0002, "elapsed_sec": 24098.435212135315, "step_time_sec": 8.236354573018616, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.471490859985352, "lr": 0.0002, "elapsed_sec": 24106.673968315125, "step_time_sec": 8.238602320983773, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.706088066101074, "lr": 0.0002, "elapsed_sec": 24114.911691904068, "step_time_sec": 8.237562306982, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.661157608032227, "lr": 0.0002, "elapsed_sec": 24123.149260282516, "step_time_sec": 8.237411768001039, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.475597381591797, "lr": 0.0002, "elapsed_sec": 24131.386405944824, "step_time_sec": 8.237013481993927, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.492853164672852, "lr": 0.0002, "elapsed_sec": 24139.623191595078, "step_time_sec": 8.236587758990936, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.558228969573975, "lr": 0.0002, "elapsed_sec": 24147.861271619797, "step_time_sec": 8.237962931016227, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.531172275543213, "lr": 0.0002, "elapsed_sec": 24156.099012613297, "step_time_sec": 8.237543349008774, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.568062782287598, "lr": 0.0002, "elapsed_sec": 24164.33888745308, "step_time_sec": 8.239707404980436, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.391548156738281, "lr": 0.0002, "elapsed_sec": 24172.577941417694, "step_time_sec": 8.238882859004661, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.555424690246582, "lr": 0.0002, "elapsed_sec": 24180.814162254333, "step_time_sec": 8.236020549986279, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.561737060546875, "lr": 0.0002, "elapsed_sec": 24189.053355932236, "step_time_sec": 8.239089597977, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.615749359130859, "lr": 0.0002, "elapsed_sec": 24197.289874076843, "step_time_sec": 8.236294625006849, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.5691046714782715, "lr": 0.0002, "elapsed_sec": 24205.529057264328, "step_time_sec": 8.239034831989557, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.595195293426514, "lr": 0.0002, "elapsed_sec": 24213.76754975319, "step_time_sec": 8.238379287009593, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.581728458404541, "lr": 0.0002, "elapsed_sec": 24222.005929470062, "step_time_sec": 8.23816918101511, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.700037479400635, "lr": 0.0002, "elapsed_sec": 24230.243844270706, "step_time_sec": 8.237764701974811, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.534769058227539, "lr": 0.0002, "elapsed_sec": 24238.48141503334, "step_time_sec": 8.237454157002503, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.474280834197998, "lr": 0.0002, "elapsed_sec": 24246.72033762932, "step_time_sec": 8.238752313016448, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.550085544586182, "lr": 0.0002, "elapsed_sec": 24254.957869291306, "step_time_sec": 8.237367488996824, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.5946431159973145, "lr": 0.0002, "elapsed_sec": 24263.19595360756, "step_time_sec": 8.23789474699879, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.711957931518555, "lr": 0.0002, "elapsed_sec": 24271.434658527374, "step_time_sec": 8.238555291987723, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.279528617858887, "lr": 0.0002, "elapsed_sec": 24279.671255350113, "step_time_sec": 8.236405579984421, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.6658453941345215, "lr": 0.0002, "elapsed_sec": 24287.91004562378, "step_time_sec": 8.238665548007702, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.562016487121582, "lr": 0.0002, "elapsed_sec": 24296.148831367493, "step_time_sec": 8.238599984004395, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.557554244995117, "lr": 0.0002, "elapsed_sec": 24304.387729406357, "step_time_sec": 8.238697883003624, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.44597053527832, "lr": 0.0002, "elapsed_sec": 24312.62579011917, "step_time_sec": 8.237891193013638, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.456723213195801, "lr": 0.0002, "elapsed_sec": 24320.862959623337, "step_time_sec": 8.237034056015546, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.608489036560059, "lr": 0.0002, "elapsed_sec": 24329.101345062256, "step_time_sec": 8.238240236009005, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.577304363250732, "lr": 0.0002, "elapsed_sec": 24337.33754992485, "step_time_sec": 8.235989862005226, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.497570991516113, "lr": 0.0002, "elapsed_sec": 24345.575179576874, "step_time_sec": 8.237466065998888, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.400341987609863, "lr": 0.0002, "elapsed_sec": 24353.81227684021, "step_time_sec": 8.2369512639998, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.399156093597412, "lr": 0.0002, "elapsed_sec": 24362.051080703735, "step_time_sec": 8.23859344999073, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.358447551727295, "lr": 0.0002, "elapsed_sec": 24370.288642406464, "step_time_sec": 8.237416048999876, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.6674275398254395, "lr": 0.0002, "elapsed_sec": 24378.52579140663, "step_time_sec": 8.236968409997644, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.584185600280762, "lr": 0.0002, "elapsed_sec": 24386.76504778862, "step_time_sec": 8.23909480200382, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.612545013427734, "lr": 0.0002, "elapsed_sec": 24395.001940727234, "step_time_sec": 8.236738533974858, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.470470428466797, "lr": 0.0002, "elapsed_sec": 24403.23987364769, "step_time_sec": 8.237757235998288, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.492236614227295, "lr": 0.0002, "elapsed_sec": 24411.478794813156, "step_time_sec": 8.238786611007527, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.44482946395874, "lr": 0.0002, "elapsed_sec": 24419.71603679657, "step_time_sec": 8.237091347022215, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.556719779968262, "lr": 0.0002, "elapsed_sec": 24427.95428800583, "step_time_sec": 8.23809647501912, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.466104984283447, "lr": 0.0002, "elapsed_sec": 24436.191386938095, "step_time_sec": 8.23688060400309, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.374627590179443, "lr": 0.0002, "elapsed_sec": 24444.4295065403, "step_time_sec": 8.237941359024262, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.556313991546631, "lr": 0.0002, "elapsed_sec": 24452.66777586937, "step_time_sec": 8.238136672007386, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.532583713531494, "lr": 0.0002, "elapsed_sec": 24460.905534744263, "step_time_sec": 8.237568623997504, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.406681537628174, "lr": 0.0002, "elapsed_sec": 24469.142973423004, "step_time_sec": 8.237325170979602, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.520255088806152, "lr": 0.0002, "elapsed_sec": 24477.380741119385, "step_time_sec": 8.237542129994836, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.507522106170654, "lr": 0.0002, "elapsed_sec": 24485.61883997917, "step_time_sec": 8.237941975996364, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.577378273010254, "lr": 0.0002, "elapsed_sec": 24493.85620045662, "step_time_sec": 8.237203830998624, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.5517706871032715, "lr": 0.0002, "elapsed_sec": 24502.09232854843, "step_time_sec": 8.235964079009136, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.495532512664795, "lr": 0.0002, "elapsed_sec": 24510.329626083374, "step_time_sec": 8.23713830599445, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.405518531799316, "lr": 0.0002, "elapsed_sec": 24518.566885232925, "step_time_sec": 8.237048325012438, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.600934028625488, "lr": 0.0002, "elapsed_sec": 24526.80442595482, "step_time_sec": 8.23738902999321, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.53218936920166, "lr": 0.0002, "elapsed_sec": 24535.040463924408, "step_time_sec": 8.235903197986772, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.483406066894531, "lr": 0.0002, "elapsed_sec": 24543.27871155739, "step_time_sec": 8.238052505999804, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.5319366455078125, "lr": 0.0002, "elapsed_sec": 24551.514499664307, "step_time_sec": 8.235625578992767, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.5917158126831055, "lr": 0.0002, "elapsed_sec": 24559.75311589241, "step_time_sec": 8.2384857639845, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.518779754638672, "lr": 0.0002, "elapsed_sec": 24567.991190195084, "step_time_sec": 8.23791501802043, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.425802707672119, "lr": 0.0002, "elapsed_sec": 24576.22765493393, "step_time_sec": 8.236282272002427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.440704345703125, "lr": 0.0002, "elapsed_sec": 24584.465161323547, "step_time_sec": 8.237330946984002, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.433200359344482, "lr": 0.0002, "elapsed_sec": 24592.701862335205, "step_time_sec": 8.236549385997932, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.625509738922119, "lr": 0.0002, "elapsed_sec": 24600.94123196602, "step_time_sec": 8.239192425011424, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.732565879821777, "lr": 0.0002, "elapsed_sec": 24609.17829823494, "step_time_sec": 8.236930505023338, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.503669261932373, "lr": 0.0002, "elapsed_sec": 24617.415754556656, "step_time_sec": 8.237256730993977, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.632139205932617, "lr": 0.0002, "elapsed_sec": 24625.652634382248, "step_time_sec": 8.236715467995964, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.5001325607299805, "lr": 0.0002, "elapsed_sec": 24633.890663146973, "step_time_sec": 8.237874135986203, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.49492073059082, "lr": 0.0002, "elapsed_sec": 24642.12749195099, "step_time_sec": 8.236648433987284, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.408182144165039, "lr": 0.0002, "elapsed_sec": 24650.36495232582, "step_time_sec": 8.237286358023994, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.583301067352295, "lr": 0.0002, "elapsed_sec": 24658.604156970978, "step_time_sec": 8.239065793022746, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.557981491088867, "lr": 0.0002, "elapsed_sec": 24666.841989278793, "step_time_sec": 8.237674764008261, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.531000137329102, "lr": 0.0002, "elapsed_sec": 24675.081196784973, "step_time_sec": 8.239070545998402, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.483301639556885, "lr": 0.0002, "elapsed_sec": 24683.319448709488, "step_time_sec": 8.238057701993966, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.659860610961914, "lr": 0.0002, "elapsed_sec": 24691.5557513237, "step_time_sec": 8.236124763992848, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.5004448890686035, "lr": 0.0002, "elapsed_sec": 24699.79346370697, "step_time_sec": 8.237591182987671, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.55046272277832, "lr": 0.0002, "elapsed_sec": 24708.03077173233, "step_time_sec": 8.237172408000333, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.459100246429443, "lr": 0.0002, "elapsed_sec": 24716.269787311554, "step_time_sec": 8.238860574987484, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.512470245361328, "lr": 0.0002, "elapsed_sec": 24724.50798845291, "step_time_sec": 8.238011467998149, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.38498067855835, "lr": 0.0002, "elapsed_sec": 24732.746260404587, "step_time_sec": 8.238135473977309, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.648223876953125, "lr": 0.0002, "elapsed_sec": 24740.984088897705, "step_time_sec": 8.237644169013947, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.766979694366455, "lr": 0.0002, "elapsed_sec": 24749.22051525116, "step_time_sec": 8.236287881009048, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.433403968811035, "lr": 0.0002, "elapsed_sec": 24757.458112478256, "step_time_sec": 8.237392118986463, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.573748588562012, "lr": 0.0002, "elapsed_sec": 24765.69490456581, "step_time_sec": 8.23662691100617, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.547804832458496, "lr": 0.0002, "elapsed_sec": 24773.933432102203, "step_time_sec": 8.238422165974043, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.497898101806641, "lr": 0.0002, "elapsed_sec": 24782.17108774185, "step_time_sec": 8.237439452990657, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.579730033874512, "lr": 0.0002, "elapsed_sec": 24790.412904262543, "step_time_sec": 8.235265005001565, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.5086774826049805, "lr": 0.0002, "elapsed_sec": 24798.65148448944, "step_time_sec": 8.23843859799672, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.462841987609863, "lr": 0.0002, "elapsed_sec": 24806.88832449913, "step_time_sec": 8.236684642004548, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4861016273498535, "lr": 0.0002, "elapsed_sec": 24815.126697063446, "step_time_sec": 8.238237040000968, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.408799171447754, "lr": 0.0002, "elapsed_sec": 24823.364673137665, "step_time_sec": 8.237803051015362, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.442012310028076, "lr": 0.0002, "elapsed_sec": 24831.601144075394, "step_time_sec": 8.236292909015901, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.5218963623046875, "lr": 0.0002, "elapsed_sec": 24839.839165449142, "step_time_sec": 8.237878417014144, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.464473247528076, "lr": 0.0002, "elapsed_sec": 24848.076218366623, "step_time_sec": 8.236886694008717, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.427703380584717, "lr": 0.0002, "elapsed_sec": 24856.31532239914, "step_time_sec": 8.23892917399644, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.550291061401367, "lr": 0.0002, "elapsed_sec": 24864.551486968994, "step_time_sec": 8.236030675994698, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.431862831115723, "lr": 0.0002, "elapsed_sec": 24872.789484739304, "step_time_sec": 8.2378068259859, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.563920021057129, "lr": 0.0002, "elapsed_sec": 24881.02703809738, "step_time_sec": 8.237393813993549, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.6051201820373535, "lr": 0.0002, "elapsed_sec": 24889.264118433, "step_time_sec": 8.236951027996838, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.428435802459717, "lr": 0.0002, "elapsed_sec": 24897.50198340416, "step_time_sec": 8.237645676010288, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.395368576049805, "lr": 0.0002, "elapsed_sec": 24905.738914251328, "step_time_sec": 51.640169226011494, "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": 0.789690564997727, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 3001, "loss": 4.456040859222412, "lr": 0.0002, "elapsed_sec": 24957.388786792755, "step_time_sec": 8.24638945001061, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.455554962158203, "lr": 0.0002, "elapsed_sec": 24965.620810985565, "step_time_sec": 8.23185579199344, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.454599857330322, "lr": 0.0002, "elapsed_sec": 24973.85192990303, "step_time_sec": 8.230958431988256, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.557061672210693, "lr": 0.0002, "elapsed_sec": 24982.088777780533, "step_time_sec": 8.236681323003722, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.7086896896362305, "lr": 0.0002, "elapsed_sec": 24990.32525396347, "step_time_sec": 8.236334070010344, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.532684803009033, "lr": 0.0002, "elapsed_sec": 24998.563475608826, "step_time_sec": 8.238025604980066, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.517436981201172, "lr": 0.0002, "elapsed_sec": 25006.80085515976, "step_time_sec": 8.237227403995348, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.454596042633057, "lr": 0.0002, "elapsed_sec": 25015.03787636757, "step_time_sec": 8.236919062008383, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.546297550201416, "lr": 0.0002, "elapsed_sec": 25023.276425123215, "step_time_sec": 8.238402550981846, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.592620372772217, "lr": 0.0002, "elapsed_sec": 25031.51184487343, "step_time_sec": 8.23518688097829, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.572185516357422, "lr": 0.0002, "elapsed_sec": 25039.749630212784, "step_time_sec": 8.237660496000899, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.422998428344727, "lr": 0.0002, "elapsed_sec": 25047.986000537872, "step_time_sec": 8.236169748997781, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4884185791015625, "lr": 0.0002, "elapsed_sec": 25056.224757909775, "step_time_sec": 8.238578809978208, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.50895357131958, "lr": 0.0002, "elapsed_sec": 25064.46235394478, "step_time_sec": 8.237460516014835, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.423157691955566, "lr": 0.0002, "elapsed_sec": 25072.700432300568, "step_time_sec": 8.237883413996315, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.54414701461792, "lr": 0.0002, "elapsed_sec": 25080.93963074684, "step_time_sec": 8.239035442005843, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.460766315460205, "lr": 0.0002, "elapsed_sec": 25089.17666912079, "step_time_sec": 8.236876937997295, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.592531204223633, "lr": 0.0002, "elapsed_sec": 25097.413806676865, "step_time_sec": 8.236956948006991, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.570296764373779, "lr": 0.0002, "elapsed_sec": 25105.650779247284, "step_time_sec": 8.236803651991067, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.320363521575928, "lr": 0.0002, "elapsed_sec": 25113.88849401474, "step_time_sec": 8.237602332024835, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4661102294921875, "lr": 0.0002, "elapsed_sec": 25122.126707315445, "step_time_sec": 8.238033527013613, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4245405197143555, "lr": 0.0002, "elapsed_sec": 25130.364859819412, "step_time_sec": 8.237979483004892, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.402142524719238, "lr": 0.0002, "elapsed_sec": 25138.604811906815, "step_time_sec": 8.239771766006015, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.414217948913574, "lr": 0.0002, "elapsed_sec": 25146.84242463112, "step_time_sec": 8.237448255997151, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.75179386138916, "lr": 0.0002, "elapsed_sec": 25155.08027935028, "step_time_sec": 8.237710496003274, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.470695495605469, "lr": 0.0002, "elapsed_sec": 25163.317514657974, "step_time_sec": 8.237069570022868, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.369957447052002, "lr": 0.0002, "elapsed_sec": 25171.554318904877, "step_time_sec": 8.236616340000182, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.412142753601074, "lr": 0.0002, "elapsed_sec": 25179.79365849495, "step_time_sec": 8.239210828003706, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.481240749359131, "lr": 0.0002, "elapsed_sec": 25188.030924081802, "step_time_sec": 8.237087874993449, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.560093879699707, "lr": 0.0002, "elapsed_sec": 25196.26943206787, "step_time_sec": 8.238372302992502, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.530736446380615, "lr": 0.0002, "elapsed_sec": 25204.507215976715, "step_time_sec": 8.237606076989323, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.486911296844482, "lr": 0.0002, "elapsed_sec": 25212.744652986526, "step_time_sec": 8.237259472021833, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.591207027435303, "lr": 0.0002, "elapsed_sec": 25220.980813980103, "step_time_sec": 8.235970023000846, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4041748046875, "lr": 0.0002, "elapsed_sec": 25229.21694302559, "step_time_sec": 8.235996873001568, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.485774517059326, "lr": 0.0002, "elapsed_sec": 25237.4549973011, "step_time_sec": 8.237890437012538, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.656661510467529, "lr": 0.0002, "elapsed_sec": 25245.692267417908, "step_time_sec": 8.237069496011827, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.541219234466553, "lr": 0.0002, "elapsed_sec": 25253.930924892426, "step_time_sec": 8.238570465007797, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.517388343811035, "lr": 0.0002, "elapsed_sec": 25262.16774368286, "step_time_sec": 8.236596786999144, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.512321472167969, "lr": 0.0002, "elapsed_sec": 25270.40622830391, "step_time_sec": 8.238325634010835, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.490423202514648, "lr": 0.0002, "elapsed_sec": 25278.64413166046, "step_time_sec": 8.237701815000037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.328341960906982, "lr": 0.0002, "elapsed_sec": 25286.879930496216, "step_time_sec": 8.235647093999432, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4867353439331055, "lr": 0.0002, "elapsed_sec": 25295.117822408676, "step_time_sec": 8.237757508002687, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.377402305603027, "lr": 0.0002, "elapsed_sec": 25303.354373931885, "step_time_sec": 8.23639578098664, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.452484130859375, "lr": 0.0002, "elapsed_sec": 25311.59209227562, "step_time_sec": 8.237561254005414, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.559521198272705, "lr": 0.0002, "elapsed_sec": 25319.832109212875, "step_time_sec": 8.239877012994839, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.453537464141846, "lr": 0.0002, "elapsed_sec": 25328.070509195328, "step_time_sec": 8.238233072013827, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.423774242401123, "lr": 0.0002, "elapsed_sec": 25336.3092148304, "step_time_sec": 8.238522702013142, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.500321865081787, "lr": 0.0002, "elapsed_sec": 25344.54562473297, "step_time_sec": 8.23628548902343, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.597530841827393, "lr": 0.0002, "elapsed_sec": 25352.785793304443, "step_time_sec": 8.240006607986288, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.484714508056641, "lr": 0.0002, "elapsed_sec": 25361.023638248444, "step_time_sec": 8.237613110017264, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.596338748931885, "lr": 0.0002, "elapsed_sec": 25369.26186490059, "step_time_sec": 8.23807901298278, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.456890106201172, "lr": 0.0002, "elapsed_sec": 25377.50048804283, "step_time_sec": 8.238461738015758, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.366283893585205, "lr": 0.0002, "elapsed_sec": 25385.737360477448, "step_time_sec": 8.23673241600045, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.54675817489624, "lr": 0.0002, "elapsed_sec": 25393.973680734634, "step_time_sec": 8.236138830980053, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4114532470703125, "lr": 0.0002, "elapsed_sec": 25402.210861682892, "step_time_sec": 8.237004316004459, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.560720443725586, "lr": 0.0002, "elapsed_sec": 25410.4490339756, "step_time_sec": 8.238024649006547, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.435818672180176, "lr": 0.0002, "elapsed_sec": 25418.686131954193, "step_time_sec": 8.23692253298941, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.56576681137085, "lr": 0.0002, "elapsed_sec": 25426.923269033432, "step_time_sec": 8.236956378998002, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4681396484375, "lr": 0.0002, "elapsed_sec": 25435.161064386368, "step_time_sec": 8.237642336025601, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.353123664855957, "lr": 0.0002, "elapsed_sec": 25443.39658522606, "step_time_sec": 8.23537099501118, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.289690971374512, "lr": 0.0002, "elapsed_sec": 25451.634373426437, "step_time_sec": 8.237614946003305, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.459129810333252, "lr": 0.0002, "elapsed_sec": 25459.8714094162, "step_time_sec": 8.236904976016376, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.496551513671875, "lr": 0.0002, "elapsed_sec": 25468.10930466652, "step_time_sec": 8.23771905200556, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.550938129425049, "lr": 0.0002, "elapsed_sec": 25476.346202611923, "step_time_sec": 8.23672039899975, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.469448089599609, "lr": 0.0002, "elapsed_sec": 25484.583882331848, "step_time_sec": 8.237544657982653, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.473217010498047, "lr": 0.0002, "elapsed_sec": 25492.82250070572, "step_time_sec": 8.238435251027113, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.662267208099365, "lr": 0.0002, "elapsed_sec": 25501.059386730194, "step_time_sec": 8.236717724008486, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.506896018981934, "lr": 0.0002, "elapsed_sec": 25509.296681404114, "step_time_sec": 8.237121573998593, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2494659423828125, "lr": 0.0002, "elapsed_sec": 25517.53423166275, "step_time_sec": 8.237395650998224, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.357293605804443, "lr": 0.0002, "elapsed_sec": 25525.77322268486, "step_time_sec": 8.238835398980882, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.545226573944092, "lr": 0.0002, "elapsed_sec": 25534.010758638382, "step_time_sec": 8.237335546989925, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.493772983551025, "lr": 0.0002, "elapsed_sec": 25542.247241973877, "step_time_sec": 8.236322153010406, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.480663776397705, "lr": 0.0002, "elapsed_sec": 25550.486473083496, "step_time_sec": 8.239080399012892, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.479281425476074, "lr": 0.0002, "elapsed_sec": 25558.722722530365, "step_time_sec": 8.236119097011397, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.363195896148682, "lr": 0.0002, "elapsed_sec": 25566.961004972458, "step_time_sec": 8.238058342016302, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.362518787384033, "lr": 0.0002, "elapsed_sec": 25575.197459459305, "step_time_sec": 8.23632364498917, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.428353309631348, "lr": 0.0002, "elapsed_sec": 25583.43459868431, "step_time_sec": 8.236970847996417, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.407493591308594, "lr": 0.0002, "elapsed_sec": 25591.671547412872, "step_time_sec": 8.236793704010779, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.453769207000732, "lr": 0.0002, "elapsed_sec": 25599.90906739235, "step_time_sec": 8.237390413996764, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.38165283203125, "lr": 0.0002, "elapsed_sec": 25608.14732503891, "step_time_sec": 8.238088701007655, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.493917942047119, "lr": 0.0002, "elapsed_sec": 25616.383764505386, "step_time_sec": 8.236284305981826, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.535855293273926, "lr": 0.0002, "elapsed_sec": 25624.621205806732, "step_time_sec": 8.23722118200385, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.544424057006836, "lr": 0.0002, "elapsed_sec": 25632.8587474823, "step_time_sec": 8.237393829011125, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.377197742462158, "lr": 0.0002, "elapsed_sec": 25641.097692251205, "step_time_sec": 8.238771958014695, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.542360782623291, "lr": 0.0002, "elapsed_sec": 25649.335047245026, "step_time_sec": 8.237229785998352, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.234752178192139, "lr": 0.0002, "elapsed_sec": 25657.571996688843, "step_time_sec": 8.236784644977888, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.545304775238037, "lr": 0.0002, "elapsed_sec": 25665.810653448105, "step_time_sec": 8.238497804995859, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4132537841796875, "lr": 0.0002, "elapsed_sec": 25674.047229766846, "step_time_sec": 8.236436150007648, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4363555908203125, "lr": 0.0002, "elapsed_sec": 25682.285753011703, "step_time_sec": 8.238320729025872, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.313643455505371, "lr": 0.0002, "elapsed_sec": 25690.523857593536, "step_time_sec": 8.237922557018464, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.52100944519043, "lr": 0.0002, "elapsed_sec": 25698.761474847794, "step_time_sec": 8.23752423998667, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.419403553009033, "lr": 0.0002, "elapsed_sec": 25706.999516010284, "step_time_sec": 8.237833733990556, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.48928689956665, "lr": 0.0002, "elapsed_sec": 25715.23728108406, "step_time_sec": 8.237619571998948, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4856462478637695, "lr": 0.0002, "elapsed_sec": 25723.475450992584, "step_time_sec": 8.237962467974285, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.488290309906006, "lr": 0.0002, "elapsed_sec": 25731.713045358658, "step_time_sec": 8.237431481975364, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.440281391143799, "lr": 0.0002, "elapsed_sec": 25739.9519739151, "step_time_sec": 8.238804566004546, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.463126182556152, "lr": 0.0002, "elapsed_sec": 25748.18966269493, "step_time_sec": 8.237532224011375, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.450754165649414, "lr": 0.0002, "elapsed_sec": 25756.426731348038, "step_time_sec": 8.236885215010261, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.397463321685791, "lr": 0.0002, "elapsed_sec": 25764.664111852646, "step_time_sec": 8.237197494017892, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.518917083740234, "lr": 0.0002, "elapsed_sec": 25772.90040254593, "step_time_sec": 8.23614421900129, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.485229969024658, "lr": 0.0002, "elapsed_sec": 25781.139457702637, "step_time_sec": 8.238909270003205, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.496048450469971, "lr": 0.0002, "elapsed_sec": 25789.377611637115, "step_time_sec": 8.237980984995374, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.408238410949707, "lr": 0.0002, "elapsed_sec": 25797.6168320179, "step_time_sec": 8.239035566977691, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.450734615325928, "lr": 0.0002, "elapsed_sec": 25805.854432344437, "step_time_sec": 8.237470322987065, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.33574104309082, "lr": 0.0002, "elapsed_sec": 25814.092175483704, "step_time_sec": 8.23755476201768, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.553284645080566, "lr": 0.0002, "elapsed_sec": 25822.331033229828, "step_time_sec": 8.238683699979447, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.403740882873535, "lr": 0.0002, "elapsed_sec": 25830.56905412674, "step_time_sec": 8.23790015297709, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4163031578063965, "lr": 0.0002, "elapsed_sec": 25838.807128190994, "step_time_sec": 8.237937897007214, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4251556396484375, "lr": 0.0002, "elapsed_sec": 25847.046120882034, "step_time_sec": 8.238773279997986, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.484511375427246, "lr": 0.0002, "elapsed_sec": 25855.28436613083, "step_time_sec": 8.238102589006303, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.518689155578613, "lr": 0.0002, "elapsed_sec": 25863.53008890152, "step_time_sec": 8.23851046198979, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.37257194519043, "lr": 0.0002, "elapsed_sec": 25871.76815223694, "step_time_sec": 8.237910130992532, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.444761276245117, "lr": 0.0002, "elapsed_sec": 25880.005335330963, "step_time_sec": 8.23704710698803, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.365533828735352, "lr": 0.0002, "elapsed_sec": 25888.242044448853, "step_time_sec": 8.236517728015315, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3374810218811035, "lr": 0.0002, "elapsed_sec": 25896.47926092148, "step_time_sec": 8.23706211001263, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.427613258361816, "lr": 0.0002, "elapsed_sec": 25904.717606067657, "step_time_sec": 8.238162337016547, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.573192119598389, "lr": 0.0002, "elapsed_sec": 25912.955576896667, "step_time_sec": 8.237807298020925, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.484570503234863, "lr": 0.0002, "elapsed_sec": 25921.19546341896, "step_time_sec": 8.239773287990829, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3432793617248535, "lr": 0.0002, "elapsed_sec": 25929.433965444565, "step_time_sec": 8.238304510974558, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.524179935455322, "lr": 0.0002, "elapsed_sec": 25937.67050099373, "step_time_sec": 8.236395784013439, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4429168701171875, "lr": 0.0002, "elapsed_sec": 25945.909529685974, "step_time_sec": 8.238892530993326, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.595638275146484, "lr": 0.0002, "elapsed_sec": 25954.146061182022, "step_time_sec": 8.23634923499776, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.530573844909668, "lr": 0.0002, "elapsed_sec": 25962.384380817413, "step_time_sec": 8.23816799800261, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.560223579406738, "lr": 0.0002, "elapsed_sec": 25970.62159347534, "step_time_sec": 8.237007835006807, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.454755783081055, "lr": 0.0002, "elapsed_sec": 25978.86117720604, "step_time_sec": 8.239487410988659, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.415994167327881, "lr": 0.0002, "elapsed_sec": 25987.09934282303, "step_time_sec": 8.23797528698924, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4028639793396, "lr": 0.0002, "elapsed_sec": 25995.3356590271, "step_time_sec": 8.236152408993803, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.451972484588623, "lr": 0.0002, "elapsed_sec": 26003.574798583984, "step_time_sec": 8.23893951100763, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.423138618469238, "lr": 0.0002, "elapsed_sec": 26011.812375068665, "step_time_sec": 8.237416153016966, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.443631649017334, "lr": 0.0002, "elapsed_sec": 26020.05033349991, "step_time_sec": 8.237793823005632, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.513118743896484, "lr": 0.0002, "elapsed_sec": 26028.289434432983, "step_time_sec": 8.238938015012536, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.421885013580322, "lr": 0.0002, "elapsed_sec": 26036.52671313286, "step_time_sec": 8.237134880997473, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.380263328552246, "lr": 0.0002, "elapsed_sec": 26044.76394891739, "step_time_sec": 8.237090144015383, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.304130554199219, "lr": 0.0002, "elapsed_sec": 26053.001598358154, "step_time_sec": 8.237429353990592, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.479930877685547, "lr": 0.0002, "elapsed_sec": 26061.23826265335, "step_time_sec": 8.236516603996279, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.453913688659668, "lr": 0.0002, "elapsed_sec": 26069.477231264114, "step_time_sec": 8.238772073003929, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.433154582977295, "lr": 0.0002, "elapsed_sec": 26077.714636564255, "step_time_sec": 8.237268399010645, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.380215644836426, "lr": 0.0002, "elapsed_sec": 26085.953809261322, "step_time_sec": 8.239025296992622, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.455608367919922, "lr": 0.0002, "elapsed_sec": 26094.190411567688, "step_time_sec": 8.236461637017783, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.462083339691162, "lr": 0.0002, "elapsed_sec": 26102.42843580246, "step_time_sec": 8.237823521980317, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.431964874267578, "lr": 0.0002, "elapsed_sec": 26110.66681599617, "step_time_sec": 8.23819671999081, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4374260902404785, "lr": 0.0002, "elapsed_sec": 26118.90535068512, "step_time_sec": 8.238434557017172, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4458231925964355, "lr": 0.0002, "elapsed_sec": 26127.143325328827, "step_time_sec": 8.237769928993657, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.445958614349365, "lr": 0.0002, "elapsed_sec": 26135.380110025406, "step_time_sec": 8.236667854012921, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.460577011108398, "lr": 0.0002, "elapsed_sec": 26143.61761021614, "step_time_sec": 8.237265226984164, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.408641815185547, "lr": 0.0002, "elapsed_sec": 26151.85376238823, "step_time_sec": 8.236000813980354, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.691962242126465, "lr": 0.0002, "elapsed_sec": 26160.092866659164, "step_time_sec": 8.238971006008796, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.375848770141602, "lr": 0.0002, "elapsed_sec": 26168.330706596375, "step_time_sec": 8.237646511988714, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.224295616149902, "lr": 0.0002, "elapsed_sec": 26176.5701482296, "step_time_sec": 8.239295550010866, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.486769676208496, "lr": 0.0002, "elapsed_sec": 26184.80910873413, "step_time_sec": 8.2387397029961, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.404260635375977, "lr": 0.0002, "elapsed_sec": 26193.045827150345, "step_time_sec": 8.236578871990787, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.583556175231934, "lr": 0.0002, "elapsed_sec": 26201.284683942795, "step_time_sec": 8.238711720012361, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.447826862335205, "lr": 0.0002, "elapsed_sec": 26209.5225417614, "step_time_sec": 8.237652717973106, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.486355781555176, "lr": 0.0002, "elapsed_sec": 26217.760259866714, "step_time_sec": 8.237617615988711, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4600019454956055, "lr": 0.0002, "elapsed_sec": 26225.99928021431, "step_time_sec": 8.238797071011504, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.484230995178223, "lr": 0.0002, "elapsed_sec": 26234.23670077324, "step_time_sec": 8.237248134013498, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3396315574646, "lr": 0.0002, "elapsed_sec": 26242.474537611008, "step_time_sec": 8.237677901022835, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.525191307067871, "lr": 0.0002, "elapsed_sec": 26250.71152472496, "step_time_sec": 8.236822907987516, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.431751728057861, "lr": 0.0002, "elapsed_sec": 26258.949177503586, "step_time_sec": 8.2375443209894, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.58610200881958, "lr": 0.0002, "elapsed_sec": 26267.186681985855, "step_time_sec": 8.23728338399087, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2977681159973145, "lr": 0.0002, "elapsed_sec": 26275.4247443676, "step_time_sec": 8.237921573017957, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.510735511779785, "lr": 0.0002, "elapsed_sec": 26283.662932634354, "step_time_sec": 8.238004840008216, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4872517585754395, "lr": 0.0002, "elapsed_sec": 26291.900213241577, "step_time_sec": 8.23710685200058, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.470458030700684, "lr": 0.0002, "elapsed_sec": 26300.138221025467, "step_time_sec": 8.237868241005344, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.39858865737915, "lr": 0.0002, "elapsed_sec": 26308.37554717064, "step_time_sec": 8.237205829005688, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.504742622375488, "lr": 0.0002, "elapsed_sec": 26316.613509655, "step_time_sec": 8.237779051996768, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.499757289886475, "lr": 0.0002, "elapsed_sec": 26324.851951360703, "step_time_sec": 8.238265472988132, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.38951301574707, "lr": 0.0002, "elapsed_sec": 26333.089565992355, "step_time_sec": 8.237393635994522, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.214654922485352, "lr": 0.0002, "elapsed_sec": 26341.327968358994, "step_time_sec": 8.238288336986443, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3277482986450195, "lr": 0.0002, "elapsed_sec": 26349.56542944908, "step_time_sec": 8.237288831995102, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.534128189086914, "lr": 0.0002, "elapsed_sec": 26357.80298113823, "step_time_sec": 8.237403450999409, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.467008590698242, "lr": 0.0002, "elapsed_sec": 26366.04135990143, "step_time_sec": 8.23820223097573, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3858795166015625, "lr": 0.0002, "elapsed_sec": 26374.279717683792, "step_time_sec": 8.238202459993772, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.5078253746032715, "lr": 0.0002, "elapsed_sec": 26382.51733970642, "step_time_sec": 8.237469696003245, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.498784065246582, "lr": 0.0002, "elapsed_sec": 26390.753954410553, "step_time_sec": 8.236446030001389, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.399644374847412, "lr": 0.0002, "elapsed_sec": 26398.99175643921, "step_time_sec": 8.237661028979346, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.411772727966309, "lr": 0.0002, "elapsed_sec": 26407.230367422104, "step_time_sec": 8.23842226099805, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.559114933013916, "lr": 0.0002, "elapsed_sec": 26415.467993736267, "step_time_sec": 8.237456908012973, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.352529525756836, "lr": 0.0002, "elapsed_sec": 26423.70534992218, "step_time_sec": 8.237164185004076, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.451603412628174, "lr": 0.0002, "elapsed_sec": 26431.94276380539, "step_time_sec": 8.237233452004148, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.538821697235107, "lr": 0.0002, "elapsed_sec": 26440.181467294693, "step_time_sec": 8.238572424015729, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4848222732543945, "lr": 0.0002, "elapsed_sec": 26448.41793179512, "step_time_sec": 8.236309840984177, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.544643878936768, "lr": 0.0002, "elapsed_sec": 26456.655779123306, "step_time_sec": 8.237696292984765, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4149065017700195, "lr": 0.0002, "elapsed_sec": 26464.893043279648, "step_time_sec": 8.237062713014893, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3483076095581055, "lr": 0.0002, "elapsed_sec": 26473.13149166107, "step_time_sec": 8.238295161980204, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.364552974700928, "lr": 0.0002, "elapsed_sec": 26481.369428873062, "step_time_sec": 8.237786072975723, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.490888595581055, "lr": 0.0002, "elapsed_sec": 26489.606016874313, "step_time_sec": 8.236409970995737, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.369624614715576, "lr": 0.0002, "elapsed_sec": 26497.84343981743, "step_time_sec": 8.237273741979152, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.463871955871582, "lr": 0.0002, "elapsed_sec": 26506.079202651978, "step_time_sec": 8.235589750023792, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.400969982147217, "lr": 0.0002, "elapsed_sec": 26514.317314863205, "step_time_sec": 8.238006405998021, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.47036075592041, "lr": 0.0002, "elapsed_sec": 26522.55581831932, "step_time_sec": 8.238298748998204, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.441559314727783, "lr": 0.0002, "elapsed_sec": 26530.794784784317, "step_time_sec": 8.23882568499539, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.391253471374512, "lr": 0.0002, "elapsed_sec": 26539.03321456909, "step_time_sec": 8.238280681980541, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.53718900680542, "lr": 0.0002, "elapsed_sec": 26547.26853632927, "step_time_sec": 8.235146735009039, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4629621505737305, "lr": 0.0002, "elapsed_sec": 26555.507039308548, "step_time_sec": 8.238327728991862, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.441528797149658, "lr": 0.0002, "elapsed_sec": 26563.74374771118, "step_time_sec": 8.236536084994441, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.434421539306641, "lr": 0.0002, "elapsed_sec": 26571.98231601715, "step_time_sec": 8.238400900998386, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.409146308898926, "lr": 0.0002, "elapsed_sec": 26580.220103740692, "step_time_sec": 8.237636075005867, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.458143711090088, "lr": 0.0002, "elapsed_sec": 26588.457135915756, "step_time_sec": 8.236843443010002, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.402343273162842, "lr": 0.0002, "elapsed_sec": 26596.693659305573, "step_time_sec": 8.23636309299036, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.527154922485352, "lr": 0.0002, "elapsed_sec": 26604.930780649185, "step_time_sec": 8.236933195992606, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.562844276428223, "lr": 0.0002, "elapsed_sec": 26613.16916656494, "step_time_sec": 8.23823303199606, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.45238733291626, "lr": 0.0002, "elapsed_sec": 26621.405964374542, "step_time_sec": 8.236658945999807, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.384117603302002, "lr": 0.0002, "elapsed_sec": 26629.643356323242, "step_time_sec": 8.237207822996425, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.493810176849365, "lr": 0.0002, "elapsed_sec": 26637.88236784935, "step_time_sec": 8.238863245001994, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.350779056549072, "lr": 0.0002, "elapsed_sec": 26646.119730472565, "step_time_sec": 8.237177906004945, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.336005687713623, "lr": 0.0002, "elapsed_sec": 26654.358698129654, "step_time_sec": 8.238850608991925, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.287287712097168, "lr": 0.0002, "elapsed_sec": 26662.595507144928, "step_time_sec": 8.23663144200691, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.415332317352295, "lr": 0.0002, "elapsed_sec": 26670.833112955093, "step_time_sec": 8.237438322976232, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.496626377105713, "lr": 0.0002, "elapsed_sec": 26679.07036972046, "step_time_sec": 8.237106045009568, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.397985935211182, "lr": 0.0002, "elapsed_sec": 26687.309198141098, "step_time_sec": 8.238621522003086, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.370641708374023, "lr": 0.0002, "elapsed_sec": 26695.54821896553, "step_time_sec": 8.238858716998948, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3499531745910645, "lr": 0.0002, "elapsed_sec": 26703.783498764038, "step_time_sec": 8.235111910005799, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.418497562408447, "lr": 0.0002, "elapsed_sec": 26712.022307395935, "step_time_sec": 8.238640249997843, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.336768627166748, "lr": 0.0002, "elapsed_sec": 26720.25980901718, "step_time_sec": 8.23730350099504, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.417385578155518, "lr": 0.0002, "elapsed_sec": 26728.499366760254, "step_time_sec": 8.239396641001804, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.436134338378906, "lr": 0.0002, "elapsed_sec": 26736.737398147583, "step_time_sec": 8.23789240300539, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4660515785217285, "lr": 0.0002, "elapsed_sec": 26744.975967168808, "step_time_sec": 8.238375557004474, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.423351287841797, "lr": 0.0002, "elapsed_sec": 26753.21252655983, "step_time_sec": 8.23638390598353, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.418529033660889, "lr": 0.0002, "elapsed_sec": 26761.450223207474, "step_time_sec": 8.237533298000926, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.361666202545166, "lr": 0.0002, "elapsed_sec": 26769.68951201439, "step_time_sec": 8.239132857008372, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.531149387359619, "lr": 0.0002, "elapsed_sec": 26777.927098035812, "step_time_sec": 8.237450432992773, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.523433208465576, "lr": 0.0002, "elapsed_sec": 26786.165885448456, "step_time_sec": 8.238560801983112, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.494561195373535, "lr": 0.0002, "elapsed_sec": 26794.404517889023, "step_time_sec": 8.238490496994928, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4250335693359375, "lr": 0.0002, "elapsed_sec": 26802.641674041748, "step_time_sec": 8.236999198008562, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.565781593322754, "lr": 0.0002, "elapsed_sec": 26810.879236221313, "step_time_sec": 8.237435423012357, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.494845867156982, "lr": 0.0002, "elapsed_sec": 26819.11659359932, "step_time_sec": 8.237200837000273, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.406239032745361, "lr": 0.0002, "elapsed_sec": 26827.35549211502, "step_time_sec": 8.238681529997848, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.391535758972168, "lr": 0.0002, "elapsed_sec": 26835.592791080475, "step_time_sec": 8.237143887992715, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.608323574066162, "lr": 0.0002, "elapsed_sec": 26843.830241918564, "step_time_sec": 8.2372893760039, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.407115936279297, "lr": 0.0002, "elapsed_sec": 26852.069066524506, "step_time_sec": 8.238650229002815, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.603643417358398, "lr": 0.0002, "elapsed_sec": 26860.305870056152, "step_time_sec": 8.236669426987646, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.426272392272949, "lr": 0.0002, "elapsed_sec": 26868.543888807297, "step_time_sec": 8.237849404016742, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.407775402069092, "lr": 0.0002, "elapsed_sec": 26876.780596017838, "step_time_sec": 8.236534652998671, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.413766384124756, "lr": 0.0002, "elapsed_sec": 26885.01901125908, "step_time_sec": 8.238233131007291, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.523046016693115, "lr": 0.0002, "elapsed_sec": 26893.257591485977, "step_time_sec": 8.238455770013388, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.455669403076172, "lr": 0.0002, "elapsed_sec": 26901.49441719055, "step_time_sec": 8.236610498977825, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.354883193969727, "lr": 0.0002, "elapsed_sec": 26909.73294854164, "step_time_sec": 8.23837182801799, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.462370872497559, "lr": 0.0002, "elapsed_sec": 26917.96901154518, "step_time_sec": 8.23597280101967, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.495983600616455, "lr": 0.0002, "elapsed_sec": 26926.207223176956, "step_time_sec": 8.23802952002734, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.469102382659912, "lr": 0.0002, "elapsed_sec": 26934.44487428665, "step_time_sec": 8.237535595981171, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.387712478637695, "lr": 0.0002, "elapsed_sec": 26942.684385299683, "step_time_sec": 8.239283873001114, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.375400066375732, "lr": 0.0002, "elapsed_sec": 26950.97129559517, "step_time_sec": 8.286746312020114, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.401611804962158, "lr": 0.0002, "elapsed_sec": 26959.208681583405, "step_time_sec": 8.237229421007214, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.461455345153809, "lr": 0.0002, "elapsed_sec": 26972.721284866333, "step_time_sec": 13.512458204000723, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.418726921081543, "lr": 0.0002, "elapsed_sec": 26980.956069469452, "step_time_sec": 8.234570066997549, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.393199443817139, "lr": 0.0002, "elapsed_sec": 26989.1916577816, "step_time_sec": 8.23544359600055, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3763933181762695, "lr": 0.0002, "elapsed_sec": 26997.429698228836, "step_time_sec": 8.237828423996689, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.532382965087891, "lr": 0.0002, "elapsed_sec": 27005.665098905563, "step_time_sec": 8.23520560999168, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.356425762176514, "lr": 0.0002, "elapsed_sec": 27013.90220928192, "step_time_sec": 8.237004506023368, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.357426166534424, "lr": 0.0002, "elapsed_sec": 27022.139505147934, "step_time_sec": 8.237090260983678, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.483375549316406, "lr": 0.0002, "elapsed_sec": 27030.37793469429, "step_time_sec": 8.238262119004503, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.566679954528809, "lr": 0.0002, "elapsed_sec": 27038.616154670715, "step_time_sec": 8.238090615021065, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.480968475341797, "lr": 0.0002, "elapsed_sec": 27046.855728149414, "step_time_sec": 8.239369171991711, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.589336395263672, "lr": 0.0002, "elapsed_sec": 27055.094147205353, "step_time_sec": 8.238234440010274, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.381689548492432, "lr": 0.0002, "elapsed_sec": 27063.33123421669, "step_time_sec": 8.236918803013396, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.573489665985107, "lr": 0.0002, "elapsed_sec": 27071.56973719597, "step_time_sec": 8.23833014201955, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.485418319702148, "lr": 0.0002, "elapsed_sec": 27079.80725646019, "step_time_sec": 8.23736826798995, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.632343769073486, "lr": 0.0002, "elapsed_sec": 27088.043857097626, "step_time_sec": 8.236455567996018, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4348978996276855, "lr": 0.0002, "elapsed_sec": 27096.28117084503, "step_time_sec": 8.237170406995574, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.412834167480469, "lr": 0.0002, "elapsed_sec": 27104.51944255829, "step_time_sec": 8.238053030014271, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.445314884185791, "lr": 0.0002, "elapsed_sec": 27112.757104873657, "step_time_sec": 8.237521106988424, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.446825981140137, "lr": 0.0002, "elapsed_sec": 27120.994539022446, "step_time_sec": 8.237264417984989, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.288583755493164, "lr": 0.0002, "elapsed_sec": 27129.232035160065, "step_time_sec": 8.2373430239968, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.483143329620361, "lr": 0.0002, "elapsed_sec": 27137.46914744377, "step_time_sec": 8.236900894000428, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.349587917327881, "lr": 0.0002, "elapsed_sec": 27145.70764350891, "step_time_sec": 8.238311214983696, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.504515647888184, "lr": 0.0002, "elapsed_sec": 27153.946395874023, "step_time_sec": 8.238630534004187, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.442590236663818, "lr": 0.0002, "elapsed_sec": 27162.18393445015, "step_time_sec": 8.237363719003042, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.469409465789795, "lr": 0.0002, "elapsed_sec": 27170.421139240265, "step_time_sec": 8.237049893999938, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.473818302154541, "lr": 0.0002, "elapsed_sec": 27178.658163785934, "step_time_sec": 8.236868225998478, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.40060567855835, "lr": 0.0002, "elapsed_sec": 27186.89662694931, "step_time_sec": 8.23833114898298, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4386067390441895, "lr": 0.0002, "elapsed_sec": 27195.13462662697, "step_time_sec": 8.23783821699908, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.367623329162598, "lr": 0.0002, "elapsed_sec": 27203.373415231705, "step_time_sec": 8.238584658014588, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.396994590759277, "lr": 0.0002, "elapsed_sec": 27211.6114962101, "step_time_sec": 8.237930639996193, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.394608020782471, "lr": 0.0002, "elapsed_sec": 27219.8475151062, "step_time_sec": 8.235848178999731, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.296870708465576, "lr": 0.0002, "elapsed_sec": 27228.084712982178, "step_time_sec": 8.237067287001992, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.370778560638428, "lr": 0.0002, "elapsed_sec": 27236.32180571556, "step_time_sec": 8.236944334988948, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.597590923309326, "lr": 0.0002, "elapsed_sec": 27244.561145544052, "step_time_sec": 8.239137903990922, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.610589504241943, "lr": 0.0002, "elapsed_sec": 27252.80030965805, "step_time_sec": 8.239011043013306, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.524425506591797, "lr": 0.0002, "elapsed_sec": 27261.038428544998, "step_time_sec": 8.237979632016504, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.398649215698242, "lr": 0.0002, "elapsed_sec": 27269.277401685715, "step_time_sec": 8.238833712995984, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.401309490203857, "lr": 0.0002, "elapsed_sec": 27277.513115882874, "step_time_sec": 8.235514069994679, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.447784900665283, "lr": 0.0002, "elapsed_sec": 27285.750559568405, "step_time_sec": 8.237253400002373, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.446515083312988, "lr": 0.0002, "elapsed_sec": 27293.986518383026, "step_time_sec": 8.235846789990319, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.398046493530273, "lr": 0.0002, "elapsed_sec": 27302.22691512108, "step_time_sec": 8.240183129993966, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.600822448730469, "lr": 0.0002, "elapsed_sec": 27310.465109586716, "step_time_sec": 8.237976502015954, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3386712074279785, "lr": 0.0002, "elapsed_sec": 27318.701645851135, "step_time_sec": 8.23636811299366, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.349923610687256, "lr": 0.0002, "elapsed_sec": 27326.94019651413, "step_time_sec": 8.238400139991427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.409414768218994, "lr": 0.0002, "elapsed_sec": 27335.177186250687, "step_time_sec": 8.236829676985508, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.474842071533203, "lr": 0.0002, "elapsed_sec": 27343.414956092834, "step_time_sec": 8.237587501003873, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.479452610015869, "lr": 0.0002, "elapsed_sec": 27351.65217733383, "step_time_sec": 8.237045042013051, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.509182929992676, "lr": 0.0002, "elapsed_sec": 27359.89076590538, "step_time_sec": 8.238421398011269, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.420836925506592, "lr": 0.0002, "elapsed_sec": 27368.12945318222, "step_time_sec": 8.238555080984952, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.498167514801025, "lr": 0.0002, "elapsed_sec": 27376.36670446396, "step_time_sec": 8.237117273005424, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.501334190368652, "lr": 0.0002, "elapsed_sec": 27384.604714632034, "step_time_sec": 8.237780663010199, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.323940277099609, "lr": 0.0002, "elapsed_sec": 27392.84146118164, "step_time_sec": 8.236583135992987, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.375256061553955, "lr": 0.0002, "elapsed_sec": 27401.0782828331, "step_time_sec": 8.236704634997295, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.529500484466553, "lr": 0.0002, "elapsed_sec": 27409.316239118576, "step_time_sec": 8.237777293019462, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.5744709968566895, "lr": 0.0002, "elapsed_sec": 27417.553693771362, "step_time_sec": 8.237327316979645, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.303946495056152, "lr": 0.0002, "elapsed_sec": 27425.790662765503, "step_time_sec": 8.23673224702361, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2923784255981445, "lr": 0.0002, "elapsed_sec": 27434.02738928795, "step_time_sec": 8.236585142003605, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.487064838409424, "lr": 0.0002, "elapsed_sec": 27442.265913009644, "step_time_sec": 8.238387245015474, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4903388023376465, "lr": 0.0002, "elapsed_sec": 27450.50347685814, "step_time_sec": 8.237394791998668, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.463159084320068, "lr": 0.0002, "elapsed_sec": 27458.741568803787, "step_time_sec": 8.237916565994965, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.396214962005615, "lr": 0.0002, "elapsed_sec": 27466.979731798172, "step_time_sec": 8.238011957000708, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.552412033081055, "lr": 0.0002, "elapsed_sec": 27475.216181993484, "step_time_sec": 8.23629821499344, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3814897537231445, "lr": 0.0002, "elapsed_sec": 27483.4549844265, "step_time_sec": 8.238598603988066, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.367743492126465, "lr": 0.0002, "elapsed_sec": 27491.692130565643, "step_time_sec": 8.23700053000357, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.377546787261963, "lr": 0.0002, "elapsed_sec": 27499.931035280228, "step_time_sec": 8.238771890988573, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.429981708526611, "lr": 0.0002, "elapsed_sec": 27508.169155597687, "step_time_sec": 8.237946818990167, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.333934307098389, "lr": 0.0002, "elapsed_sec": 27516.406242370605, "step_time_sec": 8.236879814008716, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.421767711639404, "lr": 0.0002, "elapsed_sec": 27524.64455962181, "step_time_sec": 8.23811549297534, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.445779323577881, "lr": 0.0002, "elapsed_sec": 27532.88098502159, "step_time_sec": 8.236252956005046, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.313858985900879, "lr": 0.0002, "elapsed_sec": 27541.120141744614, "step_time_sec": 8.239013709011488, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.502713680267334, "lr": 0.0002, "elapsed_sec": 27549.35648560524, "step_time_sec": 8.236193904012907, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.342052936553955, "lr": 0.0002, "elapsed_sec": 27557.59429860115, "step_time_sec": 8.237648188980529, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.465911388397217, "lr": 0.0002, "elapsed_sec": 27565.831357479095, "step_time_sec": 8.23689024298801, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.385291576385498, "lr": 0.0002, "elapsed_sec": 27574.070534706116, "step_time_sec": 8.238966067990987, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.490553855895996, "lr": 0.0002, "elapsed_sec": 27582.308240890503, "step_time_sec": 8.23757431001286, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.454298496246338, "lr": 0.0002, "elapsed_sec": 27590.54403591156, "step_time_sec": 8.23561643800349, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.491069793701172, "lr": 0.0002, "elapsed_sec": 27598.78527736664, "step_time_sec": 8.241132713010302, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4281229972839355, "lr": 0.0002, "elapsed_sec": 27607.022124767303, "step_time_sec": 8.236620715004392, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.463908672332764, "lr": 0.0002, "elapsed_sec": 27615.259625196457, "step_time_sec": 8.23735877397121, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.463930130004883, "lr": 0.0002, "elapsed_sec": 27623.496885299683, "step_time_sec": 8.237043670000276, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.38904333114624, "lr": 0.0002, "elapsed_sec": 27631.73353075981, "step_time_sec": 8.23651962299482, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.363987922668457, "lr": 0.0002, "elapsed_sec": 27639.971996068954, "step_time_sec": 8.23831210201024, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.5042724609375, "lr": 0.0002, "elapsed_sec": 27648.208577871323, "step_time_sec": 8.236384144984186, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.281394958496094, "lr": 0.0002, "elapsed_sec": 27656.447463989258, "step_time_sec": 8.238751541997772, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.485844135284424, "lr": 0.0002, "elapsed_sec": 27664.68452692032, "step_time_sec": 8.236846403015079, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.465339183807373, "lr": 0.0002, "elapsed_sec": 27672.922015428543, "step_time_sec": 8.237340845022118, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.538693428039551, "lr": 0.0002, "elapsed_sec": 27681.160924434662, "step_time_sec": 8.238757898012409, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.458910942077637, "lr": 0.0002, "elapsed_sec": 27689.398841381073, "step_time_sec": 8.237753597000847, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.38608980178833, "lr": 0.0002, "elapsed_sec": 27697.636487960815, "step_time_sec": 8.23750303601264, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.285592555999756, "lr": 0.0002, "elapsed_sec": 27705.872252225876, "step_time_sec": 8.235571285011247, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.350069999694824, "lr": 0.0002, "elapsed_sec": 27714.110730171204, "step_time_sec": 8.238363515003584, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.353332996368408, "lr": 0.0002, "elapsed_sec": 27722.349141836166, "step_time_sec": 8.238229723006953, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.348150253295898, "lr": 0.0002, "elapsed_sec": 27730.58639883995, "step_time_sec": 8.237101990991505, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.235888481140137, "lr": 0.0002, "elapsed_sec": 27738.825530290604, "step_time_sec": 8.238925962010399, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.309754371643066, "lr": 0.0002, "elapsed_sec": 27747.06225514412, "step_time_sec": 8.23656189799658, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.304317474365234, "lr": 0.0002, "elapsed_sec": 27755.300472974777, "step_time_sec": 8.238063523021992, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.430187702178955, "lr": 0.0002, "elapsed_sec": 27763.537425994873, "step_time_sec": 8.236826509004459, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.467700481414795, "lr": 0.0002, "elapsed_sec": 27771.777703523636, "step_time_sec": 8.240088460006518, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3351640701293945, "lr": 0.0002, "elapsed_sec": 27780.015717744827, "step_time_sec": 8.237822192022577, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3316240310668945, "lr": 0.0002, "elapsed_sec": 27788.2533288002, "step_time_sec": 8.237471013999311, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.413088321685791, "lr": 0.0002, "elapsed_sec": 27796.49175453186, "step_time_sec": 8.238269750028849, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.452693462371826, "lr": 0.0002, "elapsed_sec": 27804.728432893753, "step_time_sec": 8.236497764999513, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.361361026763916, "lr": 0.0002, "elapsed_sec": 27812.96412587166, "step_time_sec": 8.235507673001848, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.363038539886475, "lr": 0.0002, "elapsed_sec": 27821.20102930069, "step_time_sec": 8.23676279297797, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.500751495361328, "lr": 0.0002, "elapsed_sec": 27829.43916606903, "step_time_sec": 8.237946197972633, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.373816013336182, "lr": 0.0002, "elapsed_sec": 27837.677526474, "step_time_sec": 8.238215216988465, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.475759029388428, "lr": 0.0002, "elapsed_sec": 27845.91530942917, "step_time_sec": 8.237591053009965, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.416328430175781, "lr": 0.0002, "elapsed_sec": 27854.153479337692, "step_time_sec": 8.238013769994723, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.370091438293457, "lr": 0.0002, "elapsed_sec": 27862.391870737076, "step_time_sec": 8.238248667010339, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.440957069396973, "lr": 0.0002, "elapsed_sec": 27870.630972862244, "step_time_sec": 8.238968452002155, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.421606540679932, "lr": 0.0002, "elapsed_sec": 27878.87037587166, "step_time_sec": 8.23926553101046, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.360955238342285, "lr": 0.0002, "elapsed_sec": 27887.107447624207, "step_time_sec": 8.23683282500133, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.372060298919678, "lr": 0.0002, "elapsed_sec": 27895.346578121185, "step_time_sec": 8.23898173199268, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.5335283279418945, "lr": 0.0002, "elapsed_sec": 27903.583953142166, "step_time_sec": 8.237208338017808, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.29362678527832, "lr": 0.0002, "elapsed_sec": 27911.822295188904, "step_time_sec": 8.238207527989289, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.40012788772583, "lr": 0.0002, "elapsed_sec": 27920.059827566147, "step_time_sec": 8.237336600985145, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.452895641326904, "lr": 0.0002, "elapsed_sec": 27928.297731637955, "step_time_sec": 8.237727950006956, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.416973114013672, "lr": 0.0002, "elapsed_sec": 27936.53599333763, "step_time_sec": 8.238094524014741, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.407390117645264, "lr": 0.0002, "elapsed_sec": 27944.773193359375, "step_time_sec": 8.237056729994947, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.537154674530029, "lr": 0.0002, "elapsed_sec": 27953.01119184494, "step_time_sec": 8.237810421996983, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.394865036010742, "lr": 0.0002, "elapsed_sec": 27961.250117778778, "step_time_sec": 8.238793199008796, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.432243824005127, "lr": 0.0002, "elapsed_sec": 27969.488700151443, "step_time_sec": 8.238406162010506, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.545205593109131, "lr": 0.0002, "elapsed_sec": 27977.725836277008, "step_time_sec": 8.236961692979094, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.442507266998291, "lr": 0.0002, "elapsed_sec": 27985.963417053223, "step_time_sec": 8.237369973008754, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.285163402557373, "lr": 0.0002, "elapsed_sec": 27994.20107650757, "step_time_sec": 8.237475946021732, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.414118766784668, "lr": 0.0002, "elapsed_sec": 28002.438912153244, "step_time_sec": 8.237655056000222, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.330008029937744, "lr": 0.0002, "elapsed_sec": 28010.67891216278, "step_time_sec": 8.239836376014864, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.39773416519165, "lr": 0.0002, "elapsed_sec": 28018.918192863464, "step_time_sec": 8.23916955300956, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.313744068145752, "lr": 0.0002, "elapsed_sec": 28027.155400037766, "step_time_sec": 8.23699936401681, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.502163887023926, "lr": 0.0002, "elapsed_sec": 28035.39429283142, "step_time_sec": 8.238739137013908, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.361230850219727, "lr": 0.0002, "elapsed_sec": 28043.6301176548, "step_time_sec": 8.235666336026043, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.46663761138916, "lr": 0.0002, "elapsed_sec": 28051.8679959774, "step_time_sec": 8.237726271996507, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.45913553237915, "lr": 0.0002, "elapsed_sec": 28060.10594201088, "step_time_sec": 8.237754731992027, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.397537708282471, "lr": 0.0002, "elapsed_sec": 28068.34305024147, "step_time_sec": 8.236949586018454, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.31104040145874, "lr": 0.0002, "elapsed_sec": 28076.581729650497, "step_time_sec": 8.23856583898305, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.473084926605225, "lr": 0.0002, "elapsed_sec": 28084.819522619247, "step_time_sec": 8.237590144999558, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.441870212554932, "lr": 0.0002, "elapsed_sec": 28093.057865858078, "step_time_sec": 8.23820460299612, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4173126220703125, "lr": 0.0002, "elapsed_sec": 28101.294929504395, "step_time_sec": 8.23689388600178, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.403237342834473, "lr": 0.0002, "elapsed_sec": 28109.534175395966, "step_time_sec": 8.239079939987278, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.254983901977539, "lr": 0.0002, "elapsed_sec": 28117.773317098618, "step_time_sec": 8.238965893979184, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.543445110321045, "lr": 0.0002, "elapsed_sec": 28126.009934425354, "step_time_sec": 8.236421596026048, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3370490074157715, "lr": 0.0002, "elapsed_sec": 28134.248561143875, "step_time_sec": 8.23855510901194, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.339327812194824, "lr": 0.0002, "elapsed_sec": 28142.486535787582, "step_time_sec": 8.237922175991116, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.292534351348877, "lr": 0.0002, "elapsed_sec": 28150.727106571198, "step_time_sec": 8.240252880990738, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4637274742126465, "lr": 0.0002, "elapsed_sec": 28158.96643781662, "step_time_sec": 8.239159712014953, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4584856033325195, "lr": 0.0002, "elapsed_sec": 28167.20393180847, "step_time_sec": 8.23730833400623, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.356819152832031, "lr": 0.0002, "elapsed_sec": 28175.442291736603, "step_time_sec": 8.238193797995336, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3325958251953125, "lr": 0.0002, "elapsed_sec": 28183.679726839066, "step_time_sec": 8.237264477997087, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.520047664642334, "lr": 0.0002, "elapsed_sec": 28191.91879582405, "step_time_sec": 8.238922405987978, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.411490440368652, "lr": 0.0002, "elapsed_sec": 28200.15739583969, "step_time_sec": 8.238419522996992, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.316840171813965, "lr": 0.0002, "elapsed_sec": 28208.395594358444, "step_time_sec": 8.238038790004794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.387576580047607, "lr": 0.0002, "elapsed_sec": 28216.63409805298, "step_time_sec": 8.238340876006987, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.414055347442627, "lr": 0.0002, "elapsed_sec": 28224.870058774948, "step_time_sec": 8.235837631014874, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.367653846740723, "lr": 0.0002, "elapsed_sec": 28233.108972787857, "step_time_sec": 8.238778241007822, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.551064491271973, "lr": 0.0002, "elapsed_sec": 28241.346297979355, "step_time_sec": 8.237159779004287, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4676194190979, "lr": 0.0002, "elapsed_sec": 28249.584283590317, "step_time_sec": 8.23779645600007, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.327559471130371, "lr": 0.0002, "elapsed_sec": 28257.823530435562, "step_time_sec": 8.239130565023515, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4092116355896, "lr": 0.0002, "elapsed_sec": 28266.06023645401, "step_time_sec": 8.236492209980497, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.378715991973877, "lr": 0.0002, "elapsed_sec": 28274.298718214035, "step_time_sec": 8.238292574998923, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.474677562713623, "lr": 0.0002, "elapsed_sec": 28282.53652191162, "step_time_sec": 8.237662658008048, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.375843048095703, "lr": 0.0002, "elapsed_sec": 28290.77543449402, "step_time_sec": 8.238752976001706, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.292738437652588, "lr": 0.0002, "elapsed_sec": 28299.01425218582, "step_time_sec": 8.238722781010438, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3773417472839355, "lr": 0.0002, "elapsed_sec": 28307.25200009346, "step_time_sec": 8.23754942099913, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.459635257720947, "lr": 0.0002, "elapsed_sec": 28315.489649772644, "step_time_sec": 8.237484640005277, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.428633213043213, "lr": 0.0002, "elapsed_sec": 28323.727924108505, "step_time_sec": 8.238074620021507, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.383819103240967, "lr": 0.0002, "elapsed_sec": 28331.966291427612, "step_time_sec": 8.238246758992318, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.543961524963379, "lr": 0.0002, "elapsed_sec": 28340.203971862793, "step_time_sec": 8.237474836001638, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.30269718170166, "lr": 0.0002, "elapsed_sec": 28348.442655086517, "step_time_sec": 8.238530934992013, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.417938709259033, "lr": 0.0002, "elapsed_sec": 28356.6800262928, "step_time_sec": 8.237180247990182, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.417739391326904, "lr": 0.0002, "elapsed_sec": 28364.916053295135, "step_time_sec": 8.235912343981909, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.361861705780029, "lr": 0.0002, "elapsed_sec": 28373.153304100037, "step_time_sec": 8.237104917003307, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.285589694976807, "lr": 0.0002, "elapsed_sec": 28381.3902695179, "step_time_sec": 8.236783134023426, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.386926651000977, "lr": 0.0002, "elapsed_sec": 28389.62837743759, "step_time_sec": 8.237912873999448, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2520623207092285, "lr": 0.0002, "elapsed_sec": 28397.86516070366, "step_time_sec": 8.236650092003401, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.438873767852783, "lr": 0.0002, "elapsed_sec": 28406.103945970535, "step_time_sec": 8.238606617000187, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.505037307739258, "lr": 0.0002, "elapsed_sec": 28414.342332839966, "step_time_sec": 8.23822009502328, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.384033679962158, "lr": 0.0002, "elapsed_sec": 28422.580245256424, "step_time_sec": 8.237794472981477, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.343587875366211, "lr": 0.0002, "elapsed_sec": 28430.81795310974, "step_time_sec": 8.23748211900238, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.308926105499268, "lr": 0.0002, "elapsed_sec": 28439.05534219742, "step_time_sec": 8.237299700005678, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.414224624633789, "lr": 0.0002, "elapsed_sec": 28447.293665409088, "step_time_sec": 8.238134503015317, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.519930839538574, "lr": 0.0002, "elapsed_sec": 28455.531538963318, "step_time_sec": 8.237669823982287, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4107794761657715, "lr": 0.0002, "elapsed_sec": 28463.769142389297, "step_time_sec": 8.237446691986406, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.413518905639648, "lr": 0.0002, "elapsed_sec": 28472.007666826248, "step_time_sec": 8.238402220013086, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.334877014160156, "lr": 0.0002, "elapsed_sec": 28480.24512529373, "step_time_sec": 8.237295947998064, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.492735862731934, "lr": 0.0002, "elapsed_sec": 28488.483542203903, "step_time_sec": 8.238243005005643, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.391277313232422, "lr": 0.0002, "elapsed_sec": 28496.72142481804, "step_time_sec": 8.237671053997474, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.315216541290283, "lr": 0.0002, "elapsed_sec": 28504.957526683807, "step_time_sec": 8.235900438012322, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.219366073608398, "lr": 0.0002, "elapsed_sec": 28513.19609951973, "step_time_sec": 8.238474095007405, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.487746715545654, "lr": 0.0002, "elapsed_sec": 28521.433233737946, "step_time_sec": 8.236928305006586, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.224540710449219, "lr": 0.0002, "elapsed_sec": 28529.671115398407, "step_time_sec": 8.237696325988509, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.37473726272583, "lr": 0.0002, "elapsed_sec": 28537.908813238144, "step_time_sec": 8.237574152997695, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.270147323608398, "lr": 0.0002, "elapsed_sec": 28546.146951675415, "step_time_sec": 8.237963377992855, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.564642429351807, "lr": 0.0002, "elapsed_sec": 28554.38436484337, "step_time_sec": 8.237248737976188, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.335887432098389, "lr": 0.0002, "elapsed_sec": 28562.62199282646, "step_time_sec": 8.237499832001049, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.428789138793945, "lr": 0.0002, "elapsed_sec": 28570.85974764824, "step_time_sec": 8.237573891005013, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.5275726318359375, "lr": 0.0002, "elapsed_sec": 28579.096526622772, "step_time_sec": 8.236684423987754, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.259210586547852, "lr": 0.0002, "elapsed_sec": 28587.334854602814, "step_time_sec": 8.238128867000341, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.434814929962158, "lr": 0.0002, "elapsed_sec": 28595.571716070175, "step_time_sec": 8.236654493026435, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.300678253173828, "lr": 0.0002, "elapsed_sec": 28603.81024169922, "step_time_sec": 8.238413123995997, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.347679138183594, "lr": 0.0002, "elapsed_sec": 28612.048179388046, "step_time_sec": 8.237726347986609, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.372876167297363, "lr": 0.0002, "elapsed_sec": 28620.285133600235, "step_time_sec": 8.236782033985946, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.33601713180542, "lr": 0.0002, "elapsed_sec": 28628.522985219955, "step_time_sec": 8.237693075992865, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.454608917236328, "lr": 0.0002, "elapsed_sec": 28636.759531497955, "step_time_sec": 8.236402533017099, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.418458938598633, "lr": 0.0002, "elapsed_sec": 28644.999013900757, "step_time_sec": 8.239302183996188, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.537985801696777, "lr": 0.0002, "elapsed_sec": 28653.236689805984, "step_time_sec": 8.237539358990034, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.411245822906494, "lr": 0.0002, "elapsed_sec": 28661.47503733635, "step_time_sec": 8.23816799800261, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.236522197723389, "lr": 0.0002, "elapsed_sec": 28669.71317100525, "step_time_sec": 8.23797131300671, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.360296249389648, "lr": 0.0002, "elapsed_sec": 28677.95097231865, "step_time_sec": 8.237617848993978, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.254251480102539, "lr": 0.0002, "elapsed_sec": 28686.18886899948, "step_time_sec": 8.237751409993507, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.367908000946045, "lr": 0.0002, "elapsed_sec": 28694.427473545074, "step_time_sec": 8.238490762014408, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.460081100463867, "lr": 0.0002, "elapsed_sec": 28702.666532039642, "step_time_sec": 8.238894762005657, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.521881580352783, "lr": 0.0002, "elapsed_sec": 28710.906071662903, "step_time_sec": 8.239333575998899, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.325178623199463, "lr": 0.0002, "elapsed_sec": 28719.143023729324, "step_time_sec": 8.236845862003975, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.288949489593506, "lr": 0.0002, "elapsed_sec": 28727.381225585938, "step_time_sec": 8.238006454979768, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.311233997344971, "lr": 0.0002, "elapsed_sec": 28735.617739915848, "step_time_sec": 8.236334813002031, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.411513328552246, "lr": 0.0002, "elapsed_sec": 28743.8571972847, "step_time_sec": 8.239303672016831, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.390308380126953, "lr": 0.0002, "elapsed_sec": 28752.095563173294, "step_time_sec": 8.238197458995273, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.323287487030029, "lr": 0.0002, "elapsed_sec": 28760.333532571793, "step_time_sec": 8.23781086501549, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.519815444946289, "lr": 0.0002, "elapsed_sec": 28768.571577310562, "step_time_sec": 8.237889872980304, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.35936975479126, "lr": 0.0002, "elapsed_sec": 28776.808176279068, "step_time_sec": 8.236413372011157, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.333994388580322, "lr": 0.0002, "elapsed_sec": 28785.048037290573, "step_time_sec": 8.239775750000263, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.389078617095947, "lr": 0.0002, "elapsed_sec": 28793.286080121994, "step_time_sec": 8.237828843004536, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.265275955200195, "lr": 0.0002, "elapsed_sec": 28801.5236890316, "step_time_sec": 8.2374779600068, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2274861335754395, "lr": 0.0002, "elapsed_sec": 28809.76269197464, "step_time_sec": 8.23878837400116, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.382589340209961, "lr": 0.0002, "elapsed_sec": 28817.998988628387, "step_time_sec": 8.236156940023648, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.381465911865234, "lr": 0.0002, "elapsed_sec": 28826.23811030388, "step_time_sec": 8.238977183005773, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.306281089782715, "lr": 0.0002, "elapsed_sec": 28834.475583791733, "step_time_sec": 8.23727210398647, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.26032829284668, "lr": 0.0002, "elapsed_sec": 28842.713464975357, "step_time_sec": 8.237731026019901, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.451468467712402, "lr": 0.0002, "elapsed_sec": 28850.952734470367, "step_time_sec": 8.239090585993836, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.41701078414917, "lr": 0.0002, "elapsed_sec": 28859.189683437347, "step_time_sec": 8.236842625017744, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.233287334442139, "lr": 0.0002, "elapsed_sec": 28867.428597688675, "step_time_sec": 8.238764894980704, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.435937881469727, "lr": 0.0002, "elapsed_sec": 28875.66518521309, "step_time_sec": 8.236372721003136, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.280813694000244, "lr": 0.0002, "elapsed_sec": 28883.902071475983, "step_time_sec": 8.236708718992304, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.341541767120361, "lr": 0.0002, "elapsed_sec": 28892.139721632004, "step_time_sec": 8.2374970120145, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4146270751953125, "lr": 0.0002, "elapsed_sec": 28900.376842737198, "step_time_sec": 8.236970981000923, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.22993278503418, "lr": 0.0002, "elapsed_sec": 28908.615236759186, "step_time_sec": 8.238250481983414, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.365472316741943, "lr": 0.0002, "elapsed_sec": 28916.85281777382, "step_time_sec": 8.23738389997743, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.378816604614258, "lr": 0.0002, "elapsed_sec": 28925.09102678299, "step_time_sec": 8.238087796984473, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.283634185791016, "lr": 0.0002, "elapsed_sec": 28933.328911542892, "step_time_sec": 8.237746319005964, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.512610912322998, "lr": 0.0002, "elapsed_sec": 28941.56782269478, "step_time_sec": 8.238746187998913, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.400058269500732, "lr": 0.0002, "elapsed_sec": 28949.806936979294, "step_time_sec": 8.238924105011392, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.400516510009766, "lr": 0.0002, "elapsed_sec": 28958.044225931168, "step_time_sec": 8.237149549997412, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.456083297729492, "lr": 0.0002, "elapsed_sec": 28966.283034801483, "step_time_sec": 8.238640310999472, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.469172954559326, "lr": 0.0002, "elapsed_sec": 28974.520178318024, "step_time_sec": 8.236959191999631, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.291409969329834, "lr": 0.0002, "elapsed_sec": 28982.760716676712, "step_time_sec": 8.240406141005224, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.457159519195557, "lr": 0.0002, "elapsed_sec": 28990.99760913849, "step_time_sec": 8.236676547996467, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.479078769683838, "lr": 0.0002, "elapsed_sec": 28999.236134290695, "step_time_sec": 8.238435292005306, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3694353103637695, "lr": 0.0002, "elapsed_sec": 29007.473442077637, "step_time_sec": 8.237137400981737, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.400837421417236, "lr": 0.0002, "elapsed_sec": 29015.710983753204, "step_time_sec": 8.237353431992233, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.522779941558838, "lr": 0.0002, "elapsed_sec": 29023.949605941772, "step_time_sec": 8.238452464982402, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.257938861846924, "lr": 0.0002, "elapsed_sec": 29032.185586452484, "step_time_sec": 8.235818798013497, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3666672706604, "lr": 0.0002, "elapsed_sec": 29040.425089120865, "step_time_sec": 8.239352014992619, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.254823207855225, "lr": 0.0002, "elapsed_sec": 29048.662570476532, "step_time_sec": 8.23736063198885, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.304654121398926, "lr": 0.0002, "elapsed_sec": 29056.900020837784, "step_time_sec": 8.237268815020798, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.343841552734375, "lr": 0.0002, "elapsed_sec": 29065.137730121613, "step_time_sec": 8.237533311999869, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.34325647354126, "lr": 0.0002, "elapsed_sec": 29073.374651908875, "step_time_sec": 29.59365595600684, "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": 4.46828031539917, "lr": 0.0002, "elapsed_sec": 29102.987804174423, "step_time_sec": 8.255781026004115, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.429767608642578, "lr": 0.0002, "elapsed_sec": 29111.219160079956, "step_time_sec": 8.231178474001354, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.286291122436523, "lr": 0.0002, "elapsed_sec": 29119.450799942017, "step_time_sec": 8.231520511995768, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.348874092102051, "lr": 0.0002, "elapsed_sec": 29127.683038949966, "step_time_sec": 8.23206393700093, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.452743053436279, "lr": 0.0002, "elapsed_sec": 29135.921364307404, "step_time_sec": 8.23814729199512, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.401007652282715, "lr": 0.0002, "elapsed_sec": 29144.159870624542, "step_time_sec": 8.238335533998907, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.439055442810059, "lr": 0.0002, "elapsed_sec": 29152.398440361023, "step_time_sec": 8.238430235011037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.314476013183594, "lr": 0.0002, "elapsed_sec": 29160.637085437775, "step_time_sec": 8.238474136014702, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.303840160369873, "lr": 0.0002, "elapsed_sec": 29168.8733420372, "step_time_sec": 8.236120813991874, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.438956260681152, "lr": 0.0002, "elapsed_sec": 29177.111481428146, "step_time_sec": 8.237967882974772, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.349648475646973, "lr": 0.0002, "elapsed_sec": 29185.349977731705, "step_time_sec": 8.238326318009058, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.305601119995117, "lr": 0.0002, "elapsed_sec": 29193.586665391922, "step_time_sec": 8.23659154298366, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.322262763977051, "lr": 0.0002, "elapsed_sec": 29201.82662463188, "step_time_sec": 8.239758517011069, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.400176048278809, "lr": 0.0002, "elapsed_sec": 29210.06375169754, "step_time_sec": 8.236978506989544, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.271637916564941, "lr": 0.0002, "elapsed_sec": 29218.303121089935, "step_time_sec": 8.23919623097754, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.382606506347656, "lr": 0.0002, "elapsed_sec": 29226.539476156235, "step_time_sec": 8.23628827900393, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.399238109588623, "lr": 0.0002, "elapsed_sec": 29234.777544736862, "step_time_sec": 8.237859446991934, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1587700843811035, "lr": 0.0002, "elapsed_sec": 29243.015219449997, "step_time_sec": 8.237510564998956, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.360640525817871, "lr": 0.0002, "elapsed_sec": 29251.252315044403, "step_time_sec": 8.236894572008168, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4335479736328125, "lr": 0.0002, "elapsed_sec": 29259.491055965424, "step_time_sec": 8.238562286016531, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.243926048278809, "lr": 0.0002, "elapsed_sec": 29267.72763991356, "step_time_sec": 8.23645218499587, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.313496112823486, "lr": 0.0002, "elapsed_sec": 29275.966395378113, "step_time_sec": 8.238572859991109, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.27386999130249, "lr": 0.0002, "elapsed_sec": 29284.204889774323, "step_time_sec": 8.238350574014476, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.135430335998535, "lr": 0.0002, "elapsed_sec": 29292.443454265594, "step_time_sec": 8.238402520975797, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.380985736846924, "lr": 0.0002, "elapsed_sec": 29300.680846214294, "step_time_sec": 8.237196060013957, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2666239738464355, "lr": 0.0002, "elapsed_sec": 29308.918209791183, "step_time_sec": 8.237206005986081, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.408627033233643, "lr": 0.0002, "elapsed_sec": 29317.157313108444, "step_time_sec": 8.238917384995148, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.253599166870117, "lr": 0.0002, "elapsed_sec": 29325.395773410797, "step_time_sec": 8.238276292016963, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.310755252838135, "lr": 0.0002, "elapsed_sec": 29333.634147644043, "step_time_sec": 8.238276885997038, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.52099084854126, "lr": 0.0002, "elapsed_sec": 29341.87275671959, "step_time_sec": 8.238401310023619, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.41063928604126, "lr": 0.0002, "elapsed_sec": 29350.110145807266, "step_time_sec": 8.237330726988148, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3032636642456055, "lr": 0.0002, "elapsed_sec": 29358.356902122498, "step_time_sec": 8.246492781006964, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2212066650390625, "lr": 0.0002, "elapsed_sec": 29366.595151424408, "step_time_sec": 8.238062560994877, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1941022872924805, "lr": 0.0002, "elapsed_sec": 29374.830447673798, "step_time_sec": 8.235122114012484, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.349311828613281, "lr": 0.0002, "elapsed_sec": 29383.0693089962, "step_time_sec": 8.238726868003141, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.287959098815918, "lr": 0.0002, "elapsed_sec": 29391.305629253387, "step_time_sec": 8.236125419003656, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.394532203674316, "lr": 0.0002, "elapsed_sec": 29399.545115232468, "step_time_sec": 8.239370297989808, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2735981941223145, "lr": 0.0002, "elapsed_sec": 29407.782886743546, "step_time_sec": 8.237584191985661, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2259039878845215, "lr": 0.0002, "elapsed_sec": 29416.0211789608, "step_time_sec": 8.238100179005414, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.35712194442749, "lr": 0.0002, "elapsed_sec": 29424.260140895844, "step_time_sec": 8.238831800001208, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.456433296203613, "lr": 0.0002, "elapsed_sec": 29432.496566057205, "step_time_sec": 8.236223145009717, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.300111770629883, "lr": 0.0002, "elapsed_sec": 29440.735025644302, "step_time_sec": 8.238301853009034, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.389754772186279, "lr": 0.0002, "elapsed_sec": 29448.9729783535, "step_time_sec": 8.23783403498237, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.333188533782959, "lr": 0.0002, "elapsed_sec": 29457.21121573448, "step_time_sec": 8.238036530994577, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.324831485748291, "lr": 0.0002, "elapsed_sec": 29465.450328588486, "step_time_sec": 8.238957884022966, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3289875984191895, "lr": 0.0002, "elapsed_sec": 29473.686488866806, "step_time_sec": 8.236024481011555, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.240740776062012, "lr": 0.0002, "elapsed_sec": 29481.92603969574, "step_time_sec": 8.239368120004656, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.400956153869629, "lr": 0.0002, "elapsed_sec": 29490.163514137268, "step_time_sec": 8.237315253005363, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.497768878936768, "lr": 0.0002, "elapsed_sec": 29498.401064634323, "step_time_sec": 8.237408346001757, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.383245468139648, "lr": 0.0002, "elapsed_sec": 29506.63857984543, "step_time_sec": 8.237318919011159, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.427738666534424, "lr": 0.0002, "elapsed_sec": 29514.876448869705, "step_time_sec": 8.237781411997275, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4222235679626465, "lr": 0.0002, "elapsed_sec": 29523.115116119385, "step_time_sec": 8.238490033982089, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.362160682678223, "lr": 0.0002, "elapsed_sec": 29531.351643800735, "step_time_sec": 8.236328424012754, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.349854946136475, "lr": 0.0002, "elapsed_sec": 29539.590675592422, "step_time_sec": 8.23893472799682, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.352668285369873, "lr": 0.0002, "elapsed_sec": 29547.82799720764, "step_time_sec": 8.237124621024122, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.335269927978516, "lr": 0.0002, "elapsed_sec": 29556.067138433456, "step_time_sec": 8.238963934010826, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2340216636657715, "lr": 0.0002, "elapsed_sec": 29564.306022167206, "step_time_sec": 8.238727839983767, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.187987327575684, "lr": 0.0002, "elapsed_sec": 29572.544100284576, "step_time_sec": 8.237898591003614, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.242432594299316, "lr": 0.0002, "elapsed_sec": 29580.78227162361, "step_time_sec": 8.238030704000266, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.340466499328613, "lr": 0.0002, "elapsed_sec": 29589.020379543304, "step_time_sec": 8.23798178700963, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.413679599761963, "lr": 0.0002, "elapsed_sec": 29597.25998711586, "step_time_sec": 8.23944382998161, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.394239902496338, "lr": 0.0002, "elapsed_sec": 29605.499027252197, "step_time_sec": 8.238882536010351, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.38295841217041, "lr": 0.0002, "elapsed_sec": 29613.736658334732, "step_time_sec": 8.237452626985032, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.230544090270996, "lr": 0.0002, "elapsed_sec": 29621.975784778595, "step_time_sec": 8.238980494003044, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.248073577880859, "lr": 0.0002, "elapsed_sec": 29630.213183641434, "step_time_sec": 8.237249520025216, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.288527011871338, "lr": 0.0002, "elapsed_sec": 29638.451469659805, "step_time_sec": 8.238134273007745, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.246128559112549, "lr": 0.0002, "elapsed_sec": 29646.689477920532, "step_time_sec": 8.23788813399733, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.449016094207764, "lr": 0.0002, "elapsed_sec": 29654.926952123642, "step_time_sec": 8.237313116987934, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.315877914428711, "lr": 0.0002, "elapsed_sec": 29663.164466142654, "step_time_sec": 8.237356775993248, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.427804470062256, "lr": 0.0002, "elapsed_sec": 29671.401529073715, "step_time_sec": 8.236832619004417, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.337061882019043, "lr": 0.0002, "elapsed_sec": 29679.639590263367, "step_time_sec": 8.237974816991482, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.282601356506348, "lr": 0.0002, "elapsed_sec": 29687.878078222275, "step_time_sec": 8.238277968979673, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.248013496398926, "lr": 0.0002, "elapsed_sec": 29696.1160633564, "step_time_sec": 8.237831959006144, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.500490665435791, "lr": 0.0002, "elapsed_sec": 29704.356685876846, "step_time_sec": 8.240463310008636, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.395130157470703, "lr": 0.0002, "elapsed_sec": 29712.59236383438, "step_time_sec": 8.235499911999796, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.438777446746826, "lr": 0.0002, "elapsed_sec": 29720.830884218216, "step_time_sec": 8.238355123001384, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.377900123596191, "lr": 0.0002, "elapsed_sec": 29729.06895661354, "step_time_sec": 8.237953232019208, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.297933101654053, "lr": 0.0002, "elapsed_sec": 29737.307865142822, "step_time_sec": 8.23872403401765, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.390320777893066, "lr": 0.0002, "elapsed_sec": 29745.54644012451, "step_time_sec": 8.238417441025376, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.566832065582275, "lr": 0.0002, "elapsed_sec": 29753.784260988235, "step_time_sec": 8.237643517990364, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3320417404174805, "lr": 0.0002, "elapsed_sec": 29762.02409863472, "step_time_sec": 8.239663601998473, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.403951168060303, "lr": 0.0002, "elapsed_sec": 29770.262134313583, "step_time_sec": 8.23790687599103, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.495944499969482, "lr": 0.0002, "elapsed_sec": 29778.501758813858, "step_time_sec": 8.239485357975354, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.408505916595459, "lr": 0.0002, "elapsed_sec": 29786.7411403656, "step_time_sec": 8.239187553990632, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.342982769012451, "lr": 0.0002, "elapsed_sec": 29794.97834444046, "step_time_sec": 8.237112428003456, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.388031482696533, "lr": 0.0002, "elapsed_sec": 29803.21655201912, "step_time_sec": 8.238019445998361, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.482406139373779, "lr": 0.0002, "elapsed_sec": 29811.454824209213, "step_time_sec": 8.238062720018206, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.427165508270264, "lr": 0.0002, "elapsed_sec": 29819.693344831467, "step_time_sec": 8.238411193015054, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.346136569976807, "lr": 0.0002, "elapsed_sec": 29827.93256831169, "step_time_sec": 8.239064848981798, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.307560443878174, "lr": 0.0002, "elapsed_sec": 29836.169815540314, "step_time_sec": 8.237062096013688, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.437350749969482, "lr": 0.0002, "elapsed_sec": 29844.408742904663, "step_time_sec": 8.238785053021275, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.424604892730713, "lr": 0.0002, "elapsed_sec": 29852.64718079567, "step_time_sec": 8.238306587008992, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.329212188720703, "lr": 0.0002, "elapsed_sec": 29860.88578414917, "step_time_sec": 8.238384156982647, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.273758411407471, "lr": 0.0002, "elapsed_sec": 29869.124464035034, "step_time_sec": 8.23858278800617, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.383314609527588, "lr": 0.0002, "elapsed_sec": 29877.361135721207, "step_time_sec": 8.236498162994394, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.308589458465576, "lr": 0.0002, "elapsed_sec": 29885.598994731903, "step_time_sec": 8.237664411019068, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.332305431365967, "lr": 0.0002, "elapsed_sec": 29893.835520505905, "step_time_sec": 8.236380438989727, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.447888374328613, "lr": 0.0002, "elapsed_sec": 29902.073736429214, "step_time_sec": 8.238047472987091, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.425334930419922, "lr": 0.0002, "elapsed_sec": 29910.30984067917, "step_time_sec": 8.235947290988406, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.369475364685059, "lr": 0.0002, "elapsed_sec": 29918.547961473465, "step_time_sec": 8.237984615989262, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.354649066925049, "lr": 0.0002, "elapsed_sec": 29926.786729335785, "step_time_sec": 8.23857261100784, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.37288236618042, "lr": 0.0002, "elapsed_sec": 29935.022797584534, "step_time_sec": 8.235922759020468, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.272086143493652, "lr": 0.0002, "elapsed_sec": 29943.26115131378, "step_time_sec": 8.238229459995637, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.36978006362915, "lr": 0.0002, "elapsed_sec": 29951.498158931732, "step_time_sec": 8.23679861301207, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.231175899505615, "lr": 0.0002, "elapsed_sec": 29959.735934972763, "step_time_sec": 8.237629760988057, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.565993785858154, "lr": 0.0002, "elapsed_sec": 29967.974673509598, "step_time_sec": 8.238565906998701, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.366423606872559, "lr": 0.0002, "elapsed_sec": 29976.212818861008, "step_time_sec": 8.237967019987991, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3287177085876465, "lr": 0.0002, "elapsed_sec": 29984.451643705368, "step_time_sec": 8.23868710300303, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.353262901306152, "lr": 0.0002, "elapsed_sec": 29992.68886756897, "step_time_sec": 8.23705220100237, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.307557582855225, "lr": 0.0002, "elapsed_sec": 30000.928948640823, "step_time_sec": 8.239906292990781, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.316878318786621, "lr": 0.0002, "elapsed_sec": 30009.167471170425, "step_time_sec": 8.238358196016634, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.407021522521973, "lr": 0.0002, "elapsed_sec": 30017.407020568848, "step_time_sec": 8.239410549023887, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.227360725402832, "lr": 0.0002, "elapsed_sec": 30025.644850730896, "step_time_sec": 8.237704293977004, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.306209087371826, "lr": 0.0002, "elapsed_sec": 30033.882145166397, "step_time_sec": 8.237084943015361, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3005690574646, "lr": 0.0002, "elapsed_sec": 30042.121454000473, "step_time_sec": 8.239168789994437, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3221049308776855, "lr": 0.0002, "elapsed_sec": 30050.35989165306, "step_time_sec": 8.238240095000947, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.422794342041016, "lr": 0.0002, "elapsed_sec": 30058.598267793655, "step_time_sec": 8.238270999980159, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.426102161407471, "lr": 0.0002, "elapsed_sec": 30066.837268829346, "step_time_sec": 8.238821195991477, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.391790866851807, "lr": 0.0002, "elapsed_sec": 30075.073806524277, "step_time_sec": 8.236349876009626, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.346864223480225, "lr": 0.0002, "elapsed_sec": 30083.311707496643, "step_time_sec": 8.237790521990974, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.386427402496338, "lr": 0.0002, "elapsed_sec": 30091.54992890358, "step_time_sec": 8.238022356003057, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.25115966796875, "lr": 0.0002, "elapsed_sec": 30099.78748703003, "step_time_sec": 8.237432267982513, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.397669315338135, "lr": 0.0002, "elapsed_sec": 30108.025666713715, "step_time_sec": 8.238034829992102, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1950154304504395, "lr": 0.0002, "elapsed_sec": 30116.262622356415, "step_time_sec": 8.236814421019517, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3993940353393555, "lr": 0.0002, "elapsed_sec": 30124.501887083054, "step_time_sec": 8.23907185101416, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.408727645874023, "lr": 0.0002, "elapsed_sec": 30132.73838043213, "step_time_sec": 8.236315457004821, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.287301540374756, "lr": 0.0002, "elapsed_sec": 30140.977963209152, "step_time_sec": 8.23939134000102, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.261500835418701, "lr": 0.0002, "elapsed_sec": 30149.216136455536, "step_time_sec": 8.238071987987496, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3952531814575195, "lr": 0.0002, "elapsed_sec": 30157.454221963882, "step_time_sec": 8.237855112995021, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.321798324584961, "lr": 0.0002, "elapsed_sec": 30165.692249774933, "step_time_sec": 8.237857723026536, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.34439754486084, "lr": 0.0002, "elapsed_sec": 30173.929599761963, "step_time_sec": 8.237221010000212, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.394031524658203, "lr": 0.0002, "elapsed_sec": 30182.167885541916, "step_time_sec": 8.238078478985699, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.29960298538208, "lr": 0.0002, "elapsed_sec": 30190.40499854088, "step_time_sec": 8.236957402026746, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.293110370635986, "lr": 0.0002, "elapsed_sec": 30198.64486002922, "step_time_sec": 8.239739823999116, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.392670631408691, "lr": 0.0002, "elapsed_sec": 30206.883357048035, "step_time_sec": 8.238311173015973, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.237158298492432, "lr": 0.0002, "elapsed_sec": 30215.120637893677, "step_time_sec": 8.237108948000241, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3363542556762695, "lr": 0.0002, "elapsed_sec": 30223.358840703964, "step_time_sec": 8.238093202002347, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.249436855316162, "lr": 0.0002, "elapsed_sec": 30231.596026182175, "step_time_sec": 8.236956524982816, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.238667011260986, "lr": 0.0002, "elapsed_sec": 30239.83432984352, "step_time_sec": 8.238195326994173, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.376083850860596, "lr": 0.0002, "elapsed_sec": 30248.07183122635, "step_time_sec": 8.237310278025689, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.498410224914551, "lr": 0.0002, "elapsed_sec": 30256.30954194069, "step_time_sec": 8.237568189011654, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.318254470825195, "lr": 0.0002, "elapsed_sec": 30264.547664880753, "step_time_sec": 8.237906013004249, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.385929584503174, "lr": 0.0002, "elapsed_sec": 30272.784821033478, "step_time_sec": 8.23700933798682, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.538455486297607, "lr": 0.0002, "elapsed_sec": 30281.02394104004, "step_time_sec": 8.238957136985846, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.161925792694092, "lr": 0.0002, "elapsed_sec": 30289.261559963226, "step_time_sec": 8.237508699006867, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3847761154174805, "lr": 0.0002, "elapsed_sec": 30297.500172138214, "step_time_sec": 8.238402835006127, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.364043712615967, "lr": 0.0002, "elapsed_sec": 30305.73908996582, "step_time_sec": 8.238768162991619, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.380936622619629, "lr": 0.0002, "elapsed_sec": 30313.976416826248, "step_time_sec": 8.237171020999085, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1492791175842285, "lr": 0.0002, "elapsed_sec": 30322.215707063675, "step_time_sec": 8.239117354009068, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.349788188934326, "lr": 0.0002, "elapsed_sec": 30330.453369140625, "step_time_sec": 8.237488198996289, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.403958320617676, "lr": 0.0002, "elapsed_sec": 30338.690984487534, "step_time_sec": 8.237500965013169, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.437175750732422, "lr": 0.0002, "elapsed_sec": 30346.930269241333, "step_time_sec": 8.239130631001899, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.361569404602051, "lr": 0.0002, "elapsed_sec": 30355.167178869247, "step_time_sec": 8.236704314011149, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.293140888214111, "lr": 0.0002, "elapsed_sec": 30363.406888723373, "step_time_sec": 8.239556874003028, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.312507152557373, "lr": 0.0002, "elapsed_sec": 30371.64492058754, "step_time_sec": 8.23786181898322, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.286087989807129, "lr": 0.0002, "elapsed_sec": 30379.88468170166, "step_time_sec": 8.239632356009679, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.397895812988281, "lr": 0.0002, "elapsed_sec": 30388.121716737747, "step_time_sec": 8.23688978899736, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.381776809692383, "lr": 0.0002, "elapsed_sec": 30396.359749555588, "step_time_sec": 8.237898375024088, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.410858154296875, "lr": 0.0002, "elapsed_sec": 30404.59915614128, "step_time_sec": 8.23918197400053, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.361931800842285, "lr": 0.0002, "elapsed_sec": 30412.837055683136, "step_time_sec": 8.23771859597764, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3662028312683105, "lr": 0.0002, "elapsed_sec": 30421.076459407806, "step_time_sec": 8.239261883019935, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.472776412963867, "lr": 0.0002, "elapsed_sec": 30429.314552545547, "step_time_sec": 8.237983364000684, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.410768032073975, "lr": 0.0002, "elapsed_sec": 30437.55280828476, "step_time_sec": 8.238022901001386, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.368664741516113, "lr": 0.0002, "elapsed_sec": 30445.791523456573, "step_time_sec": 8.238597489980748, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.48378324508667, "lr": 0.0002, "elapsed_sec": 30454.028191566467, "step_time_sec": 8.236502648011083, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3929443359375, "lr": 0.0002, "elapsed_sec": 30462.26751923561, "step_time_sec": 8.23915643300279, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.331309795379639, "lr": 0.0002, "elapsed_sec": 30470.506677865982, "step_time_sec": 8.238980608992279, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.341609001159668, "lr": 0.0002, "elapsed_sec": 30478.745950222015, "step_time_sec": 8.239197086979402, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.410431861877441, "lr": 0.0002, "elapsed_sec": 30486.984268188477, "step_time_sec": 8.238160200009588, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.618744373321533, "lr": 0.0002, "elapsed_sec": 30495.22154736519, "step_time_sec": 8.237073311000131, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.376124382019043, "lr": 0.0002, "elapsed_sec": 30503.460309505463, "step_time_sec": 8.238629574974766, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.409319877624512, "lr": 0.0002, "elapsed_sec": 30511.69785094261, "step_time_sec": 8.237326681002742, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.368844509124756, "lr": 0.0002, "elapsed_sec": 30519.936455249786, "step_time_sec": 8.238479317020392, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.459313869476318, "lr": 0.0002, "elapsed_sec": 30528.175963163376, "step_time_sec": 8.239353290991858, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.359630107879639, "lr": 0.0002, "elapsed_sec": 30536.414984703064, "step_time_sec": 8.238902329991106, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.341731548309326, "lr": 0.0002, "elapsed_sec": 30544.662172317505, "step_time_sec": 8.246979800984263, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.260354518890381, "lr": 0.0002, "elapsed_sec": 30552.901144504547, "step_time_sec": 8.238847129017813, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.275266647338867, "lr": 0.0002, "elapsed_sec": 30561.138255357742, "step_time_sec": 8.2368787759915, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.32251501083374, "lr": 0.0002, "elapsed_sec": 30569.376945972443, "step_time_sec": 8.238583194004605, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2771525382995605, "lr": 0.0002, "elapsed_sec": 30577.61368727684, "step_time_sec": 8.236547671986045, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.291356563568115, "lr": 0.0002, "elapsed_sec": 30585.85263133049, "step_time_sec": 8.238742624991573, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.326085090637207, "lr": 0.0002, "elapsed_sec": 30594.090485095978, "step_time_sec": 8.237745705991983, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.354647636413574, "lr": 0.0002, "elapsed_sec": 30602.32808971405, "step_time_sec": 8.23738370501087, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.338208198547363, "lr": 0.0002, "elapsed_sec": 30610.566730976105, "step_time_sec": 8.238481903012143, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.358354091644287, "lr": 0.0002, "elapsed_sec": 30618.803790330887, "step_time_sec": 8.236935244989581, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.222380638122559, "lr": 0.0002, "elapsed_sec": 30627.04233264923, "step_time_sec": 8.238411522004753, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.185197353363037, "lr": 0.0002, "elapsed_sec": 30635.281680107117, "step_time_sec": 8.23915418901015, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.402924537658691, "lr": 0.0002, "elapsed_sec": 30643.5202395916, "step_time_sec": 8.238366051984485, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.283945560455322, "lr": 0.0002, "elapsed_sec": 30651.759354114532, "step_time_sec": 8.238992121012416, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.328866004943848, "lr": 0.0002, "elapsed_sec": 30659.998236894608, "step_time_sec": 8.238694195984863, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.450779438018799, "lr": 0.0002, "elapsed_sec": 30668.236884593964, "step_time_sec": 8.23850125700119, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.212610244750977, "lr": 0.0002, "elapsed_sec": 30676.475280046463, "step_time_sec": 8.238154887018027, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.362677574157715, "lr": 0.0002, "elapsed_sec": 30684.713008403778, "step_time_sec": 8.237593461002689, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.220289707183838, "lr": 0.0002, "elapsed_sec": 30692.951666116714, "step_time_sec": 8.238481670006877, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.480444431304932, "lr": 0.0002, "elapsed_sec": 30701.18794465065, "step_time_sec": 8.236132657999406, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.306623458862305, "lr": 0.0002, "elapsed_sec": 30709.428184747696, "step_time_sec": 8.240109596023103, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.33247709274292, "lr": 0.0002, "elapsed_sec": 30717.667697668076, "step_time_sec": 8.239286447991617, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.315585136413574, "lr": 0.0002, "elapsed_sec": 30725.905612945557, "step_time_sec": 8.237781105999602, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3417158126831055, "lr": 0.0002, "elapsed_sec": 30734.14514517784, "step_time_sec": 8.23940390397911, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.384271621704102, "lr": 0.0002, "elapsed_sec": 30742.38144803047, "step_time_sec": 8.236096725973766, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.39431619644165, "lr": 0.0002, "elapsed_sec": 30750.620035886765, "step_time_sec": 8.238437509979121, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3570332527160645, "lr": 0.0002, "elapsed_sec": 30758.857297182083, "step_time_sec": 8.237131915986538, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.350486755371094, "lr": 0.0002, "elapsed_sec": 30767.096116304398, "step_time_sec": 8.238650863990188, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.239217758178711, "lr": 0.0002, "elapsed_sec": 30775.33462715149, "step_time_sec": 8.238345328019932, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.386196136474609, "lr": 0.0002, "elapsed_sec": 30783.571761846542, "step_time_sec": 8.236940750008216, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.343501091003418, "lr": 0.0002, "elapsed_sec": 30791.809864521027, "step_time_sec": 8.237947703019017, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3015265464782715, "lr": 0.0002, "elapsed_sec": 30800.047919511795, "step_time_sec": 8.237926549976692, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.313004016876221, "lr": 0.0002, "elapsed_sec": 30808.287222623825, "step_time_sec": 8.239134789997479, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.434473991394043, "lr": 0.0002, "elapsed_sec": 30816.52783179283, "step_time_sec": 8.24043345998507, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.347704887390137, "lr": 0.0002, "elapsed_sec": 30824.765992879868, "step_time_sec": 8.23798211599933, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.279545783996582, "lr": 0.0002, "elapsed_sec": 30833.004549980164, "step_time_sec": 8.238416329986649, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.23440408706665, "lr": 0.0002, "elapsed_sec": 30841.24120926857, "step_time_sec": 8.236485291999998, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.358019828796387, "lr": 0.0002, "elapsed_sec": 30849.479375123978, "step_time_sec": 8.238036395981908, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.344114303588867, "lr": 0.0002, "elapsed_sec": 30857.717682361603, "step_time_sec": 8.238138417014852, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.408824920654297, "lr": 0.0002, "elapsed_sec": 30865.955229759216, "step_time_sec": 8.237337574013509, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.165398120880127, "lr": 0.0002, "elapsed_sec": 30874.194272994995, "step_time_sec": 8.238966864999384, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.397724628448486, "lr": 0.0002, "elapsed_sec": 30882.431544542313, "step_time_sec": 8.23703959499835, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.388077259063721, "lr": 0.0002, "elapsed_sec": 30890.67061805725, "step_time_sec": 8.238895224989392, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2931036949157715, "lr": 0.0002, "elapsed_sec": 30898.91028380394, "step_time_sec": 8.239601323002717, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.382881164550781, "lr": 0.0002, "elapsed_sec": 30907.14820241928, "step_time_sec": 8.237683068000479, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.297619342803955, "lr": 0.0002, "elapsed_sec": 30915.38669848442, "step_time_sec": 8.238366382982349, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.542304515838623, "lr": 0.0002, "elapsed_sec": 30923.62432575226, "step_time_sec": 8.237431319983443, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.393881797790527, "lr": 0.0002, "elapsed_sec": 30931.862918376923, "step_time_sec": 8.238398208020953, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.332931995391846, "lr": 0.0002, "elapsed_sec": 30940.100164413452, "step_time_sec": 8.237091316987062, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.493195056915283, "lr": 0.0002, "elapsed_sec": 30948.33844256401, "step_time_sec": 8.238124561001314, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2736663818359375, "lr": 0.0002, "elapsed_sec": 30956.57743358612, "step_time_sec": 8.238857201009523, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.39303731918335, "lr": 0.0002, "elapsed_sec": 30964.814952611923, "step_time_sec": 8.237305532005848, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.414438724517822, "lr": 0.0002, "elapsed_sec": 30973.05361366272, "step_time_sec": 8.238494307006476, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.437811374664307, "lr": 0.0002, "elapsed_sec": 30981.291897058487, "step_time_sec": 8.238145362003706, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.301793575286865, "lr": 0.0002, "elapsed_sec": 30989.53255391121, "step_time_sec": 8.240526843001135, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3124589920043945, "lr": 0.0002, "elapsed_sec": 30997.771666049957, "step_time_sec": 8.238950385013595, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.200011730194092, "lr": 0.0002, "elapsed_sec": 31006.008849859238, "step_time_sec": 8.236961078015156, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.336522102355957, "lr": 0.0002, "elapsed_sec": 31014.248792886734, "step_time_sec": 8.239888294017874, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.416440010070801, "lr": 0.0002, "elapsed_sec": 31022.48931980133, "step_time_sec": 8.240296086994931, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.205130100250244, "lr": 0.0002, "elapsed_sec": 31030.72777414322, "step_time_sec": 8.238277105992893, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.404598712921143, "lr": 0.0002, "elapsed_sec": 31039.03232693672, "step_time_sec": 8.248271306045353, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.301882266998291, "lr": 0.0002, "elapsed_sec": 31047.27114701271, "step_time_sec": 8.238600858021528, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.248377799987793, "lr": 0.0002, "elapsed_sec": 31055.508734464645, "step_time_sec": 8.237477874965407, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.388154029846191, "lr": 0.0002, "elapsed_sec": 31063.748551607132, "step_time_sec": 8.239623263012618, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.378807067871094, "lr": 0.0002, "elapsed_sec": 31071.986212968826, "step_time_sec": 8.237508316989988, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1932759284973145, "lr": 0.0002, "elapsed_sec": 31080.224141836166, "step_time_sec": 8.237765662954189, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.321776390075684, "lr": 0.0002, "elapsed_sec": 31088.46124434471, "step_time_sec": 8.236957260989584, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3004608154296875, "lr": 0.0002, "elapsed_sec": 31096.69912815094, "step_time_sec": 8.237683042010758, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.294757843017578, "lr": 0.0002, "elapsed_sec": 31104.937403440475, "step_time_sec": 8.238145512994379, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.256836891174316, "lr": 0.0002, "elapsed_sec": 31113.174028873444, "step_time_sec": 8.236437526997179, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.303531169891357, "lr": 0.0002, "elapsed_sec": 31121.412314653397, "step_time_sec": 8.238141928974073, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.387462615966797, "lr": 0.0002, "elapsed_sec": 31129.651227474213, "step_time_sec": 8.238807525020093, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.235381603240967, "lr": 0.0002, "elapsed_sec": 31137.889909267426, "step_time_sec": 8.238475790945813, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.237895965576172, "lr": 0.0002, "elapsed_sec": 31146.128604650497, "step_time_sec": 8.23853065498406, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.392631530761719, "lr": 0.0002, "elapsed_sec": 31154.365666151047, "step_time_sec": 8.236943787022028, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.274448871612549, "lr": 0.0002, "elapsed_sec": 31162.604298830032, "step_time_sec": 8.23843939800281, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.381223678588867, "lr": 0.0002, "elapsed_sec": 31170.842625379562, "step_time_sec": 8.238160522014368, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.32876443862915, "lr": 0.0002, "elapsed_sec": 31179.08059835434, "step_time_sec": 8.23781300499104, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.280463218688965, "lr": 0.0002, "elapsed_sec": 31187.319353103638, "step_time_sec": 8.238611528999172, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.321142196655273, "lr": 0.0002, "elapsed_sec": 31195.556283950806, "step_time_sec": 8.23677806701744, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.37672233581543, "lr": 0.0002, "elapsed_sec": 31203.79510831833, "step_time_sec": 8.238655399996787, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.339879989624023, "lr": 0.0002, "elapsed_sec": 31212.03249692917, "step_time_sec": 8.2372374709812, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.304082870483398, "lr": 0.0002, "elapsed_sec": 31220.271305322647, "step_time_sec": 8.238683376985136, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3472723960876465, "lr": 0.0002, "elapsed_sec": 31228.510135650635, "step_time_sec": 8.238673718005884, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.361141681671143, "lr": 0.0002, "elapsed_sec": 31236.74797463417, "step_time_sec": 8.23762181197526, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.185876369476318, "lr": 0.0002, "elapsed_sec": 31244.98719716072, "step_time_sec": 8.239070240000729, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.147497177124023, "lr": 0.0002, "elapsed_sec": 31253.22595191002, "step_time_sec": 8.238603261997923, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.322505474090576, "lr": 0.0002, "elapsed_sec": 31261.464888095856, "step_time_sec": 8.238818273996003, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.5298848152160645, "lr": 0.0002, "elapsed_sec": 31269.703815221786, "step_time_sec": 8.23877525201533, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.297025680541992, "lr": 0.0002, "elapsed_sec": 31277.940988063812, "step_time_sec": 8.23698389402125, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.358012676239014, "lr": 0.0002, "elapsed_sec": 31286.179122924805, "step_time_sec": 8.23794642399298, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.351353645324707, "lr": 0.0002, "elapsed_sec": 31294.416338920593, "step_time_sec": 8.23709610500373, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.308080196380615, "lr": 0.0002, "elapsed_sec": 31302.656072854996, "step_time_sec": 8.239550404017791, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.312357425689697, "lr": 0.0002, "elapsed_sec": 31310.894142627716, "step_time_sec": 8.237941869010683, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.318914413452148, "lr": 0.0002, "elapsed_sec": 31319.133120298386, "step_time_sec": 8.238837658020202, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.416834354400635, "lr": 0.0002, "elapsed_sec": 31327.372017860413, "step_time_sec": 8.238677469955292, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2722296714782715, "lr": 0.0002, "elapsed_sec": 31335.610478162766, "step_time_sec": 8.238321123993956, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.217820167541504, "lr": 0.0002, "elapsed_sec": 31343.849077939987, "step_time_sec": 8.238397092034575, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.176516056060791, "lr": 0.0002, "elapsed_sec": 31352.087609767914, "step_time_sec": 8.238396923989058, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.290335178375244, "lr": 0.0002, "elapsed_sec": 31360.32479929924, "step_time_sec": 8.237059323000722, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.366627216339111, "lr": 0.0002, "elapsed_sec": 31368.5623254776, "step_time_sec": 8.23736283899052, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.291501522064209, "lr": 0.0002, "elapsed_sec": 31376.798158168793, "step_time_sec": 8.235634440963622, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.389039039611816, "lr": 0.0002, "elapsed_sec": 31385.036494731903, "step_time_sec": 8.238162927969825, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.262106895446777, "lr": 0.0002, "elapsed_sec": 31393.274443387985, "step_time_sec": 8.237781342992093, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.439878940582275, "lr": 0.0002, "elapsed_sec": 31401.51354074478, "step_time_sec": 8.238924233010039, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.567384243011475, "lr": 0.0002, "elapsed_sec": 31409.752793312073, "step_time_sec": 8.239138729986735, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.362504005432129, "lr": 0.0002, "elapsed_sec": 31417.99017882347, "step_time_sec": 8.237224530021194, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.362929344177246, "lr": 0.0002, "elapsed_sec": 31426.22910094261, "step_time_sec": 8.238726015028078, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.246171474456787, "lr": 0.0002, "elapsed_sec": 31434.4678068161, "step_time_sec": 8.23859005200211, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.228638172149658, "lr": 0.0002, "elapsed_sec": 31442.707322597504, "step_time_sec": 8.239337605948094, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.498332500457764, "lr": 0.0002, "elapsed_sec": 31450.946906089783, "step_time_sec": 8.239438335003797, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.306432247161865, "lr": 0.0002, "elapsed_sec": 31459.183610916138, "step_time_sec": 8.236524746986106, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.292108058929443, "lr": 0.0002, "elapsed_sec": 31467.422055006027, "step_time_sec": 8.238259072008077, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2033610343933105, "lr": 0.0002, "elapsed_sec": 31475.65815758705, "step_time_sec": 8.236037300957832, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.27983283996582, "lr": 0.0002, "elapsed_sec": 31483.89728498459, "step_time_sec": 8.238865295948926, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.227691650390625, "lr": 0.0002, "elapsed_sec": 31492.13581609726, "step_time_sec": 8.23841013101628, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.457766532897949, "lr": 0.0002, "elapsed_sec": 31500.373980283737, "step_time_sec": 8.238029776024632, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.260645866394043, "lr": 0.0002, "elapsed_sec": 31508.61282634735, "step_time_sec": 8.23865980899427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.420899391174316, "lr": 0.0002, "elapsed_sec": 31516.84978532791, "step_time_sec": 8.236796774959657, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.193614959716797, "lr": 0.0002, "elapsed_sec": 31525.088516950607, "step_time_sec": 8.238600536016747, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2854905128479, "lr": 0.0002, "elapsed_sec": 31533.327445983887, "step_time_sec": 8.238776967977174, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.193363189697266, "lr": 0.0002, "elapsed_sec": 31541.56539130211, "step_time_sec": 8.237809935992118, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3749918937683105, "lr": 0.0002, "elapsed_sec": 31549.80463218689, "step_time_sec": 8.239064875990152, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2330780029296875, "lr": 0.0002, "elapsed_sec": 31558.041776180267, "step_time_sec": 8.236952911014669, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.27131986618042, "lr": 0.0002, "elapsed_sec": 31566.280254602432, "step_time_sec": 8.238308904052246, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.383771896362305, "lr": 0.0002, "elapsed_sec": 31574.51864886284, "step_time_sec": 8.238255848002154, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3199968338012695, "lr": 0.0002, "elapsed_sec": 31582.757292985916, "step_time_sec": 8.238452738034539, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.230175018310547, "lr": 0.0002, "elapsed_sec": 31590.995862722397, "step_time_sec": 8.238423950970173, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.329941272735596, "lr": 0.0002, "elapsed_sec": 31599.23348045349, "step_time_sec": 8.237471338012256, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.355108261108398, "lr": 0.0002, "elapsed_sec": 31607.472293138504, "step_time_sec": 8.238592009991407, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.222870826721191, "lr": 0.0002, "elapsed_sec": 31615.711919546127, "step_time_sec": 8.239502259995788, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.348118305206299, "lr": 0.0002, "elapsed_sec": 31623.95245552063, "step_time_sec": 8.240371868014336, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.369987964630127, "lr": 0.0002, "elapsed_sec": 31632.19157743454, "step_time_sec": 8.238924251054414, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.456779956817627, "lr": 0.0002, "elapsed_sec": 31640.429053783417, "step_time_sec": 8.23727595701348, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.34929895401001, "lr": 0.0002, "elapsed_sec": 31648.669362068176, "step_time_sec": 8.24016379902605, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.297998905181885, "lr": 0.0002, "elapsed_sec": 31656.907430648804, "step_time_sec": 8.237921938998625, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.343082904815674, "lr": 0.0002, "elapsed_sec": 31665.145344734192, "step_time_sec": 8.237715358962305, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.265287399291992, "lr": 0.0002, "elapsed_sec": 31673.384702205658, "step_time_sec": 8.239226340025198, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.324635982513428, "lr": 0.0002, "elapsed_sec": 31681.62194275856, "step_time_sec": 8.237112071015872, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.176480293273926, "lr": 0.0002, "elapsed_sec": 31689.86071896553, "step_time_sec": 8.238632369961124, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.282201290130615, "lr": 0.0002, "elapsed_sec": 31698.098928928375, "step_time_sec": 8.238006603030954, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.37953519821167, "lr": 0.0002, "elapsed_sec": 31706.337072134018, "step_time_sec": 8.238015399954747, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.305086135864258, "lr": 0.0002, "elapsed_sec": 31714.575778722763, "step_time_sec": 8.238540941034444, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.165152072906494, "lr": 0.0002, "elapsed_sec": 31722.812407016754, "step_time_sec": 8.236423912050668, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.151882171630859, "lr": 0.0002, "elapsed_sec": 31731.05110669136, "step_time_sec": 8.23859183199238, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.226212501525879, "lr": 0.0002, "elapsed_sec": 31739.291216611862, "step_time_sec": 8.239925127010792, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.31797456741333, "lr": 0.0002, "elapsed_sec": 31747.531306266785, "step_time_sec": 8.239928026974667, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.188072204589844, "lr": 0.0002, "elapsed_sec": 31755.77087044716, "step_time_sec": 8.239348846953362, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.348066329956055, "lr": 0.0002, "elapsed_sec": 31764.008347272873, "step_time_sec": 8.237338464998174, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2271904945373535, "lr": 0.0002, "elapsed_sec": 31772.247678279877, "step_time_sec": 8.239166538987774, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.325568675994873, "lr": 0.0002, "elapsed_sec": 31780.485922574997, "step_time_sec": 8.23802919400623, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3653388023376465, "lr": 0.0002, "elapsed_sec": 31788.724046945572, "step_time_sec": 8.238000341050792, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.173856735229492, "lr": 0.0002, "elapsed_sec": 31796.962873458862, "step_time_sec": 8.238670857972465, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.38436222076416, "lr": 0.0002, "elapsed_sec": 31805.19940161705, "step_time_sec": 8.23632585600717, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.232585906982422, "lr": 0.0002, "elapsed_sec": 31813.437090158463, "step_time_sec": 8.237542605027556, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2775774002075195, "lr": 0.0002, "elapsed_sec": 31821.675146579742, "step_time_sec": 8.237857159983832, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.289498805999756, "lr": 0.0002, "elapsed_sec": 31829.913162469864, "step_time_sec": 8.237925347988494, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4596171379089355, "lr": 0.0002, "elapsed_sec": 31838.150994300842, "step_time_sec": 8.237639254017267, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.305646896362305, "lr": 0.0002, "elapsed_sec": 31846.38866829872, "step_time_sec": 8.237497082038317, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2880377769470215, "lr": 0.0002, "elapsed_sec": 31854.6271007061, "step_time_sec": 8.238265239051543, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.182316780090332, "lr": 0.0002, "elapsed_sec": 31862.864604234695, "step_time_sec": 8.237302746973, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.388816833496094, "lr": 0.0002, "elapsed_sec": 31871.102724790573, "step_time_sec": 8.23799269500887, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2926225662231445, "lr": 0.0002, "elapsed_sec": 31879.34088754654, "step_time_sec": 8.238021887023933, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.209106922149658, "lr": 0.0002, "elapsed_sec": 31887.578543663025, "step_time_sec": 8.23744114098372, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2383599281311035, "lr": 0.0002, "elapsed_sec": 31895.817801475525, "step_time_sec": 8.239168651984073, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.132901191711426, "lr": 0.0002, "elapsed_sec": 31904.056309223175, "step_time_sec": 8.23828643100569, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.335829734802246, "lr": 0.0002, "elapsed_sec": 31912.295443058014, "step_time_sec": 8.238968517980538, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.365062236785889, "lr": 0.0002, "elapsed_sec": 31920.535041570663, "step_time_sec": 8.239437117998023, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.206733226776123, "lr": 0.0002, "elapsed_sec": 31928.772518634796, "step_time_sec": 8.23730052204337, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.253090858459473, "lr": 0.0002, "elapsed_sec": 31937.010960817337, "step_time_sec": 8.238326028978918, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.19545316696167, "lr": 0.0002, "elapsed_sec": 31945.247319221497, "step_time_sec": 8.236167064984329, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.216259479522705, "lr": 0.0002, "elapsed_sec": 31953.487720489502, "step_time_sec": 8.240223057975527, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.316081523895264, "lr": 0.0002, "elapsed_sec": 31961.726293325424, "step_time_sec": 8.238405796990264, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.973722219467163, "lr": 0.0002, "elapsed_sec": 31969.965227127075, "step_time_sec": 8.23880489898147, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.404300212860107, "lr": 0.0002, "elapsed_sec": 31978.205825567245, "step_time_sec": 8.240417386987247, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.264021396636963, "lr": 0.0002, "elapsed_sec": 31986.443237543106, "step_time_sec": 8.237272528000176, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.323495388031006, "lr": 0.0002, "elapsed_sec": 31994.68195104599, "step_time_sec": 8.238489038019907, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.234013557434082, "lr": 0.0002, "elapsed_sec": 32002.920735120773, "step_time_sec": 8.238652199972421, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.232490062713623, "lr": 0.0002, "elapsed_sec": 32011.158922433853, "step_time_sec": 8.238007364037912, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.38837194442749, "lr": 0.0002, "elapsed_sec": 32019.397679805756, "step_time_sec": 8.23863567499211, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.273307800292969, "lr": 0.0002, "elapsed_sec": 32027.635472536087, "step_time_sec": 8.23761910002213, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.244423866271973, "lr": 0.0002, "elapsed_sec": 32035.875923395157, "step_time_sec": 8.24024496896891, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.225288391113281, "lr": 0.0002, "elapsed_sec": 32044.115313768387, "step_time_sec": 8.239248564990703, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.277928829193115, "lr": 0.0002, "elapsed_sec": 32052.35442852974, "step_time_sec": 8.239016879990231, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.320356845855713, "lr": 0.0002, "elapsed_sec": 32060.593227624893, "step_time_sec": 8.238611559034325, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.240963459014893, "lr": 0.0002, "elapsed_sec": 32068.830454587936, "step_time_sec": 8.237020548956934, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.265176296234131, "lr": 0.0002, "elapsed_sec": 32077.069984197617, "step_time_sec": 8.239428049011622, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.114509582519531, "lr": 0.0002, "elapsed_sec": 32085.30756831169, "step_time_sec": 8.237373567011673, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2275800704956055, "lr": 0.0002, "elapsed_sec": 32093.545462846756, "step_time_sec": 8.237730754015502, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.394952297210693, "lr": 0.0002, "elapsed_sec": 32101.784802913666, "step_time_sec": 8.239179608994164, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.269394397735596, "lr": 0.0002, "elapsed_sec": 32110.022344827652, "step_time_sec": 8.237452494038735, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.204465866088867, "lr": 0.0002, "elapsed_sec": 32118.260910987854, "step_time_sec": 8.238369258993771, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.207573413848877, "lr": 0.0002, "elapsed_sec": 32126.498532295227, "step_time_sec": 8.237450082029682, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.284860610961914, "lr": 0.0002, "elapsed_sec": 32134.737977027893, "step_time_sec": 8.239272773033008, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.289801120758057, "lr": 0.0002, "elapsed_sec": 32142.975025892258, "step_time_sec": 8.236899418989196, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3670196533203125, "lr": 0.0002, "elapsed_sec": 32151.213394403458, "step_time_sec": 8.238192309974693, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.261650085449219, "lr": 0.0002, "elapsed_sec": 32159.453268051147, "step_time_sec": 8.23975348501699, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.32681131362915, "lr": 0.0002, "elapsed_sec": 32167.691975831985, "step_time_sec": 8.238483556022402, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.225473403930664, "lr": 0.0002, "elapsed_sec": 32175.931531190872, "step_time_sec": 8.23937569197733, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.357000350952148, "lr": 0.0002, "elapsed_sec": 32184.17157292366, "step_time_sec": 8.239871496974956, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.066219806671143, "lr": 0.0002, "elapsed_sec": 32192.408575057983, "step_time_sec": 8.236852900998201, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0920634269714355, "lr": 0.0002, "elapsed_sec": 32200.647943019867, "step_time_sec": 8.239215996989515, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.292657375335693, "lr": 0.0002, "elapsed_sec": 32208.885159015656, "step_time_sec": 8.23704621498473, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.347927570343018, "lr": 0.0002, "elapsed_sec": 32217.124191761017, "step_time_sec": 8.238872589019593, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.197340488433838, "lr": 0.0002, "elapsed_sec": 32225.362244606018, "step_time_sec": 8.237860905006528, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.104312419891357, "lr": 0.0002, "elapsed_sec": 32233.599633932114, "step_time_sec": 8.237206978024915, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2811737060546875, "lr": 0.0002, "elapsed_sec": 32241.838830709457, "step_time_sec": 8.239024520036764, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.223499298095703, "lr": 0.0002, "elapsed_sec": 32250.076572418213, "step_time_sec": 8.237617984996177, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.362332820892334, "lr": 0.0002, "elapsed_sec": 32258.315844535828, "step_time_sec": 8.239061819040217, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.280002593994141, "lr": 0.0002, "elapsed_sec": 32266.554978132248, "step_time_sec": 8.238981098984368, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.310129165649414, "lr": 0.0002, "elapsed_sec": 32274.79252433777, "step_time_sec": 8.237369612965267, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.13959264755249, "lr": 0.0002, "elapsed_sec": 32283.032371282578, "step_time_sec": 8.239691829949152, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.257194995880127, "lr": 0.0002, "elapsed_sec": 32291.270545482635, "step_time_sec": 8.238017149968073, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.198663711547852, "lr": 0.0002, "elapsed_sec": 32299.51034641266, "step_time_sec": 8.23967973398976, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.298747539520264, "lr": 0.0002, "elapsed_sec": 32307.75011062622, "step_time_sec": 8.239533276995644, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.290701389312744, "lr": 0.0002, "elapsed_sec": 32315.98733854294, "step_time_sec": 8.237127113970928, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.297969341278076, "lr": 0.0002, "elapsed_sec": 32324.22665786743, "step_time_sec": 8.239126825006679, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.271695137023926, "lr": 0.0002, "elapsed_sec": 32332.46460723877, "step_time_sec": 8.237782632000744, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.297766208648682, "lr": 0.0002, "elapsed_sec": 32340.704505205154, "step_time_sec": 8.239747535029892, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.279171943664551, "lr": 0.0002, "elapsed_sec": 32348.94332933426, "step_time_sec": 8.238644135009963, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.210034370422363, "lr": 0.0002, "elapsed_sec": 32357.18100118637, "step_time_sec": 8.237553311977535, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.386277198791504, "lr": 0.0002, "elapsed_sec": 32365.418220043182, "step_time_sec": 8.237043071014341, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.339313507080078, "lr": 0.0002, "elapsed_sec": 32373.655835151672, "step_time_sec": 8.237403273000382, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.284798622131348, "lr": 0.0002, "elapsed_sec": 32381.895483493805, "step_time_sec": 8.239500493975356, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.286204814910889, "lr": 0.0002, "elapsed_sec": 32390.13386464119, "step_time_sec": 8.238268479995895, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2246222496032715, "lr": 0.0002, "elapsed_sec": 32398.37267756462, "step_time_sec": 8.238608826010022, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2861647605896, "lr": 0.0002, "elapsed_sec": 32406.611662626266, "step_time_sec": 8.238864424987696, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.456000328063965, "lr": 0.0002, "elapsed_sec": 32414.848373174667, "step_time_sec": 8.236508161993697, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3222551345825195, "lr": 0.0002, "elapsed_sec": 32423.087052345276, "step_time_sec": 8.238518105994444, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.189384460449219, "lr": 0.0002, "elapsed_sec": 32431.324673891068, "step_time_sec": 8.237488026963547, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.27869176864624, "lr": 0.0002, "elapsed_sec": 32439.562500476837, "step_time_sec": 8.237673396011814, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.174830913543701, "lr": 0.0002, "elapsed_sec": 32447.800452709198, "step_time_sec": 8.237752224958967, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.129342079162598, "lr": 0.0002, "elapsed_sec": 32456.038737773895, "step_time_sec": 8.238120081019588, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.095276832580566, "lr": 0.0002, "elapsed_sec": 32464.278088092804, "step_time_sec": 8.239182001969311, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.356851577758789, "lr": 0.0002, "elapsed_sec": 32472.516736507416, "step_time_sec": 8.238506752008107, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.268824577331543, "lr": 0.0002, "elapsed_sec": 32480.754880428314, "step_time_sec": 8.23796500096796, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.166153907775879, "lr": 0.0002, "elapsed_sec": 32488.993141651154, "step_time_sec": 8.238108698977157, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.170119285583496, "lr": 0.0002, "elapsed_sec": 32497.230170965195, "step_time_sec": 8.236899713985622, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.274721622467041, "lr": 0.0002, "elapsed_sec": 32505.469811439514, "step_time_sec": 8.239465866005048, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.200294494628906, "lr": 0.0002, "elapsed_sec": 32513.706596136093, "step_time_sec": 8.236595491995104, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.246576309204102, "lr": 0.0002, "elapsed_sec": 32521.94603562355, "step_time_sec": 8.239249481004663, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.346077919006348, "lr": 0.0002, "elapsed_sec": 32530.184948682785, "step_time_sec": 8.238792326010298, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.327313423156738, "lr": 0.0002, "elapsed_sec": 32538.423337697983, "step_time_sec": 8.238192554039415, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1512885093688965, "lr": 0.0002, "elapsed_sec": 32546.661329746246, "step_time_sec": 8.237885326961987, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3461222648620605, "lr": 0.0002, "elapsed_sec": 32554.898711919785, "step_time_sec": 8.237184400029946, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.369561195373535, "lr": 0.0002, "elapsed_sec": 32563.137835264206, "step_time_sec": 8.238985000993125, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.274317741394043, "lr": 0.0002, "elapsed_sec": 32571.376103401184, "step_time_sec": 8.238072698994074, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.167466640472412, "lr": 0.0002, "elapsed_sec": 32579.614625692368, "step_time_sec": 8.23839534400031, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.471753120422363, "lr": 0.0002, "elapsed_sec": 32587.853085279465, "step_time_sec": 8.23828297003638, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.391533851623535, "lr": 0.0002, "elapsed_sec": 32596.090135335922, "step_time_sec": 8.236868053034414, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.441435813903809, "lr": 0.0002, "elapsed_sec": 32604.32860159874, "step_time_sec": 8.238293014001101, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.120610237121582, "lr": 0.0002, "elapsed_sec": 32612.565881967545, "step_time_sec": 8.237114857998677, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3146281242370605, "lr": 0.0002, "elapsed_sec": 32620.804232120514, "step_time_sec": 8.238184084999375, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.255404472351074, "lr": 0.0002, "elapsed_sec": 32629.04207420349, "step_time_sec": 8.237699195975438, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.210740089416504, "lr": 0.0002, "elapsed_sec": 32637.280071020126, "step_time_sec": 8.237835233972874, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.308083534240723, "lr": 0.0002, "elapsed_sec": 32645.518192768097, "step_time_sec": 8.23795089701889, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.414462566375732, "lr": 0.0002, "elapsed_sec": 32653.7560839653, "step_time_sec": 8.237744374957401, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.334253311157227, "lr": 0.0002, "elapsed_sec": 32661.995944023132, "step_time_sec": 8.239688680041581, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.277858734130859, "lr": 0.0002, "elapsed_sec": 32670.234958171844, "step_time_sec": 8.238802582025528, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.267975807189941, "lr": 0.0002, "elapsed_sec": 32678.47136116028, "step_time_sec": 8.236231248010881, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.224398612976074, "lr": 0.0002, "elapsed_sec": 32686.71022582054, "step_time_sec": 8.23874565801816, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3030853271484375, "lr": 0.0002, "elapsed_sec": 32694.947985887527, "step_time_sec": 8.237560235953424, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.246865749359131, "lr": 0.0002, "elapsed_sec": 32703.187515735626, "step_time_sec": 8.239375712000765, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.303539276123047, "lr": 0.0002, "elapsed_sec": 32711.426611185074, "step_time_sec": 8.238919173018076, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.21943473815918, "lr": 0.0002, "elapsed_sec": 32719.6632874012, "step_time_sec": 8.236580826051068, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.194304943084717, "lr": 0.0002, "elapsed_sec": 32727.901225566864, "step_time_sec": 8.237738871015608, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.255188941955566, "lr": 0.0002, "elapsed_sec": 32736.138360261917, "step_time_sec": 8.236962033028249, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.19265604019165, "lr": 0.0002, "elapsed_sec": 32744.377039194107, "step_time_sec": 8.23850015498465, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.482476711273193, "lr": 0.0002, "elapsed_sec": 32752.61425590515, "step_time_sec": 8.237039798987098, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.267163276672363, "lr": 0.0002, "elapsed_sec": 32760.853142499924, "step_time_sec": 8.23872596397996, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.336217880249023, "lr": 0.0002, "elapsed_sec": 32769.09008336067, "step_time_sec": 8.236773297016043, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.23851203918457, "lr": 0.0002, "elapsed_sec": 32777.327676296234, "step_time_sec": 8.237450999964494, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.293518543243408, "lr": 0.0002, "elapsed_sec": 32785.56479430199, "step_time_sec": 8.236941164010204, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2242841720581055, "lr": 0.0002, "elapsed_sec": 32793.803285360336, "step_time_sec": 8.238319089985453, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.287110805511475, "lr": 0.0002, "elapsed_sec": 32802.041751384735, "step_time_sec": 8.238346024998464, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.421627998352051, "lr": 0.0002, "elapsed_sec": 32810.2793533802, "step_time_sec": 8.237400844984222, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.405486583709717, "lr": 0.0002, "elapsed_sec": 32818.517973184586, "step_time_sec": 8.23846807499649, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.184892654418945, "lr": 0.0002, "elapsed_sec": 32826.755338430405, "step_time_sec": 8.237252696999349, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.249536991119385, "lr": 0.0002, "elapsed_sec": 32834.99215531349, "step_time_sec": 8.236626780009829, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.200733184814453, "lr": 0.0002, "elapsed_sec": 32843.23166394234, "step_time_sec": 8.239327688002959, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.419333457946777, "lr": 0.0002, "elapsed_sec": 32851.46884393692, "step_time_sec": 8.236990149016492, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.236555576324463, "lr": 0.0002, "elapsed_sec": 32859.70782208443, "step_time_sec": 8.238877131021582, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.392820835113525, "lr": 0.0002, "elapsed_sec": 32867.94565010071, "step_time_sec": 8.237659089965746, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2541890144348145, "lr": 0.0002, "elapsed_sec": 32876.184017419815, "step_time_sec": 8.238198296981864, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.337652683258057, "lr": 0.0002, "elapsed_sec": 32884.422626018524, "step_time_sec": 8.238398881978355, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.291143417358398, "lr": 0.0002, "elapsed_sec": 32892.66043186188, "step_time_sec": 8.23766995902406, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3128581047058105, "lr": 0.0002, "elapsed_sec": 32900.89981365204, "step_time_sec": 8.239196436014026, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.251355171203613, "lr": 0.0002, "elapsed_sec": 32909.13874030113, "step_time_sec": 8.238813080999535, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.212435722351074, "lr": 0.0002, "elapsed_sec": 32917.3762049675, "step_time_sec": 8.237271215999499, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.322213649749756, "lr": 0.0002, "elapsed_sec": 32925.61479330063, "step_time_sec": 8.23844112403458, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2618408203125, "lr": 0.0002, "elapsed_sec": 32933.85234332085, "step_time_sec": 8.237447411054745, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.190688610076904, "lr": 0.0002, "elapsed_sec": 32942.092599868774, "step_time_sec": 8.240051358996425, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.186803817749023, "lr": 0.0002, "elapsed_sec": 32950.33156251907, "step_time_sec": 8.238756120961625, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.269508361816406, "lr": 0.0002, "elapsed_sec": 32958.56792449951, "step_time_sec": 8.236168211966287, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.250082492828369, "lr": 0.0002, "elapsed_sec": 32966.80710530281, "step_time_sec": 8.23902741499478, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.369337558746338, "lr": 0.0002, "elapsed_sec": 32975.045214891434, "step_time_sec": 8.238013753958512, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3027663230896, "lr": 0.0002, "elapsed_sec": 32983.28506374359, "step_time_sec": 8.239666932960972, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.328153610229492, "lr": 0.0002, "elapsed_sec": 32991.523584365845, "step_time_sec": 8.23837266600458, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.288069725036621, "lr": 0.0002, "elapsed_sec": 32999.76188850403, "step_time_sec": 8.238078457012307, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.177012920379639, "lr": 0.0002, "elapsed_sec": 33008.00011777878, "step_time_sec": 8.238118287990801, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.307654857635498, "lr": 0.0002, "elapsed_sec": 33016.23803329468, "step_time_sec": 8.237719429016579, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.265021324157715, "lr": 0.0002, "elapsed_sec": 33024.47738242149, "step_time_sec": 8.239189100975636, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.322574138641357, "lr": 0.0002, "elapsed_sec": 33032.71629357338, "step_time_sec": 8.23879183497047, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.236466884613037, "lr": 0.0002, "elapsed_sec": 33040.95355296135, "step_time_sec": 8.237121128011495, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.304058074951172, "lr": 0.0002, "elapsed_sec": 33049.192307949066, "step_time_sec": 8.238558203971479, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.399951457977295, "lr": 0.0002, "elapsed_sec": 33057.43034172058, "step_time_sec": 8.23786618898157, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.201830863952637, "lr": 0.0002, "elapsed_sec": 33065.670711278915, "step_time_sec": 8.240189846022986, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.232452392578125, "lr": 0.0002, "elapsed_sec": 33073.9099919796, "step_time_sec": 8.239088352012914, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.177659034729004, "lr": 0.0002, "elapsed_sec": 33082.14745736122, "step_time_sec": 8.237319517007563, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.156391620635986, "lr": 0.0002, "elapsed_sec": 33090.385343551636, "step_time_sec": 8.23771133396076, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.41533088684082, "lr": 0.0002, "elapsed_sec": 33098.62278676033, "step_time_sec": 8.237278455984779, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.289748668670654, "lr": 0.0002, "elapsed_sec": 33106.86088705063, "step_time_sec": 8.237946823006496, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.33282995223999, "lr": 0.0002, "elapsed_sec": 33115.099358320236, "step_time_sec": 8.238349688996095, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.22662878036499, "lr": 0.0002, "elapsed_sec": 33123.337705135345, "step_time_sec": 8.238150328979827, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3116774559021, "lr": 0.0002, "elapsed_sec": 33131.57665872574, "step_time_sec": 8.238753631012514, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.307544231414795, "lr": 0.0002, "elapsed_sec": 33139.813192367554, "step_time_sec": 8.23636715603061, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3525190353393555, "lr": 0.0002, "elapsed_sec": 33148.05100440979, "step_time_sec": 8.237671336974017, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.050108909606934, "lr": 0.0002, "elapsed_sec": 33156.28887462616, "step_time_sec": 8.237695874995552, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.230567455291748, "lr": 0.0002, "elapsed_sec": 33164.52673602104, "step_time_sec": 8.23770992201753, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.181527614593506, "lr": 0.0002, "elapsed_sec": 33172.7646753788, "step_time_sec": 8.23777023103321, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.286514759063721, "lr": 0.0002, "elapsed_sec": 33181.00248122215, "step_time_sec": 8.237715233990457, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.220928192138672, "lr": 0.0002, "elapsed_sec": 33189.23956441879, "step_time_sec": 8.236912544001825, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.274075031280518, "lr": 0.0002, "elapsed_sec": 33197.47640681267, "step_time_sec": 8.23665858298773, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.483989238739014, "lr": 0.0002, "elapsed_sec": 33205.71426844597, "step_time_sec": 8.237734046997502, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.317939281463623, "lr": 0.0002, "elapsed_sec": 33213.951446294785, "step_time_sec": 51.50038216699613, "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": 0.8423655309597962, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 4001, "loss": 4.332525253295898, "lr": 0.0002, "elapsed_sec": 33265.456033706665, "step_time_sec": 8.241082986991387, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.216709613800049, "lr": 0.0002, "elapsed_sec": 33273.687683820724, "step_time_sec": 8.231454393011518, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.402665615081787, "lr": 0.0002, "elapsed_sec": 33281.92375946045, "step_time_sec": 8.235947320994455, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.363664627075195, "lr": 0.0002, "elapsed_sec": 33290.161146879196, "step_time_sec": 8.23723211698234, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.407895565032959, "lr": 0.0002, "elapsed_sec": 33298.4002430439, "step_time_sec": 8.238938872003928, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.269865036010742, "lr": 0.0002, "elapsed_sec": 33306.63818860054, "step_time_sec": 8.237811177968979, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.228655815124512, "lr": 0.0002, "elapsed_sec": 33314.87721180916, "step_time_sec": 8.23878373199841, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.347040176391602, "lr": 0.0002, "elapsed_sec": 33323.11579537392, "step_time_sec": 8.238420801004395, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.567370891571045, "lr": 0.0002, "elapsed_sec": 33331.35446381569, "step_time_sec": 8.238536972028669, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.242109775543213, "lr": 0.0002, "elapsed_sec": 33339.5922293663, "step_time_sec": 8.237556650012266, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4741530418396, "lr": 0.0002, "elapsed_sec": 33347.83001613617, "step_time_sec": 8.237587073992472, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.313460350036621, "lr": 0.0002, "elapsed_sec": 33356.06909227371, "step_time_sec": 8.238955614040606, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.33235502243042, "lr": 0.0002, "elapsed_sec": 33364.30582070351, "step_time_sec": 8.236546037020162, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2974114418029785, "lr": 0.0002, "elapsed_sec": 33372.54451465607, "step_time_sec": 8.238559313002042, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3408522605896, "lr": 0.0002, "elapsed_sec": 33380.782623529434, "step_time_sec": 8.237908931972925, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.269992828369141, "lr": 0.0002, "elapsed_sec": 33389.019325494766, "step_time_sec": 8.236593117006123, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.381808280944824, "lr": 0.0002, "elapsed_sec": 33397.25711941719, "step_time_sec": 8.237606509996112, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.314624309539795, "lr": 0.0002, "elapsed_sec": 33405.494463682175, "step_time_sec": 8.237164644990116, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.323312759399414, "lr": 0.0002, "elapsed_sec": 33413.73294591904, "step_time_sec": 8.238341592019424, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.177632808685303, "lr": 0.0002, "elapsed_sec": 33421.97013092041, "step_time_sec": 8.237027812981978, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.358078956604004, "lr": 0.0002, "elapsed_sec": 33430.2069914341, "step_time_sec": 8.236651866987813, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.337465763092041, "lr": 0.0002, "elapsed_sec": 33438.444910764694, "step_time_sec": 8.23776348598767, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.252438545227051, "lr": 0.0002, "elapsed_sec": 33446.68258905411, "step_time_sec": 8.237485203018878, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.212049961090088, "lr": 0.0002, "elapsed_sec": 33454.92037510872, "step_time_sec": 8.237628744973335, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.316171646118164, "lr": 0.0002, "elapsed_sec": 33463.15721273422, "step_time_sec": 8.236687733966392, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.408568859100342, "lr": 0.0002, "elapsed_sec": 33471.39534687996, "step_time_sec": 8.237980818958022, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.29196310043335, "lr": 0.0002, "elapsed_sec": 33479.63329076767, "step_time_sec": 8.237776194990147, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.293582439422607, "lr": 0.0002, "elapsed_sec": 33487.86944389343, "step_time_sec": 8.235974462004378, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.27224063873291, "lr": 0.0002, "elapsed_sec": 33496.1055495739, "step_time_sec": 8.235947109991685, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.241580009460449, "lr": 0.0002, "elapsed_sec": 33504.34328198433, "step_time_sec": 8.23755970201455, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.176394939422607, "lr": 0.0002, "elapsed_sec": 33512.58244752884, "step_time_sec": 8.239016701991204, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.141231536865234, "lr": 0.0002, "elapsed_sec": 33520.81869959831, "step_time_sec": 8.236065295001026, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.221315860748291, "lr": 0.0002, "elapsed_sec": 33529.05776286125, "step_time_sec": 8.23891801002901, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.294743061065674, "lr": 0.0002, "elapsed_sec": 33537.29552555084, "step_time_sec": 8.237598614010494, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1932573318481445, "lr": 0.0002, "elapsed_sec": 33545.53363990784, "step_time_sec": 8.237948850961402, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.309041976928711, "lr": 0.0002, "elapsed_sec": 33553.77133631706, "step_time_sec": 8.237588488962501, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.184100151062012, "lr": 0.0002, "elapsed_sec": 33562.00873017311, "step_time_sec": 8.23719226103276, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.28866720199585, "lr": 0.0002, "elapsed_sec": 33570.247354745865, "step_time_sec": 8.238475841004401, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2675557136535645, "lr": 0.0002, "elapsed_sec": 33578.48372507095, "step_time_sec": 8.236207249981817, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.175292015075684, "lr": 0.0002, "elapsed_sec": 33586.72187709808, "step_time_sec": 8.237972964008804, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.218161582946777, "lr": 0.0002, "elapsed_sec": 33594.957691669464, "step_time_sec": 8.235689705994446, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.174448490142822, "lr": 0.0002, "elapsed_sec": 33603.19629073143, "step_time_sec": 8.23842185101239, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3944621086120605, "lr": 0.0002, "elapsed_sec": 33611.434473752975, "step_time_sec": 8.238045959034935, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.410107612609863, "lr": 0.0002, "elapsed_sec": 33619.67138051987, "step_time_sec": 8.236684083996806, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.173929691314697, "lr": 0.0002, "elapsed_sec": 33627.90977978706, "step_time_sec": 8.23824362497544, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.282301425933838, "lr": 0.0002, "elapsed_sec": 33636.14615535736, "step_time_sec": 8.236223515006714, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.361879348754883, "lr": 0.0002, "elapsed_sec": 33644.38561487198, "step_time_sec": 8.239294926985167, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.261289596557617, "lr": 0.0002, "elapsed_sec": 33652.623747348785, "step_time_sec": 8.237937699013855, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2960991859436035, "lr": 0.0002, "elapsed_sec": 33660.86314344406, "step_time_sec": 8.23922607494751, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.291438102722168, "lr": 0.0002, "elapsed_sec": 33669.1012942791, "step_time_sec": 8.238012937945314, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.250870704650879, "lr": 0.0002, "elapsed_sec": 33677.33886599541, "step_time_sec": 8.237356821016874, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.325672626495361, "lr": 0.0002, "elapsed_sec": 33685.57735490799, "step_time_sec": 8.23834432195872, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.219729900360107, "lr": 0.0002, "elapsed_sec": 33693.81497788429, "step_time_sec": 8.237448151980061, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.394983291625977, "lr": 0.0002, "elapsed_sec": 33702.05251407623, "step_time_sec": 8.237368349044118, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.320186614990234, "lr": 0.0002, "elapsed_sec": 33710.29168462753, "step_time_sec": 8.239004239963833, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.328876972198486, "lr": 0.0002, "elapsed_sec": 33729.8540289402, "step_time_sec": 19.562194771016948, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2226643562316895, "lr": 0.0002, "elapsed_sec": 33738.08876991272, "step_time_sec": 8.234594537003431, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.239120006561279, "lr": 0.0002, "elapsed_sec": 33746.32674217224, "step_time_sec": 8.237839991983492, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.370445251464844, "lr": 0.0002, "elapsed_sec": 33754.563866853714, "step_time_sec": 8.236960040056147, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2322163581848145, "lr": 0.0002, "elapsed_sec": 33762.79957270622, "step_time_sec": 8.235516852000728, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.196381568908691, "lr": 0.0002, "elapsed_sec": 33771.03806948662, "step_time_sec": 8.23833022499457, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.248366355895996, "lr": 0.0002, "elapsed_sec": 33779.27598261833, "step_time_sec": 8.237789048987906, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.305052280426025, "lr": 0.0002, "elapsed_sec": 33787.51413178444, "step_time_sec": 8.237970751011744, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.285393714904785, "lr": 0.0002, "elapsed_sec": 33795.75404524803, "step_time_sec": 8.23978249495849, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.372483253479004, "lr": 0.0002, "elapsed_sec": 33803.9911775589, "step_time_sec": 8.23692797095282, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.328776836395264, "lr": 0.0002, "elapsed_sec": 33812.22990822792, "step_time_sec": 8.238554693991318, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4391961097717285, "lr": 0.0002, "elapsed_sec": 33820.46725344658, "step_time_sec": 8.237157705007121, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1974663734436035, "lr": 0.0002, "elapsed_sec": 33828.70478391647, "step_time_sec": 8.237435235001612, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.283110618591309, "lr": 0.0002, "elapsed_sec": 33836.943289756775, "step_time_sec": 8.238299926975742, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.259085178375244, "lr": 0.0002, "elapsed_sec": 33845.18125319481, "step_time_sec": 8.237792302039452, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.12278938293457, "lr": 0.0002, "elapsed_sec": 33853.42003488541, "step_time_sec": 8.238632357970346, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.370359420776367, "lr": 0.0002, "elapsed_sec": 33861.65746426582, "step_time_sec": 8.237257477012463, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.293613433837891, "lr": 0.0002, "elapsed_sec": 33869.8961918354, "step_time_sec": 8.238560213998426, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.210999965667725, "lr": 0.0002, "elapsed_sec": 33878.13447022438, "step_time_sec": 8.238140613015275, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2811503410339355, "lr": 0.0002, "elapsed_sec": 33886.37285518646, "step_time_sec": 8.238220894010738, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.220047473907471, "lr": 0.0002, "elapsed_sec": 33894.61041069031, "step_time_sec": 8.237395484989975, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.310911178588867, "lr": 0.0002, "elapsed_sec": 33902.84814333916, "step_time_sec": 8.23757607495645, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2801194190979, "lr": 0.0002, "elapsed_sec": 33911.0858502388, "step_time_sec": 8.237490304978564, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.125253677368164, "lr": 0.0002, "elapsed_sec": 33919.32332921028, "step_time_sec": 8.237334525969345, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.216183185577393, "lr": 0.0002, "elapsed_sec": 33927.561609745026, "step_time_sec": 8.23813945997972, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3554301261901855, "lr": 0.0002, "elapsed_sec": 33935.79924368858, "step_time_sec": 8.237485388002824, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1045613288879395, "lr": 0.0002, "elapsed_sec": 33944.03718543053, "step_time_sec": 8.237753635970876, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.251073360443115, "lr": 0.0002, "elapsed_sec": 33952.275223493576, "step_time_sec": 8.23788206197787, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.321389675140381, "lr": 0.0002, "elapsed_sec": 33960.51216006279, "step_time_sec": 8.236760500003584, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.278934001922607, "lr": 0.0002, "elapsed_sec": 33968.75057888031, "step_time_sec": 8.238268173008692, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.255404472351074, "lr": 0.0002, "elapsed_sec": 33976.98903465271, "step_time_sec": 8.238286739971954, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.377612590789795, "lr": 0.0002, "elapsed_sec": 33985.22751784325, "step_time_sec": 8.238328900013585, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.319643020629883, "lr": 0.0002, "elapsed_sec": 33993.46619939804, "step_time_sec": 8.238483111024834, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.194125175476074, "lr": 0.0002, "elapsed_sec": 34001.70444536209, "step_time_sec": 8.238106962002348, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.320428371429443, "lr": 0.0002, "elapsed_sec": 34009.94213461876, "step_time_sec": 8.237512368010357, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.152115821838379, "lr": 0.0002, "elapsed_sec": 34018.18150424957, "step_time_sec": 8.239211731008254, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.095613479614258, "lr": 0.0002, "elapsed_sec": 34026.41919660568, "step_time_sec": 8.237576734973118, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.087966442108154, "lr": 0.0002, "elapsed_sec": 34034.65826869011, "step_time_sec": 8.238847839005757, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.137373924255371, "lr": 0.0002, "elapsed_sec": 34042.89507961273, "step_time_sec": 8.23665014602011, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.338861465454102, "lr": 0.0002, "elapsed_sec": 34051.133293390274, "step_time_sec": 8.238083393021952, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.190206050872803, "lr": 0.0002, "elapsed_sec": 34059.37110877037, "step_time_sec": 8.237602412002161, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.276089668273926, "lr": 0.0002, "elapsed_sec": 34067.60931825638, "step_time_sec": 8.238038994953968, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3993120193481445, "lr": 0.0002, "elapsed_sec": 34075.848604917526, "step_time_sec": 8.2391245450126, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.286609649658203, "lr": 0.0002, "elapsed_sec": 34084.08664035797, "step_time_sec": 8.237885862996336, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.293829441070557, "lr": 0.0002, "elapsed_sec": 34092.32566142082, "step_time_sec": 8.238887470972259, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.231761455535889, "lr": 0.0002, "elapsed_sec": 34100.562757730484, "step_time_sec": 8.236876701004803, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.135360240936279, "lr": 0.0002, "elapsed_sec": 34108.8018951416, "step_time_sec": 8.239055418001954, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3738603591918945, "lr": 0.0002, "elapsed_sec": 34117.04118561745, "step_time_sec": 8.239087084017228, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.356015682220459, "lr": 0.0002, "elapsed_sec": 34125.279270887375, "step_time_sec": 8.237941428960767, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.179507255554199, "lr": 0.0002, "elapsed_sec": 34133.517800569534, "step_time_sec": 8.238380958966445, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.493074893951416, "lr": 0.0002, "elapsed_sec": 34141.756093263626, "step_time_sec": 8.23817671497818, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.27776575088501, "lr": 0.0002, "elapsed_sec": 34149.99649834633, "step_time_sec": 8.24014836398419, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.216855049133301, "lr": 0.0002, "elapsed_sec": 34158.23377466202, "step_time_sec": 8.237135287956335, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.348235130310059, "lr": 0.0002, "elapsed_sec": 34166.470848321915, "step_time_sec": 8.23690992599586, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.260838985443115, "lr": 0.0002, "elapsed_sec": 34174.70898270607, "step_time_sec": 8.237909839022905, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.402682781219482, "lr": 0.0002, "elapsed_sec": 34182.94583487511, "step_time_sec": 8.236720105982386, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.298460006713867, "lr": 0.0002, "elapsed_sec": 34191.18410778046, "step_time_sec": 8.238139282038901, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.331075668334961, "lr": 0.0002, "elapsed_sec": 34199.421921014786, "step_time_sec": 8.237582030007616, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.189430236816406, "lr": 0.0002, "elapsed_sec": 34207.66097307205, "step_time_sec": 8.238902110024355, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.228514194488525, "lr": 0.0002, "elapsed_sec": 34215.898873090744, "step_time_sec": 8.237724085047375, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.144121170043945, "lr": 0.0002, "elapsed_sec": 34224.13590955734, "step_time_sec": 8.236877396004274, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.262341022491455, "lr": 0.0002, "elapsed_sec": 34232.37508034706, "step_time_sec": 8.23901964299148, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.234873294830322, "lr": 0.0002, "elapsed_sec": 34240.61234116554, "step_time_sec": 8.237094960990362, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.285311222076416, "lr": 0.0002, "elapsed_sec": 34248.852053403854, "step_time_sec": 8.239559206995182, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2779693603515625, "lr": 0.0002, "elapsed_sec": 34257.09052681923, "step_time_sec": 8.238293550966773, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.229325294494629, "lr": 0.0002, "elapsed_sec": 34265.32925724983, "step_time_sec": 8.238594000984449, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.358031749725342, "lr": 0.0002, "elapsed_sec": 34273.56774902344, "step_time_sec": 8.238307465973776, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.169337749481201, "lr": 0.0002, "elapsed_sec": 34281.80538034439, "step_time_sec": 8.237482509983238, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.242358207702637, "lr": 0.0002, "elapsed_sec": 34290.04522418976, "step_time_sec": 8.239648350980133, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.234091758728027, "lr": 0.0002, "elapsed_sec": 34298.28379535675, "step_time_sec": 8.238434227008838, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1415534019470215, "lr": 0.0002, "elapsed_sec": 34306.52147316933, "step_time_sec": 8.237477664952166, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.284631729125977, "lr": 0.0002, "elapsed_sec": 34314.75862836838, "step_time_sec": 8.236991719983052, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.13320779800415, "lr": 0.0002, "elapsed_sec": 34322.995987176895, "step_time_sec": 8.237216125999112, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.259149551391602, "lr": 0.0002, "elapsed_sec": 34331.23492240906, "step_time_sec": 8.238817966950592, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.391587734222412, "lr": 0.0002, "elapsed_sec": 34339.47261714935, "step_time_sec": 8.237500721996184, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.284672260284424, "lr": 0.0002, "elapsed_sec": 34347.71081352234, "step_time_sec": 8.238005299004726, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3489089012146, "lr": 0.0002, "elapsed_sec": 34355.94881057739, "step_time_sec": 8.237894910969771, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.272039890289307, "lr": 0.0002, "elapsed_sec": 34364.18567919731, "step_time_sec": 8.236664313008077, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.275091171264648, "lr": 0.0002, "elapsed_sec": 34372.42495536804, "step_time_sec": 8.239123976032715, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2605509757995605, "lr": 0.0002, "elapsed_sec": 34380.66490006447, "step_time_sec": 8.23978292196989, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.329466342926025, "lr": 0.0002, "elapsed_sec": 34388.90346431732, "step_time_sec": 8.238389346050099, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.454865455627441, "lr": 0.0002, "elapsed_sec": 34397.14201426506, "step_time_sec": 8.238429606019054, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.10579776763916, "lr": 0.0002, "elapsed_sec": 34405.3788292408, "step_time_sec": 8.236634857021272, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.246451377868652, "lr": 0.0002, "elapsed_sec": 34413.61781668663, "step_time_sec": 8.238856821961235, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.346693515777588, "lr": 0.0002, "elapsed_sec": 34421.85535645485, "step_time_sec": 8.237355835968629, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.213057041168213, "lr": 0.0002, "elapsed_sec": 34430.09321784973, "step_time_sec": 8.237661879044026, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.288091659545898, "lr": 0.0002, "elapsed_sec": 34438.33188056946, "step_time_sec": 8.238560744037386, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.14872407913208, "lr": 0.0002, "elapsed_sec": 34446.569522857666, "step_time_sec": 8.237494128989056, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.191847801208496, "lr": 0.0002, "elapsed_sec": 34454.807602882385, "step_time_sec": 8.237851000041701, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.314197063446045, "lr": 0.0002, "elapsed_sec": 34463.04468345642, "step_time_sec": 8.23693422198994, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2228522300720215, "lr": 0.0002, "elapsed_sec": 34471.28444314003, "step_time_sec": 8.239651183015667, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.465658187866211, "lr": 0.0002, "elapsed_sec": 34479.5228767395, "step_time_sec": 8.238221348030493, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.163425445556641, "lr": 0.0002, "elapsed_sec": 34487.76024675369, "step_time_sec": 8.237203151045833, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.314948081970215, "lr": 0.0002, "elapsed_sec": 34495.99713182449, "step_time_sec": 8.236689876997843, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.066415309906006, "lr": 0.0002, "elapsed_sec": 34504.2336564064, "step_time_sec": 8.236326258978806, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.232054710388184, "lr": 0.0002, "elapsed_sec": 34512.47171783447, "step_time_sec": 8.23792795900954, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2751688957214355, "lr": 0.0002, "elapsed_sec": 34520.71016216278, "step_time_sec": 8.238262550032232, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.26702356338501, "lr": 0.0002, "elapsed_sec": 34528.94914460182, "step_time_sec": 8.23885020497255, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.177634239196777, "lr": 0.0002, "elapsed_sec": 34537.188024520874, "step_time_sec": 8.238722727983259, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.219971179962158, "lr": 0.0002, "elapsed_sec": 34545.425819158554, "step_time_sec": 8.23759841202991, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.273309230804443, "lr": 0.0002, "elapsed_sec": 34553.664194345474, "step_time_sec": 8.238237175042741, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.373154640197754, "lr": 0.0002, "elapsed_sec": 34561.9010720253, "step_time_sec": 8.236720371991396, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2842206954956055, "lr": 0.0002, "elapsed_sec": 34570.13967394829, "step_time_sec": 8.238461784960236, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1671295166015625, "lr": 0.0002, "elapsed_sec": 34578.37746477127, "step_time_sec": 8.237608347961213, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2301177978515625, "lr": 0.0002, "elapsed_sec": 34586.61552357674, "step_time_sec": 8.237920260988176, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.206237316131592, "lr": 0.0002, "elapsed_sec": 34594.85458111763, "step_time_sec": 8.238864750019275, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.232614517211914, "lr": 0.0002, "elapsed_sec": 34603.09133434296, "step_time_sec": 8.236582083976828, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.174813747406006, "lr": 0.0002, "elapsed_sec": 34611.33036494255, "step_time_sec": 8.238905525009613, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.151431083679199, "lr": 0.0002, "elapsed_sec": 34619.567618370056, "step_time_sec": 8.237073794007301, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.134525775909424, "lr": 0.0002, "elapsed_sec": 34627.806549310684, "step_time_sec": 8.23876590601867, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.329133033752441, "lr": 0.0002, "elapsed_sec": 34636.04557275772, "step_time_sec": 8.238852873037104, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.136562347412109, "lr": 0.0002, "elapsed_sec": 34644.28231239319, "step_time_sec": 8.236575384973548, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.276406764984131, "lr": 0.0002, "elapsed_sec": 34652.52022981644, "step_time_sec": 8.237759559997357, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1364545822143555, "lr": 0.0002, "elapsed_sec": 34660.75600385666, "step_time_sec": 8.235644255997613, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.196192741394043, "lr": 0.0002, "elapsed_sec": 34668.99517130852, "step_time_sec": 8.238949320977554, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.204946994781494, "lr": 0.0002, "elapsed_sec": 34677.23457479477, "step_time_sec": 8.239292718993966, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.17025089263916, "lr": 0.0002, "elapsed_sec": 34685.4721467495, "step_time_sec": 8.237403373001143, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.192486763000488, "lr": 0.0002, "elapsed_sec": 34693.71045470238, "step_time_sec": 8.238165088987444, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.31148099899292, "lr": 0.0002, "elapsed_sec": 34701.94802284241, "step_time_sec": 8.237371287017595, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.078837871551514, "lr": 0.0002, "elapsed_sec": 34710.18675994873, "step_time_sec": 8.238522052008193, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.339496612548828, "lr": 0.0002, "elapsed_sec": 34718.42447471619, "step_time_sec": 8.237548056000378, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.287709712982178, "lr": 0.0002, "elapsed_sec": 34726.662700891495, "step_time_sec": 8.238080657029059, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.231797218322754, "lr": 0.0002, "elapsed_sec": 34734.90075588226, "step_time_sec": 8.237865836010315, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.29251766204834, "lr": 0.0002, "elapsed_sec": 34743.138899326324, "step_time_sec": 8.237991394009441, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.026669025421143, "lr": 0.0002, "elapsed_sec": 34751.37683939934, "step_time_sec": 8.237811931991018, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.138430595397949, "lr": 0.0002, "elapsed_sec": 34759.61336708069, "step_time_sec": 8.23633410903858, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2363176345825195, "lr": 0.0002, "elapsed_sec": 34767.85242319107, "step_time_sec": 8.238885935978033, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.212788105010986, "lr": 0.0002, "elapsed_sec": 34776.09000611305, "step_time_sec": 8.23743951099459, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.269416332244873, "lr": 0.0002, "elapsed_sec": 34784.326756477356, "step_time_sec": 8.236545430030674, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.240116596221924, "lr": 0.0002, "elapsed_sec": 34792.564826488495, "step_time_sec": 8.237942899984773, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.23109769821167, "lr": 0.0002, "elapsed_sec": 34800.802201747894, "step_time_sec": 8.237194913031999, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1104583740234375, "lr": 0.0002, "elapsed_sec": 34809.04068803787, "step_time_sec": 8.238336072012316, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.115628242492676, "lr": 0.0002, "elapsed_sec": 34817.27843618393, "step_time_sec": 8.23762936895946, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.170982837677002, "lr": 0.0002, "elapsed_sec": 34825.517159700394, "step_time_sec": 8.238520034996327, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.324882984161377, "lr": 0.0002, "elapsed_sec": 34833.75557565689, "step_time_sec": 8.238241004990414, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.221053123474121, "lr": 0.0002, "elapsed_sec": 34841.99385142326, "step_time_sec": 8.238162526977248, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.314375400543213, "lr": 0.0002, "elapsed_sec": 34850.233026742935, "step_time_sec": 8.238967737008352, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.128601551055908, "lr": 0.0002, "elapsed_sec": 34858.471547842026, "step_time_sec": 8.23834793601418, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.094912052154541, "lr": 0.0002, "elapsed_sec": 34866.70990753174, "step_time_sec": 8.23825435998151, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.231185436248779, "lr": 0.0002, "elapsed_sec": 34874.948952913284, "step_time_sec": 8.23882326495368, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.202019214630127, "lr": 0.0002, "elapsed_sec": 34883.18605494499, "step_time_sec": 8.236943644995335, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.396154403686523, "lr": 0.0002, "elapsed_sec": 34891.42422223091, "step_time_sec": 8.238053150009364, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.088714122772217, "lr": 0.0002, "elapsed_sec": 34899.66188144684, "step_time_sec": 8.237463852041401, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.227989673614502, "lr": 0.0002, "elapsed_sec": 34907.89969563484, "step_time_sec": 8.237668910005596, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.209372520446777, "lr": 0.0002, "elapsed_sec": 34916.13753890991, "step_time_sec": 8.237706259998959, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.224759578704834, "lr": 0.0002, "elapsed_sec": 34924.37528562546, "step_time_sec": 8.237557663000189, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.211297988891602, "lr": 0.0002, "elapsed_sec": 34932.612763404846, "step_time_sec": 8.237310198950581, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.287496566772461, "lr": 0.0002, "elapsed_sec": 34940.84976649284, "step_time_sec": 8.236842337006237, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9865636825561523, "lr": 0.0002, "elapsed_sec": 34949.08864212036, "step_time_sec": 8.238759408006445, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.133233070373535, "lr": 0.0002, "elapsed_sec": 34957.32613682747, "step_time_sec": 8.237340782012325, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.235281944274902, "lr": 0.0002, "elapsed_sec": 34965.56434178352, "step_time_sec": 8.237989627989009, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.243983268737793, "lr": 0.0002, "elapsed_sec": 34973.80366396904, "step_time_sec": 8.239185591985006, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.164382457733154, "lr": 0.0002, "elapsed_sec": 34982.04123330116, "step_time_sec": 8.237386350985616, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.220180034637451, "lr": 0.0002, "elapsed_sec": 34990.2788746357, "step_time_sec": 8.237469914951362, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1857590675354, "lr": 0.0002, "elapsed_sec": 34998.517359972, "step_time_sec": 8.238327170023695, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.152737140655518, "lr": 0.0002, "elapsed_sec": 35006.756039619446, "step_time_sec": 8.238500695035327, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.151899337768555, "lr": 0.0002, "elapsed_sec": 35014.993337631226, "step_time_sec": 8.237166300008539, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.099872589111328, "lr": 0.0002, "elapsed_sec": 35023.231581926346, "step_time_sec": 8.238093755033333, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.239892482757568, "lr": 0.0002, "elapsed_sec": 35031.47005319595, "step_time_sec": 8.2383448019973, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.183614730834961, "lr": 0.0002, "elapsed_sec": 35039.70765686035, "step_time_sec": 8.237420380988624, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.119120121002197, "lr": 0.0002, "elapsed_sec": 35047.946118831635, "step_time_sec": 8.238302789977752, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.23544454574585, "lr": 0.0002, "elapsed_sec": 35056.18316459656, "step_time_sec": 8.23683827399509, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.162294864654541, "lr": 0.0002, "elapsed_sec": 35064.421015024185, "step_time_sec": 8.237676689983346, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.277704238891602, "lr": 0.0002, "elapsed_sec": 35072.65884947777, "step_time_sec": 8.237688301014714, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.094055652618408, "lr": 0.0002, "elapsed_sec": 35080.89587974548, "step_time_sec": 8.236872975016013, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.126673698425293, "lr": 0.0002, "elapsed_sec": 35089.13452553749, "step_time_sec": 8.238494747958612, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2068095207214355, "lr": 0.0002, "elapsed_sec": 35097.37197160721, "step_time_sec": 8.23730997898383, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.167977809906006, "lr": 0.0002, "elapsed_sec": 35105.61118078232, "step_time_sec": 8.239001635985915, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.167577266693115, "lr": 0.0002, "elapsed_sec": 35113.84987807274, "step_time_sec": 8.238569733977783, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.024320602416992, "lr": 0.0002, "elapsed_sec": 35122.087133169174, "step_time_sec": 8.23706877802033, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.303617477416992, "lr": 0.0002, "elapsed_sec": 35130.32628774643, "step_time_sec": 8.238980122026987, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.25099515914917, "lr": 0.0002, "elapsed_sec": 35138.56288981438, "step_time_sec": 8.2364208449726, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.247211933135986, "lr": 0.0002, "elapsed_sec": 35146.80186462402, "step_time_sec": 8.238845237006899, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.391366958618164, "lr": 0.0002, "elapsed_sec": 35155.040585041046, "step_time_sec": 8.238548760011327, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2815680503845215, "lr": 0.0002, "elapsed_sec": 35163.278940439224, "step_time_sec": 8.238226908026263, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1971282958984375, "lr": 0.0002, "elapsed_sec": 35171.51730847359, "step_time_sec": 8.23825787100941, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.192706108093262, "lr": 0.0002, "elapsed_sec": 35179.754083156586, "step_time_sec": 8.236538499011658, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.109732627868652, "lr": 0.0002, "elapsed_sec": 35187.993742227554, "step_time_sec": 8.239534290973097, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.250375270843506, "lr": 0.0002, "elapsed_sec": 35196.231947898865, "step_time_sec": 8.23801359499339, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.202469825744629, "lr": 0.0002, "elapsed_sec": 35204.47054409981, "step_time_sec": 8.238428836979438, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1691741943359375, "lr": 0.0002, "elapsed_sec": 35212.70939517021, "step_time_sec": 8.238699837995227, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2150373458862305, "lr": 0.0002, "elapsed_sec": 35220.94640469551, "step_time_sec": 8.23692985798698, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.257594108581543, "lr": 0.0002, "elapsed_sec": 35229.18398785591, "step_time_sec": 8.237370065995492, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.275219440460205, "lr": 0.0002, "elapsed_sec": 35237.42164564133, "step_time_sec": 8.237487848033197, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0669264793396, "lr": 0.0002, "elapsed_sec": 35245.660707473755, "step_time_sec": 8.238879757991526, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.031993389129639, "lr": 0.0002, "elapsed_sec": 35253.89905190468, "step_time_sec": 8.238157586019952, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.24227237701416, "lr": 0.0002, "elapsed_sec": 35262.13648271561, "step_time_sec": 8.237304237030912, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.076522350311279, "lr": 0.0002, "elapsed_sec": 35270.374723911285, "step_time_sec": 8.238069035985973, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.159014701843262, "lr": 0.0002, "elapsed_sec": 35278.612117767334, "step_time_sec": 8.237229081976693, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.197451591491699, "lr": 0.0002, "elapsed_sec": 35286.85119223595, "step_time_sec": 8.238920731004328, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.054800033569336, "lr": 0.0002, "elapsed_sec": 35295.090039014816, "step_time_sec": 8.238707386946771, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9850001335144043, "lr": 0.0002, "elapsed_sec": 35303.32716679573, "step_time_sec": 8.236960551992524, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.094669342041016, "lr": 0.0002, "elapsed_sec": 35311.56430053711, "step_time_sec": 8.236996091029141, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.213955402374268, "lr": 0.0002, "elapsed_sec": 35319.80160522461, "step_time_sec": 8.237073529977351, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.12924337387085, "lr": 0.0002, "elapsed_sec": 35328.040667533875, "step_time_sec": 8.238945599994622, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.210931777954102, "lr": 0.0002, "elapsed_sec": 35336.277889966965, "step_time_sec": 8.237077139026951, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.190046787261963, "lr": 0.0002, "elapsed_sec": 35344.51720046997, "step_time_sec": 8.239161046978552, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1500444412231445, "lr": 0.0002, "elapsed_sec": 35352.75675344467, "step_time_sec": 8.239424121042248, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.222771644592285, "lr": 0.0002, "elapsed_sec": 35360.99385762215, "step_time_sec": 8.236878802999854, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.444699764251709, "lr": 0.0002, "elapsed_sec": 35369.233194351196, "step_time_sec": 8.239169936045073, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.172503471374512, "lr": 0.0002, "elapsed_sec": 35377.470863342285, "step_time_sec": 8.237504205026198, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.240043640136719, "lr": 0.0002, "elapsed_sec": 35385.70929002762, "step_time_sec": 8.238239191006869, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.094482898712158, "lr": 0.0002, "elapsed_sec": 35393.94800424576, "step_time_sec": 8.238553964998573, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.244169235229492, "lr": 0.0002, "elapsed_sec": 35402.18583583832, "step_time_sec": 8.237678926961962, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9876842498779297, "lr": 0.0002, "elapsed_sec": 35410.42448210716, "step_time_sec": 8.238541276950855, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.093581676483154, "lr": 0.0002, "elapsed_sec": 35418.66274404526, "step_time_sec": 8.238007394014858, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.24793815612793, "lr": 0.0002, "elapsed_sec": 35426.90209698677, "step_time_sec": 8.239154362003319, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.112114429473877, "lr": 0.0002, "elapsed_sec": 35435.14020371437, "step_time_sec": 8.237953425967135, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1613287925720215, "lr": 0.0002, "elapsed_sec": 35443.37805771828, "step_time_sec": 8.23769372998504, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.146435737609863, "lr": 0.0002, "elapsed_sec": 35451.61762928963, "step_time_sec": 8.239430929010268, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.277610778808594, "lr": 0.0002, "elapsed_sec": 35459.855993032455, "step_time_sec": 8.238142026006244, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.275297164916992, "lr": 0.0002, "elapsed_sec": 35468.095269203186, "step_time_sec": 8.239167520019691, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1158223152160645, "lr": 0.0002, "elapsed_sec": 35476.335527181625, "step_time_sec": 8.240053081011865, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.270660400390625, "lr": 0.0002, "elapsed_sec": 35484.57321929932, "step_time_sec": 8.237546828982886, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.264853477478027, "lr": 0.0002, "elapsed_sec": 35492.81169676781, "step_time_sec": 8.238282261998393, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.086278915405273, "lr": 0.0002, "elapsed_sec": 35501.0494222641, "step_time_sec": 8.237540672998875, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.091567039489746, "lr": 0.0002, "elapsed_sec": 35509.28867292404, "step_time_sec": 8.239103052997962, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.20521354675293, "lr": 0.0002, "elapsed_sec": 35517.527973651886, "step_time_sec": 8.2391733090044, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.084066867828369, "lr": 0.0002, "elapsed_sec": 35525.76601600647, "step_time_sec": 8.23785690200748, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.060276508331299, "lr": 0.0002, "elapsed_sec": 35534.004021167755, "step_time_sec": 8.23783513403032, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.149857521057129, "lr": 0.0002, "elapsed_sec": 35542.24255990982, "step_time_sec": 8.23839441203745, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.132849216461182, "lr": 0.0002, "elapsed_sec": 35550.48054289818, "step_time_sec": 8.23781305400189, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.232365131378174, "lr": 0.0002, "elapsed_sec": 35558.71783494949, "step_time_sec": 8.237146675994154, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.074552536010742, "lr": 0.0002, "elapsed_sec": 35566.95543766022, "step_time_sec": 8.2374180710176, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.995394229888916, "lr": 0.0002, "elapsed_sec": 35575.194956064224, "step_time_sec": 8.239343944995198, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.206185817718506, "lr": 0.0002, "elapsed_sec": 35583.432393074036, "step_time_sec": 8.237287435971666, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.150733947753906, "lr": 0.0002, "elapsed_sec": 35591.6713476181, "step_time_sec": 8.238814929034561, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.102302074432373, "lr": 0.0002, "elapsed_sec": 35599.90845775604, "step_time_sec": 8.236938618007116, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.169652462005615, "lr": 0.0002, "elapsed_sec": 35608.146320581436, "step_time_sec": 8.237742919009179, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.330347061157227, "lr": 0.0002, "elapsed_sec": 35616.38511276245, "step_time_sec": 8.238622426986694, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.250908851623535, "lr": 0.0002, "elapsed_sec": 35624.62272119522, "step_time_sec": 8.23744307202287, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.273144245147705, "lr": 0.0002, "elapsed_sec": 35632.861511707306, "step_time_sec": 8.238629145023879, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.339997291564941, "lr": 0.0002, "elapsed_sec": 35641.09882736206, "step_time_sec": 8.23714108200511, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.306777477264404, "lr": 0.0002, "elapsed_sec": 35649.33778834343, "step_time_sec": 8.238837927987333, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.283524036407471, "lr": 0.0002, "elapsed_sec": 35657.57701706886, "step_time_sec": 8.239028315991163, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.123482704162598, "lr": 0.0002, "elapsed_sec": 35665.81415462494, "step_time_sec": 8.236958186957054, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.335500717163086, "lr": 0.0002, "elapsed_sec": 35674.05295395851, "step_time_sec": 8.238618746981956, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.22018575668335, "lr": 0.0002, "elapsed_sec": 35682.28973078728, "step_time_sec": 8.236633714986965, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.123100757598877, "lr": 0.0002, "elapsed_sec": 35690.52850103378, "step_time_sec": 8.23860727698775, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.155775547027588, "lr": 0.0002, "elapsed_sec": 35698.7670814991, "step_time_sec": 8.238435849023517, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.292233467102051, "lr": 0.0002, "elapsed_sec": 35707.00481247902, "step_time_sec": 8.237522808020003, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.087250232696533, "lr": 0.0002, "elapsed_sec": 35715.24302864075, "step_time_sec": 8.238050390034914, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.030216693878174, "lr": 0.0002, "elapsed_sec": 35723.48026680946, "step_time_sec": 8.237057083984837, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.271288871765137, "lr": 0.0002, "elapsed_sec": 35731.719342947006, "step_time_sec": 8.238904680009, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.095358371734619, "lr": 0.0002, "elapsed_sec": 35739.95707345009, "step_time_sec": 8.237575834034942, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.235461235046387, "lr": 0.0002, "elapsed_sec": 35748.19521212578, "step_time_sec": 8.238019126001745, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.244289875030518, "lr": 0.0002, "elapsed_sec": 35756.4345369339, "step_time_sec": 8.23916882899357, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.191647529602051, "lr": 0.0002, "elapsed_sec": 35764.67172265053, "step_time_sec": 8.236989179975353, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.222681999206543, "lr": 0.0002, "elapsed_sec": 35772.90978050232, "step_time_sec": 8.2379012760357, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.283695697784424, "lr": 0.0002, "elapsed_sec": 35781.14716672897, "step_time_sec": 8.237186595972162, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.137146949768066, "lr": 0.0002, "elapsed_sec": 35789.385618686676, "step_time_sec": 8.238318710005842, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.188652515411377, "lr": 0.0002, "elapsed_sec": 35797.62361550331, "step_time_sec": 8.237799504015129, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.038000583648682, "lr": 0.0002, "elapsed_sec": 35805.861329078674, "step_time_sec": 8.237564499024302, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2083740234375, "lr": 0.0002, "elapsed_sec": 35814.09789729118, "step_time_sec": 8.236430565011688, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.29633903503418, "lr": 0.0002, "elapsed_sec": 35822.335881233215, "step_time_sec": 8.23778816399863, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.255521774291992, "lr": 0.0002, "elapsed_sec": 35830.574684381485, "step_time_sec": 8.238631006039213, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.213602066040039, "lr": 0.0002, "elapsed_sec": 35838.81204533577, "step_time_sec": 8.237203157972544, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.114095211029053, "lr": 0.0002, "elapsed_sec": 35847.050010204315, "step_time_sec": 8.237797394045629, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.139884948730469, "lr": 0.0002, "elapsed_sec": 35855.28743505478, "step_time_sec": 8.23729836801067, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.249284744262695, "lr": 0.0002, "elapsed_sec": 35863.52400684357, "step_time_sec": 8.236384080024436, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.406081199645996, "lr": 0.0002, "elapsed_sec": 35871.76288771629, "step_time_sec": 8.238751141005196, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1919636726379395, "lr": 0.0002, "elapsed_sec": 35880.00086688995, "step_time_sec": 8.237803595024161, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2112507820129395, "lr": 0.0002, "elapsed_sec": 35888.2394528389, "step_time_sec": 8.23844357096823, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.157942295074463, "lr": 0.0002, "elapsed_sec": 35896.477591753006, "step_time_sec": 8.237954436044674, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0868449211120605, "lr": 0.0002, "elapsed_sec": 35904.71511435509, "step_time_sec": 8.237362532003317, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.243132591247559, "lr": 0.0002, "elapsed_sec": 35912.95348358154, "step_time_sec": 8.238217903999612, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.219614028930664, "lr": 0.0002, "elapsed_sec": 35921.19079065323, "step_time_sec": 8.237168096005917, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.265698432922363, "lr": 0.0002, "elapsed_sec": 35929.430846214294, "step_time_sec": 8.239850949961692, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.179643630981445, "lr": 0.0002, "elapsed_sec": 35937.66796755791, "step_time_sec": 8.236927393008955, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.270030498504639, "lr": 0.0002, "elapsed_sec": 35945.906259059906, "step_time_sec": 8.23813455295749, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.138940811157227, "lr": 0.0002, "elapsed_sec": 35954.14570760727, "step_time_sec": 8.239257631008513, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.232596397399902, "lr": 0.0002, "elapsed_sec": 35962.38266372681, "step_time_sec": 8.236782667983789, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.129784107208252, "lr": 0.0002, "elapsed_sec": 35970.62292599678, "step_time_sec": 8.240112919011153, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.063918113708496, "lr": 0.0002, "elapsed_sec": 35978.86110472679, "step_time_sec": 8.238007617008407, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2101569175720215, "lr": 0.0002, "elapsed_sec": 35987.09867119789, "step_time_sec": 8.237413523020223, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.08259916305542, "lr": 0.0002, "elapsed_sec": 35995.337465286255, "step_time_sec": 8.238624725025147, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.222995758056641, "lr": 0.0002, "elapsed_sec": 36003.574335575104, "step_time_sec": 8.236711034027394, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.142134666442871, "lr": 0.0002, "elapsed_sec": 36011.814148426056, "step_time_sec": 8.239650867006276, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.204169273376465, "lr": 0.0002, "elapsed_sec": 36020.05139183998, "step_time_sec": 8.2371157000307, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.220112323760986, "lr": 0.0002, "elapsed_sec": 36028.28983092308, "step_time_sec": 8.238303263962734, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.427163600921631, "lr": 0.0002, "elapsed_sec": 36036.5271191597, "step_time_sec": 8.237020717002451, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.181581974029541, "lr": 0.0002, "elapsed_sec": 36044.764951467514, "step_time_sec": 8.23773285903735, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.349039554595947, "lr": 0.0002, "elapsed_sec": 36053.00349211693, "step_time_sec": 8.238398462999612, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.340701103210449, "lr": 0.0002, "elapsed_sec": 36061.24064421654, "step_time_sec": 8.23699405696243, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.201333045959473, "lr": 0.0002, "elapsed_sec": 36069.47918963432, "step_time_sec": 8.238320945994928, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.25993013381958, "lr": 0.0002, "elapsed_sec": 36077.71624493599, "step_time_sec": 8.236893208988477, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.278172016143799, "lr": 0.0002, "elapsed_sec": 36085.95530486107, "step_time_sec": 8.238918901013676, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.197564601898193, "lr": 0.0002, "elapsed_sec": 36094.19391846657, "step_time_sec": 8.238434146973304, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.014283180236816, "lr": 0.0002, "elapsed_sec": 36102.430768966675, "step_time_sec": 8.23669864796102, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.426385879516602, "lr": 0.0002, "elapsed_sec": 36110.669529914856, "step_time_sec": 8.238593690039124, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.151092052459717, "lr": 0.0002, "elapsed_sec": 36118.9086022377, "step_time_sec": 8.238927780010272, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.165097713470459, "lr": 0.0002, "elapsed_sec": 36127.14720439911, "step_time_sec": 8.238403922994621, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9411988258361816, "lr": 0.0002, "elapsed_sec": 36135.38633275032, "step_time_sec": 8.238947134988848, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.240778923034668, "lr": 0.0002, "elapsed_sec": 36143.62418293953, "step_time_sec": 8.237687579006888, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.105739593505859, "lr": 0.0002, "elapsed_sec": 36151.86244773865, "step_time_sec": 8.23811424500309, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.346912384033203, "lr": 0.0002, "elapsed_sec": 36160.0992565155, "step_time_sec": 8.236668560944963, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.236778736114502, "lr": 0.0002, "elapsed_sec": 36168.336810827255, "step_time_sec": 8.237387328001205, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.170539379119873, "lr": 0.0002, "elapsed_sec": 36176.57553744316, "step_time_sec": 8.238577259005979, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.308356761932373, "lr": 0.0002, "elapsed_sec": 36184.81382751465, "step_time_sec": 8.238113222003449, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.291208267211914, "lr": 0.0002, "elapsed_sec": 36193.051077604294, "step_time_sec": 8.237112645991147, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.280903339385986, "lr": 0.0002, "elapsed_sec": 36201.35427713394, "step_time_sec": 8.245222339988686, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.43576192855835, "lr": 0.0002, "elapsed_sec": 36209.591334819794, "step_time_sec": 8.236836212978233, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.038180828094482, "lr": 0.0002, "elapsed_sec": 36217.830770254135, "step_time_sec": 8.23926984803984, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.270641803741455, "lr": 0.0002, "elapsed_sec": 36226.069813489914, "step_time_sec": 8.238863523991313, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.07422399520874, "lr": 0.0002, "elapsed_sec": 36234.307697057724, "step_time_sec": 8.237768359016627, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.232858657836914, "lr": 0.0002, "elapsed_sec": 36242.54731845856, "step_time_sec": 8.239406883018091, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.221579074859619, "lr": 0.0002, "elapsed_sec": 36250.785276412964, "step_time_sec": 8.237865631002933, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.118222713470459, "lr": 0.0002, "elapsed_sec": 36259.0241253376, "step_time_sec": 8.238623692013789, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.144503116607666, "lr": 0.0002, "elapsed_sec": 36267.26276111603, "step_time_sec": 8.238448087009601, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1302809715271, "lr": 0.0002, "elapsed_sec": 36275.50049686432, "step_time_sec": 8.237546033982653, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.027010440826416, "lr": 0.0002, "elapsed_sec": 36283.73866534233, "step_time_sec": 8.23800737701822, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1215500831604, "lr": 0.0002, "elapsed_sec": 36291.97522711754, "step_time_sec": 8.236449116026051, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.251752853393555, "lr": 0.0002, "elapsed_sec": 36300.21469831467, "step_time_sec": 8.239286885014735, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.190617084503174, "lr": 0.0002, "elapsed_sec": 36308.45323252678, "step_time_sec": 8.23837395600276, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.267226219177246, "lr": 0.0002, "elapsed_sec": 36316.69151830673, "step_time_sec": 8.238093847001437, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1746344566345215, "lr": 0.0002, "elapsed_sec": 36324.9297349453, "step_time_sec": 8.238075243018102, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.170200347900391, "lr": 0.0002, "elapsed_sec": 36333.16647338867, "step_time_sec": 8.236590087006334, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.163989543914795, "lr": 0.0002, "elapsed_sec": 36341.40540409088, "step_time_sec": 8.238779709034134, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.323579788208008, "lr": 0.0002, "elapsed_sec": 36349.64414048195, "step_time_sec": 8.238518218975514, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.16907262802124, "lr": 0.0002, "elapsed_sec": 36357.88357138634, "step_time_sec": 8.239279493980575, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.226309299468994, "lr": 0.0002, "elapsed_sec": 36366.12222337723, "step_time_sec": 8.238472299010027, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.093650817871094, "lr": 0.0002, "elapsed_sec": 36374.35954141617, "step_time_sec": 8.237151060020551, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.110720157623291, "lr": 0.0002, "elapsed_sec": 36382.59611606598, "step_time_sec": 8.236417944950517, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.294766426086426, "lr": 0.0002, "elapsed_sec": 36390.832890987396, "step_time_sec": 8.236665848002303, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.18235969543457, "lr": 0.0002, "elapsed_sec": 36399.072315216064, "step_time_sec": 8.239220245042816, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.260263919830322, "lr": 0.0002, "elapsed_sec": 36407.30968308449, "step_time_sec": 8.23718859103974, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.253835678100586, "lr": 0.0002, "elapsed_sec": 36415.54927992821, "step_time_sec": 8.239494459005073, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.056495666503906, "lr": 0.0002, "elapsed_sec": 36423.78945827484, "step_time_sec": 8.239928624010645, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.197266101837158, "lr": 0.0002, "elapsed_sec": 36432.02735352516, "step_time_sec": 8.23777929903008, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.293363094329834, "lr": 0.0002, "elapsed_sec": 36440.26563048363, "step_time_sec": 8.23811770498287, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.322695732116699, "lr": 0.0002, "elapsed_sec": 36448.50468087196, "step_time_sec": 8.238906324957497, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.235807418823242, "lr": 0.0002, "elapsed_sec": 36456.744109392166, "step_time_sec": 8.239212354994379, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.163825035095215, "lr": 0.0002, "elapsed_sec": 36464.98147416115, "step_time_sec": 8.237225810997188, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.229636192321777, "lr": 0.0002, "elapsed_sec": 36473.217633247375, "step_time_sec": 8.236018872994464, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.186251640319824, "lr": 0.0002, "elapsed_sec": 36481.457678079605, "step_time_sec": 8.23987191496417, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.162107467651367, "lr": 0.0002, "elapsed_sec": 36489.696009874344, "step_time_sec": 8.23811817803653, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.148438930511475, "lr": 0.0002, "elapsed_sec": 36497.93457746506, "step_time_sec": 8.23840618802933, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2433342933654785, "lr": 0.0002, "elapsed_sec": 36506.17300724983, "step_time_sec": 8.23826392000774, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.220589637756348, "lr": 0.0002, "elapsed_sec": 36514.41064071655, "step_time_sec": 8.23747142998036, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.268258571624756, "lr": 0.0002, "elapsed_sec": 36522.64903259277, "step_time_sec": 8.2382257420104, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.230904579162598, "lr": 0.0002, "elapsed_sec": 36530.88659095764, "step_time_sec": 8.237447199004237, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.392730236053467, "lr": 0.0002, "elapsed_sec": 36539.127302885056, "step_time_sec": 8.240542608022224, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.156170845031738, "lr": 0.0002, "elapsed_sec": 36547.373668432236, "step_time_sec": 8.246198738983367, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.184268474578857, "lr": 0.0002, "elapsed_sec": 36555.61101913452, "step_time_sec": 8.237201728974469, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.321427822113037, "lr": 0.0002, "elapsed_sec": 36563.84908294678, "step_time_sec": 8.237850323028397, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2453293800354, "lr": 0.0002, "elapsed_sec": 36572.08806705475, "step_time_sec": 8.23881904000882, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.284922122955322, "lr": 0.0002, "elapsed_sec": 36580.32611346245, "step_time_sec": 8.23789490800118, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.395054340362549, "lr": 0.0002, "elapsed_sec": 36588.5644569397, "step_time_sec": 8.238196773978416, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.15890645980835, "lr": 0.0002, "elapsed_sec": 36596.8018655777, "step_time_sec": 8.23719718598295, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.178498268127441, "lr": 0.0002, "elapsed_sec": 36605.041043519974, "step_time_sec": 8.239020753011573, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3157057762146, "lr": 0.0002, "elapsed_sec": 36613.28119421005, "step_time_sec": 8.240030104003381, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.261221885681152, "lr": 0.0002, "elapsed_sec": 36621.51823282242, "step_time_sec": 8.23684885702096, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.140872478485107, "lr": 0.0002, "elapsed_sec": 36629.75691175461, "step_time_sec": 8.23846926697297, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.218742370605469, "lr": 0.0002, "elapsed_sec": 36637.99400019646, "step_time_sec": 8.236912686028518, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.212958335876465, "lr": 0.0002, "elapsed_sec": 36646.232949733734, "step_time_sec": 8.238774890021887, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.252966403961182, "lr": 0.0002, "elapsed_sec": 36654.471086740494, "step_time_sec": 8.238002494967077, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.201757431030273, "lr": 0.0002, "elapsed_sec": 36662.709919929504, "step_time_sec": 8.238631171989255, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.193933010101318, "lr": 0.0002, "elapsed_sec": 36670.948734760284, "step_time_sec": 8.238666945020668, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.240233898162842, "lr": 0.0002, "elapsed_sec": 36679.18626618385, "step_time_sec": 8.237374251999427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.250577926635742, "lr": 0.0002, "elapsed_sec": 36687.424728155136, "step_time_sec": 8.238298467011191, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1980390548706055, "lr": 0.0002, "elapsed_sec": 36695.662722826004, "step_time_sec": 8.237830883008428, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.202699184417725, "lr": 0.0002, "elapsed_sec": 36703.9018599987, "step_time_sec": 8.238951976003591, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.122067928314209, "lr": 0.0002, "elapsed_sec": 36712.139907836914, "step_time_sec": 8.237892817007378, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.16832160949707, "lr": 0.0002, "elapsed_sec": 36720.37727808952, "step_time_sec": 8.237236631975975, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2000412940979, "lr": 0.0002, "elapsed_sec": 36728.615723371506, "step_time_sec": 8.23829150799429, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.116688251495361, "lr": 0.0002, "elapsed_sec": 36736.85282897949, "step_time_sec": 8.236894684028812, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.201578617095947, "lr": 0.0002, "elapsed_sec": 36745.091364622116, "step_time_sec": 8.238390425976831, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.027960777282715, "lr": 0.0002, "elapsed_sec": 36753.329989910126, "step_time_sec": 8.238476709986571, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.085685729980469, "lr": 0.0002, "elapsed_sec": 36761.56875371933, "step_time_sec": 8.238601461984217, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.196449279785156, "lr": 0.0002, "elapsed_sec": 36769.807037591934, "step_time_sec": 8.238056853006128, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.434169292449951, "lr": 0.0002, "elapsed_sec": 36778.04419207573, "step_time_sec": 8.237007969990373, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.225183486938477, "lr": 0.0002, "elapsed_sec": 36786.28369617462, "step_time_sec": 8.239315684011672, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.126071929931641, "lr": 0.0002, "elapsed_sec": 36794.5214908123, "step_time_sec": 8.237655316013843, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.309789180755615, "lr": 0.0002, "elapsed_sec": 36802.75968027115, "step_time_sec": 8.238023379992228, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.319571495056152, "lr": 0.0002, "elapsed_sec": 36810.998980522156, "step_time_sec": 8.239147040003445, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.135509014129639, "lr": 0.0002, "elapsed_sec": 36819.23549199104, "step_time_sec": 8.236335689027328, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2062153816223145, "lr": 0.0002, "elapsed_sec": 36827.47372364998, "step_time_sec": 8.238067899015732, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.176890850067139, "lr": 0.0002, "elapsed_sec": 36835.71144199371, "step_time_sec": 8.237578280037269, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.211792469024658, "lr": 0.0002, "elapsed_sec": 36843.9518930912, "step_time_sec": 8.240287963009905, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.170310020446777, "lr": 0.0002, "elapsed_sec": 36852.19069623947, "step_time_sec": 8.238639199000318, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.164503574371338, "lr": 0.0002, "elapsed_sec": 36860.42875909805, "step_time_sec": 8.237874613027088, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.357630252838135, "lr": 0.0002, "elapsed_sec": 36868.66669034958, "step_time_sec": 8.237785217002966, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.212507247924805, "lr": 0.0002, "elapsed_sec": 36876.903584480286, "step_time_sec": 8.236694402992725, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3450236320495605, "lr": 0.0002, "elapsed_sec": 36885.14199233055, "step_time_sec": 8.238230054965243, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.283513069152832, "lr": 0.0002, "elapsed_sec": 36893.38090848923, "step_time_sec": 8.238810056005605, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.195075988769531, "lr": 0.0002, "elapsed_sec": 36901.617971897125, "step_time_sec": 8.23688136698911, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.124695777893066, "lr": 0.0002, "elapsed_sec": 36909.85776209831, "step_time_sec": 8.23960635100957, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2494282722473145, "lr": 0.0002, "elapsed_sec": 36918.09505176544, "step_time_sec": 8.237152744957712, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.226937770843506, "lr": 0.0002, "elapsed_sec": 36926.33374667168, "step_time_sec": 8.23852886800887, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.227321624755859, "lr": 0.0002, "elapsed_sec": 36934.57200431824, "step_time_sec": 8.238087280013133, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2045793533325195, "lr": 0.0002, "elapsed_sec": 36942.80967402458, "step_time_sec": 8.237509434984531, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.259798049926758, "lr": 0.0002, "elapsed_sec": 36951.048320531845, "step_time_sec": 8.238509199989494, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.328245162963867, "lr": 0.0002, "elapsed_sec": 36959.28587579727, "step_time_sec": 8.237430890963878, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.175269603729248, "lr": 0.0002, "elapsed_sec": 36967.52457523346, "step_time_sec": 8.238501728978008, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.163276195526123, "lr": 0.0002, "elapsed_sec": 36975.7638361454, "step_time_sec": 8.239103163999971, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.097005367279053, "lr": 0.0002, "elapsed_sec": 36984.0019159317, "step_time_sec": 8.237933958997019, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.166031360626221, "lr": 0.0002, "elapsed_sec": 36992.23991179466, "step_time_sec": 8.237790517974645, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.14982795715332, "lr": 0.0002, "elapsed_sec": 37000.47574663162, "step_time_sec": 8.235703201033175, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.259395599365234, "lr": 0.0002, "elapsed_sec": 37008.71314287186, "step_time_sec": 8.23726850602543, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.110797882080078, "lr": 0.0002, "elapsed_sec": 37016.950328826904, "step_time_sec": 8.237048432987649, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.143641471862793, "lr": 0.0002, "elapsed_sec": 37025.190126895905, "step_time_sec": 8.239600814995356, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.224063873291016, "lr": 0.0002, "elapsed_sec": 37033.42821407318, "step_time_sec": 8.237916659971233, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2302470207214355, "lr": 0.0002, "elapsed_sec": 37041.66541433334, "step_time_sec": 8.237036500009708, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.133482456207275, "lr": 0.0002, "elapsed_sec": 37049.9044713974, "step_time_sec": 8.238877116993535, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.226389408111572, "lr": 0.0002, "elapsed_sec": 37058.14125227928, "step_time_sec": 8.23664588999236, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.139624118804932, "lr": 0.0002, "elapsed_sec": 37066.37740635872, "step_time_sec": 8.2359415179817, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.159592151641846, "lr": 0.0002, "elapsed_sec": 37074.615181446075, "step_time_sec": 8.237645317974966, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.064944744110107, "lr": 0.0002, "elapsed_sec": 37082.854088783264, "step_time_sec": 8.238737478037365, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.196554660797119, "lr": 0.0002, "elapsed_sec": 37091.0926527977, "step_time_sec": 8.23838882095879, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1161723136901855, "lr": 0.0002, "elapsed_sec": 37099.33083343506, "step_time_sec": 8.238047100021504, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.25516939163208, "lr": 0.0002, "elapsed_sec": 37107.5702316761, "step_time_sec": 8.23922752396902, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.220822811126709, "lr": 0.0002, "elapsed_sec": 37115.807581186295, "step_time_sec": 8.23721052298788, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0898118019104, "lr": 0.0002, "elapsed_sec": 37124.04637694359, "step_time_sec": 8.23866045702016, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.286575794219971, "lr": 0.0002, "elapsed_sec": 37132.285638570786, "step_time_sec": 8.239056847000029, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.210989475250244, "lr": 0.0002, "elapsed_sec": 37140.52298069, "step_time_sec": 8.237151652981993, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1111860275268555, "lr": 0.0002, "elapsed_sec": 37148.76240134239, "step_time_sec": 8.239296487008687, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.127174377441406, "lr": 0.0002, "elapsed_sec": 37156.99948644638, "step_time_sec": 8.236887186998501, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9928760528564453, "lr": 0.0002, "elapsed_sec": 37165.238241910934, "step_time_sec": 8.238583525991999, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.203277587890625, "lr": 0.0002, "elapsed_sec": 37173.4767768383, "step_time_sec": 8.238371841958724, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.140213489532471, "lr": 0.0002, "elapsed_sec": 37181.71476793289, "step_time_sec": 8.237835132982582, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.195841312408447, "lr": 0.0002, "elapsed_sec": 37189.953889131546, "step_time_sec": 8.238949424994644, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1483259201049805, "lr": 0.0002, "elapsed_sec": 37198.19318199158, "step_time_sec": 8.23914508294547, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.267709732055664, "lr": 0.0002, "elapsed_sec": 37206.43234395981, "step_time_sec": 8.239077154023107, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.18519926071167, "lr": 0.0002, "elapsed_sec": 37214.67031311989, "step_time_sec": 8.237726367020514, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.217789173126221, "lr": 0.0002, "elapsed_sec": 37222.90721106529, "step_time_sec": 8.236758704006206, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.258252143859863, "lr": 0.0002, "elapsed_sec": 37231.14584112167, "step_time_sec": 8.23839836998377, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0368332862854, "lr": 0.0002, "elapsed_sec": 37239.38301920891, "step_time_sec": 8.237029369047377, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9851274490356445, "lr": 0.0002, "elapsed_sec": 37247.62116265297, "step_time_sec": 8.237992529000621, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.170616626739502, "lr": 0.0002, "elapsed_sec": 37255.8587269783, "step_time_sec": 8.23742843902437, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.275882720947266, "lr": 0.0002, "elapsed_sec": 37264.096058130264, "step_time_sec": 8.23715155001264, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.185625076293945, "lr": 0.0002, "elapsed_sec": 37272.33504009247, "step_time_sec": 8.238785464025568, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.131767272949219, "lr": 0.0002, "elapsed_sec": 37280.57217860222, "step_time_sec": 8.236980845977087, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2698211669921875, "lr": 0.0002, "elapsed_sec": 37288.81138563156, "step_time_sec": 8.239050776988734, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.158820152282715, "lr": 0.0002, "elapsed_sec": 37297.04808354378, "step_time_sec": 8.236557581985835, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.110804080963135, "lr": 0.0002, "elapsed_sec": 37305.286791324615, "step_time_sec": 8.238552651018836, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1349005699157715, "lr": 0.0002, "elapsed_sec": 37313.52402496338, "step_time_sec": 8.237040113017429, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.15975284576416, "lr": 0.0002, "elapsed_sec": 37321.761016607285, "step_time_sec": 8.236864903999958, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.208216667175293, "lr": 0.0002, "elapsed_sec": 37329.999376535416, "step_time_sec": 8.238195971993264, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.234611988067627, "lr": 0.0002, "elapsed_sec": 37338.23593020439, "step_time_sec": 8.236393760016654, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.255133628845215, "lr": 0.0002, "elapsed_sec": 37346.47469615936, "step_time_sec": 8.238604554964695, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.18194055557251, "lr": 0.0002, "elapsed_sec": 37354.71358180046, "step_time_sec": 8.238711178011727, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0378265380859375, "lr": 0.0002, "elapsed_sec": 37362.951335668564, "step_time_sec": 8.237649485003203, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.191167831420898, "lr": 0.0002, "elapsed_sec": 37371.19052863121, "step_time_sec": 8.238980133028235, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.218264579772949, "lr": 0.0002, "elapsed_sec": 37379.42650961876, "step_time_sec": 8.23584669700358, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1850762367248535, "lr": 0.0002, "elapsed_sec": 37387.664774656296, "step_time_sec": 31.11512309499085, "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": 4.277535915374756, "lr": 0.0002, "elapsed_sec": 37418.79783153534, "step_time_sec": 8.255482250999194, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.137026309967041, "lr": 0.0002, "elapsed_sec": 37427.02927517891, "step_time_sec": 8.231288211012725, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.450814723968506, "lr": 0.0002, "elapsed_sec": 37435.266767024994, "step_time_sec": 8.237310303025879, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.064167022705078, "lr": 0.0002, "elapsed_sec": 37443.50554203987, "step_time_sec": 8.238630309002474, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.289475440979004, "lr": 0.0002, "elapsed_sec": 37451.74312376976, "step_time_sec": 8.237386251974385, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.231146335601807, "lr": 0.0002, "elapsed_sec": 37459.98200464249, "step_time_sec": 8.2387077040039, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.263645648956299, "lr": 0.0002, "elapsed_sec": 37468.21989893913, "step_time_sec": 8.237774787005037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.024589538574219, "lr": 0.0002, "elapsed_sec": 37476.45686483383, "step_time_sec": 8.236775835044682, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.095203399658203, "lr": 0.0002, "elapsed_sec": 37484.69531393051, "step_time_sec": 8.23834635404637, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.168583393096924, "lr": 0.0002, "elapsed_sec": 37492.93275117874, "step_time_sec": 8.23728849698091, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.132707118988037, "lr": 0.0002, "elapsed_sec": 37501.17213129997, "step_time_sec": 8.23920051503228, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.262685775756836, "lr": 0.0002, "elapsed_sec": 37509.40942525864, "step_time_sec": 8.237097228993662, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1701459884643555, "lr": 0.0002, "elapsed_sec": 37517.64750599861, "step_time_sec": 8.23789755802136, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1802473068237305, "lr": 0.0002, "elapsed_sec": 37525.88718342781, "step_time_sec": 8.239531045022886, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1718645095825195, "lr": 0.0002, "elapsed_sec": 37534.124549627304, "step_time_sec": 8.237219237023965, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.215733528137207, "lr": 0.0002, "elapsed_sec": 37542.36221933365, "step_time_sec": 8.237463235971518, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.185635566711426, "lr": 0.0002, "elapsed_sec": 37550.59976935387, "step_time_sec": 8.237390098976903, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0661702156066895, "lr": 0.0002, "elapsed_sec": 37558.83889603615, "step_time_sec": 8.239015562052373, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1482157707214355, "lr": 0.0002, "elapsed_sec": 37567.07671332359, "step_time_sec": 8.237603040004615, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.251093864440918, "lr": 0.0002, "elapsed_sec": 37575.31488084793, "step_time_sec": 8.238015181967057, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.123724937438965, "lr": 0.0002, "elapsed_sec": 37583.55456495285, "step_time_sec": 8.239492233027704, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.137021064758301, "lr": 0.0002, "elapsed_sec": 37591.79152059555, "step_time_sec": 8.23679793899646, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1189141273498535, "lr": 0.0002, "elapsed_sec": 37600.029643297195, "step_time_sec": 8.237988299981225, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.163074970245361, "lr": 0.0002, "elapsed_sec": 37608.268162965775, "step_time_sec": 8.23837137594819, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.956494092941284, "lr": 0.0002, "elapsed_sec": 37616.506098270416, "step_time_sec": 8.23781192599563, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.133907318115234, "lr": 0.0002, "elapsed_sec": 37624.74606323242, "step_time_sec": 8.239730959990993, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.207100868225098, "lr": 0.0002, "elapsed_sec": 37632.984555244446, "step_time_sec": 8.238362920994405, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.042492866516113, "lr": 0.0002, "elapsed_sec": 37641.222744464874, "step_time_sec": 8.237989574030507, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9500274658203125, "lr": 0.0002, "elapsed_sec": 37649.461741924286, "step_time_sec": 8.238841799960937, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.182082653045654, "lr": 0.0002, "elapsed_sec": 37657.70079445839, "step_time_sec": 8.238866522966418, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.207836627960205, "lr": 0.0002, "elapsed_sec": 37665.93911218643, "step_time_sec": 8.238175760023296, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2519683837890625, "lr": 0.0002, "elapsed_sec": 37674.17644691467, "step_time_sec": 8.237188540981151, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.372382164001465, "lr": 0.0002, "elapsed_sec": 37682.41593122482, "step_time_sec": 8.239267037017271, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.202483177185059, "lr": 0.0002, "elapsed_sec": 37690.654571294785, "step_time_sec": 8.238494442019146, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.148791313171387, "lr": 0.0002, "elapsed_sec": 37698.89243412018, "step_time_sec": 8.237670654023532, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9842255115509033, "lr": 0.0002, "elapsed_sec": 37707.13149309158, "step_time_sec": 8.238919486990198, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.192275524139404, "lr": 0.0002, "elapsed_sec": 37715.369188547134, "step_time_sec": 8.237536157015711, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1193742752075195, "lr": 0.0002, "elapsed_sec": 37723.608192682266, "step_time_sec": 8.238802327017765, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.342931270599365, "lr": 0.0002, "elapsed_sec": 37731.84729599953, "step_time_sec": 8.238994868006557, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.240489482879639, "lr": 0.0002, "elapsed_sec": 37740.08643627167, "step_time_sec": 8.23893556202529, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.094580173492432, "lr": 0.0002, "elapsed_sec": 37748.32573890686, "step_time_sec": 8.239155012008268, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.106756210327148, "lr": 0.0002, "elapsed_sec": 37756.56365585327, "step_time_sec": 8.237753813969903, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.148312568664551, "lr": 0.0002, "elapsed_sec": 37764.80261540413, "step_time_sec": 8.238756624981761, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.151423931121826, "lr": 0.0002, "elapsed_sec": 37773.04006910324, "step_time_sec": 8.237349104019813, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.259347915649414, "lr": 0.0002, "elapsed_sec": 37781.278010606766, "step_time_sec": 8.237754317000508, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.368976593017578, "lr": 0.0002, "elapsed_sec": 37789.51489114761, "step_time_sec": 8.236741148983128, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.014595031738281, "lr": 0.0002, "elapsed_sec": 37797.75275158882, "step_time_sec": 8.237653135030996, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.172301292419434, "lr": 0.0002, "elapsed_sec": 37805.99219799042, "step_time_sec": 8.239287090022117, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.038549423217773, "lr": 0.0002, "elapsed_sec": 37814.22935318947, "step_time_sec": 8.237011026008986, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.080174446105957, "lr": 0.0002, "elapsed_sec": 37822.46831655502, "step_time_sec": 8.238788756949361, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.076727390289307, "lr": 0.0002, "elapsed_sec": 37830.70668339729, "step_time_sec": 8.238234860007651, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.808065414428711, "lr": 0.0002, "elapsed_sec": 37838.944877147675, "step_time_sec": 8.238037423987407, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.244051933288574, "lr": 0.0002, "elapsed_sec": 37847.18365859985, "step_time_sec": 8.238602865021676, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8934078216552734, "lr": 0.0002, "elapsed_sec": 37855.4215593338, "step_time_sec": 8.237716322008055, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.082381725311279, "lr": 0.0002, "elapsed_sec": 37863.660198926926, "step_time_sec": 8.238483087043278, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.149556636810303, "lr": 0.0002, "elapsed_sec": 37871.8987300396, "step_time_sec": 8.238360657007433, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.246063232421875, "lr": 0.0002, "elapsed_sec": 37880.137439250946, "step_time_sec": 8.238590218999889, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.169212818145752, "lr": 0.0002, "elapsed_sec": 37888.37543773651, "step_time_sec": 8.237837869965006, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.204085826873779, "lr": 0.0002, "elapsed_sec": 37896.6127011776, "step_time_sec": 8.237087448011152, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.128238677978516, "lr": 0.0002, "elapsed_sec": 37904.85093188286, "step_time_sec": 8.238070106017403, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.285471439361572, "lr": 0.0002, "elapsed_sec": 37913.08733201027, "step_time_sec": 8.23626057599904, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9796340465545654, "lr": 0.0002, "elapsed_sec": 37921.32672572136, "step_time_sec": 8.239194080000743, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.230348587036133, "lr": 0.0002, "elapsed_sec": 37929.56543636322, "step_time_sec": 8.238565040985122, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.133745193481445, "lr": 0.0002, "elapsed_sec": 37937.80188179016, "step_time_sec": 8.23624330200255, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9947946071624756, "lr": 0.0002, "elapsed_sec": 37946.04104423523, "step_time_sec": 8.238991845981218, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.087040901184082, "lr": 0.0002, "elapsed_sec": 37954.27836203575, "step_time_sec": 8.237180483993143, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.156976222991943, "lr": 0.0002, "elapsed_sec": 37962.516273736954, "step_time_sec": 8.237755462992936, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.223365783691406, "lr": 0.0002, "elapsed_sec": 37970.75522685051, "step_time_sec": 8.23878885700833, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.114789009094238, "lr": 0.0002, "elapsed_sec": 37978.99255037308, "step_time_sec": 8.237184678029735, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.056978702545166, "lr": 0.0002, "elapsed_sec": 37987.23230051994, "step_time_sec": 8.239522642048541, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.164126396179199, "lr": 0.0002, "elapsed_sec": 37995.46944308281, "step_time_sec": 8.237004917988088, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.226686000823975, "lr": 0.0002, "elapsed_sec": 38003.708721637726, "step_time_sec": 8.23915478301933, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.143733501434326, "lr": 0.0002, "elapsed_sec": 38011.94737768173, "step_time_sec": 8.23844625899801, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.004849433898926, "lr": 0.0002, "elapsed_sec": 38020.18529129028, "step_time_sec": 8.23772337497212, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.115100860595703, "lr": 0.0002, "elapsed_sec": 38028.42335629463, "step_time_sec": 8.237940739025362, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.173854351043701, "lr": 0.0002, "elapsed_sec": 38036.66096520424, "step_time_sec": 8.237494842032902, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.016769886016846, "lr": 0.0002, "elapsed_sec": 38044.90104031563, "step_time_sec": 8.239859559049364, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.185695171356201, "lr": 0.0002, "elapsed_sec": 38053.13942527771, "step_time_sec": 8.238214073004201, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.255514621734619, "lr": 0.0002, "elapsed_sec": 38061.37750887871, "step_time_sec": 8.237868408032227, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.13618278503418, "lr": 0.0002, "elapsed_sec": 38069.616251945496, "step_time_sec": 8.23862039600499, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.374534606933594, "lr": 0.0002, "elapsed_sec": 38077.853505373, "step_time_sec": 8.237050507043023, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.210512161254883, "lr": 0.0002, "elapsed_sec": 38086.091809272766, "step_time_sec": 8.23814355902141, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.103761196136475, "lr": 0.0002, "elapsed_sec": 38094.32878780365, "step_time_sec": 8.23685759701766, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.152535438537598, "lr": 0.0002, "elapsed_sec": 38102.56777024269, "step_time_sec": 8.238778368046042, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.277773857116699, "lr": 0.0002, "elapsed_sec": 38110.80546665192, "step_time_sec": 8.237545607960783, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.149586200714111, "lr": 0.0002, "elapsed_sec": 38119.044335365295, "step_time_sec": 8.238709211989772, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.240469932556152, "lr": 0.0002, "elapsed_sec": 38127.283571243286, "step_time_sec": 8.239055102982093, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.143903732299805, "lr": 0.0002, "elapsed_sec": 38135.52114415169, "step_time_sec": 8.237439683987759, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.054134845733643, "lr": 0.0002, "elapsed_sec": 38143.75973033905, "step_time_sec": 8.23840095300693, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.055593967437744, "lr": 0.0002, "elapsed_sec": 38151.9978055954, "step_time_sec": 8.237891863973346, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.989884376525879, "lr": 0.0002, "elapsed_sec": 38160.23487114906, "step_time_sec": 8.236897491966374, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.069105625152588, "lr": 0.0002, "elapsed_sec": 38168.47517681122, "step_time_sec": 8.24017713102512, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.190490245819092, "lr": 0.0002, "elapsed_sec": 38176.712866067886, "step_time_sec": 8.237531686027069, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.130584716796875, "lr": 0.0002, "elapsed_sec": 38184.95123696327, "step_time_sec": 8.238233439973556, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.084534168243408, "lr": 0.0002, "elapsed_sec": 38193.190910577774, "step_time_sec": 8.239540840033442, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.090651512145996, "lr": 0.0002, "elapsed_sec": 38201.42938017845, "step_time_sec": 8.23825175903039, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.200536251068115, "lr": 0.0002, "elapsed_sec": 38209.66813373566, "step_time_sec": 8.238637688977178, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.229975700378418, "lr": 0.0002, "elapsed_sec": 38217.904823064804, "step_time_sec": 8.236518536985386, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.142666816711426, "lr": 0.0002, "elapsed_sec": 38226.143280506134, "step_time_sec": 8.238251639995724, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.127118110656738, "lr": 0.0002, "elapsed_sec": 38234.38206195831, "step_time_sec": 8.238604875048622, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.272995471954346, "lr": 0.0002, "elapsed_sec": 38242.6200401783, "step_time_sec": 8.237805217970163, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.089298725128174, "lr": 0.0002, "elapsed_sec": 38250.85918092728, "step_time_sec": 8.238967461977154, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.954298496246338, "lr": 0.0002, "elapsed_sec": 38259.09531664848, "step_time_sec": 8.23599777504569, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.101698875427246, "lr": 0.0002, "elapsed_sec": 38267.333928346634, "step_time_sec": 8.23842510097893, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.20704984664917, "lr": 0.0002, "elapsed_sec": 38275.57097029686, "step_time_sec": 8.236962158989627, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.031403064727783, "lr": 0.0002, "elapsed_sec": 38283.81006360054, "step_time_sec": 8.238857935008127, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.208693027496338, "lr": 0.0002, "elapsed_sec": 38292.04814338684, "step_time_sec": 8.237964122963604, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.213529586791992, "lr": 0.0002, "elapsed_sec": 38300.286556720734, "step_time_sec": 8.238233928044792, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.130240440368652, "lr": 0.0002, "elapsed_sec": 38308.52629876137, "step_time_sec": 8.239622213994153, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.135151386260986, "lr": 0.0002, "elapsed_sec": 38316.76429152489, "step_time_sec": 8.23780322202947, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.134964942932129, "lr": 0.0002, "elapsed_sec": 38325.00284075737, "step_time_sec": 8.238382878014818, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.17641544342041, "lr": 0.0002, "elapsed_sec": 38333.24158859253, "step_time_sec": 8.238596461014822, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.150266170501709, "lr": 0.0002, "elapsed_sec": 38341.47948384285, "step_time_sec": 8.23766835097922, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.259943962097168, "lr": 0.0002, "elapsed_sec": 38349.71858930588, "step_time_sec": 8.238989277975634, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1518707275390625, "lr": 0.0002, "elapsed_sec": 38357.95532178879, "step_time_sec": 8.236549916968215, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.182876110076904, "lr": 0.0002, "elapsed_sec": 38366.19427609444, "step_time_sec": 8.23877822299255, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.203747272491455, "lr": 0.0002, "elapsed_sec": 38374.43341255188, "step_time_sec": 8.238965183962137, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.153195858001709, "lr": 0.0002, "elapsed_sec": 38382.67070937157, "step_time_sec": 8.237147106963675, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.106899738311768, "lr": 0.0002, "elapsed_sec": 38390.908867836, "step_time_sec": 8.238014819973614, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.177383899688721, "lr": 0.0002, "elapsed_sec": 38399.147428274155, "step_time_sec": 8.23845148802502, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.268152713775635, "lr": 0.0002, "elapsed_sec": 38407.38662433624, "step_time_sec": 8.238999287015758, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.113731384277344, "lr": 0.0002, "elapsed_sec": 38415.62601351738, "step_time_sec": 8.23923302598996, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.201530933380127, "lr": 0.0002, "elapsed_sec": 38423.86343359947, "step_time_sec": 8.237272653961554, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9422378540039062, "lr": 0.0002, "elapsed_sec": 38432.1015458107, "step_time_sec": 8.237965249980334, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.052651882171631, "lr": 0.0002, "elapsed_sec": 38440.33853459358, "step_time_sec": 8.236806894012261, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.95434308052063, "lr": 0.0002, "elapsed_sec": 38448.576530218124, "step_time_sec": 8.237861977017019, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3070068359375, "lr": 0.0002, "elapsed_sec": 38456.81469869614, "step_time_sec": 8.237989848013967, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.26159143447876, "lr": 0.0002, "elapsed_sec": 38465.05349826813, "step_time_sec": 8.238655181019567, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.143768310546875, "lr": 0.0002, "elapsed_sec": 38473.29236865044, "step_time_sec": 8.23872346902499, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2965474128723145, "lr": 0.0002, "elapsed_sec": 38481.52966284752, "step_time_sec": 8.237066055007745, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.238779067993164, "lr": 0.0002, "elapsed_sec": 38489.76838874817, "step_time_sec": 8.238563402963337, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.33662223815918, "lr": 0.0002, "elapsed_sec": 38498.007232666016, "step_time_sec": 8.238690949976444, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.099190711975098, "lr": 0.0002, "elapsed_sec": 38506.24622917175, "step_time_sec": 8.23885208799038, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.279786586761475, "lr": 0.0002, "elapsed_sec": 38514.48599934578, "step_time_sec": 8.239588338998146, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.181594371795654, "lr": 0.0002, "elapsed_sec": 38522.72403931618, "step_time_sec": 8.237869781034533, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.01067590713501, "lr": 0.0002, "elapsed_sec": 38530.962548971176, "step_time_sec": 8.238356917048804, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.137972831726074, "lr": 0.0002, "elapsed_sec": 38539.1997795105, "step_time_sec": 8.237065088003874, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.275479316711426, "lr": 0.0002, "elapsed_sec": 38547.43759036064, "step_time_sec": 8.237656185985543, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.127469062805176, "lr": 0.0002, "elapsed_sec": 38555.67744374275, "step_time_sec": 8.239686148008332, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.142709255218506, "lr": 0.0002, "elapsed_sec": 38563.91425395012, "step_time_sec": 8.236633277963847, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.132641792297363, "lr": 0.0002, "elapsed_sec": 38572.152787685394, "step_time_sec": 8.238392992992885, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1710052490234375, "lr": 0.0002, "elapsed_sec": 38580.38979244232, "step_time_sec": 8.236811203998514, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.194606304168701, "lr": 0.0002, "elapsed_sec": 38588.62902832031, "step_time_sec": 8.239068922004662, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.237257957458496, "lr": 0.0002, "elapsed_sec": 38596.86841273308, "step_time_sec": 8.239260032016318, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.294415473937988, "lr": 0.0002, "elapsed_sec": 38605.10598111153, "step_time_sec": 8.237449995009229, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.242104530334473, "lr": 0.0002, "elapsed_sec": 38613.34406042099, "step_time_sec": 8.237903464992996, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.106468677520752, "lr": 0.0002, "elapsed_sec": 38621.58056116104, "step_time_sec": 8.236299448995851, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.159144878387451, "lr": 0.0002, "elapsed_sec": 38629.81879234314, "step_time_sec": 8.238065341021866, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.076589584350586, "lr": 0.0002, "elapsed_sec": 38638.05751633644, "step_time_sec": 8.23857870901702, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.092916488647461, "lr": 0.0002, "elapsed_sec": 38646.295632362366, "step_time_sec": 8.237952655006666, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9922192096710205, "lr": 0.0002, "elapsed_sec": 38654.53425335884, "step_time_sec": 8.238478973973542, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.049256801605225, "lr": 0.0002, "elapsed_sec": 38662.770859241486, "step_time_sec": 8.236441227956675, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.100191116333008, "lr": 0.0002, "elapsed_sec": 38671.00997829437, "step_time_sec": 8.238942884025164, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.031225204467773, "lr": 0.0002, "elapsed_sec": 38679.24734187126, "step_time_sec": 8.237214286986273, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0615763664245605, "lr": 0.0002, "elapsed_sec": 38687.486288547516, "step_time_sec": 8.238805559987668, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.097773551940918, "lr": 0.0002, "elapsed_sec": 38695.72509050369, "step_time_sec": 8.238608814019244, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.027910232543945, "lr": 0.0002, "elapsed_sec": 38703.962003707886, "step_time_sec": 8.23676031595096, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.136272430419922, "lr": 0.0002, "elapsed_sec": 38712.200278520584, "step_time_sec": 8.238127965014428, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.307176113128662, "lr": 0.0002, "elapsed_sec": 38720.437714099884, "step_time_sec": 8.237300703010987, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.12824010848999, "lr": 0.0002, "elapsed_sec": 38728.67546033859, "step_time_sec": 8.237530846032314, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.266060829162598, "lr": 0.0002, "elapsed_sec": 38736.914283275604, "step_time_sec": 8.238725223054644, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.138919830322266, "lr": 0.0002, "elapsed_sec": 38745.152743816376, "step_time_sec": 8.238275928015355, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.220980644226074, "lr": 0.0002, "elapsed_sec": 38753.39198732376, "step_time_sec": 8.239098085963633, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0722270011901855, "lr": 0.0002, "elapsed_sec": 38761.62903738022, "step_time_sec": 8.236823974992149, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.101015090942383, "lr": 0.0002, "elapsed_sec": 38769.8686914444, "step_time_sec": 8.23949679499492, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.120131015777588, "lr": 0.0002, "elapsed_sec": 38778.10742807388, "step_time_sec": 8.238589357992169, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.008871078491211, "lr": 0.0002, "elapsed_sec": 38786.344888687134, "step_time_sec": 8.23731456103269, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.171273708343506, "lr": 0.0002, "elapsed_sec": 38794.58370804787, "step_time_sec": 8.238617199007422, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.103845596313477, "lr": 0.0002, "elapsed_sec": 38802.82086443901, "step_time_sec": 8.237046702997759, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0965471267700195, "lr": 0.0002, "elapsed_sec": 38811.06029844284, "step_time_sec": 8.23929086397402, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.002838611602783, "lr": 0.0002, "elapsed_sec": 38819.29782271385, "step_time_sec": 8.237361763021909, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.254896640777588, "lr": 0.0002, "elapsed_sec": 38827.53547954559, "step_time_sec": 8.23745818901807, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.059600353240967, "lr": 0.0002, "elapsed_sec": 38835.77240729332, "step_time_sec": 8.236772310978267, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.094878673553467, "lr": 0.0002, "elapsed_sec": 38844.00967955589, "step_time_sec": 8.237106554966886, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.314605236053467, "lr": 0.0002, "elapsed_sec": 38852.248036146164, "step_time_sec": 8.23820797202643, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.082945823669434, "lr": 0.0002, "elapsed_sec": 38860.485943078995, "step_time_sec": 8.23773966502631, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.274631500244141, "lr": 0.0002, "elapsed_sec": 38868.72502756119, "step_time_sec": 8.238945032004267, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.02215051651001, "lr": 0.0002, "elapsed_sec": 38876.96409153938, "step_time_sec": 8.23887848801678, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.008260250091553, "lr": 0.0002, "elapsed_sec": 38885.20158481598, "step_time_sec": 8.237314091995358, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.040772914886475, "lr": 0.0002, "elapsed_sec": 38893.441318273544, "step_time_sec": 8.239569551020395, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.244556903839111, "lr": 0.0002, "elapsed_sec": 38901.678336143494, "step_time_sec": 8.236833105038386, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0742292404174805, "lr": 0.0002, "elapsed_sec": 38909.91721820831, "step_time_sec": 8.238709828990977, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.082838535308838, "lr": 0.0002, "elapsed_sec": 38918.15551686287, "step_time_sec": 8.238138443965, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.040527820587158, "lr": 0.0002, "elapsed_sec": 38926.3939640522, "step_time_sec": 8.23827492701821, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9072346687316895, "lr": 0.0002, "elapsed_sec": 38934.63264417648, "step_time_sec": 8.238574899034575, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.270802021026611, "lr": 0.0002, "elapsed_sec": 38942.872492313385, "step_time_sec": 8.239695648953784, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.164196014404297, "lr": 0.0002, "elapsed_sec": 38951.11209821701, "step_time_sec": 8.23938470199937, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.14161491394043, "lr": 0.0002, "elapsed_sec": 38959.35014462471, "step_time_sec": 8.237824467010796, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.054760456085205, "lr": 0.0002, "elapsed_sec": 38967.58850979805, "step_time_sec": 8.238219138991553, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.189032554626465, "lr": 0.0002, "elapsed_sec": 38975.827291727066, "step_time_sec": 8.23863173997961, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.205406665802002, "lr": 0.0002, "elapsed_sec": 38984.064202308655, "step_time_sec": 8.236726998002268, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1703200340271, "lr": 0.0002, "elapsed_sec": 38992.30255198479, "step_time_sec": 8.238196473044809, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1225504875183105, "lr": 0.0002, "elapsed_sec": 39000.53984117508, "step_time_sec": 8.237147787003778, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1723761558532715, "lr": 0.0002, "elapsed_sec": 39008.77781558037, "step_time_sec": 8.237784839002416, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1332855224609375, "lr": 0.0002, "elapsed_sec": 39017.01600408554, "step_time_sec": 8.23805150797125, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.19288969039917, "lr": 0.0002, "elapsed_sec": 39025.25371623039, "step_time_sec": 8.23755841102684, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.151423931121826, "lr": 0.0002, "elapsed_sec": 39033.49266052246, "step_time_sec": 8.238745915994514, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.265441417694092, "lr": 0.0002, "elapsed_sec": 39041.729479551315, "step_time_sec": 8.236699663975742, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.096005439758301, "lr": 0.0002, "elapsed_sec": 39049.968160152435, "step_time_sec": 8.238464812049642, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1205902099609375, "lr": 0.0002, "elapsed_sec": 39058.20722579956, "step_time_sec": 8.23889733699616, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.222405433654785, "lr": 0.0002, "elapsed_sec": 39066.44340467453, "step_time_sec": 8.236045094963629, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.130985736846924, "lr": 0.0002, "elapsed_sec": 39074.68186044693, "step_time_sec": 8.238255238975398, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.081662654876709, "lr": 0.0002, "elapsed_sec": 39082.91976523399, "step_time_sec": 8.237738683994394, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.21730375289917, "lr": 0.0002, "elapsed_sec": 39091.15804982185, "step_time_sec": 8.23813697899459, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.050417900085449, "lr": 0.0002, "elapsed_sec": 39099.396817445755, "step_time_sec": 8.23858140403172, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1551194190979, "lr": 0.0002, "elapsed_sec": 39107.63563632965, "step_time_sec": 8.238701427995693, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.055342197418213, "lr": 0.0002, "elapsed_sec": 39115.87344956398, "step_time_sec": 8.23771463194862, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.103372573852539, "lr": 0.0002, "elapsed_sec": 39124.11133003235, "step_time_sec": 8.237613742996473, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.256953716278076, "lr": 0.0002, "elapsed_sec": 39132.35131597519, "step_time_sec": 8.239873181039002, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.122291088104248, "lr": 0.0002, "elapsed_sec": 39140.59064245224, "step_time_sec": 8.23914717999287, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.154661655426025, "lr": 0.0002, "elapsed_sec": 39148.8288795948, "step_time_sec": 8.238032269000541, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.324481964111328, "lr": 0.0002, "elapsed_sec": 39157.0677087307, "step_time_sec": 8.238709954021033, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.114332675933838, "lr": 0.0002, "elapsed_sec": 39165.30489063263, "step_time_sec": 8.236984957999084, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.184126377105713, "lr": 0.0002, "elapsed_sec": 39173.54322719574, "step_time_sec": 8.23820695397444, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.089258193969727, "lr": 0.0002, "elapsed_sec": 39181.781411886215, "step_time_sec": 8.23802304099081, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.006596088409424, "lr": 0.0002, "elapsed_sec": 39190.019493341446, "step_time_sec": 8.23789161699824, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.019099235534668, "lr": 0.0002, "elapsed_sec": 39198.256273031235, "step_time_sec": 8.236630877945572, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0936689376831055, "lr": 0.0002, "elapsed_sec": 39206.493799209595, "step_time_sec": 8.237371201976202, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.160923004150391, "lr": 0.0002, "elapsed_sec": 39214.73238039017, "step_time_sec": 8.238394841027912, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.138136386871338, "lr": 0.0002, "elapsed_sec": 39222.96957421303, "step_time_sec": 8.237017088045832, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.012280464172363, "lr": 0.0002, "elapsed_sec": 39231.20806765556, "step_time_sec": 8.238341650052462, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.132092475891113, "lr": 0.0002, "elapsed_sec": 39239.44763994217, "step_time_sec": 8.239389271009713, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.067625999450684, "lr": 0.0002, "elapsed_sec": 39247.68436360359, "step_time_sec": 8.236616615962703, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.011422634124756, "lr": 0.0002, "elapsed_sec": 39255.922948122025, "step_time_sec": 8.238392025989015, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.142153739929199, "lr": 0.0002, "elapsed_sec": 39264.15982222557, "step_time_sec": 8.23670792899793, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.215088844299316, "lr": 0.0002, "elapsed_sec": 39272.399626255035, "step_time_sec": 8.239631000964437, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.063607215881348, "lr": 0.0002, "elapsed_sec": 39280.638192892075, "step_time_sec": 8.238435859035235, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.148125171661377, "lr": 0.0002, "elapsed_sec": 39288.87513375282, "step_time_sec": 8.236765409994405, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.181743144989014, "lr": 0.0002, "elapsed_sec": 39297.11219382286, "step_time_sec": 8.236916516034398, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.022737503051758, "lr": 0.0002, "elapsed_sec": 39305.349573373795, "step_time_sec": 8.237195130030159, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1244893074035645, "lr": 0.0002, "elapsed_sec": 39313.58819794655, "step_time_sec": 8.238463582994882, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.122616767883301, "lr": 0.0002, "elapsed_sec": 39321.824575424194, "step_time_sec": 8.236206407018472, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.170510292053223, "lr": 0.0002, "elapsed_sec": 39330.06253147125, "step_time_sec": 8.237774660985451, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.126738548278809, "lr": 0.0002, "elapsed_sec": 39338.30121302605, "step_time_sec": 8.238550600013696, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.178616046905518, "lr": 0.0002, "elapsed_sec": 39346.5391292572, "step_time_sec": 8.237709531967994, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.20386266708374, "lr": 0.0002, "elapsed_sec": 39354.77751469612, "step_time_sec": 8.238249318965245, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.15822172164917, "lr": 0.0002, "elapsed_sec": 39363.014473199844, "step_time_sec": 8.23684902099194, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.159368515014648, "lr": 0.0002, "elapsed_sec": 39371.25355577469, "step_time_sec": 8.23884263500804, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.066781997680664, "lr": 0.0002, "elapsed_sec": 39379.491993904114, "step_time_sec": 8.238276605959982, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.411567687988281, "lr": 0.0002, "elapsed_sec": 39387.72975850105, "step_time_sec": 8.237628059985582, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.141366958618164, "lr": 0.0002, "elapsed_sec": 39395.96782064438, "step_time_sec": 8.237905620015226, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.185072422027588, "lr": 0.0002, "elapsed_sec": 39404.204307079315, "step_time_sec": 8.236276737996377, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.151291847229004, "lr": 0.0002, "elapsed_sec": 39412.44095110893, "step_time_sec": 8.236469466995914, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.146592617034912, "lr": 0.0002, "elapsed_sec": 39420.67963218689, "step_time_sec": 8.238545481988695, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.098271369934082, "lr": 0.0002, "elapsed_sec": 39428.91700863838, "step_time_sec": 8.237202000047546, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.207322120666504, "lr": 0.0002, "elapsed_sec": 39437.1550450325, "step_time_sec": 8.237890193995554, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.219666481018066, "lr": 0.0002, "elapsed_sec": 39445.39322018623, "step_time_sec": 8.237981569953263, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.316993236541748, "lr": 0.0002, "elapsed_sec": 39453.63171052933, "step_time_sec": 8.23834074701881, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1746697425842285, "lr": 0.0002, "elapsed_sec": 39461.86790537834, "step_time_sec": 8.236022209981456, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.012084484100342, "lr": 0.0002, "elapsed_sec": 39470.10532283783, "step_time_sec": 8.237253476982005, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.092646598815918, "lr": 0.0002, "elapsed_sec": 39478.3435690403, "step_time_sec": 8.23813751898706, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.066361427307129, "lr": 0.0002, "elapsed_sec": 39486.58247232437, "step_time_sec": 8.238749689015094, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.168209075927734, "lr": 0.0002, "elapsed_sec": 39494.82180094719, "step_time_sec": 8.239104131003842, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1436448097229, "lr": 0.0002, "elapsed_sec": 39503.05956315994, "step_time_sec": 8.237651953008026, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.205262660980225, "lr": 0.0002, "elapsed_sec": 39511.29753446579, "step_time_sec": 8.237751426990144, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.22307014465332, "lr": 0.0002, "elapsed_sec": 39519.53489923477, "step_time_sec": 8.237206379999407, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2275543212890625, "lr": 0.0002, "elapsed_sec": 39527.77254033089, "step_time_sec": 8.23749696998857, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.145629405975342, "lr": 0.0002, "elapsed_sec": 39536.01056647301, "step_time_sec": 8.237853404018097, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.209803581237793, "lr": 0.0002, "elapsed_sec": 39544.24893760681, "step_time_sec": 8.238198605016805, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.076907634735107, "lr": 0.0002, "elapsed_sec": 39552.488758802414, "step_time_sec": 8.239656843012199, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.066554546356201, "lr": 0.0002, "elapsed_sec": 39560.72666144371, "step_time_sec": 8.237765130004846, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.141489028930664, "lr": 0.0002, "elapsed_sec": 39568.9659178257, "step_time_sec": 8.239088297996204, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1477484703063965, "lr": 0.0002, "elapsed_sec": 39577.204075574875, "step_time_sec": 8.237966774962842, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.082592010498047, "lr": 0.0002, "elapsed_sec": 39585.442120313644, "step_time_sec": 8.237870769982692, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.233999729156494, "lr": 0.0002, "elapsed_sec": 39593.681062459946, "step_time_sec": 8.238826027954929, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.170904159545898, "lr": 0.0002, "elapsed_sec": 39601.919607400894, "step_time_sec": 8.238310008018743, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.090085029602051, "lr": 0.0002, "elapsed_sec": 39610.15873646736, "step_time_sec": 8.238971557002515, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.296651840209961, "lr": 0.0002, "elapsed_sec": 39618.39685797691, "step_time_sec": 8.237977488024626, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.038412570953369, "lr": 0.0002, "elapsed_sec": 39626.6347618103, "step_time_sec": 8.237752041022759, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.084400653839111, "lr": 0.0002, "elapsed_sec": 39634.87450695038, "step_time_sec": 8.23953282495495, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.111002445220947, "lr": 0.0002, "elapsed_sec": 39643.11180019379, "step_time_sec": 8.237211362982634, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.083620071411133, "lr": 0.0002, "elapsed_sec": 39651.35152220726, "step_time_sec": 8.23950753203826, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.139312267303467, "lr": 0.0002, "elapsed_sec": 39659.591827869415, "step_time_sec": 8.240199650987051, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.038697719573975, "lr": 0.0002, "elapsed_sec": 39667.83017063141, "step_time_sec": 8.23815074498998, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.07318115234375, "lr": 0.0002, "elapsed_sec": 39676.068816661835, "step_time_sec": 8.238483960973099, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.123652458190918, "lr": 0.0002, "elapsed_sec": 39684.306520462036, "step_time_sec": 8.237503829004709, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.131955623626709, "lr": 0.0002, "elapsed_sec": 39692.544164419174, "step_time_sec": 8.237476681999397, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.023375034332275, "lr": 0.0002, "elapsed_sec": 39700.78185558319, "step_time_sec": 8.237582449044567, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.120555877685547, "lr": 0.0002, "elapsed_sec": 39709.0208427906, "step_time_sec": 8.238792285032105, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.090997695922852, "lr": 0.0002, "elapsed_sec": 39717.25869894028, "step_time_sec": 8.237681019993033, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.210145950317383, "lr": 0.0002, "elapsed_sec": 39725.49610829353, "step_time_sec": 8.23721603496233, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1698126792907715, "lr": 0.0002, "elapsed_sec": 39733.73392057419, "step_time_sec": 8.237694031966385, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.308048725128174, "lr": 0.0002, "elapsed_sec": 39741.970983982086, "step_time_sec": 8.236925214994699, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.232709884643555, "lr": 0.0002, "elapsed_sec": 39750.21019101143, "step_time_sec": 8.238992510014214, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.055616855621338, "lr": 0.0002, "elapsed_sec": 39758.448628902435, "step_time_sec": 8.238221306004561, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.074617862701416, "lr": 0.0002, "elapsed_sec": 39766.687289476395, "step_time_sec": 8.238496671023313, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.161369800567627, "lr": 0.0002, "elapsed_sec": 39774.926018476486, "step_time_sec": 8.238590344961267, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.131813049316406, "lr": 0.0002, "elapsed_sec": 39783.16498231888, "step_time_sec": 8.238820168015081, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.156527042388916, "lr": 0.0002, "elapsed_sec": 39791.40406346321, "step_time_sec": 8.238882183970418, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.01446533203125, "lr": 0.0002, "elapsed_sec": 39799.642990112305, "step_time_sec": 8.238744942995254, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.223954200744629, "lr": 0.0002, "elapsed_sec": 39807.880247831345, "step_time_sec": 8.2371058720164, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.227327823638916, "lr": 0.0002, "elapsed_sec": 39816.11926794052, "step_time_sec": 8.238884830032475, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.244609355926514, "lr": 0.0002, "elapsed_sec": 39824.35608434677, "step_time_sec": 8.23660901596304, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.190406322479248, "lr": 0.0002, "elapsed_sec": 39832.59492492676, "step_time_sec": 8.238673054031096, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.108188629150391, "lr": 0.0002, "elapsed_sec": 39840.83360981941, "step_time_sec": 8.238508718030062, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.094362258911133, "lr": 0.0002, "elapsed_sec": 39849.072150707245, "step_time_sec": 8.238378571986686, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.270168304443359, "lr": 0.0002, "elapsed_sec": 39857.31058597565, "step_time_sec": 8.238306910963729, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1730828285217285, "lr": 0.0002, "elapsed_sec": 39865.54848694801, "step_time_sec": 8.237701775040478, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.104867458343506, "lr": 0.0002, "elapsed_sec": 39873.78772544861, "step_time_sec": 8.23909161699703, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.075407981872559, "lr": 0.0002, "elapsed_sec": 39882.02546000481, "step_time_sec": 8.237558506021742, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.175532817840576, "lr": 0.0002, "elapsed_sec": 39890.26290035248, "step_time_sec": 8.237273300997913, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.257031440734863, "lr": 0.0002, "elapsed_sec": 39898.502165317535, "step_time_sec": 8.239099595986772, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.104894638061523, "lr": 0.0002, "elapsed_sec": 39906.73943543434, "step_time_sec": 8.2371497299755, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.150005340576172, "lr": 0.0002, "elapsed_sec": 39914.97836518288, "step_time_sec": 8.238739138003439, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.247046947479248, "lr": 0.0002, "elapsed_sec": 39923.21587204933, "step_time_sec": 8.237370154995006, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.149430274963379, "lr": 0.0002, "elapsed_sec": 39931.45386385918, "step_time_sec": 8.237766653997824, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.053738117218018, "lr": 0.0002, "elapsed_sec": 39939.693544626236, "step_time_sec": 8.23956143099349, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.070993900299072, "lr": 0.0002, "elapsed_sec": 39947.93025112152, "step_time_sec": 8.236522927996702, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.11898946762085, "lr": 0.0002, "elapsed_sec": 39956.16938471794, "step_time_sec": 8.239007847965695, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.093795299530029, "lr": 0.0002, "elapsed_sec": 39964.40752530098, "step_time_sec": 8.237972625996917, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.251888751983643, "lr": 0.0002, "elapsed_sec": 39972.646872758865, "step_time_sec": 8.239159081014805, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.343179225921631, "lr": 0.0002, "elapsed_sec": 39980.88555574417, "step_time_sec": 8.238525965018198, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.158233642578125, "lr": 0.0002, "elapsed_sec": 39989.122279167175, "step_time_sec": 8.236565630999394, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.172425746917725, "lr": 0.0002, "elapsed_sec": 39997.36136174202, "step_time_sec": 8.238933145999908, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.101687908172607, "lr": 0.0002, "elapsed_sec": 40005.5978178978, "step_time_sec": 8.236260827980004, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.164266586303711, "lr": 0.0002, "elapsed_sec": 40013.83667087555, "step_time_sec": 8.2387022980256, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1618475914001465, "lr": 0.0002, "elapsed_sec": 40022.07509803772, "step_time_sec": 8.23824675596552, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.230382919311523, "lr": 0.0002, "elapsed_sec": 40030.31392598152, "step_time_sec": 8.23867504298687, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.081011772155762, "lr": 0.0002, "elapsed_sec": 40038.552356004715, "step_time_sec": 8.238259293022566, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.115556716918945, "lr": 0.0002, "elapsed_sec": 40046.79021978378, "step_time_sec": 8.237742248980794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.035170555114746, "lr": 0.0002, "elapsed_sec": 40055.02923679352, "step_time_sec": 8.238823660009075, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.181735038757324, "lr": 0.0002, "elapsed_sec": 40063.26726388931, "step_time_sec": 8.237906805996317, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.156154155731201, "lr": 0.0002, "elapsed_sec": 40071.50566983223, "step_time_sec": 8.238192435994279, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.136815547943115, "lr": 0.0002, "elapsed_sec": 40079.74325418472, "step_time_sec": 8.237427941989154, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1123762130737305, "lr": 0.0002, "elapsed_sec": 40087.98091340065, "step_time_sec": 8.23749404400587, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.12057638168335, "lr": 0.0002, "elapsed_sec": 40096.22029709816, "step_time_sec": 8.239287222968414, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.251621246337891, "lr": 0.0002, "elapsed_sec": 40104.45829153061, "step_time_sec": 8.23775446700165, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.149824619293213, "lr": 0.0002, "elapsed_sec": 40112.69583797455, "step_time_sec": 8.237382196995895, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.093118667602539, "lr": 0.0002, "elapsed_sec": 40120.933584690094, "step_time_sec": 8.237573314982, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.119990348815918, "lr": 0.0002, "elapsed_sec": 40129.17118811607, "step_time_sec": 8.237456787959673, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.084250450134277, "lr": 0.0002, "elapsed_sec": 40137.41080737114, "step_time_sec": 8.239449496963061, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.148065090179443, "lr": 0.0002, "elapsed_sec": 40145.64954042435, "step_time_sec": 8.238551404036116, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.095302104949951, "lr": 0.0002, "elapsed_sec": 40153.88872885704, "step_time_sec": 8.23900138004683, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.970902919769287, "lr": 0.0002, "elapsed_sec": 40162.1273522377, "step_time_sec": 8.23849863797659, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.11286735534668, "lr": 0.0002, "elapsed_sec": 40170.36343669891, "step_time_sec": 8.235919323051348, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.188074588775635, "lr": 0.0002, "elapsed_sec": 40178.6026391983, "step_time_sec": 8.239026402996387, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2594122886657715, "lr": 0.0002, "elapsed_sec": 40186.83906054497, "step_time_sec": 8.23622099601198, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1362175941467285, "lr": 0.0002, "elapsed_sec": 40195.07730317116, "step_time_sec": 8.238163342990447, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.084496974945068, "lr": 0.0002, "elapsed_sec": 40203.31595993042, "step_time_sec": 8.238495516008697, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.262493133544922, "lr": 0.0002, "elapsed_sec": 40211.5543949604, "step_time_sec": 8.238242627005093, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.086878299713135, "lr": 0.0002, "elapsed_sec": 40219.7924926281, "step_time_sec": 8.237910292984452, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.168858528137207, "lr": 0.0002, "elapsed_sec": 40228.02982187271, "step_time_sec": 8.237165535974782, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.118706226348877, "lr": 0.0002, "elapsed_sec": 40236.26841926575, "step_time_sec": 8.238464327005204, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.096404075622559, "lr": 0.0002, "elapsed_sec": 40244.50660252571, "step_time_sec": 8.237972403003369, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.155847072601318, "lr": 0.0002, "elapsed_sec": 40252.74549102783, "step_time_sec": 8.238746553019155, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.248122692108154, "lr": 0.0002, "elapsed_sec": 40260.98462319374, "step_time_sec": 8.238995304040145, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.223729610443115, "lr": 0.0002, "elapsed_sec": 40269.22252321243, "step_time_sec": 8.237698653014377, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.185520648956299, "lr": 0.0002, "elapsed_sec": 40277.46071434021, "step_time_sec": 8.238024073012639, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.097385883331299, "lr": 0.0002, "elapsed_sec": 40285.69804191589, "step_time_sec": 8.237160383025184, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.156153678894043, "lr": 0.0002, "elapsed_sec": 40293.93651175499, "step_time_sec": 8.238301664008759, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.158508777618408, "lr": 0.0002, "elapsed_sec": 40302.17438054085, "step_time_sec": 8.237730243999977, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.084793567657471, "lr": 0.0002, "elapsed_sec": 40310.41195988655, "step_time_sec": 8.237419883953407, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9992380142211914, "lr": 0.0002, "elapsed_sec": 40318.65093445778, "step_time_sec": 8.238797904981766, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.403267860412598, "lr": 0.0002, "elapsed_sec": 40326.888187885284, "step_time_sec": 8.237084215972573, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.28911828994751, "lr": 0.0002, "elapsed_sec": 40335.127480745316, "step_time_sec": 8.239137770957313, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2206315994262695, "lr": 0.0002, "elapsed_sec": 40343.36577153206, "step_time_sec": 8.238168173993472, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.191739082336426, "lr": 0.0002, "elapsed_sec": 40351.602860450745, "step_time_sec": 8.23690700699808, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.125054359436035, "lr": 0.0002, "elapsed_sec": 40359.84231066704, "step_time_sec": 8.239254289015662, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0991668701171875, "lr": 0.0002, "elapsed_sec": 40368.07931447029, "step_time_sec": 8.236850725021213, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.173267841339111, "lr": 0.0002, "elapsed_sec": 40376.31780076027, "step_time_sec": 8.238339876988903, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.206549167633057, "lr": 0.0002, "elapsed_sec": 40384.55499005318, "step_time_sec": 8.236985287978314, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2321953773498535, "lr": 0.0002, "elapsed_sec": 40392.793142557144, "step_time_sec": 8.237964640022255, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0590105056762695, "lr": 0.0002, "elapsed_sec": 40401.03062200546, "step_time_sec": 8.237366118002683, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.170128345489502, "lr": 0.0002, "elapsed_sec": 40409.268702983856, "step_time_sec": 8.237906058027875, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9998936653137207, "lr": 0.0002, "elapsed_sec": 40417.50648236275, "step_time_sec": 8.237608202034608, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.143731117248535, "lr": 0.0002, "elapsed_sec": 40425.743352890015, "step_time_sec": 8.236702558002435, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.182894706726074, "lr": 0.0002, "elapsed_sec": 40445.37551999092, "step_time_sec": 19.63200792297721, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.064566135406494, "lr": 0.0002, "elapsed_sec": 40453.60889005661, "step_time_sec": 8.233238279004581, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.152867317199707, "lr": 0.0002, "elapsed_sec": 40461.84530854225, "step_time_sec": 8.236221956030931, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.152390956878662, "lr": 0.0002, "elapsed_sec": 40470.084067583084, "step_time_sec": 8.238599208008964, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.030417442321777, "lr": 0.0002, "elapsed_sec": 40478.32220315933, "step_time_sec": 8.237979391997214, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.084092140197754, "lr": 0.0002, "elapsed_sec": 40486.56017136574, "step_time_sec": 8.237835485953838, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.089109897613525, "lr": 0.0002, "elapsed_sec": 40494.79715514183, "step_time_sec": 8.236758166982327, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.103480339050293, "lr": 0.0002, "elapsed_sec": 40503.03537225723, "step_time_sec": 8.238068033999298, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1848978996276855, "lr": 0.0002, "elapsed_sec": 40511.27265834808, "step_time_sec": 8.237153480993584, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.123196601867676, "lr": 0.0002, "elapsed_sec": 40519.50854206085, "step_time_sec": 8.235666981956456, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.060643672943115, "lr": 0.0002, "elapsed_sec": 40527.74731659889, "step_time_sec": 8.238630565989297, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.085572242736816, "lr": 0.0002, "elapsed_sec": 40535.9859187603, "step_time_sec": 8.238437199033797, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9786064624786377, "lr": 0.0002, "elapsed_sec": 40544.22356200218, "step_time_sec": 8.237459445022978, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.136937141418457, "lr": 0.0002, "elapsed_sec": 40552.460216999054, "step_time_sec": 8.236512712028343, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.254648685455322, "lr": 0.0002, "elapsed_sec": 40560.69859957695, "step_time_sec": 8.238184229994658, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9999377727508545, "lr": 0.0002, "elapsed_sec": 40568.93693995476, "step_time_sec": 8.238198511011433, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.042741298675537, "lr": 0.0002, "elapsed_sec": 40577.17301750183, "step_time_sec": 8.235876569990069, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.005956649780273, "lr": 0.0002, "elapsed_sec": 40585.41104602814, "step_time_sec": 8.237896931997966, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.28877067565918, "lr": 0.0002, "elapsed_sec": 40593.64840364456, "step_time_sec": 8.23720885696821, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.098435878753662, "lr": 0.0002, "elapsed_sec": 40601.88589930534, "step_time_sec": 8.237297708983533, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.12207555770874, "lr": 0.0002, "elapsed_sec": 40610.12477707863, "step_time_sec": 8.238685077987611, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.104743480682373, "lr": 0.0002, "elapsed_sec": 40618.36428165436, "step_time_sec": 8.239374600991141, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.169459342956543, "lr": 0.0002, "elapsed_sec": 40626.603867053986, "step_time_sec": 8.239423246996012, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.985013246536255, "lr": 0.0002, "elapsed_sec": 40634.84257388115, "step_time_sec": 8.238498842983972, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.16037654876709, "lr": 0.0002, "elapsed_sec": 40643.081080913544, "step_time_sec": 8.238384377968032, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.118943214416504, "lr": 0.0002, "elapsed_sec": 40651.320318222046, "step_time_sec": 8.239023717993405, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.022836208343506, "lr": 0.0002, "elapsed_sec": 40659.557789087296, "step_time_sec": 8.237291236990131, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.123659133911133, "lr": 0.0002, "elapsed_sec": 40667.797786951065, "step_time_sec": 8.239833722996991, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1051106452941895, "lr": 0.0002, "elapsed_sec": 40676.03599834442, "step_time_sec": 8.23806845699437, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.033498764038086, "lr": 0.0002, "elapsed_sec": 40684.274960041046, "step_time_sec": 8.238750891003292, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.202043533325195, "lr": 0.0002, "elapsed_sec": 40692.51600146294, "step_time_sec": 8.240909697953612, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.057281970977783, "lr": 0.0002, "elapsed_sec": 40700.75220823288, "step_time_sec": 8.236026806989685, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.18203067779541, "lr": 0.0002, "elapsed_sec": 40708.99154686928, "step_time_sec": 8.23912768199807, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.143141746520996, "lr": 0.0002, "elapsed_sec": 40717.22772073746, "step_time_sec": 8.235977951029781, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.217984676361084, "lr": 0.0002, "elapsed_sec": 40725.46597456932, "step_time_sec": 8.238085342978593, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.127089023590088, "lr": 0.0002, "elapsed_sec": 40733.704583883286, "step_time_sec": 8.238481501990464, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.293765544891357, "lr": 0.0002, "elapsed_sec": 40741.941610097885, "step_time_sec": 8.236830891983118, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.135391712188721, "lr": 0.0002, "elapsed_sec": 40750.180589437485, "step_time_sec": 8.238824034982827, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.269411087036133, "lr": 0.0002, "elapsed_sec": 40758.4179725647, "step_time_sec": 8.237190894025844, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.171034336090088, "lr": 0.0002, "elapsed_sec": 40766.6574575901, "step_time_sec": 8.239355626050383, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.053078651428223, "lr": 0.0002, "elapsed_sec": 40774.89626002312, "step_time_sec": 8.238613838970196, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.158570766448975, "lr": 0.0002, "elapsed_sec": 40783.13380408287, "step_time_sec": 8.237370894988999, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9432883262634277, "lr": 0.0002, "elapsed_sec": 40791.37135052681, "step_time_sec": 8.237378165998962, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.018770217895508, "lr": 0.0002, "elapsed_sec": 40799.60873603821, "step_time_sec": 8.237226778990589, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.035993576049805, "lr": 0.0002, "elapsed_sec": 40807.84724402428, "step_time_sec": 8.238354834960774, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.002190589904785, "lr": 0.0002, "elapsed_sec": 40816.08555436134, "step_time_sec": 8.23818664898863, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.165412425994873, "lr": 0.0002, "elapsed_sec": 40824.324041843414, "step_time_sec": 8.23826630198164, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.063661098480225, "lr": 0.0002, "elapsed_sec": 40832.562818050385, "step_time_sec": 8.238607735023834, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.161090850830078, "lr": 0.0002, "elapsed_sec": 40840.79998922348, "step_time_sec": 8.237026539980434, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.081125259399414, "lr": 0.0002, "elapsed_sec": 40849.03962612152, "step_time_sec": 8.239474867994431, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.125208377838135, "lr": 0.0002, "elapsed_sec": 40857.277923345566, "step_time_sec": 8.238158155989368, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.166604042053223, "lr": 0.0002, "elapsed_sec": 40865.51615214348, "step_time_sec": 8.2380217500031, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.11185359954834, "lr": 0.0002, "elapsed_sec": 40873.75502085686, "step_time_sec": 8.238705830008257, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.109750270843506, "lr": 0.0002, "elapsed_sec": 40881.991510629654, "step_time_sec": 8.236355150002055, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.018115997314453, "lr": 0.0002, "elapsed_sec": 40890.22940707207, "step_time_sec": 8.237681124999654, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.068024635314941, "lr": 0.0002, "elapsed_sec": 40898.46740746498, "step_time_sec": 8.23785712197423, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.041150093078613, "lr": 0.0002, "elapsed_sec": 40906.70405817032, "step_time_sec": 8.236516691977158, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.008793830871582, "lr": 0.0002, "elapsed_sec": 40914.941339969635, "step_time_sec": 8.237080286024138, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.063887596130371, "lr": 0.0002, "elapsed_sec": 40923.17910671234, "step_time_sec": 8.237591163022444, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.138852596282959, "lr": 0.0002, "elapsed_sec": 40931.41690278053, "step_time_sec": 8.237599887012038, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9267680644989014, "lr": 0.0002, "elapsed_sec": 40939.6527326107, "step_time_sec": 8.23568098800024, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.144618988037109, "lr": 0.0002, "elapsed_sec": 40947.89068198204, "step_time_sec": 8.23784631502349, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.290583610534668, "lr": 0.0002, "elapsed_sec": 40956.127640247345, "step_time_sec": 8.236760303960182, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.004245281219482, "lr": 0.0002, "elapsed_sec": 40964.36585021019, "step_time_sec": 8.238063765980769, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.092045783996582, "lr": 0.0002, "elapsed_sec": 40972.60441946983, "step_time_sec": 8.238357422000263, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.052606105804443, "lr": 0.0002, "elapsed_sec": 40980.84217143059, "step_time_sec": 8.237615243997425, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.059430122375488, "lr": 0.0002, "elapsed_sec": 40989.0804913044, "step_time_sec": 8.2381710470072, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.29556131362915, "lr": 0.0002, "elapsed_sec": 40997.315806388855, "step_time_sec": 8.235124580038246, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9653069972991943, "lr": 0.0002, "elapsed_sec": 41005.552943229675, "step_time_sec": 8.23694006795995, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.128475666046143, "lr": 0.0002, "elapsed_sec": 41013.79054784775, "step_time_sec": 8.237444706028327, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.37470817565918, "lr": 0.0002, "elapsed_sec": 41022.02922153473, "step_time_sec": 8.238562036014628, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.23886251449585, "lr": 0.0002, "elapsed_sec": 41030.26691555977, "step_time_sec": 8.237507187994197, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2483439445495605, "lr": 0.0002, "elapsed_sec": 41038.50507068634, "step_time_sec": 8.238013606984168, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.105104446411133, "lr": 0.0002, "elapsed_sec": 41046.74402594566, "step_time_sec": 8.238763243018184, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.075575828552246, "lr": 0.0002, "elapsed_sec": 41054.98062944412, "step_time_sec": 8.236429231998045, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.077264308929443, "lr": 0.0002, "elapsed_sec": 41063.2191901207, "step_time_sec": 8.238389989011921, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.154603004455566, "lr": 0.0002, "elapsed_sec": 41071.45754790306, "step_time_sec": 8.238164631999098, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.144250869750977, "lr": 0.0002, "elapsed_sec": 41079.69511508942, "step_time_sec": 8.237397575983778, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.108080863952637, "lr": 0.0002, "elapsed_sec": 41087.93305230141, "step_time_sec": 8.237744144978933, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.083272933959961, "lr": 0.0002, "elapsed_sec": 41096.17078471184, "step_time_sec": 8.237619366962463, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.235654830932617, "lr": 0.0002, "elapsed_sec": 41104.4105155468, "step_time_sec": 8.239518146961927, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.078118324279785, "lr": 0.0002, "elapsed_sec": 41112.64915585518, "step_time_sec": 8.238509345974308, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0774712562561035, "lr": 0.0002, "elapsed_sec": 41120.88755607605, "step_time_sec": 8.238201975997072, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.072124004364014, "lr": 0.0002, "elapsed_sec": 41129.12514138222, "step_time_sec": 8.237438748998102, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.063593864440918, "lr": 0.0002, "elapsed_sec": 41137.36108088493, "step_time_sec": 8.235739936004393, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.108907699584961, "lr": 0.0002, "elapsed_sec": 41145.599481105804, "step_time_sec": 8.238224018015899, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.13688850402832, "lr": 0.0002, "elapsed_sec": 41153.836547613144, "step_time_sec": 8.236909362021834, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.02551794052124, "lr": 0.0002, "elapsed_sec": 41162.07467961311, "step_time_sec": 8.237997212971095, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.138952255249023, "lr": 0.0002, "elapsed_sec": 41170.31311535835, "step_time_sec": 8.23823416698724, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.117924213409424, "lr": 0.0002, "elapsed_sec": 41178.55179142952, "step_time_sec": 8.238491274008993, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.10938024520874, "lr": 0.0002, "elapsed_sec": 41186.78985452652, "step_time_sec": 8.237884758040309, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0661773681640625, "lr": 0.0002, "elapsed_sec": 41195.027039051056, "step_time_sec": 8.237034964957274, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1593146324157715, "lr": 0.0002, "elapsed_sec": 41203.26532793045, "step_time_sec": 8.238170084019657, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.065994739532471, "lr": 0.0002, "elapsed_sec": 41211.502636671066, "step_time_sec": 8.237064848013688, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.869211196899414, "lr": 0.0002, "elapsed_sec": 41219.7408926487, "step_time_sec": 8.238104392017704, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.319993019104004, "lr": 0.0002, "elapsed_sec": 41227.9791033268, "step_time_sec": 8.23804010701133, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.116899490356445, "lr": 0.0002, "elapsed_sec": 41236.215569496155, "step_time_sec": 8.23632763599744, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.213868141174316, "lr": 0.0002, "elapsed_sec": 41244.45383167267, "step_time_sec": 8.238078172027599, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.021697998046875, "lr": 0.0002, "elapsed_sec": 41252.691623687744, "step_time_sec": 8.237654328986537, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.129363536834717, "lr": 0.0002, "elapsed_sec": 41260.93130683899, "step_time_sec": 8.239484576042742, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.783477306365967, "lr": 0.0002, "elapsed_sec": 41269.16988611221, "step_time_sec": 8.23838685901137, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.269361972808838, "lr": 0.0002, "elapsed_sec": 41277.40912795067, "step_time_sec": 8.239102969993837, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.269881248474121, "lr": 0.0002, "elapsed_sec": 41285.647394657135, "step_time_sec": 8.238065036013722, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.166187763214111, "lr": 0.0002, "elapsed_sec": 41293.88354635239, "step_time_sec": 8.235999419004656, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.012437343597412, "lr": 0.0002, "elapsed_sec": 41302.12141394615, "step_time_sec": 8.237716439005453, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.109721660614014, "lr": 0.0002, "elapsed_sec": 41310.358389377594, "step_time_sec": 8.23675772297429, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.884021282196045, "lr": 0.0002, "elapsed_sec": 41318.59624314308, "step_time_sec": 8.23770323395729, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.148698329925537, "lr": 0.0002, "elapsed_sec": 41326.833396196365, "step_time_sec": 8.236981853027828, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.084864616394043, "lr": 0.0002, "elapsed_sec": 41335.07075405121, "step_time_sec": 8.237244682037272, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.059075832366943, "lr": 0.0002, "elapsed_sec": 41343.30985021591, "step_time_sec": 8.238931332016364, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.166491508483887, "lr": 0.0002, "elapsed_sec": 41351.54658651352, "step_time_sec": 8.236576060997322, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.074453830718994, "lr": 0.0002, "elapsed_sec": 41359.78650069237, "step_time_sec": 8.239757318980992, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.083418846130371, "lr": 0.0002, "elapsed_sec": 41368.02451920509, "step_time_sec": 8.23779307003133, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0695013999938965, "lr": 0.0002, "elapsed_sec": 41376.26281952858, "step_time_sec": 8.238171678967774, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.022055625915527, "lr": 0.0002, "elapsed_sec": 41384.566388607025, "step_time_sec": 8.244598168006632, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.100009918212891, "lr": 0.0002, "elapsed_sec": 41392.80400776863, "step_time_sec": 8.23741473298287, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.097066879272461, "lr": 0.0002, "elapsed_sec": 41401.0412466526, "step_time_sec": 8.237130267021712, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0551438331604, "lr": 0.0002, "elapsed_sec": 41409.279515743256, "step_time_sec": 8.238060749950819, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1129679679870605, "lr": 0.0002, "elapsed_sec": 41417.51783943176, "step_time_sec": 8.238199982966762, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.100416660308838, "lr": 0.0002, "elapsed_sec": 41425.75584077835, "step_time_sec": 8.237787481048144, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.060953140258789, "lr": 0.0002, "elapsed_sec": 41433.99444079399, "step_time_sec": 8.238435323990416, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.182270526885986, "lr": 0.0002, "elapsed_sec": 41442.23186612129, "step_time_sec": 8.237291154975537, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9397966861724854, "lr": 0.0002, "elapsed_sec": 41450.47050666809, "step_time_sec": 8.238429119985085, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.08830451965332, "lr": 0.0002, "elapsed_sec": 41458.70720553398, "step_time_sec": 8.23653758695582, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.136861324310303, "lr": 0.0002, "elapsed_sec": 41466.94587612152, "step_time_sec": 8.238484049972612, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.014965534210205, "lr": 0.0002, "elapsed_sec": 41475.18412685394, "step_time_sec": 8.238095844979398, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.061525821685791, "lr": 0.0002, "elapsed_sec": 41483.42186355591, "step_time_sec": 8.237620515981689, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.261218070983887, "lr": 0.0002, "elapsed_sec": 41491.66014313698, "step_time_sec": 8.238059469033033, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.100178241729736, "lr": 0.0002, "elapsed_sec": 41499.896802425385, "step_time_sec": 8.236486738023814, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9323480129241943, "lr": 0.0002, "elapsed_sec": 41508.13555312157, "step_time_sec": 8.238608651969116, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.032712936401367, "lr": 0.0002, "elapsed_sec": 41516.3737449646, "step_time_sec": 8.238065348996315, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.018951416015625, "lr": 0.0002, "elapsed_sec": 41524.61250424385, "step_time_sec": 8.238563349004835, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.008364677429199, "lr": 0.0002, "elapsed_sec": 41532.85143017769, "step_time_sec": 8.23876159102656, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0575151443481445, "lr": 0.0002, "elapsed_sec": 41541.08820319176, "step_time_sec": 51.80375623202417, "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": 0.9979407929931767, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 5001, "loss": 4.080952167510986, "lr": 0.0002, "elapsed_sec": 41592.90073251724, "step_time_sec": 8.24526530096773, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9342939853668213, "lr": 0.0002, "elapsed_sec": 41601.13606238365, "step_time_sec": 8.23516770702554, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.960510492324829, "lr": 0.0002, "elapsed_sec": 41609.374349832535, "step_time_sec": 8.238140601024497, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4454169273376465, "lr": 0.0002, "elapsed_sec": 41617.61303687096, "step_time_sec": 8.238547697023023, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.039341926574707, "lr": 0.0002, "elapsed_sec": 41625.84997820854, "step_time_sec": 8.236740903987084, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.119768142700195, "lr": 0.0002, "elapsed_sec": 41634.088017463684, "step_time_sec": 8.237909228017088, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.091827392578125, "lr": 0.0002, "elapsed_sec": 41642.326083660126, "step_time_sec": 8.237892992969137, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.187719821929932, "lr": 0.0002, "elapsed_sec": 41650.56522488594, "step_time_sec": 8.23894533701241, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9197678565979004, "lr": 0.0002, "elapsed_sec": 41658.80347800255, "step_time_sec": 8.23807920300169, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.11436653137207, "lr": 0.0002, "elapsed_sec": 41667.04124522209, "step_time_sec": 8.237622204003856, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.019270896911621, "lr": 0.0002, "elapsed_sec": 41675.280729055405, "step_time_sec": 8.239305021998007, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.136500358581543, "lr": 0.0002, "elapsed_sec": 41683.51711511612, "step_time_sec": 8.236284360988066, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.195864200592041, "lr": 0.0002, "elapsed_sec": 41691.75505185127, "step_time_sec": 8.23771137202857, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.002102851867676, "lr": 0.0002, "elapsed_sec": 41699.99229884148, "step_time_sec": 8.237130704976153, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.199377059936523, "lr": 0.0002, "elapsed_sec": 41708.23165822029, "step_time_sec": 8.239172619010787, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.013152122497559, "lr": 0.0002, "elapsed_sec": 41716.47001624107, "step_time_sec": 8.238178406027146, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.209020614624023, "lr": 0.0002, "elapsed_sec": 41724.70756340027, "step_time_sec": 8.237416713032871, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.067429065704346, "lr": 0.0002, "elapsed_sec": 41732.945758104324, "step_time_sec": 8.237995589035563, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.101102352142334, "lr": 0.0002, "elapsed_sec": 41741.18185377121, "step_time_sec": 8.23594648600556, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.029889106750488, "lr": 0.0002, "elapsed_sec": 41749.42062544823, "step_time_sec": 8.23858233896317, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.083903789520264, "lr": 0.0002, "elapsed_sec": 41757.660470962524, "step_time_sec": 8.239712565031368, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1476593017578125, "lr": 0.0002, "elapsed_sec": 41765.89772891998, "step_time_sec": 8.2370610020007, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.073474884033203, "lr": 0.0002, "elapsed_sec": 41774.137212753296, "step_time_sec": 8.239356725011021, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.241556644439697, "lr": 0.0002, "elapsed_sec": 41782.37385940552, "step_time_sec": 8.236409247037955, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.045310020446777, "lr": 0.0002, "elapsed_sec": 41790.61222410202, "step_time_sec": 8.23825053498149, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.031671524047852, "lr": 0.0002, "elapsed_sec": 41798.84933233261, "step_time_sec": 8.2369087410043, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.071165084838867, "lr": 0.0002, "elapsed_sec": 41807.08759379387, "step_time_sec": 8.238111362967175, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.060286521911621, "lr": 0.0002, "elapsed_sec": 41815.324964761734, "step_time_sec": 8.23718968196772, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8354978561401367, "lr": 0.0002, "elapsed_sec": 41823.562188625336, "step_time_sec": 8.237114390998613, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.160107612609863, "lr": 0.0002, "elapsed_sec": 41831.80040359497, "step_time_sec": 8.2380111249513, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.190998077392578, "lr": 0.0002, "elapsed_sec": 41840.03738474846, "step_time_sec": 8.236797620018478, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.136758327484131, "lr": 0.0002, "elapsed_sec": 41848.275364398956, "step_time_sec": 8.237836887012236, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9719488620758057, "lr": 0.0002, "elapsed_sec": 41856.512430906296, "step_time_sec": 8.236882124037948, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.08211088180542, "lr": 0.0002, "elapsed_sec": 41864.75044775009, "step_time_sec": 8.237865131988656, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.852506160736084, "lr": 0.0002, "elapsed_sec": 41872.98845362663, "step_time_sec": 8.237824841984548, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.063809871673584, "lr": 0.0002, "elapsed_sec": 41881.22589588165, "step_time_sec": 8.23733600002015, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.008259296417236, "lr": 0.0002, "elapsed_sec": 41889.46414351463, "step_time_sec": 8.23801369703142, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2123823165893555, "lr": 0.0002, "elapsed_sec": 41897.700632572174, "step_time_sec": 8.236373626976274, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.071891784667969, "lr": 0.0002, "elapsed_sec": 41905.93978428841, "step_time_sec": 8.238938983995467, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.044096946716309, "lr": 0.0002, "elapsed_sec": 41914.17914867401, "step_time_sec": 8.23918486398179, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.061112880706787, "lr": 0.0002, "elapsed_sec": 41922.41639351845, "step_time_sec": 8.237083873013034, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.115190029144287, "lr": 0.0002, "elapsed_sec": 41930.653923511505, "step_time_sec": 8.237350542971399, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.096264839172363, "lr": 0.0002, "elapsed_sec": 41938.8909971714, "step_time_sec": 8.236947951023467, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.037415981292725, "lr": 0.0002, "elapsed_sec": 41947.13056898117, "step_time_sec": 8.239370894967578, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.106069564819336, "lr": 0.0002, "elapsed_sec": 41955.369544267654, "step_time_sec": 8.238807321002241, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.125078201293945, "lr": 0.0002, "elapsed_sec": 41963.60830163956, "step_time_sec": 8.238674997002818, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1532464027404785, "lr": 0.0002, "elapsed_sec": 41971.84586548805, "step_time_sec": 8.237342342967167, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.035763740539551, "lr": 0.0002, "elapsed_sec": 41980.08246278763, "step_time_sec": 8.236427301017102, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.166826248168945, "lr": 0.0002, "elapsed_sec": 41988.32115530968, "step_time_sec": 8.238537109980825, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.010436534881592, "lr": 0.0002, "elapsed_sec": 41996.559495687485, "step_time_sec": 8.238169235002715, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.17208194732666, "lr": 0.0002, "elapsed_sec": 42004.79764223099, "step_time_sec": 8.23798658902524, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1222333908081055, "lr": 0.0002, "elapsed_sec": 42013.03547120094, "step_time_sec": 8.237598833977245, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.103516578674316, "lr": 0.0002, "elapsed_sec": 42021.272292375565, "step_time_sec": 8.236649056023452, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.08071231842041, "lr": 0.0002, "elapsed_sec": 42029.51098442078, "step_time_sec": 8.238524304004386, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9816994667053223, "lr": 0.0002, "elapsed_sec": 42037.74640035629, "step_time_sec": 8.235298052954022, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.936256170272827, "lr": 0.0002, "elapsed_sec": 42045.98443245888, "step_time_sec": 8.237867826013826, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.045525550842285, "lr": 0.0002, "elapsed_sec": 42054.22147512436, "step_time_sec": 8.236915693967603, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.101372241973877, "lr": 0.0002, "elapsed_sec": 42062.45951962471, "step_time_sec": 8.237850094039459, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.962266683578491, "lr": 0.0002, "elapsed_sec": 42070.6979868412, "step_time_sec": 8.238281760015525, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.008145809173584, "lr": 0.0002, "elapsed_sec": 42078.935963630676, "step_time_sec": 8.237813673971687, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9444222450256348, "lr": 0.0002, "elapsed_sec": 42087.175612688065, "step_time_sec": 8.239489700004924, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.070284843444824, "lr": 0.0002, "elapsed_sec": 42095.41332554817, "step_time_sec": 8.237536531989463, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.090407848358154, "lr": 0.0002, "elapsed_sec": 42103.651908397675, "step_time_sec": 8.238494221994188, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.105003833770752, "lr": 0.0002, "elapsed_sec": 42111.88958239555, "step_time_sec": 8.23745015397435, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.972513437271118, "lr": 0.0002, "elapsed_sec": 42120.12738585472, "step_time_sec": 8.237665874999948, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8856558799743652, "lr": 0.0002, "elapsed_sec": 42128.36573553085, "step_time_sec": 8.238193127966952, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.106051445007324, "lr": 0.0002, "elapsed_sec": 42136.6022298336, "step_time_sec": 8.236306651961058, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.059381008148193, "lr": 0.0002, "elapsed_sec": 42144.8398911953, "step_time_sec": 8.237506607023533, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.212306022644043, "lr": 0.0002, "elapsed_sec": 42153.0766620636, "step_time_sec": 8.236635635024868, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.064215660095215, "lr": 0.0002, "elapsed_sec": 42161.31485080719, "step_time_sec": 8.237986949039623, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0509538650512695, "lr": 0.0002, "elapsed_sec": 42169.55338358879, "step_time_sec": 8.238355827983469, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0511956214904785, "lr": 0.0002, "elapsed_sec": 42177.79045510292, "step_time_sec": 8.236948530946393, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.05798864364624, "lr": 0.0002, "elapsed_sec": 42186.02869105339, "step_time_sec": 8.238059399009217, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.984066963195801, "lr": 0.0002, "elapsed_sec": 42194.26564121246, "step_time_sec": 8.236753199016675, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.075218677520752, "lr": 0.0002, "elapsed_sec": 42202.50460934639, "step_time_sec": 8.238797833037097, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1125807762146, "lr": 0.0002, "elapsed_sec": 42210.74295258522, "step_time_sec": 8.238173844991252, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8917832374572754, "lr": 0.0002, "elapsed_sec": 42218.980336904526, "step_time_sec": 8.23726786498446, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.175902843475342, "lr": 0.0002, "elapsed_sec": 42227.21846675873, "step_time_sec": 8.237955082033295, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1498799324035645, "lr": 0.0002, "elapsed_sec": 42235.45525741577, "step_time_sec": 8.23662742599845, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.147986888885498, "lr": 0.0002, "elapsed_sec": 42243.69453239441, "step_time_sec": 8.239118708996102, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.131357669830322, "lr": 0.0002, "elapsed_sec": 42251.93264532089, "step_time_sec": 8.237889993994031, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0092644691467285, "lr": 0.0002, "elapsed_sec": 42260.17042469978, "step_time_sec": 8.237635180004872, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.016315460205078, "lr": 0.0002, "elapsed_sec": 42268.40780568123, "step_time_sec": 8.237213569984306, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.232234954833984, "lr": 0.0002, "elapsed_sec": 42276.64512181282, "step_time_sec": 8.237109994050115, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.071267127990723, "lr": 0.0002, "elapsed_sec": 42284.88268041611, "step_time_sec": 8.237409550056327, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0661516189575195, "lr": 0.0002, "elapsed_sec": 42293.12060332298, "step_time_sec": 8.237779723014683, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.065340042114258, "lr": 0.0002, "elapsed_sec": 42301.359307050705, "step_time_sec": 8.238514172961004, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.137168884277344, "lr": 0.0002, "elapsed_sec": 42309.59731507301, "step_time_sec": 8.23783412197372, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.110866546630859, "lr": 0.0002, "elapsed_sec": 42317.8348865509, "step_time_sec": 8.237442115030717, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.121829509735107, "lr": 0.0002, "elapsed_sec": 42326.07235455513, "step_time_sec": 8.237318033003248, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.857813596725464, "lr": 0.0002, "elapsed_sec": 42334.3084628582, "step_time_sec": 8.235890308977105, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.031270980834961, "lr": 0.0002, "elapsed_sec": 42342.54430294037, "step_time_sec": 8.235663849976845, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.079507350921631, "lr": 0.0002, "elapsed_sec": 42350.78196787834, "step_time_sec": 8.237510234001093, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1247944831848145, "lr": 0.0002, "elapsed_sec": 42359.02185988426, "step_time_sec": 8.239776349975727, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0819010734558105, "lr": 0.0002, "elapsed_sec": 42367.25955581665, "step_time_sec": 8.237488735001534, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9579222202301025, "lr": 0.0002, "elapsed_sec": 42375.496923685074, "step_time_sec": 8.237213123007677, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.027462482452393, "lr": 0.0002, "elapsed_sec": 42383.73553991318, "step_time_sec": 8.238443802983966, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0405755043029785, "lr": 0.0002, "elapsed_sec": 42391.97111868858, "step_time_sec": 8.235382237995509, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9788150787353516, "lr": 0.0002, "elapsed_sec": 42400.20915269852, "step_time_sec": 8.237928700982593, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.088127136230469, "lr": 0.0002, "elapsed_sec": 42408.44637346268, "step_time_sec": 8.237036521022674, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1257171630859375, "lr": 0.0002, "elapsed_sec": 42416.68425655365, "step_time_sec": 8.237674996023998, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.017090320587158, "lr": 0.0002, "elapsed_sec": 42424.92164158821, "step_time_sec": 8.237185525009409, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.056936264038086, "lr": 0.0002, "elapsed_sec": 42433.158301353455, "step_time_sec": 8.236523988016415, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.142986297607422, "lr": 0.0002, "elapsed_sec": 42441.39617395401, "step_time_sec": 8.237672101007774, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.068072319030762, "lr": 0.0002, "elapsed_sec": 42449.63309645653, "step_time_sec": 8.236755956022535, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.336813926696777, "lr": 0.0002, "elapsed_sec": 42457.870384931564, "step_time_sec": 8.23714411596302, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.030776500701904, "lr": 0.0002, "elapsed_sec": 42466.107330560684, "step_time_sec": 8.236802378029097, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9316253662109375, "lr": 0.0002, "elapsed_sec": 42474.34471106529, "step_time_sec": 8.237215842003934, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.216373443603516, "lr": 0.0002, "elapsed_sec": 42482.582976818085, "step_time_sec": 8.23805770900799, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.084118843078613, "lr": 0.0002, "elapsed_sec": 42490.82026600838, "step_time_sec": 8.23711457598256, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.028067111968994, "lr": 0.0002, "elapsed_sec": 42499.059047698975, "step_time_sec": 8.238660762028303, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.032754421234131, "lr": 0.0002, "elapsed_sec": 42507.29507255554, "step_time_sec": 8.23584556701826, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.084836483001709, "lr": 0.0002, "elapsed_sec": 42515.53324365616, "step_time_sec": 8.23799916903954, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.200761318206787, "lr": 0.0002, "elapsed_sec": 42523.770335674286, "step_time_sec": 8.236982249014545, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.170886039733887, "lr": 0.0002, "elapsed_sec": 42532.00879430771, "step_time_sec": 8.238278857956175, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9103548526763916, "lr": 0.0002, "elapsed_sec": 42540.24909758568, "step_time_sec": 8.240096291003283, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.971414566040039, "lr": 0.0002, "elapsed_sec": 42548.487651109695, "step_time_sec": 8.238421993970405, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.190511703491211, "lr": 0.0002, "elapsed_sec": 42556.72619509697, "step_time_sec": 8.238345430989284, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.140244483947754, "lr": 0.0002, "elapsed_sec": 42564.96175599098, "step_time_sec": 8.235386073996779, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9853053092956543, "lr": 0.0002, "elapsed_sec": 42573.19934821129, "step_time_sec": 8.23746910499176, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.032797336578369, "lr": 0.0002, "elapsed_sec": 42581.437643289566, "step_time_sec": 8.238103127980139, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9666309356689453, "lr": 0.0002, "elapsed_sec": 42589.67417168617, "step_time_sec": 8.236365083954297, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.143857479095459, "lr": 0.0002, "elapsed_sec": 42597.91389346123, "step_time_sec": 8.239554119005334, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.278141021728516, "lr": 0.0002, "elapsed_sec": 42606.15192246437, "step_time_sec": 8.237899116007611, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9805893898010254, "lr": 0.0002, "elapsed_sec": 42614.391973018646, "step_time_sec": 8.239843143965118, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9818153381347656, "lr": 0.0002, "elapsed_sec": 42622.62992858887, "step_time_sec": 8.237810552003793, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.056858539581299, "lr": 0.0002, "elapsed_sec": 42630.86716580391, "step_time_sec": 8.237075995013583, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.142970085144043, "lr": 0.0002, "elapsed_sec": 42639.104652404785, "step_time_sec": 8.237337831989862, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.05332088470459, "lr": 0.0002, "elapsed_sec": 42647.34185528755, "step_time_sec": 8.237014668993652, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.092161178588867, "lr": 0.0002, "elapsed_sec": 42655.580843925476, "step_time_sec": 8.238825251988601, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.199728012084961, "lr": 0.0002, "elapsed_sec": 42663.81928396225, "step_time_sec": 8.238330136053264, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.077695369720459, "lr": 0.0002, "elapsed_sec": 42672.058446884155, "step_time_sec": 8.238938271009829, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.165023326873779, "lr": 0.0002, "elapsed_sec": 42680.29734611511, "step_time_sec": 8.238756295002531, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.091210842132568, "lr": 0.0002, "elapsed_sec": 42688.53377819061, "step_time_sec": 8.236312977038324, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0889363288879395, "lr": 0.0002, "elapsed_sec": 42696.772334098816, "step_time_sec": 8.238373517990112, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.21475887298584, "lr": 0.0002, "elapsed_sec": 42705.00900673866, "step_time_sec": 8.23654135398101, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.123498916625977, "lr": 0.0002, "elapsed_sec": 42713.24784064293, "step_time_sec": 8.238675768021494, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.914275646209717, "lr": 0.0002, "elapsed_sec": 42721.48701071739, "step_time_sec": 8.238971551007126, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.050546646118164, "lr": 0.0002, "elapsed_sec": 42729.7246735096, "step_time_sec": 8.237509967002552, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0890631675720215, "lr": 0.0002, "elapsed_sec": 42737.96349644661, "step_time_sec": 8.238647257036064, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0077314376831055, "lr": 0.0002, "elapsed_sec": 42746.19989466667, "step_time_sec": 8.236250861024018, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.226086616516113, "lr": 0.0002, "elapsed_sec": 42754.438396692276, "step_time_sec": 8.238382340001408, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.158082008361816, "lr": 0.0002, "elapsed_sec": 42762.674847602844, "step_time_sec": 8.236290594970342, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.364383220672607, "lr": 0.0002, "elapsed_sec": 42770.9128305912, "step_time_sec": 8.23778794996906, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.910083770751953, "lr": 0.0002, "elapsed_sec": 42779.151851177216, "step_time_sec": 8.238878619973548, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.070162296295166, "lr": 0.0002, "elapsed_sec": 42787.388963222504, "step_time_sec": 8.236925436998717, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9891276359558105, "lr": 0.0002, "elapsed_sec": 42795.6277115345, "step_time_sec": 8.238587675965391, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.142857074737549, "lr": 0.0002, "elapsed_sec": 42803.86537218094, "step_time_sec": 8.237540611997247, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.191635608673096, "lr": 0.0002, "elapsed_sec": 42812.10329389572, "step_time_sec": 8.237716895004269, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.262484550476074, "lr": 0.0002, "elapsed_sec": 42820.34270787239, "step_time_sec": 8.239219578972552, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.125124931335449, "lr": 0.0002, "elapsed_sec": 42828.58006429672, "step_time_sec": 8.237260224996135, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.300273895263672, "lr": 0.0002, "elapsed_sec": 42836.81680774689, "step_time_sec": 8.236568221997004, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.231332778930664, "lr": 0.0002, "elapsed_sec": 42845.053574085236, "step_time_sec": 8.236643046024255, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9038593769073486, "lr": 0.0002, "elapsed_sec": 42853.293298482895, "step_time_sec": 8.239523598051164, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.070638656616211, "lr": 0.0002, "elapsed_sec": 42861.53172111511, "step_time_sec": 8.238220536033623, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.153090000152588, "lr": 0.0002, "elapsed_sec": 42869.77017426491, "step_time_sec": 8.238285970990546, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.104682445526123, "lr": 0.0002, "elapsed_sec": 42878.00861644745, "step_time_sec": 8.238262580009177, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.011694431304932, "lr": 0.0002, "elapsed_sec": 42886.24495911598, "step_time_sec": 8.236202596977819, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0933942794799805, "lr": 0.0002, "elapsed_sec": 42894.483300447464, "step_time_sec": 8.238218540034723, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.013219356536865, "lr": 0.0002, "elapsed_sec": 42902.72138476372, "step_time_sec": 8.237891485972796, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.087450981140137, "lr": 0.0002, "elapsed_sec": 42910.959008932114, "step_time_sec": 8.237505779019557, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.178984642028809, "lr": 0.0002, "elapsed_sec": 42919.19770479202, "step_time_sec": 8.238502374035306, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.009154319763184, "lr": 0.0002, "elapsed_sec": 42927.435526371, "step_time_sec": 8.237638862978201, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.934835433959961, "lr": 0.0002, "elapsed_sec": 42935.673966646194, "step_time_sec": 8.238326700986363, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.219262599945068, "lr": 0.0002, "elapsed_sec": 42943.91194319725, "step_time_sec": 8.237800811999477, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.018511772155762, "lr": 0.0002, "elapsed_sec": 42952.151366472244, "step_time_sec": 8.239234421984293, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9703621864318848, "lr": 0.0002, "elapsed_sec": 42960.3900179863, "step_time_sec": 8.23846121202223, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.052854537963867, "lr": 0.0002, "elapsed_sec": 42968.62654519081, "step_time_sec": 8.23639924597228, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.026381969451904, "lr": 0.0002, "elapsed_sec": 42976.86489701271, "step_time_sec": 8.238195337995421, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.118804931640625, "lr": 0.0002, "elapsed_sec": 42985.10099339485, "step_time_sec": 8.235887535964139, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.979478359222412, "lr": 0.0002, "elapsed_sec": 42993.33910250664, "step_time_sec": 8.23793783702422, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.101484775543213, "lr": 0.0002, "elapsed_sec": 43001.576031923294, "step_time_sec": 8.236769096984062, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.992500066757202, "lr": 0.0002, "elapsed_sec": 43009.81535220146, "step_time_sec": 8.239158260985278, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.232422351837158, "lr": 0.0002, "elapsed_sec": 43018.05372142792, "step_time_sec": 8.238241905986797, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.113676071166992, "lr": 0.0002, "elapsed_sec": 43026.29048037529, "step_time_sec": 8.236595483962446, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1914753913879395, "lr": 0.0002, "elapsed_sec": 43034.529413700104, "step_time_sec": 8.23874787800014, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.220984935760498, "lr": 0.0002, "elapsed_sec": 43042.76557254791, "step_time_sec": 8.236008676991332, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.205135345458984, "lr": 0.0002, "elapsed_sec": 43051.00398159027, "step_time_sec": 8.238270749978255, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9974193572998047, "lr": 0.0002, "elapsed_sec": 43059.24151968956, "step_time_sec": 8.237380528997164, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.24017858505249, "lr": 0.0002, "elapsed_sec": 43067.479833841324, "step_time_sec": 8.238137628999539, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.246116638183594, "lr": 0.0002, "elapsed_sec": 43075.718665122986, "step_time_sec": 8.238629059051163, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.103309631347656, "lr": 0.0002, "elapsed_sec": 43083.95654273033, "step_time_sec": 8.237746026017703, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.073929786682129, "lr": 0.0002, "elapsed_sec": 43092.19470787048, "step_time_sec": 8.238001190999057, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.130112648010254, "lr": 0.0002, "elapsed_sec": 43100.433081150055, "step_time_sec": 8.238151592027862, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.172706604003906, "lr": 0.0002, "elapsed_sec": 43108.671241760254, "step_time_sec": 8.238083921023645, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.163802623748779, "lr": 0.0002, "elapsed_sec": 43116.90938234329, "step_time_sec": 8.237926598987542, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.076635837554932, "lr": 0.0002, "elapsed_sec": 43125.14560508728, "step_time_sec": 8.236076792993117, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.118464946746826, "lr": 0.0002, "elapsed_sec": 43133.384677410126, "step_time_sec": 8.238879347976763, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.097728729248047, "lr": 0.0002, "elapsed_sec": 43141.62173867226, "step_time_sec": 8.236880477983505, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0666823387146, "lr": 0.0002, "elapsed_sec": 43149.85926246643, "step_time_sec": 8.2373678679578, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.222337245941162, "lr": 0.0002, "elapsed_sec": 43158.096824884415, "step_time_sec": 8.237405430991203, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.039463520050049, "lr": 0.0002, "elapsed_sec": 43166.33498120308, "step_time_sec": 8.238028212042991, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.280864238739014, "lr": 0.0002, "elapsed_sec": 43174.573472738266, "step_time_sec": 8.238312262052204, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.134127140045166, "lr": 0.0002, "elapsed_sec": 43182.810574769974, "step_time_sec": 8.236921845003963, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1995930671691895, "lr": 0.0002, "elapsed_sec": 43191.04975104332, "step_time_sec": 8.239025354967453, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.059616565704346, "lr": 0.0002, "elapsed_sec": 43199.28765654564, "step_time_sec": 8.237742569996044, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.089636325836182, "lr": 0.0002, "elapsed_sec": 43207.524668216705, "step_time_sec": 8.236838290002197, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1224846839904785, "lr": 0.0002, "elapsed_sec": 43215.76322174072, "step_time_sec": 8.238348619954195, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.075918197631836, "lr": 0.0002, "elapsed_sec": 43224.00002169609, "step_time_sec": 8.236681277980097, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.028504371643066, "lr": 0.0002, "elapsed_sec": 43232.237830877304, "step_time_sec": 8.237615472986363, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.086059093475342, "lr": 0.0002, "elapsed_sec": 43240.47529196739, "step_time_sec": 8.237339219020214, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.102231025695801, "lr": 0.0002, "elapsed_sec": 43248.71344971657, "step_time_sec": 8.237991677015088, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.081364154815674, "lr": 0.0002, "elapsed_sec": 43256.95161294937, "step_time_sec": 8.237961071019527, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.115140914916992, "lr": 0.0002, "elapsed_sec": 43265.18992090225, "step_time_sec": 8.238165015995037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.087908744812012, "lr": 0.0002, "elapsed_sec": 43273.428226709366, "step_time_sec": 8.238157133979257, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.972510576248169, "lr": 0.0002, "elapsed_sec": 43281.66425085068, "step_time_sec": 8.235812018043362, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1700334548950195, "lr": 0.0002, "elapsed_sec": 43289.90160870552, "step_time_sec": 8.237196648027748, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.033212184906006, "lr": 0.0002, "elapsed_sec": 43298.13823008537, "step_time_sec": 8.236477165017277, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.182351112365723, "lr": 0.0002, "elapsed_sec": 43306.376198768616, "step_time_sec": 8.237817676970735, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.037895202636719, "lr": 0.0002, "elapsed_sec": 43314.614399671555, "step_time_sec": 8.23800489702262, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.181305408477783, "lr": 0.0002, "elapsed_sec": 43322.85341024399, "step_time_sec": 8.238867014995776, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.101391315460205, "lr": 0.0002, "elapsed_sec": 43331.09209871292, "step_time_sec": 8.238544889958575, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.023227214813232, "lr": 0.0002, "elapsed_sec": 43339.32800102234, "step_time_sec": 8.235690701985732, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.018639087677002, "lr": 0.0002, "elapsed_sec": 43347.56695175171, "step_time_sec": 8.238781049032696, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.022938251495361, "lr": 0.0002, "elapsed_sec": 43355.80407357216, "step_time_sec": 8.236964444979094, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.131568431854248, "lr": 0.0002, "elapsed_sec": 43364.04236578941, "step_time_sec": 8.238193959987257, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.037928581237793, "lr": 0.0002, "elapsed_sec": 43372.28062033653, "step_time_sec": 8.238047519000247, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.126255989074707, "lr": 0.0002, "elapsed_sec": 43380.51919269562, "step_time_sec": 8.238389307982288, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.114139556884766, "lr": 0.0002, "elapsed_sec": 43388.75773382187, "step_time_sec": 8.238419352972414, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.965965509414673, "lr": 0.0002, "elapsed_sec": 43396.99401140213, "step_time_sec": 8.236119908979163, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.093507766723633, "lr": 0.0002, "elapsed_sec": 43405.232323646545, "step_time_sec": 8.238133476988878, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3349127769470215, "lr": 0.0002, "elapsed_sec": 43413.47014403343, "step_time_sec": 8.237645264016464, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1876397132873535, "lr": 0.0002, "elapsed_sec": 43421.70635843277, "step_time_sec": 8.236063944001216, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.079740524291992, "lr": 0.0002, "elapsed_sec": 43429.9437456131, "step_time_sec": 8.23719901801087, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.065438270568848, "lr": 0.0002, "elapsed_sec": 43438.1817843914, "step_time_sec": 8.237913212971762, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.248654365539551, "lr": 0.0002, "elapsed_sec": 43446.41900086403, "step_time_sec": 8.237049918039702, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.137764930725098, "lr": 0.0002, "elapsed_sec": 43454.65667510033, "step_time_sec": 8.237455533002503, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.003637313842773, "lr": 0.0002, "elapsed_sec": 43462.89419770241, "step_time_sec": 8.237385823042132, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.147636890411377, "lr": 0.0002, "elapsed_sec": 43471.131437540054, "step_time_sec": 8.237078532984015, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.051130771636963, "lr": 0.0002, "elapsed_sec": 43479.369014024734, "step_time_sec": 8.237391175993253, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0666184425354, "lr": 0.0002, "elapsed_sec": 43487.60694646835, "step_time_sec": 8.237774309993256, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9247379302978516, "lr": 0.0002, "elapsed_sec": 43495.84447526932, "step_time_sec": 8.237359357008245, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.086363792419434, "lr": 0.0002, "elapsed_sec": 43504.08311295509, "step_time_sec": 8.238518223981373, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.086656093597412, "lr": 0.0002, "elapsed_sec": 43512.32037305832, "step_time_sec": 8.237055291014258, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.131547451019287, "lr": 0.0002, "elapsed_sec": 43520.559215545654, "step_time_sec": 8.23864416097058, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.125100135803223, "lr": 0.0002, "elapsed_sec": 43528.79653787613, "step_time_sec": 8.237179982010275, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.100745677947998, "lr": 0.0002, "elapsed_sec": 43537.03309178352, "step_time_sec": 8.236401671019848, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.147770404815674, "lr": 0.0002, "elapsed_sec": 43545.272747039795, "step_time_sec": 8.239462047989946, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.112496852874756, "lr": 0.0002, "elapsed_sec": 43553.50936841965, "step_time_sec": 8.236434767022729, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.181028366088867, "lr": 0.0002, "elapsed_sec": 43561.748470783234, "step_time_sec": 8.238922032003757, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0623345375061035, "lr": 0.0002, "elapsed_sec": 43569.984896183014, "step_time_sec": 8.236290134023875, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.097845554351807, "lr": 0.0002, "elapsed_sec": 43578.22232794762, "step_time_sec": 8.237280401983298, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9648640155792236, "lr": 0.0002, "elapsed_sec": 43586.46124076843, "step_time_sec": 8.238791761978064, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.036179065704346, "lr": 0.0002, "elapsed_sec": 43594.69791007042, "step_time_sec": 8.236456333019305, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0126051902771, "lr": 0.0002, "elapsed_sec": 43602.93674302101, "step_time_sec": 8.23865213303361, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.165648460388184, "lr": 0.0002, "elapsed_sec": 43611.17447423935, "step_time_sec": 8.237582864996511, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.169041633605957, "lr": 0.0002, "elapsed_sec": 43619.413133621216, "step_time_sec": 8.23848832497606, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0370988845825195, "lr": 0.0002, "elapsed_sec": 43627.651270627975, "step_time_sec": 8.237969039008021, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.14927864074707, "lr": 0.0002, "elapsed_sec": 43635.887567043304, "step_time_sec": 8.236119361012243, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.038569927215576, "lr": 0.0002, "elapsed_sec": 43644.12584066391, "step_time_sec": 8.238163197005633, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0313401222229, "lr": 0.0002, "elapsed_sec": 43652.363518714905, "step_time_sec": 8.237498758011498, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.085696220397949, "lr": 0.0002, "elapsed_sec": 43660.60250735283, "step_time_sec": 8.238800102029927, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.131263732910156, "lr": 0.0002, "elapsed_sec": 43668.840300798416, "step_time_sec": 8.23767207597848, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.080154895782471, "lr": 0.0002, "elapsed_sec": 43677.079577207565, "step_time_sec": 8.23910413502017, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.019993782043457, "lr": 0.0002, "elapsed_sec": 43685.317808151245, "step_time_sec": 8.238039702991955, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.146896839141846, "lr": 0.0002, "elapsed_sec": 43693.55420207977, "step_time_sec": 8.236241095000878, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9851720333099365, "lr": 0.0002, "elapsed_sec": 43701.79269480705, "step_time_sec": 8.238289866014384, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.052669048309326, "lr": 0.0002, "elapsed_sec": 43710.0312268734, "step_time_sec": 8.238366888021119, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1299967765808105, "lr": 0.0002, "elapsed_sec": 43718.26922392845, "step_time_sec": 8.237852356978692, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.035645484924316, "lr": 0.0002, "elapsed_sec": 43726.508138895035, "step_time_sec": 8.238742435001768, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.130650997161865, "lr": 0.0002, "elapsed_sec": 43734.745671749115, "step_time_sec": 8.237398575991392, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.109706401824951, "lr": 0.0002, "elapsed_sec": 43742.98395752907, "step_time_sec": 8.238089923979715, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0962910652160645, "lr": 0.0002, "elapsed_sec": 43751.22144317627, "step_time_sec": 8.237349481030833, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.138253688812256, "lr": 0.0002, "elapsed_sec": 43759.4604408741, "step_time_sec": 8.238761844986584, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.052216529846191, "lr": 0.0002, "elapsed_sec": 43767.699316978455, "step_time_sec": 8.23872337298235, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.082056522369385, "lr": 0.0002, "elapsed_sec": 43775.93598937988, "step_time_sec": 8.236505518027116, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0397772789001465, "lr": 0.0002, "elapsed_sec": 43784.174678087234, "step_time_sec": 8.238560717029031, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.974010944366455, "lr": 0.0002, "elapsed_sec": 43792.41132116318, "step_time_sec": 8.236459696956445, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9510750770568848, "lr": 0.0002, "elapsed_sec": 43800.649465084076, "step_time_sec": 8.237971336988267, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9932830333709717, "lr": 0.0002, "elapsed_sec": 43808.886070251465, "step_time_sec": 8.236489263013937, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.026957035064697, "lr": 0.0002, "elapsed_sec": 43817.124223947525, "step_time_sec": 8.237921431020368, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.942251443862915, "lr": 0.0002, "elapsed_sec": 43825.36155796051, "step_time_sec": 8.237179482995998, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.213651657104492, "lr": 0.0002, "elapsed_sec": 43833.5974214077, "step_time_sec": 8.235700576973613, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9941012859344482, "lr": 0.0002, "elapsed_sec": 43841.834891080856, "step_time_sec": 8.2373491859762, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.180769920349121, "lr": 0.0002, "elapsed_sec": 43850.071454524994, "step_time_sec": 8.23634140699869, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.210861682891846, "lr": 0.0002, "elapsed_sec": 43858.31123614311, "step_time_sec": 8.239635668986011, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.05915641784668, "lr": 0.0002, "elapsed_sec": 43866.54836392403, "step_time_sec": 8.23699259001296, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.091561794281006, "lr": 0.0002, "elapsed_sec": 43874.786256074905, "step_time_sec": 8.237697039963678, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9644968509674072, "lr": 0.0002, "elapsed_sec": 43883.024456739426, "step_time_sec": 8.238009978027549, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.186977863311768, "lr": 0.0002, "elapsed_sec": 43891.26062488556, "step_time_sec": 8.236022649973165, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.114277362823486, "lr": 0.0002, "elapsed_sec": 43899.49964737892, "step_time_sec": 8.238843900966458, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9427075386047363, "lr": 0.0002, "elapsed_sec": 43907.73779511452, "step_time_sec": 8.238005555991549, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.016139984130859, "lr": 0.0002, "elapsed_sec": 43915.97650003433, "step_time_sec": 8.23856217903085, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.08778190612793, "lr": 0.0002, "elapsed_sec": 43924.215457201004, "step_time_sec": 8.238780094950926, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.13686466217041, "lr": 0.0002, "elapsed_sec": 43932.45233988762, "step_time_sec": 8.236712355981581, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.031654357910156, "lr": 0.0002, "elapsed_sec": 43940.6903321743, "step_time_sec": 8.237806877994444, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.228846549987793, "lr": 0.0002, "elapsed_sec": 43948.92778611183, "step_time_sec": 8.237319239997305, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.979717493057251, "lr": 0.0002, "elapsed_sec": 43957.166002988815, "step_time_sec": 8.238028649997432, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.166132926940918, "lr": 0.0002, "elapsed_sec": 43965.40461587906, "step_time_sec": 8.238460208987817, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9652187824249268, "lr": 0.0002, "elapsed_sec": 43973.641659498215, "step_time_sec": 8.236904718040023, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.121528625488281, "lr": 0.0002, "elapsed_sec": 43981.881506204605, "step_time_sec": 8.239693526993506, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.149382591247559, "lr": 0.0002, "elapsed_sec": 43990.1178252697, "step_time_sec": 8.23610225203447, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.025160789489746, "lr": 0.0002, "elapsed_sec": 43998.35492658615, "step_time_sec": 8.23695999395568, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.068941593170166, "lr": 0.0002, "elapsed_sec": 44006.592242479324, "step_time_sec": 8.237186743994243, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.125993251800537, "lr": 0.0002, "elapsed_sec": 44014.830785512924, "step_time_sec": 8.238336073001847, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1675286293029785, "lr": 0.0002, "elapsed_sec": 44023.068821430206, "step_time_sec": 8.237856417021248, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.083128929138184, "lr": 0.0002, "elapsed_sec": 44031.30531311035, "step_time_sec": 8.236343366967048, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0622172355651855, "lr": 0.0002, "elapsed_sec": 44039.5427839756, "step_time_sec": 8.237335390993394, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.184752464294434, "lr": 0.0002, "elapsed_sec": 44047.77882242203, "step_time_sec": 8.235843928006943, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.207063674926758, "lr": 0.0002, "elapsed_sec": 44056.01613378525, "step_time_sec": 8.237136518000625, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.253960609436035, "lr": 0.0002, "elapsed_sec": 44064.25360965729, "step_time_sec": 8.237324765010271, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.223635196685791, "lr": 0.0002, "elapsed_sec": 44072.49250483513, "step_time_sec": 8.238707159005571, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.089238166809082, "lr": 0.0002, "elapsed_sec": 44080.729815006256, "step_time_sec": 8.237146072962787, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.038427829742432, "lr": 0.0002, "elapsed_sec": 44088.96742725372, "step_time_sec": 8.237474850029685, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.103794097900391, "lr": 0.0002, "elapsed_sec": 44097.20637965202, "step_time_sec": 8.238784013024997, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.412179470062256, "lr": 0.0002, "elapsed_sec": 44105.44369721413, "step_time_sec": 8.237160021031741, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.194885730743408, "lr": 0.0002, "elapsed_sec": 44113.681955099106, "step_time_sec": 8.238156404986512, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9242918491363525, "lr": 0.0002, "elapsed_sec": 44121.919674396515, "step_time_sec": 8.237508616002742, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1856303215026855, "lr": 0.0002, "elapsed_sec": 44130.15702056885, "step_time_sec": 8.237185493984725, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.191248416900635, "lr": 0.0002, "elapsed_sec": 44138.39535880089, "step_time_sec": 8.238204699999187, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.029068470001221, "lr": 0.0002, "elapsed_sec": 44146.6324326992, "step_time_sec": 8.236874954949599, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.122981548309326, "lr": 0.0002, "elapsed_sec": 44154.87232398987, "step_time_sec": 8.239721936988644, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.145902156829834, "lr": 0.0002, "elapsed_sec": 44163.10961937904, "step_time_sec": 8.23714069504058, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.036309719085693, "lr": 0.0002, "elapsed_sec": 44171.34828090668, "step_time_sec": 8.238530697009992, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0053277015686035, "lr": 0.0002, "elapsed_sec": 44179.58699607849, "step_time_sec": 8.238539649988525, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.119058132171631, "lr": 0.0002, "elapsed_sec": 44187.82327699661, "step_time_sec": 8.236061992007308, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.048933506011963, "lr": 0.0002, "elapsed_sec": 44196.06143951416, "step_time_sec": 8.238025529019069, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.184764385223389, "lr": 0.0002, "elapsed_sec": 44204.29744195938, "step_time_sec": 8.235873100988101, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1316609382629395, "lr": 0.0002, "elapsed_sec": 44212.53598666191, "step_time_sec": 8.238330420979764, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9340522289276123, "lr": 0.0002, "elapsed_sec": 44220.77422475815, "step_time_sec": 8.238096337998286, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.18238639831543, "lr": 0.0002, "elapsed_sec": 44229.01171898842, "step_time_sec": 8.23728428396862, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.094729423522949, "lr": 0.0002, "elapsed_sec": 44237.24986124039, "step_time_sec": 8.238028952036984, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.102606296539307, "lr": 0.0002, "elapsed_sec": 44245.48700761795, "step_time_sec": 8.236928539001383, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.115172863006592, "lr": 0.0002, "elapsed_sec": 44253.7254383564, "step_time_sec": 8.238329259038437, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0387349128723145, "lr": 0.0002, "elapsed_sec": 44261.96316552162, "step_time_sec": 8.237532036029734, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9410316944122314, "lr": 0.0002, "elapsed_sec": 44270.20145368576, "step_time_sec": 8.238177054969128, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1001386642456055, "lr": 0.0002, "elapsed_sec": 44278.44080400467, "step_time_sec": 8.239148446009494, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.039514064788818, "lr": 0.0002, "elapsed_sec": 44286.678470134735, "step_time_sec": 8.237553703016602, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.235373497009277, "lr": 0.0002, "elapsed_sec": 44294.91713643074, "step_time_sec": 8.238424400042277, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.895435333251953, "lr": 0.0002, "elapsed_sec": 44303.15448260307, "step_time_sec": 8.237176846014336, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.119272232055664, "lr": 0.0002, "elapsed_sec": 44311.3944003582, "step_time_sec": 8.239741383004002, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0051727294921875, "lr": 0.0002, "elapsed_sec": 44319.63368058205, "step_time_sec": 8.239155127957929, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1700758934021, "lr": 0.0002, "elapsed_sec": 44327.87128806114, "step_time_sec": 8.237408599001355, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.091407775878906, "lr": 0.0002, "elapsed_sec": 44336.111040353775, "step_time_sec": 8.239583368995227, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.066498279571533, "lr": 0.0002, "elapsed_sec": 44344.350183963776, "step_time_sec": 8.238966794975568, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.030264377593994, "lr": 0.0002, "elapsed_sec": 44352.58984899521, "step_time_sec": 8.239493185013998, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.157924652099609, "lr": 0.0002, "elapsed_sec": 44360.82920742035, "step_time_sec": 8.239200456009712, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.044251918792725, "lr": 0.0002, "elapsed_sec": 44369.06621026993, "step_time_sec": 8.236842876009177, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.086382865905762, "lr": 0.0002, "elapsed_sec": 44377.3059220314, "step_time_sec": 8.239575810031965, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8586628437042236, "lr": 0.0002, "elapsed_sec": 44385.54355096817, "step_time_sec": 8.237426478997804, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.006214141845703, "lr": 0.0002, "elapsed_sec": 44393.781383275986, "step_time_sec": 8.237725329003297, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.055958271026611, "lr": 0.0002, "elapsed_sec": 44402.020136356354, "step_time_sec": 8.238551532966085, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.093395233154297, "lr": 0.0002, "elapsed_sec": 44410.25790143013, "step_time_sec": 8.237594424979761, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.082529067993164, "lr": 0.0002, "elapsed_sec": 44418.496039152145, "step_time_sec": 8.237942548003048, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0254597663879395, "lr": 0.0002, "elapsed_sec": 44426.73460698128, "step_time_sec": 8.238373989006504, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.055071830749512, "lr": 0.0002, "elapsed_sec": 44434.97403717041, "step_time_sec": 8.23931440198794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.990588903427124, "lr": 0.0002, "elapsed_sec": 44443.21110153198, "step_time_sec": 8.236881052027456, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.904327869415283, "lr": 0.0002, "elapsed_sec": 44451.44944739342, "step_time_sec": 8.238212659955025, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.984790802001953, "lr": 0.0002, "elapsed_sec": 44459.68795108795, "step_time_sec": 8.238337614980992, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.016822814941406, "lr": 0.0002, "elapsed_sec": 44467.924988269806, "step_time_sec": 8.236864315986168, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.037398338317871, "lr": 0.0002, "elapsed_sec": 44476.163733005524, "step_time_sec": 8.238603504025377, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9760003089904785, "lr": 0.0002, "elapsed_sec": 44484.401288986206, "step_time_sec": 8.237339874031022, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.076251983642578, "lr": 0.0002, "elapsed_sec": 44492.63987517357, "step_time_sec": 8.238418173976243, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.125396251678467, "lr": 0.0002, "elapsed_sec": 44500.87893152237, "step_time_sec": 8.238914690969978, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.125175952911377, "lr": 0.0002, "elapsed_sec": 44509.11691355705, "step_time_sec": 8.237841750029474, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.956345796585083, "lr": 0.0002, "elapsed_sec": 44517.356464624405, "step_time_sec": 8.239435530966148, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.049799919128418, "lr": 0.0002, "elapsed_sec": 44525.594544410706, "step_time_sec": 8.237876890983898, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8051018714904785, "lr": 0.0002, "elapsed_sec": 44533.83344578743, "step_time_sec": 8.238775977981277, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.069907188415527, "lr": 0.0002, "elapsed_sec": 44542.07306313515, "step_time_sec": 8.239407963992562, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.137475490570068, "lr": 0.0002, "elapsed_sec": 44550.31015229225, "step_time_sec": 8.236924487981014, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.084944725036621, "lr": 0.0002, "elapsed_sec": 44558.54976248741, "step_time_sec": 8.239436340983957, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.107731342315674, "lr": 0.0002, "elapsed_sec": 44566.78786063194, "step_time_sec": 8.237908940995112, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.022996425628662, "lr": 0.0002, "elapsed_sec": 44575.0255856514, "step_time_sec": 8.237584521993995, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.859145164489746, "lr": 0.0002, "elapsed_sec": 44583.264595746994, "step_time_sec": 8.238841245998628, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.111030101776123, "lr": 0.0002, "elapsed_sec": 44591.502907037735, "step_time_sec": 8.238186420989223, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.029654026031494, "lr": 0.0002, "elapsed_sec": 44599.740728616714, "step_time_sec": 8.237646471010521, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.113776683807373, "lr": 0.0002, "elapsed_sec": 44607.979748249054, "step_time_sec": 8.2388122449629, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.126867294311523, "lr": 0.0002, "elapsed_sec": 44616.21935939789, "step_time_sec": 8.239482239005156, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.003154754638672, "lr": 0.0002, "elapsed_sec": 44624.4592936039, "step_time_sec": 8.23976110498188, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.082717418670654, "lr": 0.0002, "elapsed_sec": 44632.69664788246, "step_time_sec": 8.237128025968559, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8850560188293457, "lr": 0.0002, "elapsed_sec": 44640.93740391731, "step_time_sec": 8.240649861982092, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.138636589050293, "lr": 0.0002, "elapsed_sec": 44649.17673397064, "step_time_sec": 8.239129946974572, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9912140369415283, "lr": 0.0002, "elapsed_sec": 44657.41532063484, "step_time_sec": 8.238488241971936, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.104211330413818, "lr": 0.0002, "elapsed_sec": 44665.65355682373, "step_time_sec": 8.237973072973546, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9601049423217773, "lr": 0.0002, "elapsed_sec": 44673.890508413315, "step_time_sec": 8.236818558943924, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.988719940185547, "lr": 0.0002, "elapsed_sec": 44682.12968993187, "step_time_sec": 8.239002832036931, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.008472442626953, "lr": 0.0002, "elapsed_sec": 44690.36711144447, "step_time_sec": 8.237263245973736, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9884915351867676, "lr": 0.0002, "elapsed_sec": 44698.607566833496, "step_time_sec": 8.240284719970077, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.160545349121094, "lr": 0.0002, "elapsed_sec": 44706.84749960899, "step_time_sec": 8.239754311973229, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.150954723358154, "lr": 0.0002, "elapsed_sec": 44715.08478093147, "step_time_sec": 8.237167500017677, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.176431179046631, "lr": 0.0002, "elapsed_sec": 44723.32406258583, "step_time_sec": 8.239093328011222, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.147573947906494, "lr": 0.0002, "elapsed_sec": 44731.563875198364, "step_time_sec": 8.239653458003886, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9701712131500244, "lr": 0.0002, "elapsed_sec": 44739.802421569824, "step_time_sec": 8.238409737998154, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.078446388244629, "lr": 0.0002, "elapsed_sec": 44748.04249715805, "step_time_sec": 8.239872534992173, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.117279529571533, "lr": 0.0002, "elapsed_sec": 44756.281011343, "step_time_sec": 8.238400208007079, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.966555118560791, "lr": 0.0002, "elapsed_sec": 44764.519969940186, "step_time_sec": 8.238773765042424, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9478538036346436, "lr": 0.0002, "elapsed_sec": 44772.758784770966, "step_time_sec": 8.23864809499355, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9401724338531494, "lr": 0.0002, "elapsed_sec": 44780.99890470505, "step_time_sec": 8.24000618600985, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.031282424926758, "lr": 0.0002, "elapsed_sec": 44789.23780751228, "step_time_sec": 8.238712055957876, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9769537448883057, "lr": 0.0002, "elapsed_sec": 44797.47543859482, "step_time_sec": 8.23746694199508, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9585742950439453, "lr": 0.0002, "elapsed_sec": 44805.71494817734, "step_time_sec": 8.239309854980092, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9713151454925537, "lr": 0.0002, "elapsed_sec": 44813.954067230225, "step_time_sec": 8.23900623904774, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.957296371459961, "lr": 0.0002, "elapsed_sec": 44822.192994594574, "step_time_sec": 8.238715769955888, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.015642166137695, "lr": 0.0002, "elapsed_sec": 44830.43274283409, "step_time_sec": 8.239578013017308, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.059998989105225, "lr": 0.0002, "elapsed_sec": 44838.6705391407, "step_time_sec": 8.237661709950771, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9594202041625977, "lr": 0.0002, "elapsed_sec": 44846.90962052345, "step_time_sec": 8.238872997986618, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9215493202209473, "lr": 0.0002, "elapsed_sec": 44855.14848637581, "step_time_sec": 8.238750455027912, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9690330028533936, "lr": 0.0002, "elapsed_sec": 44863.38642311096, "step_time_sec": 8.237751840031706, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.96348237991333, "lr": 0.0002, "elapsed_sec": 44871.626180648804, "step_time_sec": 8.23960273899138, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.151398658752441, "lr": 0.0002, "elapsed_sec": 44879.86497473717, "step_time_sec": 8.238620275980793, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.032796382904053, "lr": 0.0002, "elapsed_sec": 44888.10487961769, "step_time_sec": 8.239724299986847, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.070352554321289, "lr": 0.0002, "elapsed_sec": 44896.34265804291, "step_time_sec": 8.237627411959693, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.044580936431885, "lr": 0.0002, "elapsed_sec": 44904.57917046547, "step_time_sec": 8.236351450032089, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.125411033630371, "lr": 0.0002, "elapsed_sec": 44912.81782865524, "step_time_sec": 8.23849328700453, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.033686637878418, "lr": 0.0002, "elapsed_sec": 44921.05544257164, "step_time_sec": 8.237441949022468, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.933933734893799, "lr": 0.0002, "elapsed_sec": 44929.295365571976, "step_time_sec": 8.239815962966532, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.935023069381714, "lr": 0.0002, "elapsed_sec": 44937.53469967842, "step_time_sec": 8.239128386019729, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.030294418334961, "lr": 0.0002, "elapsed_sec": 44945.77396059036, "step_time_sec": 8.23910301597789, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.015948295593262, "lr": 0.0002, "elapsed_sec": 44954.0134267807, "step_time_sec": 8.239337239007, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.040946006774902, "lr": 0.0002, "elapsed_sec": 44962.25155353546, "step_time_sec": 8.237898153020069, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.090763092041016, "lr": 0.0002, "elapsed_sec": 44970.48987650871, "step_time_sec": 8.238130548037589, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.013285160064697, "lr": 0.0002, "elapsed_sec": 44978.72804021835, "step_time_sec": 8.23800841200864, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.874539375305176, "lr": 0.0002, "elapsed_sec": 44986.964612960815, "step_time_sec": 8.236397731991019, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.15023946762085, "lr": 0.0002, "elapsed_sec": 44995.20098781586, "step_time_sec": 8.236266238964163, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.086859226226807, "lr": 0.0002, "elapsed_sec": 45003.43999218941, "step_time_sec": 8.2388051149901, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.09937047958374, "lr": 0.0002, "elapsed_sec": 45011.67912697792, "step_time_sec": 8.23901684395969, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.076521396636963, "lr": 0.0002, "elapsed_sec": 45019.916669130325, "step_time_sec": 8.23731877701357, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.058349609375, "lr": 0.0002, "elapsed_sec": 45028.1545419693, "step_time_sec": 8.237715310999192, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.06975793838501, "lr": 0.0002, "elapsed_sec": 45036.39317035675, "step_time_sec": 8.238495683996007, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.074459552764893, "lr": 0.0002, "elapsed_sec": 45044.6300637722, "step_time_sec": 8.23668909596745, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.161808967590332, "lr": 0.0002, "elapsed_sec": 45052.86831378937, "step_time_sec": 8.23810943798162, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.028165340423584, "lr": 0.0002, "elapsed_sec": 45061.10573554039, "step_time_sec": 8.23725442495197, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.062748432159424, "lr": 0.0002, "elapsed_sec": 45069.344581604004, "step_time_sec": 8.238689301011618, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9715681076049805, "lr": 0.0002, "elapsed_sec": 45077.58263373375, "step_time_sec": 8.237927998998202, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.018869876861572, "lr": 0.0002, "elapsed_sec": 45085.82042145729, "step_time_sec": 8.237606670998503, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.096563816070557, "lr": 0.0002, "elapsed_sec": 45094.058601140976, "step_time_sec": 8.23806617804803, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9249298572540283, "lr": 0.0002, "elapsed_sec": 45102.295410871506, "step_time_sec": 8.236613292014226, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.922743320465088, "lr": 0.0002, "elapsed_sec": 45110.53470778465, "step_time_sec": 8.239111336006317, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9325082302093506, "lr": 0.0002, "elapsed_sec": 45118.774128198624, "step_time_sec": 8.239292465033941, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.064513683319092, "lr": 0.0002, "elapsed_sec": 45127.01276803017, "step_time_sec": 8.238436245010234, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.039268493652344, "lr": 0.0002, "elapsed_sec": 45135.2520737648, "step_time_sec": 8.239163125050254, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.058730125427246, "lr": 0.0002, "elapsed_sec": 45143.48893547058, "step_time_sec": 8.236670218000654, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.988010883331299, "lr": 0.0002, "elapsed_sec": 45151.72867274284, "step_time_sec": 8.23956526798429, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.101035118103027, "lr": 0.0002, "elapsed_sec": 45159.96861577034, "step_time_sec": 8.239797131041996, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8727633953094482, "lr": 0.0002, "elapsed_sec": 45168.20695376396, "step_time_sec": 8.238186330010649, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9779105186462402, "lr": 0.0002, "elapsed_sec": 45176.44641828537, "step_time_sec": 8.23935095401248, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9508938789367676, "lr": 0.0002, "elapsed_sec": 45184.68379640579, "step_time_sec": 8.237175556016155, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.09435510635376, "lr": 0.0002, "elapsed_sec": 45192.92234611511, "step_time_sec": 8.238395689986646, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.116748332977295, "lr": 0.0002, "elapsed_sec": 45201.16130590439, "step_time_sec": 8.238795664045028, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8924310207366943, "lr": 0.0002, "elapsed_sec": 45209.39946484566, "step_time_sec": 8.237979061028454, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8860809803009033, "lr": 0.0002, "elapsed_sec": 45217.63792490959, "step_time_sec": 8.238338415976614, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9887897968292236, "lr": 0.0002, "elapsed_sec": 45225.87588882446, "step_time_sec": 8.237772906955797, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.065738201141357, "lr": 0.0002, "elapsed_sec": 45234.11606645584, "step_time_sec": 8.240031267981976, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.021927356719971, "lr": 0.0002, "elapsed_sec": 45242.354467868805, "step_time_sec": 8.238229912007228, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.147515296936035, "lr": 0.0002, "elapsed_sec": 45250.5930659771, "step_time_sec": 8.238400075002573, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9949545860290527, "lr": 0.0002, "elapsed_sec": 45258.83308887482, "step_time_sec": 8.239910534000956, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8998453617095947, "lr": 0.0002, "elapsed_sec": 45267.070138931274, "step_time_sec": 8.236857080017217, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.982961654663086, "lr": 0.0002, "elapsed_sec": 45275.30910086632, "step_time_sec": 8.238834949966986, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9938266277313232, "lr": 0.0002, "elapsed_sec": 45283.5469956398, "step_time_sec": 8.23768266994739, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.087568283081055, "lr": 0.0002, "elapsed_sec": 45291.785236120224, "step_time_sec": 8.238082247960847, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.044704914093018, "lr": 0.0002, "elapsed_sec": 45300.0264647007, "step_time_sec": 8.241071755008306, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.843034505844116, "lr": 0.0002, "elapsed_sec": 45308.26483082771, "step_time_sec": 8.238182748958934, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.177486419677734, "lr": 0.0002, "elapsed_sec": 45316.5046081543, "step_time_sec": 8.239613111014478, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.053934574127197, "lr": 0.0002, "elapsed_sec": 45324.74286746979, "step_time_sec": 8.238106253033038, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.012391567230225, "lr": 0.0002, "elapsed_sec": 45332.98121738434, "step_time_sec": 8.238183749024756, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.840360403060913, "lr": 0.0002, "elapsed_sec": 45341.21960759163, "step_time_sec": 8.238210765994154, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.986985445022583, "lr": 0.0002, "elapsed_sec": 45349.45692062378, "step_time_sec": 8.237127630971372, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.044724464416504, "lr": 0.0002, "elapsed_sec": 45357.69605731964, "step_time_sec": 8.238972530001774, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9576146602630615, "lr": 0.0002, "elapsed_sec": 45365.93425679207, "step_time_sec": 8.238045708974823, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0448503494262695, "lr": 0.0002, "elapsed_sec": 45374.172602415085, "step_time_sec": 8.238160870037973, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.917152166366577, "lr": 0.0002, "elapsed_sec": 45382.412977933884, "step_time_sec": 8.24021493899636, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.14599609375, "lr": 0.0002, "elapsed_sec": 45390.64987921715, "step_time_sec": 8.23674005700741, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.061285495758057, "lr": 0.0002, "elapsed_sec": 45398.890254974365, "step_time_sec": 8.240219807019457, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9682695865631104, "lr": 0.0002, "elapsed_sec": 45407.12892603874, "step_time_sec": 8.238487795984838, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.019341945648193, "lr": 0.0002, "elapsed_sec": 45415.36797499657, "step_time_sec": 8.238949644030072, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.276937484741211, "lr": 0.0002, "elapsed_sec": 45423.60793161392, "step_time_sec": 8.239753284025937, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.795382499694824, "lr": 0.0002, "elapsed_sec": 45431.84587955475, "step_time_sec": 8.237782469019294, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.071252346038818, "lr": 0.0002, "elapsed_sec": 45440.08432674408, "step_time_sec": 8.23830133600859, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.167531490325928, "lr": 0.0002, "elapsed_sec": 45448.32175254822, "step_time_sec": 8.237270537007134, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.002408504486084, "lr": 0.0002, "elapsed_sec": 45456.561524391174, "step_time_sec": 8.23961208097171, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.236574172973633, "lr": 0.0002, "elapsed_sec": 45464.799949645996, "step_time_sec": 8.23826639301842, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.05597448348999, "lr": 0.0002, "elapsed_sec": 45473.03731608391, "step_time_sec": 8.237245004042052, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.103815078735352, "lr": 0.0002, "elapsed_sec": 45481.278601408005, "step_time_sec": 8.241090964002069, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.138671398162842, "lr": 0.0002, "elapsed_sec": 45489.51712059975, "step_time_sec": 8.238344030047301, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.176310062408447, "lr": 0.0002, "elapsed_sec": 45497.75539064407, "step_time_sec": 8.238126148004085, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.217746257781982, "lr": 0.0002, "elapsed_sec": 45505.99522399902, "step_time_sec": 8.239677548001055, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.173986911773682, "lr": 0.0002, "elapsed_sec": 45514.2327272892, "step_time_sec": 8.23733320197789, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.053465366363525, "lr": 0.0002, "elapsed_sec": 45522.47305846214, "step_time_sec": 8.240130814956501, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.17471981048584, "lr": 0.0002, "elapsed_sec": 45530.71317863464, "step_time_sec": 8.239994872012176, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.970207452774048, "lr": 0.0002, "elapsed_sec": 45538.950595378876, "step_time_sec": 8.237275977036916, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.106350898742676, "lr": 0.0002, "elapsed_sec": 45547.191910505295, "step_time_sec": 8.24114262900548, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.916874647140503, "lr": 0.0002, "elapsed_sec": 45555.43134903908, "step_time_sec": 8.239273271989077, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.042069435119629, "lr": 0.0002, "elapsed_sec": 45563.67063379288, "step_time_sec": 8.239135320996866, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.990478992462158, "lr": 0.0002, "elapsed_sec": 45571.910425662994, "step_time_sec": 8.23960273998091, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9452507495880127, "lr": 0.0002, "elapsed_sec": 45580.147939920425, "step_time_sec": 8.237343152984977, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.106581211090088, "lr": 0.0002, "elapsed_sec": 45588.38867235184, "step_time_sec": 8.240586783969775, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.963778018951416, "lr": 0.0002, "elapsed_sec": 45596.62798023224, "step_time_sec": 8.239131006004754, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0607380867004395, "lr": 0.0002, "elapsed_sec": 45604.866958618164, "step_time_sec": 8.238780846993905, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.013725757598877, "lr": 0.0002, "elapsed_sec": 45613.107838869095, "step_time_sec": 8.240762667963281, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0005621910095215, "lr": 0.0002, "elapsed_sec": 45621.34567141533, "step_time_sec": 8.237629402021412, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.998384714126587, "lr": 0.0002, "elapsed_sec": 45629.586081027985, "step_time_sec": 8.240243098989595, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.299689769744873, "lr": 0.0002, "elapsed_sec": 45637.82494401932, "step_time_sec": 8.238710988021921, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.034742832183838, "lr": 0.0002, "elapsed_sec": 45646.062863111496, "step_time_sec": 8.237755942973308, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.123429775238037, "lr": 0.0002, "elapsed_sec": 45654.30111551285, "step_time_sec": 8.238132559985388, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9677345752716064, "lr": 0.0002, "elapsed_sec": 45662.538813352585, "step_time_sec": 8.237452114000916, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9708707332611084, "lr": 0.0002, "elapsed_sec": 45670.7775952816, "step_time_sec": 8.238633721019141, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9637227058410645, "lr": 0.0002, "elapsed_sec": 45679.01576209068, "step_time_sec": 8.237998443015385, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.018721103668213, "lr": 0.0002, "elapsed_sec": 45687.254009485245, "step_time_sec": 8.23810846899869, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.067065715789795, "lr": 0.0002, "elapsed_sec": 45695.49220943451, "step_time_sec": 8.23806780698942, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9601805210113525, "lr": 0.0002, "elapsed_sec": 45703.728726387024, "step_time_sec": 30.62945150298765, "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": 3.9936654567718506, "lr": 0.0002, "elapsed_sec": 45734.37751555443, "step_time_sec": 8.255159858963452, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.128000736236572, "lr": 0.0002, "elapsed_sec": 45742.61554455757, "step_time_sec": 8.237819242000114, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.961207866668701, "lr": 0.0002, "elapsed_sec": 45750.85200548172, "step_time_sec": 8.236323675955646, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.012259483337402, "lr": 0.0002, "elapsed_sec": 45759.08981466293, "step_time_sec": 8.237616683007218, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.023586750030518, "lr": 0.0002, "elapsed_sec": 45767.3276591301, "step_time_sec": 8.237689819012303, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0698113441467285, "lr": 0.0002, "elapsed_sec": 45775.56555604935, "step_time_sec": 8.237746130966116, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9805197715759277, "lr": 0.0002, "elapsed_sec": 45783.80451178551, "step_time_sec": 8.23880793194985, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.173211574554443, "lr": 0.0002, "elapsed_sec": 45792.04148721695, "step_time_sec": 8.236744774971157, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.039617538452148, "lr": 0.0002, "elapsed_sec": 45800.28005385399, "step_time_sec": 8.23842531000264, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.106337547302246, "lr": 0.0002, "elapsed_sec": 45808.517973184586, "step_time_sec": 8.237813637009822, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.169365882873535, "lr": 0.0002, "elapsed_sec": 45816.75688457489, "step_time_sec": 8.238683485949878, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.095675945281982, "lr": 0.0002, "elapsed_sec": 45824.99542307854, "step_time_sec": 8.23839960503392, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8277907371520996, "lr": 0.0002, "elapsed_sec": 45833.233211517334, "step_time_sec": 8.237615308025852, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9287846088409424, "lr": 0.0002, "elapsed_sec": 45841.47260046005, "step_time_sec": 8.239206468046177, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.148542881011963, "lr": 0.0002, "elapsed_sec": 45849.71195816994, "step_time_sec": 8.239197132003028, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.987320899963379, "lr": 0.0002, "elapsed_sec": 45857.95123839378, "step_time_sec": 8.23911472799955, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.042338848114014, "lr": 0.0002, "elapsed_sec": 45866.19038915634, "step_time_sec": 8.238996896950994, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.03355073928833, "lr": 0.0002, "elapsed_sec": 45874.42689847946, "step_time_sec": 8.236343075986952, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.119361400604248, "lr": 0.0002, "elapsed_sec": 45882.6657371521, "step_time_sec": 8.23869599797763, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.068878173828125, "lr": 0.0002, "elapsed_sec": 45890.90317893028, "step_time_sec": 8.237325540976599, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.956228733062744, "lr": 0.0002, "elapsed_sec": 45899.14419054985, "step_time_sec": 8.240841778984759, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.144902229309082, "lr": 0.0002, "elapsed_sec": 45907.38387179375, "step_time_sec": 8.239471644046716, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.042187690734863, "lr": 0.0002, "elapsed_sec": 45915.621180295944, "step_time_sec": 8.23711121303495, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.044527053833008, "lr": 0.0002, "elapsed_sec": 45923.860565662384, "step_time_sec": 8.2392569980002, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0620198249816895, "lr": 0.0002, "elapsed_sec": 45932.099769592285, "step_time_sec": 8.23899684799835, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.121598243713379, "lr": 0.0002, "elapsed_sec": 45940.33781385422, "step_time_sec": 8.23789097397821, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.97636079788208, "lr": 0.0002, "elapsed_sec": 45948.576530218124, "step_time_sec": 8.238574151007924, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.141158580780029, "lr": 0.0002, "elapsed_sec": 45956.814802885056, "step_time_sec": 8.238087813020684, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.072977066040039, "lr": 0.0002, "elapsed_sec": 45965.05459880829, "step_time_sec": 8.239629881980363, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.415694713592529, "lr": 0.0002, "elapsed_sec": 45973.292946100235, "step_time_sec": 8.238190951000433, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.215582847595215, "lr": 0.0002, "elapsed_sec": 45981.53216314316, "step_time_sec": 8.239057883038186, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.125635623931885, "lr": 0.0002, "elapsed_sec": 45989.77040934563, "step_time_sec": 8.2380747749703, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.050057888031006, "lr": 0.0002, "elapsed_sec": 45998.00805282593, "step_time_sec": 8.237467187980656, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.043858051300049, "lr": 0.0002, "elapsed_sec": 46006.24726653099, "step_time_sec": 8.239085725042969, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.026516914367676, "lr": 0.0002, "elapsed_sec": 46014.4850730896, "step_time_sec": 8.237644887994975, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.079141139984131, "lr": 0.0002, "elapsed_sec": 46022.72410130501, "step_time_sec": 8.238815220014658, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.086946964263916, "lr": 0.0002, "elapsed_sec": 46030.96233129501, "step_time_sec": 8.238110305042937, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.043464183807373, "lr": 0.0002, "elapsed_sec": 46039.19999861717, "step_time_sec": 8.23744383698795, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.110718250274658, "lr": 0.0002, "elapsed_sec": 46047.439341545105, "step_time_sec": 8.239193664980121, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9960997104644775, "lr": 0.0002, "elapsed_sec": 46055.67719888687, "step_time_sec": 8.23767575097736, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.037060737609863, "lr": 0.0002, "elapsed_sec": 46063.91532564163, "step_time_sec": 8.238008160027675, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.009443759918213, "lr": 0.0002, "elapsed_sec": 46072.154103279114, "step_time_sec": 8.238575792987831, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0421977043151855, "lr": 0.0002, "elapsed_sec": 46080.39166069031, "step_time_sec": 8.2374443730223, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.107380390167236, "lr": 0.0002, "elapsed_sec": 46088.63033294678, "step_time_sec": 8.238516871992033, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.08229398727417, "lr": 0.0002, "elapsed_sec": 46096.86777806282, "step_time_sec": 8.237223946955055, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.08209228515625, "lr": 0.0002, "elapsed_sec": 46105.10757637024, "step_time_sec": 8.239643918990623, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.075145244598389, "lr": 0.0002, "elapsed_sec": 46113.34783697128, "step_time_sec": 8.240140353969764, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.152971267700195, "lr": 0.0002, "elapsed_sec": 46121.58461523056, "step_time_sec": 8.236576005001552, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.157851219177246, "lr": 0.0002, "elapsed_sec": 46129.8243060112, "step_time_sec": 8.23958551499527, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.151512622833252, "lr": 0.0002, "elapsed_sec": 46138.06365060806, "step_time_sec": 8.239238816022407, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.133360385894775, "lr": 0.0002, "elapsed_sec": 46146.30471158028, "step_time_sec": 8.24086909397738, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.060877799987793, "lr": 0.0002, "elapsed_sec": 46154.55974364281, "step_time_sec": 8.254834296996705, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.015816688537598, "lr": 0.0002, "elapsed_sec": 46162.79930090904, "step_time_sec": 8.239342047017999, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.079280376434326, "lr": 0.0002, "elapsed_sec": 46171.03738069534, "step_time_sec": 8.237965709995478, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.027078628540039, "lr": 0.0002, "elapsed_sec": 46179.27427530289, "step_time_sec": 8.236702817957848, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.06351375579834, "lr": 0.0002, "elapsed_sec": 46187.51227378845, "step_time_sec": 8.23783357301727, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.084133148193359, "lr": 0.0002, "elapsed_sec": 46195.75189495087, "step_time_sec": 8.239426081010606, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.038568496704102, "lr": 0.0002, "elapsed_sec": 46203.99070882797, "step_time_sec": 8.23870499804616, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.108774662017822, "lr": 0.0002, "elapsed_sec": 46212.22912168503, "step_time_sec": 8.238213195989374, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.048689365386963, "lr": 0.0002, "elapsed_sec": 46220.46791815758, "step_time_sec": 8.23858314997051, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.116130828857422, "lr": 0.0002, "elapsed_sec": 46228.7057993412, "step_time_sec": 8.237765371974092, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.860865354537964, "lr": 0.0002, "elapsed_sec": 46236.945848703384, "step_time_sec": 8.239892227051314, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.135509014129639, "lr": 0.0002, "elapsed_sec": 46245.186495780945, "step_time_sec": 8.240436332998797, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.011013984680176, "lr": 0.0002, "elapsed_sec": 46253.42359375954, "step_time_sec": 8.236952497973107, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.196418762207031, "lr": 0.0002, "elapsed_sec": 46261.666450977325, "step_time_sec": 8.242691208026372, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.08104944229126, "lr": 0.0002, "elapsed_sec": 46269.90626883507, "step_time_sec": 8.239663881016895, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1321330070495605, "lr": 0.0002, "elapsed_sec": 46278.14502429962, "step_time_sec": 8.23855349700898, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.221017837524414, "lr": 0.0002, "elapsed_sec": 46286.38396191597, "step_time_sec": 8.238824678992387, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.985668897628784, "lr": 0.0002, "elapsed_sec": 46294.62089347839, "step_time_sec": 8.236757270002272, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.093317031860352, "lr": 0.0002, "elapsed_sec": 46302.86342692375, "step_time_sec": 8.242359730007593, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.165695667266846, "lr": 0.0002, "elapsed_sec": 46311.10224890709, "step_time_sec": 8.238684623967856, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.058137893676758, "lr": 0.0002, "elapsed_sec": 46319.339896678925, "step_time_sec": 8.23746295098681, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8901784420013428, "lr": 0.0002, "elapsed_sec": 46327.579330444336, "step_time_sec": 8.239275341969915, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.992460012435913, "lr": 0.0002, "elapsed_sec": 46335.81576061249, "step_time_sec": 8.236248506000265, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.163873195648193, "lr": 0.0002, "elapsed_sec": 46344.05469202995, "step_time_sec": 8.2388460340444, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.07382345199585, "lr": 0.0002, "elapsed_sec": 46352.293647527695, "step_time_sec": 8.23870668100426, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.112142086029053, "lr": 0.0002, "elapsed_sec": 46360.53220772743, "step_time_sec": 8.238401993992738, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0658860206604, "lr": 0.0002, "elapsed_sec": 46368.77086853981, "step_time_sec": 8.238527236040682, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.316068172454834, "lr": 0.0002, "elapsed_sec": 46377.0092215538, "step_time_sec": 8.238200132967904, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.149226188659668, "lr": 0.0002, "elapsed_sec": 46385.25025343895, "step_time_sec": 8.24089679500321, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.006844997406006, "lr": 0.0002, "elapsed_sec": 46393.490337371826, "step_time_sec": 8.239914485020563, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.049746036529541, "lr": 0.0002, "elapsed_sec": 46401.7279689312, "step_time_sec": 8.23744195298059, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.013989448547363, "lr": 0.0002, "elapsed_sec": 46409.966628074646, "step_time_sec": 8.238494971010368, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.088929653167725, "lr": 0.0002, "elapsed_sec": 46418.2039706707, "step_time_sec": 8.237186470010784, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1309967041015625, "lr": 0.0002, "elapsed_sec": 46426.44361329079, "step_time_sec": 8.239529643033165, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.055381774902344, "lr": 0.0002, "elapsed_sec": 46434.68241381645, "step_time_sec": 8.238582954974845, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.09625768661499, "lr": 0.0002, "elapsed_sec": 46442.92024087906, "step_time_sec": 8.237671529990621, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.144355297088623, "lr": 0.0002, "elapsed_sec": 46451.160094976425, "step_time_sec": 8.23972385097295, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.027351379394531, "lr": 0.0002, "elapsed_sec": 46459.396349191666, "step_time_sec": 8.236030463012867, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9814233779907227, "lr": 0.0002, "elapsed_sec": 46467.63683009148, "step_time_sec": 8.240320848999545, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.968233108520508, "lr": 0.0002, "elapsed_sec": 46475.875375032425, "step_time_sec": 8.23841240699403, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.136630535125732, "lr": 0.0002, "elapsed_sec": 46484.11364173889, "step_time_sec": 8.238131531048566, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.075613021850586, "lr": 0.0002, "elapsed_sec": 46492.35255527496, "step_time_sec": 8.238714425999206, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.125372886657715, "lr": 0.0002, "elapsed_sec": 46500.5905687809, "step_time_sec": 8.237834426981863, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0533928871154785, "lr": 0.0002, "elapsed_sec": 46508.83138823509, "step_time_sec": 8.240693428029772, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.268414497375488, "lr": 0.0002, "elapsed_sec": 46517.07064175606, "step_time_sec": 8.239124089013785, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.123129844665527, "lr": 0.0002, "elapsed_sec": 46525.308114528656, "step_time_sec": 8.237314478028566, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0482497215271, "lr": 0.0002, "elapsed_sec": 46533.545345544815, "step_time_sec": 8.237012970028445, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9651873111724854, "lr": 0.0002, "elapsed_sec": 46541.78228473663, "step_time_sec": 8.236817112017889, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.126597881317139, "lr": 0.0002, "elapsed_sec": 46550.02445268631, "step_time_sec": 8.241983496991452, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9846198558807373, "lr": 0.0002, "elapsed_sec": 46558.27140331268, "step_time_sec": 8.23912168102106, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.157541751861572, "lr": 0.0002, "elapsed_sec": 46566.50892019272, "step_time_sec": 8.237408227985725, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.045502662658691, "lr": 0.0002, "elapsed_sec": 46574.74822664261, "step_time_sec": 8.23908017098438, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8703179359436035, "lr": 0.0002, "elapsed_sec": 46582.98732495308, "step_time_sec": 8.238928854989354, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.039676666259766, "lr": 0.0002, "elapsed_sec": 46591.2255423069, "step_time_sec": 8.238080512965098, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.020901203155518, "lr": 0.0002, "elapsed_sec": 46599.46441745758, "step_time_sec": 8.238665070035495, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.935884714126587, "lr": 0.0002, "elapsed_sec": 46607.70101737976, "step_time_sec": 8.236466712027323, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.055030822753906, "lr": 0.0002, "elapsed_sec": 46615.93981003761, "step_time_sec": 8.238678412046283, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9838883876800537, "lr": 0.0002, "elapsed_sec": 46624.17992544174, "step_time_sec": 8.239999687997624, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.063645839691162, "lr": 0.0002, "elapsed_sec": 46632.41817498207, "step_time_sec": 8.237995416973718, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.02728271484375, "lr": 0.0002, "elapsed_sec": 46640.65855431557, "step_time_sec": 8.240196844970342, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.990967035293579, "lr": 0.0002, "elapsed_sec": 46648.896807193756, "step_time_sec": 8.23811970098177, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.038244247436523, "lr": 0.0002, "elapsed_sec": 46657.137505054474, "step_time_sec": 8.240542154002469, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9200639724731445, "lr": 0.0002, "elapsed_sec": 46665.37440299988, "step_time_sec": 8.236708640994038, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.051065921783447, "lr": 0.0002, "elapsed_sec": 46673.61365938187, "step_time_sec": 8.239095143042505, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.175920009613037, "lr": 0.0002, "elapsed_sec": 46681.854768276215, "step_time_sec": 8.240943850018084, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.026732921600342, "lr": 0.0002, "elapsed_sec": 46690.093702316284, "step_time_sec": 8.238772881974, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.004151344299316, "lr": 0.0002, "elapsed_sec": 46698.33311128616, "step_time_sec": 8.239263187046163, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.062562942504883, "lr": 0.0002, "elapsed_sec": 46706.5716586113, "step_time_sec": 8.238381226023193, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0104780197143555, "lr": 0.0002, "elapsed_sec": 46714.80900454521, "step_time_sec": 8.237149060994852, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.149368762969971, "lr": 0.0002, "elapsed_sec": 46723.04881310463, "step_time_sec": 8.239656989986543, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.071677207946777, "lr": 0.0002, "elapsed_sec": 46731.2858941555, "step_time_sec": 8.23694555403199, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0857954025268555, "lr": 0.0002, "elapsed_sec": 46739.52685546875, "step_time_sec": 8.240748228039593, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9366366863250732, "lr": 0.0002, "elapsed_sec": 46747.766399145126, "step_time_sec": 8.239401405036915, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.08575963973999, "lr": 0.0002, "elapsed_sec": 46756.00374555588, "step_time_sec": 8.237205268000253, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.183071136474609, "lr": 0.0002, "elapsed_sec": 46764.243546009064, "step_time_sec": 8.239626384049188, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.949610948562622, "lr": 0.0002, "elapsed_sec": 46772.481634140015, "step_time_sec": 8.237900390988216, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.872706651687622, "lr": 0.0002, "elapsed_sec": 46780.721252679825, "step_time_sec": 8.23944884998491, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.070369243621826, "lr": 0.0002, "elapsed_sec": 46788.960799217224, "step_time_sec": 8.239453724992927, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.010461807250977, "lr": 0.0002, "elapsed_sec": 46797.19818520546, "step_time_sec": 8.237211735977326, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0350117683410645, "lr": 0.0002, "elapsed_sec": 46805.44118499756, "step_time_sec": 8.242838862992357, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1183624267578125, "lr": 0.0002, "elapsed_sec": 46813.68031787872, "step_time_sec": 8.238973626983352, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1685357093811035, "lr": 0.0002, "elapsed_sec": 46821.91788959503, "step_time_sec": 8.237358695012517, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.014770030975342, "lr": 0.0002, "elapsed_sec": 46830.15781259537, "step_time_sec": 8.239764051977545, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.988258123397827, "lr": 0.0002, "elapsed_sec": 46838.39512181282, "step_time_sec": 8.23716981097823, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.031576156616211, "lr": 0.0002, "elapsed_sec": 46846.63417387009, "step_time_sec": 8.238904390018433, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.027462959289551, "lr": 0.0002, "elapsed_sec": 46854.87426280975, "step_time_sec": 8.239872050005943, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0369415283203125, "lr": 0.0002, "elapsed_sec": 46863.11261987686, "step_time_sec": 8.238209425006062, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.969357967376709, "lr": 0.0002, "elapsed_sec": 46871.35211801529, "step_time_sec": 8.239344196976162, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.014369964599609, "lr": 0.0002, "elapsed_sec": 46879.59014296532, "step_time_sec": 8.23785172594944, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.949735641479492, "lr": 0.0002, "elapsed_sec": 46887.83134675026, "step_time_sec": 8.24102729500737, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9989678859710693, "lr": 0.0002, "elapsed_sec": 46896.06953501701, "step_time_sec": 8.238013551977929, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.02817964553833, "lr": 0.0002, "elapsed_sec": 46904.3068549633, "step_time_sec": 8.237166911014356, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.051020622253418, "lr": 0.0002, "elapsed_sec": 46912.54615902901, "step_time_sec": 8.239099797036033, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.11342191696167, "lr": 0.0002, "elapsed_sec": 46920.78484106064, "step_time_sec": 8.238502558961045, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.00063943862915, "lr": 0.0002, "elapsed_sec": 46929.02582645416, "step_time_sec": 8.240822863997892, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.164213180541992, "lr": 0.0002, "elapsed_sec": 46937.26555252075, "step_time_sec": 8.239591463992838, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.142879486083984, "lr": 0.0002, "elapsed_sec": 46945.50232601166, "step_time_sec": 8.236589912965428, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.272161960601807, "lr": 0.0002, "elapsed_sec": 46953.74062347412, "step_time_sec": 8.238195830956101, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.938706874847412, "lr": 0.0002, "elapsed_sec": 46961.97787857056, "step_time_sec": 8.237018001032993, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.824327230453491, "lr": 0.0002, "elapsed_sec": 46970.218029499054, "step_time_sec": 8.240003785991576, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.251416206359863, "lr": 0.0002, "elapsed_sec": 46978.45783996582, "step_time_sec": 8.239671627001371, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.072520732879639, "lr": 0.0002, "elapsed_sec": 46986.69448471069, "step_time_sec": 8.236483025015332, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.191497802734375, "lr": 0.0002, "elapsed_sec": 46994.934022426605, "step_time_sec": 8.239331595017575, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1723737716674805, "lr": 0.0002, "elapsed_sec": 47003.171078681946, "step_time_sec": 8.236912745982409, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9937996864318848, "lr": 0.0002, "elapsed_sec": 47011.411640405655, "step_time_sec": 8.240399093017913, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.995222568511963, "lr": 0.0002, "elapsed_sec": 47019.65106201172, "step_time_sec": 8.23926493502222, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.14972448348999, "lr": 0.0002, "elapsed_sec": 47027.88909173012, "step_time_sec": 8.23785171896452, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.05887508392334, "lr": 0.0002, "elapsed_sec": 47036.1289460659, "step_time_sec": 8.239709863031749, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.09116268157959, "lr": 0.0002, "elapsed_sec": 47044.36705780029, "step_time_sec": 8.237928460002877, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.061446666717529, "lr": 0.0002, "elapsed_sec": 47052.606270074844, "step_time_sec": 8.23901456099702, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.057257175445557, "lr": 0.0002, "elapsed_sec": 47060.846054553986, "step_time_sec": 8.239624231006019, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.238011837005615, "lr": 0.0002, "elapsed_sec": 47069.08390402794, "step_time_sec": 8.237722280959133, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.893998146057129, "lr": 0.0002, "elapsed_sec": 47077.32302522659, "step_time_sec": 8.238970426027663, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.091519355773926, "lr": 0.0002, "elapsed_sec": 47085.56101822853, "step_time_sec": 8.237795279012062, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9768152236938477, "lr": 0.0002, "elapsed_sec": 47093.80057001114, "step_time_sec": 8.239464042999316, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.105100631713867, "lr": 0.0002, "elapsed_sec": 47102.0401160717, "step_time_sec": 8.239305477996822, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.146347999572754, "lr": 0.0002, "elapsed_sec": 47110.27714753151, "step_time_sec": 8.236913146974985, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.132713794708252, "lr": 0.0002, "elapsed_sec": 47118.517731666565, "step_time_sec": 8.240443142945878, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.851449489593506, "lr": 0.0002, "elapsed_sec": 47126.75610995293, "step_time_sec": 8.238228309026454, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.010964870452881, "lr": 0.0002, "elapsed_sec": 47134.99513840675, "step_time_sec": 8.238833990995772, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.16169548034668, "lr": 0.0002, "elapsed_sec": 47143.23447775841, "step_time_sec": 8.239141157013364, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.050552845001221, "lr": 0.0002, "elapsed_sec": 47151.47287416458, "step_time_sec": 8.238226301968098, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.019128799438477, "lr": 0.0002, "elapsed_sec": 47159.71290540695, "step_time_sec": 8.239907337992918, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.06695032119751, "lr": 0.0002, "elapsed_sec": 47167.95181632042, "step_time_sec": 8.238717250002082, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1424689292907715, "lr": 0.0002, "elapsed_sec": 47176.19134807587, "step_time_sec": 8.239383017993532, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.219759464263916, "lr": 0.0002, "elapsed_sec": 47184.43125605583, "step_time_sec": 8.2397599699907, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8457794189453125, "lr": 0.0002, "elapsed_sec": 47204.218445301056, "step_time_sec": 19.786998121009674, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.187292575836182, "lr": 0.0002, "elapsed_sec": 47212.45583200455, "step_time_sec": 8.237228940008208, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.07698392868042, "lr": 0.0002, "elapsed_sec": 47220.69666647911, "step_time_sec": 8.240655087982304, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.143789768218994, "lr": 0.0002, "elapsed_sec": 47228.93462944031, "step_time_sec": 8.237768684979528, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.11376428604126, "lr": 0.0002, "elapsed_sec": 47237.17173957825, "step_time_sec": 8.237003984977491, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1167731285095215, "lr": 0.0002, "elapsed_sec": 47245.411868810654, "step_time_sec": 8.239917655009776, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9880237579345703, "lr": 0.0002, "elapsed_sec": 47253.650456905365, "step_time_sec": 8.238436446990818, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.086757183074951, "lr": 0.0002, "elapsed_sec": 47261.88856101036, "step_time_sec": 8.237958971993066, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.063708782196045, "lr": 0.0002, "elapsed_sec": 47270.12866687775, "step_time_sec": 8.239923438988626, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.020572185516357, "lr": 0.0002, "elapsed_sec": 47278.36613488197, "step_time_sec": 8.237322052009404, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.516891002655029, "lr": 0.0002, "elapsed_sec": 47286.605021476746, "step_time_sec": 8.23873718699906, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.115200519561768, "lr": 0.0002, "elapsed_sec": 47294.84368419647, "step_time_sec": 8.238458384003025, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.939133405685425, "lr": 0.0002, "elapsed_sec": 47303.08174943924, "step_time_sec": 8.237863393034786, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.000925540924072, "lr": 0.0002, "elapsed_sec": 47311.320942640305, "step_time_sec": 8.23904204100836, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.96683669090271, "lr": 0.0002, "elapsed_sec": 47319.55881690979, "step_time_sec": 8.23774151597172, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.220573425292969, "lr": 0.0002, "elapsed_sec": 47327.797699689865, "step_time_sec": 8.23869005398592, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.024223327636719, "lr": 0.0002, "elapsed_sec": 47336.035012960434, "step_time_sec": 8.237202265008818, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.088996410369873, "lr": 0.0002, "elapsed_sec": 47344.274898052216, "step_time_sec": 8.239681389008183, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9565703868865967, "lr": 0.0002, "elapsed_sec": 47352.51359295845, "step_time_sec": 8.238530670001637, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.994913339614868, "lr": 0.0002, "elapsed_sec": 47360.75018572807, "step_time_sec": 8.236419883032795, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.994966745376587, "lr": 0.0002, "elapsed_sec": 47368.988708257675, "step_time_sec": 8.238369197002612, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0375165939331055, "lr": 0.0002, "elapsed_sec": 47377.22669172287, "step_time_sec": 8.237824495008681, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.030101299285889, "lr": 0.0002, "elapsed_sec": 47385.46570277214, "step_time_sec": 8.238780475978274, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0354156494140625, "lr": 0.0002, "elapsed_sec": 47393.70506763458, "step_time_sec": 8.239301508991048, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.073573112487793, "lr": 0.0002, "elapsed_sec": 47401.94412446022, "step_time_sec": 8.238867719017435, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9589297771453857, "lr": 0.0002, "elapsed_sec": 47410.18363118172, "step_time_sec": 8.239276567997877, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.015246391296387, "lr": 0.0002, "elapsed_sec": 47418.42138314247, "step_time_sec": 8.237635077966843, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.997873544692993, "lr": 0.0002, "elapsed_sec": 47426.66096043587, "step_time_sec": 8.239427745051216, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9554710388183594, "lr": 0.0002, "elapsed_sec": 47434.89976978302, "step_time_sec": 8.238622372038662, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.015567779541016, "lr": 0.0002, "elapsed_sec": 47443.13746190071, "step_time_sec": 8.237493458960671, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.892979621887207, "lr": 0.0002, "elapsed_sec": 47451.37749195099, "step_time_sec": 8.239906753005926, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.950063943862915, "lr": 0.0002, "elapsed_sec": 47459.61709737778, "step_time_sec": 8.23944921803195, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.956773281097412, "lr": 0.0002, "elapsed_sec": 47467.8560526371, "step_time_sec": 8.238798471051268, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.073165416717529, "lr": 0.0002, "elapsed_sec": 47476.09416604042, "step_time_sec": 8.23793335701339, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.048300743103027, "lr": 0.0002, "elapsed_sec": 47484.33132004738, "step_time_sec": 8.236979889974464, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0531158447265625, "lr": 0.0002, "elapsed_sec": 47492.57210063934, "step_time_sec": 8.240624277037568, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.115231037139893, "lr": 0.0002, "elapsed_sec": 47500.81047081947, "step_time_sec": 8.23820242396323, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9319703578948975, "lr": 0.0002, "elapsed_sec": 47509.04848527908, "step_time_sec": 8.237847868003882, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.020305156707764, "lr": 0.0002, "elapsed_sec": 47517.288856983185, "step_time_sec": 8.240252073970623, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0990800857543945, "lr": 0.0002, "elapsed_sec": 47525.525878190994, "step_time_sec": 8.236842757032719, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.966892957687378, "lr": 0.0002, "elapsed_sec": 47533.76517367363, "step_time_sec": 8.23915100004524, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.008157730102539, "lr": 0.0002, "elapsed_sec": 47542.00357723236, "step_time_sec": 8.238220135041047, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.993788719177246, "lr": 0.0002, "elapsed_sec": 47550.242376327515, "step_time_sec": 8.238580522011034, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.977189302444458, "lr": 0.0002, "elapsed_sec": 47558.481004953384, "step_time_sec": 8.238490900024772, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.006121635437012, "lr": 0.0002, "elapsed_sec": 47566.71919584274, "step_time_sec": 8.238034198991954, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.768256425857544, "lr": 0.0002, "elapsed_sec": 47574.958642959595, "step_time_sec": 8.239282955008093, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9224674701690674, "lr": 0.0002, "elapsed_sec": 47583.19762587547, "step_time_sec": 8.238803207990713, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.024724960327148, "lr": 0.0002, "elapsed_sec": 47591.436692237854, "step_time_sec": 8.238914046960417, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1096649169921875, "lr": 0.0002, "elapsed_sec": 47599.67687082291, "step_time_sec": 8.240016809955705, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9705159664154053, "lr": 0.0002, "elapsed_sec": 47607.91475582123, "step_time_sec": 8.237745955004357, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.020232677459717, "lr": 0.0002, "elapsed_sec": 47616.15459394455, "step_time_sec": 8.239655887009576, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.007874011993408, "lr": 0.0002, "elapsed_sec": 47624.39310145378, "step_time_sec": 8.238344759971369, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.005704879760742, "lr": 0.0002, "elapsed_sec": 47632.63120651245, "step_time_sec": 8.237943062966224, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.084695339202881, "lr": 0.0002, "elapsed_sec": 47640.871534347534, "step_time_sec": 8.240157296997495, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.920344591140747, "lr": 0.0002, "elapsed_sec": 47649.10859084129, "step_time_sec": 8.236881924967747, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.030285835266113, "lr": 0.0002, "elapsed_sec": 47657.34734940529, "step_time_sec": 8.238643190998118, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9661905765533447, "lr": 0.0002, "elapsed_sec": 47665.586439847946, "step_time_sec": 8.238870910019614, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.121699333190918, "lr": 0.0002, "elapsed_sec": 47673.82405805588, "step_time_sec": 8.23750484304037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.103182315826416, "lr": 0.0002, "elapsed_sec": 47682.06413054466, "step_time_sec": 8.239904434012715, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.032042503356934, "lr": 0.0002, "elapsed_sec": 47690.302757024765, "step_time_sec": 8.238386982993688, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9298853874206543, "lr": 0.0002, "elapsed_sec": 47698.543110370636, "step_time_sec": 8.240215127007104, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.974383592605591, "lr": 0.0002, "elapsed_sec": 47706.78181505203, "step_time_sec": 8.238521124003455, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0293498039245605, "lr": 0.0002, "elapsed_sec": 47715.0194401741, "step_time_sec": 8.237459681986365, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9710638523101807, "lr": 0.0002, "elapsed_sec": 47723.25803685188, "step_time_sec": 8.23843249795027, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.02649450302124, "lr": 0.0002, "elapsed_sec": 47731.49569296837, "step_time_sec": 8.237455858034082, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.018442153930664, "lr": 0.0002, "elapsed_sec": 47739.73396205902, "step_time_sec": 8.238142310990952, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0410614013671875, "lr": 0.0002, "elapsed_sec": 47747.97351360321, "step_time_sec": 8.239437314041425, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.068087577819824, "lr": 0.0002, "elapsed_sec": 47756.21124005318, "step_time_sec": 8.23755241802428, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.171940803527832, "lr": 0.0002, "elapsed_sec": 47764.4501953125, "step_time_sec": 8.23872606799705, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.148131370544434, "lr": 0.0002, "elapsed_sec": 47772.68634533882, "step_time_sec": 8.236009664949961, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.015791893005371, "lr": 0.0002, "elapsed_sec": 47780.924954891205, "step_time_sec": 8.238451043027453, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.930176258087158, "lr": 0.0002, "elapsed_sec": 47789.16246700287, "step_time_sec": 8.237352269992698, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.111962795257568, "lr": 0.0002, "elapsed_sec": 47797.4016392231, "step_time_sec": 8.23899533401709, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9654946327209473, "lr": 0.0002, "elapsed_sec": 47805.64100933075, "step_time_sec": 8.239165558014065, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.204833507537842, "lr": 0.0002, "elapsed_sec": 47813.87868452072, "step_time_sec": 8.237586716015358, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.934361219406128, "lr": 0.0002, "elapsed_sec": 47822.117302417755, "step_time_sec": 8.238428468990605, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.097044944763184, "lr": 0.0002, "elapsed_sec": 47830.35564661026, "step_time_sec": 8.23818961303914, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.986833333969116, "lr": 0.0002, "elapsed_sec": 47838.59482264519, "step_time_sec": 8.23898264701711, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9351747035980225, "lr": 0.0002, "elapsed_sec": 47846.83418059349, "step_time_sec": 8.239215351000894, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.922872304916382, "lr": 0.0002, "elapsed_sec": 47855.07107615471, "step_time_sec": 8.236727126000915, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.077457427978516, "lr": 0.0002, "elapsed_sec": 47863.30912232399, "step_time_sec": 8.237847110955045, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.001914024353027, "lr": 0.0002, "elapsed_sec": 47871.54693388939, "step_time_sec": 8.237638330028858, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.018575191497803, "lr": 0.0002, "elapsed_sec": 47879.7857966423, "step_time_sec": 8.238722098991275, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.077467441558838, "lr": 0.0002, "elapsed_sec": 47888.02417254448, "step_time_sec": 8.238206738024019, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.932497978210449, "lr": 0.0002, "elapsed_sec": 47896.26205325127, "step_time_sec": 8.237730625958648, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.990147113800049, "lr": 0.0002, "elapsed_sec": 47904.4997947216, "step_time_sec": 8.237558135995641, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.968677520751953, "lr": 0.0002, "elapsed_sec": 47912.73764848709, "step_time_sec": 8.237701145000756, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.05281925201416, "lr": 0.0002, "elapsed_sec": 47920.97634291649, "step_time_sec": 8.238582890015095, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.982165813446045, "lr": 0.0002, "elapsed_sec": 47929.215544462204, "step_time_sec": 8.238960381015204, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9219295978546143, "lr": 0.0002, "elapsed_sec": 47937.454538583755, "step_time_sec": 8.238882196019404, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7471673488616943, "lr": 0.0002, "elapsed_sec": 47945.69388580322, "step_time_sec": 8.239141061960254, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9345498085021973, "lr": 0.0002, "elapsed_sec": 47953.93090057373, "step_time_sec": 8.236937014036812, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.810926675796509, "lr": 0.0002, "elapsed_sec": 47962.16929149628, "step_time_sec": 8.238178731990047, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9465725421905518, "lr": 0.0002, "elapsed_sec": 47970.40781331062, "step_time_sec": 8.23834333900595, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.152656555175781, "lr": 0.0002, "elapsed_sec": 47978.64543032646, "step_time_sec": 8.23748470499413, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.010437488555908, "lr": 0.0002, "elapsed_sec": 47986.883049964905, "step_time_sec": 8.237435653980356, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.113124847412109, "lr": 0.0002, "elapsed_sec": 47995.12081193924, "step_time_sec": 8.237614680023398, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.078123569488525, "lr": 0.0002, "elapsed_sec": 48003.35946035385, "step_time_sec": 8.238470411044545, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7424964904785156, "lr": 0.0002, "elapsed_sec": 48011.59913563728, "step_time_sec": 8.23951111396309, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9636728763580322, "lr": 0.0002, "elapsed_sec": 48019.83811020851, "step_time_sec": 8.238785800000187, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.994664192199707, "lr": 0.0002, "elapsed_sec": 48028.0768904686, "step_time_sec": 8.238670953025576, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.988847255706787, "lr": 0.0002, "elapsed_sec": 48036.31396126747, "step_time_sec": 8.236874283989891, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.839416980743408, "lr": 0.0002, "elapsed_sec": 48044.55309152603, "step_time_sec": 8.238941163988784, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9955618381500244, "lr": 0.0002, "elapsed_sec": 48052.79172348976, "step_time_sec": 8.238527184003033, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.090468406677246, "lr": 0.0002, "elapsed_sec": 48061.030343294144, "step_time_sec": 8.238420414971188, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.01161003112793, "lr": 0.0002, "elapsed_sec": 48069.27033829689, "step_time_sec": 8.239824571006466, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8467745780944824, "lr": 0.0002, "elapsed_sec": 48077.50806641579, "step_time_sec": 8.237586723000277, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.111824989318848, "lr": 0.0002, "elapsed_sec": 48085.74761009216, "step_time_sec": 8.239334877987858, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.880434513092041, "lr": 0.0002, "elapsed_sec": 48093.98602986336, "step_time_sec": 8.238334507972468, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.069981098175049, "lr": 0.0002, "elapsed_sec": 48102.224848032, "step_time_sec": 8.238610522006638, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9741291999816895, "lr": 0.0002, "elapsed_sec": 48110.464376688, "step_time_sec": 8.239385144028347, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9540605545043945, "lr": 0.0002, "elapsed_sec": 48118.70278811455, "step_time_sec": 8.238210400973912, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9763591289520264, "lr": 0.0002, "elapsed_sec": 48126.94338107109, "step_time_sec": 8.240434498991817, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.002151012420654, "lr": 0.0002, "elapsed_sec": 48135.182891607285, "step_time_sec": 8.239375062985346, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.065115928649902, "lr": 0.0002, "elapsed_sec": 48143.42156243324, "step_time_sec": 8.238467296992894, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9960336685180664, "lr": 0.0002, "elapsed_sec": 48151.66015982628, "step_time_sec": 8.23840960796224, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9593451023101807, "lr": 0.0002, "elapsed_sec": 48159.89772415161, "step_time_sec": 8.237412538961507, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.950002908706665, "lr": 0.0002, "elapsed_sec": 48168.13748264313, "step_time_sec": 8.23959226597799, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.906949281692505, "lr": 0.0002, "elapsed_sec": 48176.37574625015, "step_time_sec": 8.238147280004341, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9883694648742676, "lr": 0.0002, "elapsed_sec": 48184.61373877525, "step_time_sec": 8.23778163298266, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8855037689208984, "lr": 0.0002, "elapsed_sec": 48192.85307741165, "step_time_sec": 8.239185156999156, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.024261474609375, "lr": 0.0002, "elapsed_sec": 48201.08983707428, "step_time_sec": 8.236561944999266, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9787821769714355, "lr": 0.0002, "elapsed_sec": 48209.327984809875, "step_time_sec": 8.238036477996502, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.002213001251221, "lr": 0.0002, "elapsed_sec": 48217.56480550766, "step_time_sec": 8.23666398995556, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9814577102661133, "lr": 0.0002, "elapsed_sec": 48225.8035736084, "step_time_sec": 8.238569782988634, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.931330919265747, "lr": 0.0002, "elapsed_sec": 48234.04280781746, "step_time_sec": 8.239094358985312, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.185442924499512, "lr": 0.0002, "elapsed_sec": 48242.28102993965, "step_time_sec": 8.238029188010842, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.932410717010498, "lr": 0.0002, "elapsed_sec": 48250.51876163483, "step_time_sec": 8.237561965011992, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.007445335388184, "lr": 0.0002, "elapsed_sec": 48258.75573205948, "step_time_sec": 8.236823006998748, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9570980072021484, "lr": 0.0002, "elapsed_sec": 48266.99393200874, "step_time_sec": 8.237996454001404, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8968095779418945, "lr": 0.0002, "elapsed_sec": 48275.23289203644, "step_time_sec": 8.238792847958393, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.105360984802246, "lr": 0.0002, "elapsed_sec": 48283.46969985962, "step_time_sec": 8.236639562004711, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.006598949432373, "lr": 0.0002, "elapsed_sec": 48291.708570718765, "step_time_sec": 8.238703525974415, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.937826156616211, "lr": 0.0002, "elapsed_sec": 48299.94703865051, "step_time_sec": 8.238318459014408, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.085955619812012, "lr": 0.0002, "elapsed_sec": 48308.18606233597, "step_time_sec": 8.238867550040595, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.906390428543091, "lr": 0.0002, "elapsed_sec": 48316.42469263077, "step_time_sec": 8.238472196040675, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.94079327583313, "lr": 0.0002, "elapsed_sec": 48324.6630730629, "step_time_sec": 8.238212208962068, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2381768226623535, "lr": 0.0002, "elapsed_sec": 48332.90163731575, "step_time_sec": 8.238441650988534, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.038821220397949, "lr": 0.0002, "elapsed_sec": 48341.138912439346, "step_time_sec": 8.237084580003284, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.040476322174072, "lr": 0.0002, "elapsed_sec": 48349.37918782234, "step_time_sec": 8.240176462975796, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.029268264770508, "lr": 0.0002, "elapsed_sec": 48357.61791014671, "step_time_sec": 8.238551146001555, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.865792989730835, "lr": 0.0002, "elapsed_sec": 48365.85604929924, "step_time_sec": 8.23793725098949, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.017739295959473, "lr": 0.0002, "elapsed_sec": 48374.09543323517, "step_time_sec": 8.239261350012384, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9738101959228516, "lr": 0.0002, "elapsed_sec": 48382.331857442856, "step_time_sec": 8.236196399957407, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.060883522033691, "lr": 0.0002, "elapsed_sec": 48390.570578575134, "step_time_sec": 8.238584731996525, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.912496328353882, "lr": 0.0002, "elapsed_sec": 48398.80779623985, "step_time_sec": 8.23705425899243, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9649479389190674, "lr": 0.0002, "elapsed_sec": 48407.045323848724, "step_time_sec": 8.23735771002248, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.02716064453125, "lr": 0.0002, "elapsed_sec": 48415.28274130821, "step_time_sec": 8.237279273045715, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.971311330795288, "lr": 0.0002, "elapsed_sec": 48423.52015352249, "step_time_sec": 8.23724155902164, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.974905014038086, "lr": 0.0002, "elapsed_sec": 48431.758746385574, "step_time_sec": 8.238400187983643, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.069695949554443, "lr": 0.0002, "elapsed_sec": 48439.99673271179, "step_time_sec": 8.237797203997616, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9431068897247314, "lr": 0.0002, "elapsed_sec": 48448.23495721817, "step_time_sec": 8.238116207998246, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.024615287780762, "lr": 0.0002, "elapsed_sec": 48456.47490310669, "step_time_sec": 8.23973584297346, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9286904335021973, "lr": 0.0002, "elapsed_sec": 48464.71079134941, "step_time_sec": 8.235731557011604, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9940507411956787, "lr": 0.0002, "elapsed_sec": 48472.94985079765, "step_time_sec": 8.238912195025478, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9542794227600098, "lr": 0.0002, "elapsed_sec": 48481.18722033501, "step_time_sec": 8.237197870039381, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9036667346954346, "lr": 0.0002, "elapsed_sec": 48489.42533445358, "step_time_sec": 8.23794108000584, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9553844928741455, "lr": 0.0002, "elapsed_sec": 48497.66367292404, "step_time_sec": 8.238201740023214, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.000990867614746, "lr": 0.0002, "elapsed_sec": 48505.90074110031, "step_time_sec": 8.236866994993761, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.009284019470215, "lr": 0.0002, "elapsed_sec": 48514.139191389084, "step_time_sec": 8.238344002980739, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9147517681121826, "lr": 0.0002, "elapsed_sec": 48522.37580895424, "step_time_sec": 8.236382753995713, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.040948390960693, "lr": 0.0002, "elapsed_sec": 48530.613911151886, "step_time_sec": 8.238001749035902, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8786535263061523, "lr": 0.0002, "elapsed_sec": 48538.85247993469, "step_time_sec": 8.238399572030175, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.892502784729004, "lr": 0.0002, "elapsed_sec": 48547.09155845642, "step_time_sec": 8.23892289295327, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.841339588165283, "lr": 0.0002, "elapsed_sec": 48555.3294904232, "step_time_sec": 8.237756670976523, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9021103382110596, "lr": 0.0002, "elapsed_sec": 48563.566612005234, "step_time_sec": 8.236933239968494, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.072285175323486, "lr": 0.0002, "elapsed_sec": 48571.804718494415, "step_time_sec": 8.237961261998862, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.942033052444458, "lr": 0.0002, "elapsed_sec": 48580.04192876816, "step_time_sec": 8.237066539993975, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.064085006713867, "lr": 0.0002, "elapsed_sec": 48588.28173780441, "step_time_sec": 8.239629646996036, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.989964723587036, "lr": 0.0002, "elapsed_sec": 48596.521210193634, "step_time_sec": 8.239345851005055, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.938528537750244, "lr": 0.0002, "elapsed_sec": 48604.758009672165, "step_time_sec": 8.236543422972318, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9875335693359375, "lr": 0.0002, "elapsed_sec": 48612.996385097504, "step_time_sec": 8.238271380017977, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.835489511489868, "lr": 0.0002, "elapsed_sec": 48621.23389649391, "step_time_sec": 8.237307030998636, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.920576810836792, "lr": 0.0002, "elapsed_sec": 48629.47363901138, "step_time_sec": 8.239561241003685, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.983367919921875, "lr": 0.0002, "elapsed_sec": 48637.71119594574, "step_time_sec": 8.23741008801153, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9772441387176514, "lr": 0.0002, "elapsed_sec": 48645.94983172417, "step_time_sec": 8.238468960975297, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.058597564697266, "lr": 0.0002, "elapsed_sec": 48654.18848514557, "step_time_sec": 8.238522508007009, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.000033855438232, "lr": 0.0002, "elapsed_sec": 48662.427026987076, "step_time_sec": 8.238324997015297, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.818005323410034, "lr": 0.0002, "elapsed_sec": 48670.66646003723, "step_time_sec": 8.239279716974124, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9996049404144287, "lr": 0.0002, "elapsed_sec": 48678.90571141243, "step_time_sec": 8.239064057997894, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.035641670227051, "lr": 0.0002, "elapsed_sec": 48687.14497208595, "step_time_sec": 8.239119506964926, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.028597354888916, "lr": 0.0002, "elapsed_sec": 48695.38339352608, "step_time_sec": 8.23825528100133, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9317429065704346, "lr": 0.0002, "elapsed_sec": 48703.62069153786, "step_time_sec": 8.237136787036434, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9775633811950684, "lr": 0.0002, "elapsed_sec": 48711.859907627106, "step_time_sec": 8.239044448011555, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.04594087600708, "lr": 0.0002, "elapsed_sec": 48720.10016870499, "step_time_sec": 8.240101676958147, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9717066287994385, "lr": 0.0002, "elapsed_sec": 48728.338626623154, "step_time_sec": 8.238312947039958, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.010009288787842, "lr": 0.0002, "elapsed_sec": 48736.57713294029, "step_time_sec": 8.238321887038182, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.087745189666748, "lr": 0.0002, "elapsed_sec": 48744.81472039223, "step_time_sec": 8.23745754500851, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.957772731781006, "lr": 0.0002, "elapsed_sec": 48753.05424952507, "step_time_sec": 8.23934917198494, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.014848709106445, "lr": 0.0002, "elapsed_sec": 48761.292924165726, "step_time_sec": 8.238468892988749, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.056085109710693, "lr": 0.0002, "elapsed_sec": 48769.5304312706, "step_time_sec": 8.237337992002722, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.07775354385376, "lr": 0.0002, "elapsed_sec": 48777.770022153854, "step_time_sec": 8.23947892000433, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.034864902496338, "lr": 0.0002, "elapsed_sec": 48786.00728178024, "step_time_sec": 8.237090023991186, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.054718017578125, "lr": 0.0002, "elapsed_sec": 48794.24553823471, "step_time_sec": 8.238084738957696, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.940046787261963, "lr": 0.0002, "elapsed_sec": 48802.48441410065, "step_time_sec": 8.238704885006882, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8864264488220215, "lr": 0.0002, "elapsed_sec": 48810.723902225494, "step_time_sec": 8.239364259003196, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.059573173522949, "lr": 0.0002, "elapsed_sec": 48818.96205663681, "step_time_sec": 8.237978423014283, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.103692531585693, "lr": 0.0002, "elapsed_sec": 48827.19929242134, "step_time_sec": 8.237067593028769, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9344136714935303, "lr": 0.0002, "elapsed_sec": 48835.438794374466, "step_time_sec": 8.239318654988892, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8762474060058594, "lr": 0.0002, "elapsed_sec": 48843.67729306221, "step_time_sec": 8.238365885044914, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.962402582168579, "lr": 0.0002, "elapsed_sec": 48851.91505217552, "step_time_sec": 8.23761640000157, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.037371635437012, "lr": 0.0002, "elapsed_sec": 48860.15462064743, "step_time_sec": 8.239367123984266, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8962557315826416, "lr": 0.0002, "elapsed_sec": 48868.39114713669, "step_time_sec": 8.23640252102632, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.065304756164551, "lr": 0.0002, "elapsed_sec": 48876.629885435104, "step_time_sec": 8.238525705004577, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.978369951248169, "lr": 0.0002, "elapsed_sec": 48884.86733865738, "step_time_sec": 8.23728104797192, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.036038875579834, "lr": 0.0002, "elapsed_sec": 48893.10726475716, "step_time_sec": 8.239815506967716, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.114806652069092, "lr": 0.0002, "elapsed_sec": 48901.34498882294, "step_time_sec": 8.237531695980579, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.11598539352417, "lr": 0.0002, "elapsed_sec": 48909.58231830597, "step_time_sec": 8.237172093009576, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9078917503356934, "lr": 0.0002, "elapsed_sec": 48917.821033000946, "step_time_sec": 8.238541106984485, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.039956569671631, "lr": 0.0002, "elapsed_sec": 48926.05741071701, "step_time_sec": 8.236263386963401, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.799381732940674, "lr": 0.0002, "elapsed_sec": 48934.296345710754, "step_time_sec": 8.2387418349972, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.115947723388672, "lr": 0.0002, "elapsed_sec": 48942.53410077095, "step_time_sec": 8.237643104977906, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.153665542602539, "lr": 0.0002, "elapsed_sec": 48950.77243185043, "step_time_sec": 8.238126579963136, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0400471687316895, "lr": 0.0002, "elapsed_sec": 48959.01203632355, "step_time_sec": 8.239458731026389, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0273051261901855, "lr": 0.0002, "elapsed_sec": 48967.24946808815, "step_time_sec": 8.237246422970202, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.967064142227173, "lr": 0.0002, "elapsed_sec": 48975.48764705658, "step_time_sec": 8.238002989965025, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9367220401763916, "lr": 0.0002, "elapsed_sec": 48983.72585082054, "step_time_sec": 8.238069867016748, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.80257511138916, "lr": 0.0002, "elapsed_sec": 48991.96429133415, "step_time_sec": 8.238280892022885, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7797932624816895, "lr": 0.0002, "elapsed_sec": 49000.20301556587, "step_time_sec": 8.238517244986724, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8850624561309814, "lr": 0.0002, "elapsed_sec": 49008.43976068497, "step_time_sec": 8.236627092002891, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.023852348327637, "lr": 0.0002, "elapsed_sec": 49016.67935323715, "step_time_sec": 8.239413378003519, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9811975955963135, "lr": 0.0002, "elapsed_sec": 49024.917086839676, "step_time_sec": 8.23757066699909, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9084596633911133, "lr": 0.0002, "elapsed_sec": 49033.15378022194, "step_time_sec": 8.236562862992287, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.874173402786255, "lr": 0.0002, "elapsed_sec": 49041.392844200134, "step_time_sec": 8.238904306956101, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.035162448883057, "lr": 0.0002, "elapsed_sec": 49049.630202531815, "step_time_sec": 8.237182573997416, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9003958702087402, "lr": 0.0002, "elapsed_sec": 49057.86741447449, "step_time_sec": 8.23704998800531, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9929425716400146, "lr": 0.0002, "elapsed_sec": 49066.104825258255, "step_time_sec": 8.23725624801591, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9423983097076416, "lr": 0.0002, "elapsed_sec": 49074.344403743744, "step_time_sec": 8.239455533970613, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.079373836517334, "lr": 0.0002, "elapsed_sec": 49082.58316612244, "step_time_sec": 8.238512173003983, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.934952735900879, "lr": 0.0002, "elapsed_sec": 49090.8215239048, "step_time_sec": 8.238193055032752, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9239492416381836, "lr": 0.0002, "elapsed_sec": 49099.060906887054, "step_time_sec": 8.239235174027272, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.104811668395996, "lr": 0.0002, "elapsed_sec": 49107.2985458374, "step_time_sec": 8.23753847903572, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.904597282409668, "lr": 0.0002, "elapsed_sec": 49115.53747057915, "step_time_sec": 8.23867223400157, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.969313144683838, "lr": 0.0002, "elapsed_sec": 49123.77570748329, "step_time_sec": 8.238132803002372, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.02488899230957, "lr": 0.0002, "elapsed_sec": 49132.013145685196, "step_time_sec": 8.237219335045666, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.021864891052246, "lr": 0.0002, "elapsed_sec": 49140.25261878967, "step_time_sec": 8.23931585799437, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9463765621185303, "lr": 0.0002, "elapsed_sec": 49148.48961162567, "step_time_sec": 8.236869823012967, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.942941665649414, "lr": 0.0002, "elapsed_sec": 49156.728271484375, "step_time_sec": 8.238443853973877, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.084433555603027, "lr": 0.0002, "elapsed_sec": 49164.96557474136, "step_time_sec": 8.237153725000098, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.017011642456055, "lr": 0.0002, "elapsed_sec": 49173.20472240448, "step_time_sec": 8.239025568997022, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.092900276184082, "lr": 0.0002, "elapsed_sec": 49181.443640232086, "step_time_sec": 8.23873907001689, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.174548149108887, "lr": 0.0002, "elapsed_sec": 49189.68055391312, "step_time_sec": 8.236726957024075, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.059939861297607, "lr": 0.0002, "elapsed_sec": 49197.92078948021, "step_time_sec": 8.240046716993675, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.120079517364502, "lr": 0.0002, "elapsed_sec": 49206.15843987465, "step_time_sec": 8.237511051003821, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.061838626861572, "lr": 0.0002, "elapsed_sec": 49214.396320581436, "step_time_sec": 8.23769328003982, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1995368003845215, "lr": 0.0002, "elapsed_sec": 49222.63318014145, "step_time_sec": 8.236716699961107, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9593348503112793, "lr": 0.0002, "elapsed_sec": 49230.87126302719, "step_time_sec": 8.237918111961335, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.422661781311035, "lr": 0.0002, "elapsed_sec": 49239.10915637016, "step_time_sec": 8.237723494006786, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.016028881072998, "lr": 0.0002, "elapsed_sec": 49247.34644269943, "step_time_sec": 8.237127661006525, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.079434394836426, "lr": 0.0002, "elapsed_sec": 49255.58537507057, "step_time_sec": 8.238727561023552, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.097416877746582, "lr": 0.0002, "elapsed_sec": 49263.824719667435, "step_time_sec": 8.239252711995505, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.100216865539551, "lr": 0.0002, "elapsed_sec": 49272.06262969971, "step_time_sec": 8.237699291028548, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.011446475982666, "lr": 0.0002, "elapsed_sec": 49280.30118870735, "step_time_sec": 8.238382225041278, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.03135871887207, "lr": 0.0002, "elapsed_sec": 49288.537523031235, "step_time_sec": 8.23620015400229, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1347270011901855, "lr": 0.0002, "elapsed_sec": 49296.773696899414, "step_time_sec": 8.235956047021318, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1192145347595215, "lr": 0.0002, "elapsed_sec": 49305.011291742325, "step_time_sec": 8.237441480974667, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.064297199249268, "lr": 0.0002, "elapsed_sec": 49313.25006318092, "step_time_sec": 8.238609394000378, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9769277572631836, "lr": 0.0002, "elapsed_sec": 49321.49101352692, "step_time_sec": 8.2408068459481, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0046491622924805, "lr": 0.0002, "elapsed_sec": 49329.72955703735, "step_time_sec": 8.238376975001302, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.932407855987549, "lr": 0.0002, "elapsed_sec": 49337.96937036514, "step_time_sec": 8.239628173992969, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.947444200515747, "lr": 0.0002, "elapsed_sec": 49346.20826435089, "step_time_sec": 8.23875737003982, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.030518531799316, "lr": 0.0002, "elapsed_sec": 49354.44802904129, "step_time_sec": 8.23960203898605, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.057743072509766, "lr": 0.0002, "elapsed_sec": 49362.68802380562, "step_time_sec": 8.239834058040287, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.000481605529785, "lr": 0.0002, "elapsed_sec": 49370.925626039505, "step_time_sec": 8.237458440999035, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.133023738861084, "lr": 0.0002, "elapsed_sec": 49379.16512012482, "step_time_sec": 8.239350698015187, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.016977787017822, "lr": 0.0002, "elapsed_sec": 49387.404027462006, "step_time_sec": 8.238717257976532, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0319695472717285, "lr": 0.0002, "elapsed_sec": 49395.64274740219, "step_time_sec": 8.238524109008722, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.944268226623535, "lr": 0.0002, "elapsed_sec": 49403.882855176926, "step_time_sec": 8.240000621008221, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0920491218566895, "lr": 0.0002, "elapsed_sec": 49412.119983673096, "step_time_sec": 8.236966253025457, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.869988441467285, "lr": 0.0002, "elapsed_sec": 49420.35882997513, "step_time_sec": 8.238610933010932, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9413702487945557, "lr": 0.0002, "elapsed_sec": 49428.59812831879, "step_time_sec": 8.239197468967177, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.071742534637451, "lr": 0.0002, "elapsed_sec": 49436.83667469025, "step_time_sec": 8.238333482993767, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.032363414764404, "lr": 0.0002, "elapsed_sec": 49445.07561302185, "step_time_sec": 8.238760154985357, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9530293941497803, "lr": 0.0002, "elapsed_sec": 49453.312883377075, "step_time_sec": 8.237108161963988, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.899461269378662, "lr": 0.0002, "elapsed_sec": 49461.55089879036, "step_time_sec": 8.237878591986373, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.01154899597168, "lr": 0.0002, "elapsed_sec": 49469.78931427002, "step_time_sec": 8.238262792991009, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.100590229034424, "lr": 0.0002, "elapsed_sec": 49478.029047727585, "step_time_sec": 8.239527394995093, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.151957035064697, "lr": 0.0002, "elapsed_sec": 49486.26782083511, "step_time_sec": 8.23862805502722, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.102319240570068, "lr": 0.0002, "elapsed_sec": 49494.50532126427, "step_time_sec": 8.237377231998835, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.112441062927246, "lr": 0.0002, "elapsed_sec": 49502.74464058876, "step_time_sec": 8.23909560602624, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.056798458099365, "lr": 0.0002, "elapsed_sec": 49510.98429584503, "step_time_sec": 8.239489307976328, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8676180839538574, "lr": 0.0002, "elapsed_sec": 49519.222043037415, "step_time_sec": 8.237586679984815, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9637391567230225, "lr": 0.0002, "elapsed_sec": 49527.461261987686, "step_time_sec": 8.239119436999317, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.825775623321533, "lr": 0.0002, "elapsed_sec": 49535.69789671898, "step_time_sec": 8.236404467956163, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.272625923156738, "lr": 0.0002, "elapsed_sec": 49543.9359061718, "step_time_sec": 8.23789756197948, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.949110269546509, "lr": 0.0002, "elapsed_sec": 49552.17331123352, "step_time_sec": 8.237223959993571, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.990144729614258, "lr": 0.0002, "elapsed_sec": 49560.41457629204, "step_time_sec": 8.241043417016044, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.133775234222412, "lr": 0.0002, "elapsed_sec": 49568.65283536911, "step_time_sec": 8.238143978989683, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.982320547103882, "lr": 0.0002, "elapsed_sec": 49576.89121174812, "step_time_sec": 8.238191548967734, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.97294545173645, "lr": 0.0002, "elapsed_sec": 49585.13103222847, "step_time_sec": 8.239638414001092, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.031068325042725, "lr": 0.0002, "elapsed_sec": 49593.3711335659, "step_time_sec": 8.239996675984003, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9737136363983154, "lr": 0.0002, "elapsed_sec": 49601.61092185974, "step_time_sec": 8.23957589402562, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.965296506881714, "lr": 0.0002, "elapsed_sec": 49609.85147547722, "step_time_sec": 8.240382993011735, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.918896198272705, "lr": 0.0002, "elapsed_sec": 49618.088881492615, "step_time_sec": 8.237276689033024, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.059528350830078, "lr": 0.0002, "elapsed_sec": 49626.329092502594, "step_time_sec": 8.240063113975339, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.056776523590088, "lr": 0.0002, "elapsed_sec": 49634.567608594894, "step_time_sec": 8.238287284038961, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9175472259521484, "lr": 0.0002, "elapsed_sec": 49642.80572605133, "step_time_sec": 8.237995390954893, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9829282760620117, "lr": 0.0002, "elapsed_sec": 49651.045094013214, "step_time_sec": 8.239170826040208, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.025261878967285, "lr": 0.0002, "elapsed_sec": 49659.28274726868, "step_time_sec": 8.237494895001873, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9005675315856934, "lr": 0.0002, "elapsed_sec": 49667.52153849602, "step_time_sec": 8.238614411035087, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0329155921936035, "lr": 0.0002, "elapsed_sec": 49675.76097512245, "step_time_sec": 8.239270120044239, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0154709815979, "lr": 0.0002, "elapsed_sec": 49684.000655412674, "step_time_sec": 8.239582572015934, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0036702156066895, "lr": 0.0002, "elapsed_sec": 49692.24128818512, "step_time_sec": 8.240450615994632, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8721723556518555, "lr": 0.0002, "elapsed_sec": 49700.47894334793, "step_time_sec": 8.23747752798954, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0544047355651855, "lr": 0.0002, "elapsed_sec": 49708.71852326393, "step_time_sec": 8.239431334019173, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9837095737457275, "lr": 0.0002, "elapsed_sec": 49716.95767688751, "step_time_sec": 8.238975854008459, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.986783266067505, "lr": 0.0002, "elapsed_sec": 49725.196149110794, "step_time_sec": 8.238305256003514, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.101683139801025, "lr": 0.0002, "elapsed_sec": 49733.4354159832, "step_time_sec": 8.239137762982864, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9576632976531982, "lr": 0.0002, "elapsed_sec": 49741.675998687744, "step_time_sec": 8.2404017560184, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9333889484405518, "lr": 0.0002, "elapsed_sec": 49749.91932177544, "step_time_sec": 8.24314791301731, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.913879632949829, "lr": 0.0002, "elapsed_sec": 49758.1599445343, "step_time_sec": 8.240423233015463, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.023718357086182, "lr": 0.0002, "elapsed_sec": 49766.3987300396, "step_time_sec": 8.238646756974049, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.902930498123169, "lr": 0.0002, "elapsed_sec": 49774.63864374161, "step_time_sec": 8.239711183006875, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8729407787323, "lr": 0.0002, "elapsed_sec": 49782.87884449959, "step_time_sec": 8.240087912010495, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9807751178741455, "lr": 0.0002, "elapsed_sec": 49791.116394758224, "step_time_sec": 8.237347325019073, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.006364822387695, "lr": 0.0002, "elapsed_sec": 49799.35574507713, "step_time_sec": 8.239229423983488, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8270418643951416, "lr": 0.0002, "elapsed_sec": 49807.59481883049, "step_time_sec": 8.238865210034419, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9551074504852295, "lr": 0.0002, "elapsed_sec": 49815.83417534828, "step_time_sec": 8.23919933300931, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.054593086242676, "lr": 0.0002, "elapsed_sec": 49824.07408618927, "step_time_sec": 8.239777601032984, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.048666954040527, "lr": 0.0002, "elapsed_sec": 49832.311729192734, "step_time_sec": 8.237430913955905, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.970790147781372, "lr": 0.0002, "elapsed_sec": 49840.55404543877, "step_time_sec": 8.242136385000776, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.956267833709717, "lr": 0.0002, "elapsed_sec": 49848.79309248924, "step_time_sec": 8.238923898956273, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9469540119171143, "lr": 0.0002, "elapsed_sec": 49857.03290963173, "step_time_sec": 51.727074677008204, "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": 0.9839632519870065, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 6001, "loss": 4.057270526885986, "lr": 0.0002, "elapsed_sec": 49908.76573085785, "step_time_sec": 8.245285100012552, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.159326553344727, "lr": 0.0002, "elapsed_sec": 49917.00468444824, "step_time_sec": 8.238796213001478, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0757246017456055, "lr": 0.0002, "elapsed_sec": 49925.24213767052, "step_time_sec": 8.237240932008717, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.04993200302124, "lr": 0.0002, "elapsed_sec": 49933.48061919212, "step_time_sec": 8.238371216983069, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8123505115509033, "lr": 0.0002, "elapsed_sec": 49941.71971201897, "step_time_sec": 8.238888350024354, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.977602243423462, "lr": 0.0002, "elapsed_sec": 49949.95901989937, "step_time_sec": 8.239130842033774, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.978914737701416, "lr": 0.0002, "elapsed_sec": 49958.197637319565, "step_time_sec": 8.23847898497479, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.89188289642334, "lr": 0.0002, "elapsed_sec": 49966.43536448479, "step_time_sec": 8.237569343997166, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.901660203933716, "lr": 0.0002, "elapsed_sec": 49974.67477297783, "step_time_sec": 8.23924937704578, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0289082527160645, "lr": 0.0002, "elapsed_sec": 49982.914268016815, "step_time_sec": 8.239373170014005, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.858839273452759, "lr": 0.0002, "elapsed_sec": 49991.153079748154, "step_time_sec": 8.238601433986332, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0310587882995605, "lr": 0.0002, "elapsed_sec": 49999.39241170883, "step_time_sec": 8.239200437034015, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9270620346069336, "lr": 0.0002, "elapsed_sec": 50007.62903523445, "step_time_sec": 8.236426176968962, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.037565231323242, "lr": 0.0002, "elapsed_sec": 50015.868525505066, "step_time_sec": 8.239370182040147, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.103697299957275, "lr": 0.0002, "elapsed_sec": 50024.107421875, "step_time_sec": 8.238748167990707, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.941545009613037, "lr": 0.0002, "elapsed_sec": 50032.345643520355, "step_time_sec": 8.238013238005806, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.920921802520752, "lr": 0.0002, "elapsed_sec": 50040.586735486984, "step_time_sec": 8.240894797025248, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.120968818664551, "lr": 0.0002, "elapsed_sec": 50048.825756788254, "step_time_sec": 8.238869938009884, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.111947059631348, "lr": 0.0002, "elapsed_sec": 50057.0669260025, "step_time_sec": 8.241053722042125, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.006844520568848, "lr": 0.0002, "elapsed_sec": 50065.30530643463, "step_time_sec": 8.238182369968854, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.006898880004883, "lr": 0.0002, "elapsed_sec": 50073.54507637024, "step_time_sec": 8.23963977699168, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.053910732269287, "lr": 0.0002, "elapsed_sec": 50081.785945653915, "step_time_sec": 8.2406759459991, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.028522968292236, "lr": 0.0002, "elapsed_sec": 50090.02538228035, "step_time_sec": 8.239286174008157, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9962093830108643, "lr": 0.0002, "elapsed_sec": 50098.26531410217, "step_time_sec": 8.239804331969935, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.960968255996704, "lr": 0.0002, "elapsed_sec": 50106.506296634674, "step_time_sec": 8.240835710952524, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.083446979522705, "lr": 0.0002, "elapsed_sec": 50114.745015621185, "step_time_sec": 8.238484785018954, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.022958755493164, "lr": 0.0002, "elapsed_sec": 50122.98510789871, "step_time_sec": 8.239942243031692, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.028339862823486, "lr": 0.0002, "elapsed_sec": 50131.22360610962, "step_time_sec": 8.238328178995289, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.017040729522705, "lr": 0.0002, "elapsed_sec": 50139.462270736694, "step_time_sec": 8.238487905007787, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.004744529724121, "lr": 0.0002, "elapsed_sec": 50147.70209670067, "step_time_sec": 8.239660036982968, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9387974739074707, "lr": 0.0002, "elapsed_sec": 50155.94132256508, "step_time_sec": 8.239095932978671, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8196496963500977, "lr": 0.0002, "elapsed_sec": 50164.18218970299, "step_time_sec": 8.24074026098242, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.018322467803955, "lr": 0.0002, "elapsed_sec": 50172.42307496071, "step_time_sec": 8.240679359005298, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.043954849243164, "lr": 0.0002, "elapsed_sec": 50180.661417245865, "step_time_sec": 8.238161187968217, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.043300628662109, "lr": 0.0002, "elapsed_sec": 50188.90161323547, "step_time_sec": 8.240078780043405, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.86665678024292, "lr": 0.0002, "elapsed_sec": 50197.14149069786, "step_time_sec": 8.239713950024452, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7360410690307617, "lr": 0.0002, "elapsed_sec": 50205.38078594208, "step_time_sec": 8.239133024995681, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.965510606765747, "lr": 0.0002, "elapsed_sec": 50213.62058329582, "step_time_sec": 8.239631066040602, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.992069959640503, "lr": 0.0002, "elapsed_sec": 50221.859214782715, "step_time_sec": 8.238445611030329, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9050707817077637, "lr": 0.0002, "elapsed_sec": 50230.09953427315, "step_time_sec": 8.24015656800475, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.146146297454834, "lr": 0.0002, "elapsed_sec": 50238.338509082794, "step_time_sec": 8.238814552954864, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9551618099212646, "lr": 0.0002, "elapsed_sec": 50246.57718682289, "step_time_sec": 8.23852974694455, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9521641731262207, "lr": 0.0002, "elapsed_sec": 50254.81803011894, "step_time_sec": 8.240643750003073, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.016513347625732, "lr": 0.0002, "elapsed_sec": 50263.058169841766, "step_time_sec": 8.239990197005682, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.056820869445801, "lr": 0.0002, "elapsed_sec": 50271.29683279991, "step_time_sec": 8.238500831008423, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.043396949768066, "lr": 0.0002, "elapsed_sec": 50279.536541461945, "step_time_sec": 8.23953782004537, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9148342609405518, "lr": 0.0002, "elapsed_sec": 50287.77418041229, "step_time_sec": 8.237511332030408, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.083773612976074, "lr": 0.0002, "elapsed_sec": 50296.01523423195, "step_time_sec": 8.24087584298104, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.00526762008667, "lr": 0.0002, "elapsed_sec": 50304.254271268845, "step_time_sec": 8.23892637703102, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.017520904541016, "lr": 0.0002, "elapsed_sec": 50312.49228429794, "step_time_sec": 8.237820644979365, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8495075702667236, "lr": 0.0002, "elapsed_sec": 50320.7316839695, "step_time_sec": 8.239196646027267, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.017087459564209, "lr": 0.0002, "elapsed_sec": 50328.9693646431, "step_time_sec": 8.237544392992277, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9817280769348145, "lr": 0.0002, "elapsed_sec": 50337.21209216118, "step_time_sec": 8.242580702004489, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.93608021736145, "lr": 0.0002, "elapsed_sec": 50345.45256590843, "step_time_sec": 8.240264079999179, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.019742488861084, "lr": 0.0002, "elapsed_sec": 50353.690578222275, "step_time_sec": 8.237879934022203, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.050694942474365, "lr": 0.0002, "elapsed_sec": 50361.930543899536, "step_time_sec": 8.239775939029641, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.043130874633789, "lr": 0.0002, "elapsed_sec": 50370.16868638992, "step_time_sec": 8.23798373300815, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.121400833129883, "lr": 0.0002, "elapsed_sec": 50378.40761423111, "step_time_sec": 8.238775995036121, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.02437686920166, "lr": 0.0002, "elapsed_sec": 50386.64632105827, "step_time_sec": 8.23853400797816, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.150078296661377, "lr": 0.0002, "elapsed_sec": 50394.88401174545, "step_time_sec": 8.23752213298576, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9638943672180176, "lr": 0.0002, "elapsed_sec": 50403.12466883659, "step_time_sec": 8.24050688999705, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.030642032623291, "lr": 0.0002, "elapsed_sec": 50411.364263772964, "step_time_sec": 8.239414442970883, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.915131092071533, "lr": 0.0002, "elapsed_sec": 50419.60279083252, "step_time_sec": 8.238409026002046, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.038399696350098, "lr": 0.0002, "elapsed_sec": 50427.84367823601, "step_time_sec": 8.240702692011837, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.948032855987549, "lr": 0.0002, "elapsed_sec": 50436.08167719841, "step_time_sec": 8.237840415036771, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.995117425918579, "lr": 0.0002, "elapsed_sec": 50444.321945905685, "step_time_sec": 8.24013059400022, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.923285961151123, "lr": 0.0002, "elapsed_sec": 50452.56150889397, "step_time_sec": 8.23938604700379, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.126391887664795, "lr": 0.0002, "elapsed_sec": 50460.80034518242, "step_time_sec": 8.238675406959374, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9796829223632812, "lr": 0.0002, "elapsed_sec": 50469.04093456268, "step_time_sec": 8.240445572999306, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.977579116821289, "lr": 0.0002, "elapsed_sec": 50477.280889987946, "step_time_sec": 8.239790298976004, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.891460418701172, "lr": 0.0002, "elapsed_sec": 50485.520275354385, "step_time_sec": 8.23920127202291, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.010936737060547, "lr": 0.0002, "elapsed_sec": 50493.75936412811, "step_time_sec": 8.238897952018306, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.050721168518066, "lr": 0.0002, "elapsed_sec": 50501.997811317444, "step_time_sec": 8.238289374043234, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8805668354034424, "lr": 0.0002, "elapsed_sec": 50510.23803806305, "step_time_sec": 8.240044117963407, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9831244945526123, "lr": 0.0002, "elapsed_sec": 50518.47716879845, "step_time_sec": 8.239007088996004, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9431540966033936, "lr": 0.0002, "elapsed_sec": 50526.71686887741, "step_time_sec": 8.239550252037589, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9254848957061768, "lr": 0.0002, "elapsed_sec": 50534.95691180229, "step_time_sec": 8.239884210983291, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9211912155151367, "lr": 0.0002, "elapsed_sec": 50543.196469545364, "step_time_sec": 8.239370765979402, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.954724073410034, "lr": 0.0002, "elapsed_sec": 50551.43659424782, "step_time_sec": 8.239988305023871, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9915175437927246, "lr": 0.0002, "elapsed_sec": 50559.67675065994, "step_time_sec": 8.23997930099722, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.03182315826416, "lr": 0.0002, "elapsed_sec": 50567.91600108147, "step_time_sec": 8.239040539017878, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.010885715484619, "lr": 0.0002, "elapsed_sec": 50576.15757036209, "step_time_sec": 8.2414493309916, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9397008419036865, "lr": 0.0002, "elapsed_sec": 50584.3976354599, "step_time_sec": 8.239883582980838, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8115179538726807, "lr": 0.0002, "elapsed_sec": 50592.636704683304, "step_time_sec": 8.238879739015829, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9815726280212402, "lr": 0.0002, "elapsed_sec": 50600.87734794617, "step_time_sec": 8.24046325701056, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.117669582366943, "lr": 0.0002, "elapsed_sec": 50609.11574292183, "step_time_sec": 8.238285753992386, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.921984910964966, "lr": 0.0002, "elapsed_sec": 50617.3563811779, "step_time_sec": 8.240478176041506, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.976637363433838, "lr": 0.0002, "elapsed_sec": 50625.59590101242, "step_time_sec": 8.23931650602026, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.126977920532227, "lr": 0.0002, "elapsed_sec": 50633.83425951004, "step_time_sec": 8.238249723042827, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.850403070449829, "lr": 0.0002, "elapsed_sec": 50642.07594847679, "step_time_sec": 8.24145645799581, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9488296508789062, "lr": 0.0002, "elapsed_sec": 50650.31540846825, "step_time_sec": 8.239289276010823, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9271862506866455, "lr": 0.0002, "elapsed_sec": 50658.55516290665, "step_time_sec": 8.239608741016127, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9086670875549316, "lr": 0.0002, "elapsed_sec": 50666.794523239136, "step_time_sec": 8.239179698983207, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.962968587875366, "lr": 0.0002, "elapsed_sec": 50675.0342540741, "step_time_sec": 8.239571880025323, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.035634517669678, "lr": 0.0002, "elapsed_sec": 50683.27556371689, "step_time_sec": 8.241131504997611, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.831439971923828, "lr": 0.0002, "elapsed_sec": 50691.51795005798, "step_time_sec": 8.242220347980037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.97092342376709, "lr": 0.0002, "elapsed_sec": 50699.7562828064, "step_time_sec": 8.238215880002826, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.049100875854492, "lr": 0.0002, "elapsed_sec": 50707.996134996414, "step_time_sec": 8.239652298972942, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1194682121276855, "lr": 0.0002, "elapsed_sec": 50716.23540139198, "step_time_sec": 8.239062638022006, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7824230194091797, "lr": 0.0002, "elapsed_sec": 50724.474100112915, "step_time_sec": 8.238552882976364, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8965601921081543, "lr": 0.0002, "elapsed_sec": 50732.7138607502, "step_time_sec": 8.239570598991122, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0157270431518555, "lr": 0.0002, "elapsed_sec": 50740.95287132263, "step_time_sec": 8.238815055054147, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.166753768920898, "lr": 0.0002, "elapsed_sec": 50749.19381117821, "step_time_sec": 8.240761784021743, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.823092222213745, "lr": 0.0002, "elapsed_sec": 50757.43424677849, "step_time_sec": 8.240296475007199, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.895190477371216, "lr": 0.0002, "elapsed_sec": 50765.67187714577, "step_time_sec": 8.237505782977678, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.001948833465576, "lr": 0.0002, "elapsed_sec": 50773.912232875824, "step_time_sec": 8.24021779600298, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.90262508392334, "lr": 0.0002, "elapsed_sec": 50782.151641607285, "step_time_sec": 8.239199406991247, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.111177921295166, "lr": 0.0002, "elapsed_sec": 50790.39034986496, "step_time_sec": 8.23853427299764, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.152266979217529, "lr": 0.0002, "elapsed_sec": 50798.628964185715, "step_time_sec": 8.238473978010006, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.089871883392334, "lr": 0.0002, "elapsed_sec": 50806.86606788635, "step_time_sec": 8.236932800034992, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0113677978515625, "lr": 0.0002, "elapsed_sec": 50815.106760025024, "step_time_sec": 8.240508642978966, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.904430627822876, "lr": 0.0002, "elapsed_sec": 50823.34612607956, "step_time_sec": 8.239232941006776, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.002968788146973, "lr": 0.0002, "elapsed_sec": 50831.58386206627, "step_time_sec": 8.237549801007845, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.180135726928711, "lr": 0.0002, "elapsed_sec": 50839.82388949394, "step_time_sec": 8.239894750004169, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.012788772583008, "lr": 0.0002, "elapsed_sec": 50848.061666965485, "step_time_sec": 8.23758270603139, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0454301834106445, "lr": 0.0002, "elapsed_sec": 50856.30246424675, "step_time_sec": 8.240641768963542, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.978940010070801, "lr": 0.0002, "elapsed_sec": 50864.541352272034, "step_time_sec": 8.23875403002603, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.008322238922119, "lr": 0.0002, "elapsed_sec": 50872.77909564972, "step_time_sec": 8.237576163024642, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.001063823699951, "lr": 0.0002, "elapsed_sec": 50881.02154421806, "step_time_sec": 8.242297908000182, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.931368827819824, "lr": 0.0002, "elapsed_sec": 50889.261105298996, "step_time_sec": 8.239376829995308, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.987190008163452, "lr": 0.0002, "elapsed_sec": 50897.50014472008, "step_time_sec": 8.238888577034231, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.015636920928955, "lr": 0.0002, "elapsed_sec": 50905.74137926102, "step_time_sec": 8.241082519001793, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9548633098602295, "lr": 0.0002, "elapsed_sec": 50913.97970414162, "step_time_sec": 8.238168709969614, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8769683837890625, "lr": 0.0002, "elapsed_sec": 50922.22159910202, "step_time_sec": 8.241662193031516, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.002112865447998, "lr": 0.0002, "elapsed_sec": 50930.460795879364, "step_time_sec": 8.239079744031187, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.997910976409912, "lr": 0.0002, "elapsed_sec": 50938.70168709755, "step_time_sec": 8.240718741959427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.904869556427002, "lr": 0.0002, "elapsed_sec": 50946.945385694504, "step_time_sec": 8.243574168009218, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.027311325073242, "lr": 0.0002, "elapsed_sec": 50955.18491125107, "step_time_sec": 8.239321467000991, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.01260232925415, "lr": 0.0002, "elapsed_sec": 50963.423285245895, "step_time_sec": 8.238171709002927, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.939910411834717, "lr": 0.0002, "elapsed_sec": 50971.66519117355, "step_time_sec": 8.241731591988355, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9147119522094727, "lr": 0.0002, "elapsed_sec": 50979.904546022415, "step_time_sec": 8.239223229989875, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8823180198669434, "lr": 0.0002, "elapsed_sec": 50988.14267706871, "step_time_sec": 8.237917577032931, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8753318786621094, "lr": 0.0002, "elapsed_sec": 50996.383157253265, "step_time_sec": 8.240367653022986, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.004187107086182, "lr": 0.0002, "elapsed_sec": 51004.62075138092, "step_time_sec": 8.237401477003004, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9330875873565674, "lr": 0.0002, "elapsed_sec": 51012.860577344894, "step_time_sec": 8.239623989036772, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9999570846557617, "lr": 0.0002, "elapsed_sec": 51021.09826159477, "step_time_sec": 8.237529819016345, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9293861389160156, "lr": 0.0002, "elapsed_sec": 51029.33831310272, "step_time_sec": 8.23991775198374, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7752230167388916, "lr": 0.0002, "elapsed_sec": 51037.57771611214, "step_time_sec": 8.2391991080367, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0018134117126465, "lr": 0.0002, "elapsed_sec": 51045.816039800644, "step_time_sec": 8.238152639009058, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.027954578399658, "lr": 0.0002, "elapsed_sec": 51054.0558514595, "step_time_sec": 8.239697176031768, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.788205862045288, "lr": 0.0002, "elapsed_sec": 51062.2961204052, "step_time_sec": 8.240068219020031, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.973123788833618, "lr": 0.0002, "elapsed_sec": 51070.534459114075, "step_time_sec": 8.238121975038666, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.906414031982422, "lr": 0.0002, "elapsed_sec": 51078.773468732834, "step_time_sec": 8.23888163198717, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.072862148284912, "lr": 0.0002, "elapsed_sec": 51087.01071882248, "step_time_sec": 8.237100865982939, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.890817880630493, "lr": 0.0002, "elapsed_sec": 51095.250629901886, "step_time_sec": 8.23975781397894, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.022284984588623, "lr": 0.0002, "elapsed_sec": 51103.489706754684, "step_time_sec": 8.238871671026573, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.019394397735596, "lr": 0.0002, "elapsed_sec": 51111.72757768631, "step_time_sec": 8.237719402997755, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9326608180999756, "lr": 0.0002, "elapsed_sec": 51119.966522455215, "step_time_sec": 8.238811071030796, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.949328660964966, "lr": 0.0002, "elapsed_sec": 51128.2044634819, "step_time_sec": 8.23779102199478, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8896660804748535, "lr": 0.0002, "elapsed_sec": 51136.44579744339, "step_time_sec": 8.24115456501022, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.872392177581787, "lr": 0.0002, "elapsed_sec": 51144.68443250656, "step_time_sec": 8.238439379027113, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.946965456008911, "lr": 0.0002, "elapsed_sec": 51152.92150592804, "step_time_sec": 8.23697056400124, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.035192012786865, "lr": 0.0002, "elapsed_sec": 51161.161309719086, "step_time_sec": 8.239606262010057, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.965026378631592, "lr": 0.0002, "elapsed_sec": 51169.398099422455, "step_time_sec": 8.236590699991211, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.905261993408203, "lr": 0.0002, "elapsed_sec": 51177.638882637024, "step_time_sec": 8.240631642984226, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.022128582000732, "lr": 0.0002, "elapsed_sec": 51185.87786889076, "step_time_sec": 8.23883805499645, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.996994972229004, "lr": 0.0002, "elapsed_sec": 51194.11619591713, "step_time_sec": 8.238204294990283, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.952850818634033, "lr": 0.0002, "elapsed_sec": 51202.35494971275, "step_time_sec": 8.23855494800955, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9978690147399902, "lr": 0.0002, "elapsed_sec": 51210.59235596657, "step_time_sec": 8.237234721018467, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.021662712097168, "lr": 0.0002, "elapsed_sec": 51218.833159446716, "step_time_sec": 8.240633763023652, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.993729829788208, "lr": 0.0002, "elapsed_sec": 51227.07128858566, "step_time_sec": 8.237985693034716, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.866828441619873, "lr": 0.0002, "elapsed_sec": 51235.30932378769, "step_time_sec": 8.23793551203562, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9815833568573, "lr": 0.0002, "elapsed_sec": 51243.54928803444, "step_time_sec": 8.239739806973375, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.761073589324951, "lr": 0.0002, "elapsed_sec": 51251.78891706467, "step_time_sec": 8.239471225009765, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.003175735473633, "lr": 0.0002, "elapsed_sec": 51260.02797245979, "step_time_sec": 8.23892854899168, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9108667373657227, "lr": 0.0002, "elapsed_sec": 51268.26692056656, "step_time_sec": 8.238791916985065, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9612584114074707, "lr": 0.0002, "elapsed_sec": 51276.50526833534, "step_time_sec": 8.238166451978032, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.05549955368042, "lr": 0.0002, "elapsed_sec": 51284.74577188492, "step_time_sec": 8.240307166008279, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9142189025878906, "lr": 0.0002, "elapsed_sec": 51292.98430418968, "step_time_sec": 8.238405030977447, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9329419136047363, "lr": 0.0002, "elapsed_sec": 51301.22517824173, "step_time_sec": 8.24069431098178, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.963942527770996, "lr": 0.0002, "elapsed_sec": 51309.46484708786, "step_time_sec": 8.239486449980177, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.042166233062744, "lr": 0.0002, "elapsed_sec": 51317.70202231407, "step_time_sec": 8.237029078009073, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9009501934051514, "lr": 0.0002, "elapsed_sec": 51325.941371679306, "step_time_sec": 8.239192673005164, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8939175605773926, "lr": 0.0002, "elapsed_sec": 51334.18177628517, "step_time_sec": 8.240280692989472, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.925339937210083, "lr": 0.0002, "elapsed_sec": 51342.42075943947, "step_time_sec": 8.23883599799592, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.042463302612305, "lr": 0.0002, "elapsed_sec": 51350.66078233719, "step_time_sec": 8.239805258985143, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.009212493896484, "lr": 0.0002, "elapsed_sec": 51358.90029144287, "step_time_sec": 8.239413951989263, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0298919677734375, "lr": 0.0002, "elapsed_sec": 51367.14193320274, "step_time_sec": 8.241459708020557, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8449881076812744, "lr": 0.0002, "elapsed_sec": 51375.38185882568, "step_time_sec": 8.239708815002814, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.034445762634277, "lr": 0.0002, "elapsed_sec": 51383.61910843849, "step_time_sec": 8.237097054021433, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7579774856567383, "lr": 0.0002, "elapsed_sec": 51391.8586537838, "step_time_sec": 8.239367795002181, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.716885566711426, "lr": 0.0002, "elapsed_sec": 51400.09623479843, "step_time_sec": 8.237467252998613, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.997300386428833, "lr": 0.0002, "elapsed_sec": 51408.33620715141, "step_time_sec": 8.239793471992016, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8992364406585693, "lr": 0.0002, "elapsed_sec": 51416.575897455215, "step_time_sec": 8.239534479973372, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.896594285964966, "lr": 0.0002, "elapsed_sec": 51424.814603328705, "step_time_sec": 8.238511975039728, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.870574474334717, "lr": 0.0002, "elapsed_sec": 51433.05492448807, "step_time_sec": 8.24016369401943, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8555099964141846, "lr": 0.0002, "elapsed_sec": 51441.2938272953, "step_time_sec": 8.238719757995568, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.875243902206421, "lr": 0.0002, "elapsed_sec": 51449.53310036659, "step_time_sec": 8.239095816970803, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8194386959075928, "lr": 0.0002, "elapsed_sec": 51457.773767232895, "step_time_sec": 8.240515373006929, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9124879837036133, "lr": 0.0002, "elapsed_sec": 51466.01385974884, "step_time_sec": 8.239920460968278, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.952867269515991, "lr": 0.0002, "elapsed_sec": 51474.25293183327, "step_time_sec": 8.238969672995154, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.857590436935425, "lr": 0.0002, "elapsed_sec": 51482.49349999428, "step_time_sec": 8.240363607008476, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.898017168045044, "lr": 0.0002, "elapsed_sec": 51490.73138809204, "step_time_sec": 8.237722480960656, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.868757963180542, "lr": 0.0002, "elapsed_sec": 51498.97469997406, "step_time_sec": 8.243154702999163, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9891021251678467, "lr": 0.0002, "elapsed_sec": 51507.21423006058, "step_time_sec": 8.239408756024204, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8541975021362305, "lr": 0.0002, "elapsed_sec": 51515.45102548599, "step_time_sec": 8.236601809039712, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.05007791519165, "lr": 0.0002, "elapsed_sec": 51523.69736480713, "step_time_sec": 8.246162727999035, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.957247734069824, "lr": 0.0002, "elapsed_sec": 51531.93617081642, "step_time_sec": 8.238637623027898, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9273765087127686, "lr": 0.0002, "elapsed_sec": 51540.17528986931, "step_time_sec": 8.23900516401045, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.860959529876709, "lr": 0.0002, "elapsed_sec": 51548.41556954384, "step_time_sec": 8.240098426991608, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.943329095840454, "lr": 0.0002, "elapsed_sec": 51556.65216779709, "step_time_sec": 8.236457048973534, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.899091958999634, "lr": 0.0002, "elapsed_sec": 51564.89390492439, "step_time_sec": 8.241501781973056, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.873248815536499, "lr": 0.0002, "elapsed_sec": 51573.13289833069, "step_time_sec": 8.238840490987059, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8652195930480957, "lr": 0.0002, "elapsed_sec": 51581.371029376984, "step_time_sec": 8.237978304969147, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.813340187072754, "lr": 0.0002, "elapsed_sec": 51589.61006116867, "step_time_sec": 8.238912777975202, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9044294357299805, "lr": 0.0002, "elapsed_sec": 51597.847504377365, "step_time_sec": 8.237274264974985, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9394047260284424, "lr": 0.0002, "elapsed_sec": 51606.08639335632, "step_time_sec": 8.238714513019659, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8206241130828857, "lr": 0.0002, "elapsed_sec": 51614.326333761215, "step_time_sec": 8.239789092971478, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.845482349395752, "lr": 0.0002, "elapsed_sec": 51622.564791440964, "step_time_sec": 8.238276620977558, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.848754644393921, "lr": 0.0002, "elapsed_sec": 51630.805884838104, "step_time_sec": 8.240947909012903, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.950956344604492, "lr": 0.0002, "elapsed_sec": 51639.04512548447, "step_time_sec": 8.239058438979555, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1238179206848145, "lr": 0.0002, "elapsed_sec": 51647.283752441406, "step_time_sec": 8.238467356015462, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.953679323196411, "lr": 0.0002, "elapsed_sec": 51655.52267098427, "step_time_sec": 8.238767290022224, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9105584621429443, "lr": 0.0002, "elapsed_sec": 51663.759278059006, "step_time_sec": 8.236431708035525, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.976874351501465, "lr": 0.0002, "elapsed_sec": 51671.99923443794, "step_time_sec": 8.2398073800141, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.137082099914551, "lr": 0.0002, "elapsed_sec": 51680.23764157295, "step_time_sec": 8.238269070978276, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.118045806884766, "lr": 0.0002, "elapsed_sec": 51688.4768974781, "step_time_sec": 8.239088084024843, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8289804458618164, "lr": 0.0002, "elapsed_sec": 51696.71698641777, "step_time_sec": 8.23993151501054, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.11722993850708, "lr": 0.0002, "elapsed_sec": 51704.95450735092, "step_time_sec": 8.237293659010902, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.245630741119385, "lr": 0.0002, "elapsed_sec": 51713.19382238388, "step_time_sec": 8.239173022971954, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.784658670425415, "lr": 0.0002, "elapsed_sec": 51721.50448060036, "step_time_sec": 8.250678928976413, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.776416301727295, "lr": 0.0002, "elapsed_sec": 51729.744348526, "step_time_sec": 8.239744819002226, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.003169536590576, "lr": 0.0002, "elapsed_sec": 51737.98248934746, "step_time_sec": 8.237993865972385, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9593493938446045, "lr": 0.0002, "elapsed_sec": 51746.22196030617, "step_time_sec": 8.239282783993986, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9010109901428223, "lr": 0.0002, "elapsed_sec": 51754.458478450775, "step_time_sec": 8.23644437402254, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.676842451095581, "lr": 0.0002, "elapsed_sec": 51762.70000958443, "step_time_sec": 8.24129765102407, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7133162021636963, "lr": 0.0002, "elapsed_sec": 51770.93915247917, "step_time_sec": 8.239013980026357, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.956446647644043, "lr": 0.0002, "elapsed_sec": 51779.17708158493, "step_time_sec": 8.237753362976946, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9468250274658203, "lr": 0.0002, "elapsed_sec": 51787.41598677635, "step_time_sec": 8.238709021010436, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8832335472106934, "lr": 0.0002, "elapsed_sec": 51795.65509414673, "step_time_sec": 8.238967142009642, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.91117525100708, "lr": 0.0002, "elapsed_sec": 51803.89377975464, "step_time_sec": 8.238534559961408, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0378546714782715, "lr": 0.0002, "elapsed_sec": 51812.132400751114, "step_time_sec": 8.238423576985952, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7463226318359375, "lr": 0.0002, "elapsed_sec": 51820.37004613876, "step_time_sec": 8.23752234899439, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.016586780548096, "lr": 0.0002, "elapsed_sec": 51828.60956144333, "step_time_sec": 8.239362545020413, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.171802997589111, "lr": 0.0002, "elapsed_sec": 51836.84749150276, "step_time_sec": 8.237731502973475, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.908823013305664, "lr": 0.0002, "elapsed_sec": 51845.08804273605, "step_time_sec": 8.240413354011253, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9248039722442627, "lr": 0.0002, "elapsed_sec": 51853.32684326172, "step_time_sec": 8.238625775964465, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9902966022491455, "lr": 0.0002, "elapsed_sec": 51861.56500291824, "step_time_sec": 8.237983056984376, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7765679359436035, "lr": 0.0002, "elapsed_sec": 51869.80340600014, "step_time_sec": 8.238276686985046, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8300774097442627, "lr": 0.0002, "elapsed_sec": 51878.04147052765, "step_time_sec": 8.237866590963677, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.03633451461792, "lr": 0.0002, "elapsed_sec": 51886.28114557266, "step_time_sec": 8.23951288900571, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.903587579727173, "lr": 0.0002, "elapsed_sec": 51894.52104473114, "step_time_sec": 8.239779488998465, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.005638599395752, "lr": 0.0002, "elapsed_sec": 51902.75911068916, "step_time_sec": 8.237863318994641, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.084004878997803, "lr": 0.0002, "elapsed_sec": 51910.99856710434, "step_time_sec": 8.239292427024338, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.052949905395508, "lr": 0.0002, "elapsed_sec": 51919.237743377686, "step_time_sec": 8.23900943400804, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.038268089294434, "lr": 0.0002, "elapsed_sec": 51927.4768910408, "step_time_sec": 8.238988708006218, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.087406158447266, "lr": 0.0002, "elapsed_sec": 51935.71581363678, "step_time_sec": 8.238818695012014, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9601707458496094, "lr": 0.0002, "elapsed_sec": 51943.95431923866, "step_time_sec": 8.238349464023486, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9184341430664062, "lr": 0.0002, "elapsed_sec": 51952.19448304176, "step_time_sec": 8.239957668993156, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9843862056732178, "lr": 0.0002, "elapsed_sec": 51960.434566020966, "step_time_sec": 8.23995911795646, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.976653575897217, "lr": 0.0002, "elapsed_sec": 51968.67434716225, "step_time_sec": 8.23962616699282, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8666038513183594, "lr": 0.0002, "elapsed_sec": 51976.913217782974, "step_time_sec": 8.238684468960855, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8917737007141113, "lr": 0.0002, "elapsed_sec": 51985.1508808136, "step_time_sec": 8.237466565973591, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9898362159729004, "lr": 0.0002, "elapsed_sec": 51993.390530109406, "step_time_sec": 8.239503860007972, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.016758918762207, "lr": 0.0002, "elapsed_sec": 52001.62918305397, "step_time_sec": 8.23849928699201, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9782111644744873, "lr": 0.0002, "elapsed_sec": 52009.86785745621, "step_time_sec": 8.238542434992269, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.197925090789795, "lr": 0.0002, "elapsed_sec": 52018.10886287689, "step_time_sec": 8.240801692008972, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.257174015045166, "lr": 0.0002, "elapsed_sec": 52026.34742760658, "step_time_sec": 8.238458841049578, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9287750720977783, "lr": 0.0002, "elapsed_sec": 52034.587243556976, "step_time_sec": 8.239609163021669, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9481847286224365, "lr": 0.0002, "elapsed_sec": 52042.82387590408, "step_time_sec": 8.23648778302595, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0689167976379395, "lr": 0.0002, "elapsed_sec": 52051.06186914444, "step_time_sec": 8.237790094979573, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0340752601623535, "lr": 0.0002, "elapsed_sec": 52059.3024559021, "step_time_sec": 8.24044430302456, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9319283962249756, "lr": 0.0002, "elapsed_sec": 52067.540335178375, "step_time_sec": 8.237714093003888, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9436070919036865, "lr": 0.0002, "elapsed_sec": 52075.77938771248, "step_time_sec": 8.238922653021291, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9541232585906982, "lr": 0.0002, "elapsed_sec": 52084.01944112778, "step_time_sec": 8.239916510006879, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.028792381286621, "lr": 0.0002, "elapsed_sec": 52092.25949406624, "step_time_sec": 8.239881782967132, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.889343500137329, "lr": 0.0002, "elapsed_sec": 52100.499635219574, "step_time_sec": 8.239959082973655, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.011490345001221, "lr": 0.0002, "elapsed_sec": 52108.73787999153, "step_time_sec": 8.238080319017172, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.988325595855713, "lr": 0.0002, "elapsed_sec": 52116.97739100456, "step_time_sec": 8.239345189009327, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9365103244781494, "lr": 0.0002, "elapsed_sec": 52125.21861791611, "step_time_sec": 8.241030140023213, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.935415029525757, "lr": 0.0002, "elapsed_sec": 52133.45756459236, "step_time_sec": 8.238830720016267, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.039113521575928, "lr": 0.0002, "elapsed_sec": 52141.69742488861, "step_time_sec": 8.239633425022475, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.042166233062744, "lr": 0.0002, "elapsed_sec": 52149.94664502144, "step_time_sec": 8.249116411956493, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9233362674713135, "lr": 0.0002, "elapsed_sec": 52158.18615818024, "step_time_sec": 8.239343894994818, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.034975051879883, "lr": 0.0002, "elapsed_sec": 52166.42754435539, "step_time_sec": 8.241218248032965, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9718244075775146, "lr": 0.0002, "elapsed_sec": 52174.66766166687, "step_time_sec": 8.23994680098258, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.021556854248047, "lr": 0.0002, "elapsed_sec": 52182.90575647354, "step_time_sec": 8.237905889982358, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.837843179702759, "lr": 0.0002, "elapsed_sec": 52191.14659309387, "step_time_sec": 8.240703680028673, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9055986404418945, "lr": 0.0002, "elapsed_sec": 52199.38461470604, "step_time_sec": 8.237880052009132, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.186463832855225, "lr": 0.0002, "elapsed_sec": 52207.62349677086, "step_time_sec": 8.238675725995563, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.003018856048584, "lr": 0.0002, "elapsed_sec": 52215.86307644844, "step_time_sec": 8.239435811992735, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8877577781677246, "lr": 0.0002, "elapsed_sec": 52224.10081624985, "step_time_sec": 8.237595825979952, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.032478332519531, "lr": 0.0002, "elapsed_sec": 52232.341692209244, "step_time_sec": 8.240673616994172, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.958352565765381, "lr": 0.0002, "elapsed_sec": 52240.582969903946, "step_time_sec": 8.241148656990845, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0954790115356445, "lr": 0.0002, "elapsed_sec": 52248.820936203, "step_time_sec": 8.237781826988794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.079258441925049, "lr": 0.0002, "elapsed_sec": 52257.05984902382, "step_time_sec": 8.238725400005933, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.10654878616333, "lr": 0.0002, "elapsed_sec": 52265.29725575447, "step_time_sec": 8.23721566295717, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9735920429229736, "lr": 0.0002, "elapsed_sec": 52273.537821531296, "step_time_sec": 8.240435107960366, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9067652225494385, "lr": 0.0002, "elapsed_sec": 52281.77834057808, "step_time_sec": 8.240337322989944, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9780020713806152, "lr": 0.0002, "elapsed_sec": 52290.01656627655, "step_time_sec": 8.23807190201478, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.064399242401123, "lr": 0.0002, "elapsed_sec": 52298.25790786743, "step_time_sec": 8.24119543802226, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8922550678253174, "lr": 0.0002, "elapsed_sec": 52306.49763274193, "step_time_sec": 8.239549599995371, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.878570079803467, "lr": 0.0002, "elapsed_sec": 52314.73585820198, "step_time_sec": 8.23808276699856, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8499598503112793, "lr": 0.0002, "elapsed_sec": 52322.976127147675, "step_time_sec": 8.240086467005312, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.113680839538574, "lr": 0.0002, "elapsed_sec": 52331.21416044235, "step_time_sec": 8.23784556798637, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.920015811920166, "lr": 0.0002, "elapsed_sec": 52339.45530509949, "step_time_sec": 8.2409752390231, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.881126880645752, "lr": 0.0002, "elapsed_sec": 52347.69524025917, "step_time_sec": 8.239768038038164, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9572699069976807, "lr": 0.0002, "elapsed_sec": 52355.93331027031, "step_time_sec": 8.23798470501788, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.932528257369995, "lr": 0.0002, "elapsed_sec": 52364.17234873772, "step_time_sec": 8.238857176038437, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9262146949768066, "lr": 0.0002, "elapsed_sec": 52372.40997719765, "step_time_sec": 8.237405040999874, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.089937686920166, "lr": 0.0002, "elapsed_sec": 52380.650876283646, "step_time_sec": 8.240742230031174, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0273590087890625, "lr": 0.0002, "elapsed_sec": 52388.89108252525, "step_time_sec": 8.240050772961695, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.069422245025635, "lr": 0.0002, "elapsed_sec": 52397.12964916229, "step_time_sec": 8.2383948629722, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.018158435821533, "lr": 0.0002, "elapsed_sec": 52405.37104320526, "step_time_sec": 8.241234390996397, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9182865619659424, "lr": 0.0002, "elapsed_sec": 52413.60867834091, "step_time_sec": 8.237462015997153, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9405460357666016, "lr": 0.0002, "elapsed_sec": 52421.84856247902, "step_time_sec": 8.239722670987248, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.860370635986328, "lr": 0.0002, "elapsed_sec": 52430.08801960945, "step_time_sec": 8.239324694033712, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.054671764373779, "lr": 0.0002, "elapsed_sec": 52438.32562875748, "step_time_sec": 8.237429864006117, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.081533908843994, "lr": 0.0002, "elapsed_sec": 52446.566076517105, "step_time_sec": 8.240313332993537, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.043523788452148, "lr": 0.0002, "elapsed_sec": 52454.80560588837, "step_time_sec": 8.239313367987052, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.080489635467529, "lr": 0.0002, "elapsed_sec": 52463.04383230209, "step_time_sec": 8.238068637030665, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.026498317718506, "lr": 0.0002, "elapsed_sec": 52471.28334450722, "step_time_sec": 8.23934674699558, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.093191146850586, "lr": 0.0002, "elapsed_sec": 52479.5210082531, "step_time_sec": 8.237501517985947, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.08491325378418, "lr": 0.0002, "elapsed_sec": 52487.76060771942, "step_time_sec": 8.239418695971835, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0009379386901855, "lr": 0.0002, "elapsed_sec": 52495.999846458435, "step_time_sec": 8.239131433016155, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.892324686050415, "lr": 0.0002, "elapsed_sec": 52504.23798274994, "step_time_sec": 8.237977027951274, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.907888174057007, "lr": 0.0002, "elapsed_sec": 52512.47778201103, "step_time_sec": 8.239585173956584, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.890672445297241, "lr": 0.0002, "elapsed_sec": 52520.71511363983, "step_time_sec": 8.237184835015796, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.04309606552124, "lr": 0.0002, "elapsed_sec": 52528.95600652695, "step_time_sec": 8.240738110034727, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.129114627838135, "lr": 0.0002, "elapsed_sec": 52537.194700717926, "step_time_sec": 8.238541922008153, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7884576320648193, "lr": 0.0002, "elapsed_sec": 52545.43270254135, "step_time_sec": 8.237797104986385, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.016526699066162, "lr": 0.0002, "elapsed_sec": 52553.67191171646, "step_time_sec": 8.239041380991694, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.872119188308716, "lr": 0.0002, "elapsed_sec": 52561.91094136238, "step_time_sec": 8.238860812969506, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.021841049194336, "lr": 0.0002, "elapsed_sec": 52570.15039730072, "step_time_sec": 8.239339214982465, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.995903491973877, "lr": 0.0002, "elapsed_sec": 52578.38940405846, "step_time_sec": 8.238833608978894, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8789427280426025, "lr": 0.0002, "elapsed_sec": 52586.6276845932, "step_time_sec": 8.238123708986677, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.01687479019165, "lr": 0.0002, "elapsed_sec": 52594.86805510521, "step_time_sec": 8.240169425029308, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9407734870910645, "lr": 0.0002, "elapsed_sec": 52603.10841917992, "step_time_sec": 8.240203388035297, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7043113708496094, "lr": 0.0002, "elapsed_sec": 52611.34681558609, "step_time_sec": 8.238261074002367, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9793410301208496, "lr": 0.0002, "elapsed_sec": 52619.58599328995, "step_time_sec": 8.238987164979335, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.980377674102783, "lr": 0.0002, "elapsed_sec": 52627.82364320755, "step_time_sec": 8.237526085984427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.875603437423706, "lr": 0.0002, "elapsed_sec": 52636.06579661369, "step_time_sec": 8.241970292991027, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.029937267303467, "lr": 0.0002, "elapsed_sec": 52644.30413222313, "step_time_sec": 8.23821839201264, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.068027496337891, "lr": 0.0002, "elapsed_sec": 52652.541949510574, "step_time_sec": 8.237569556978997, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9307732582092285, "lr": 0.0002, "elapsed_sec": 52660.78173828125, "step_time_sec": 8.2396949629765, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.963113784790039, "lr": 0.0002, "elapsed_sec": 52669.01899409294, "step_time_sec": 8.23706754395971, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.910696029663086, "lr": 0.0002, "elapsed_sec": 52677.25972867012, "step_time_sec": 8.240579090022948, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.061275482177734, "lr": 0.0002, "elapsed_sec": 52685.49992322922, "step_time_sec": 8.240030350978486, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9730684757232666, "lr": 0.0002, "elapsed_sec": 52693.73798751831, "step_time_sec": 8.237883596972097, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.01279878616333, "lr": 0.0002, "elapsed_sec": 52701.97865319252, "step_time_sec": 8.24049203004688, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.92049241065979, "lr": 0.0002, "elapsed_sec": 52710.21642065048, "step_time_sec": 8.237583770998754, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.966481924057007, "lr": 0.0002, "elapsed_sec": 52718.45551919937, "step_time_sec": 8.238974370993674, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.977004289627075, "lr": 0.0002, "elapsed_sec": 52726.695845365524, "step_time_sec": 8.240117513982113, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.143856048583984, "lr": 0.0002, "elapsed_sec": 52734.93327021599, "step_time_sec": 8.237290926044807, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.118819236755371, "lr": 0.0002, "elapsed_sec": 52743.17406916618, "step_time_sec": 8.240655065979809, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9915432929992676, "lr": 0.0002, "elapsed_sec": 52751.41255021095, "step_time_sec": 8.238296539988369, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.02669620513916, "lr": 0.0002, "elapsed_sec": 52759.65211081505, "step_time_sec": 8.23941391397966, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.010769844055176, "lr": 0.0002, "elapsed_sec": 52767.89216709137, "step_time_sec": 8.239859413995873, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.937499523162842, "lr": 0.0002, "elapsed_sec": 52776.1301138401, "step_time_sec": 8.237831822014414, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.904017925262451, "lr": 0.0002, "elapsed_sec": 52784.36925268173, "step_time_sec": 8.238961464026943, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.191861152648926, "lr": 0.0002, "elapsed_sec": 52792.60965228081, "step_time_sec": 8.240216644015163, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.900101661682129, "lr": 0.0002, "elapsed_sec": 52800.84847283363, "step_time_sec": 8.238716073974501, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.08247184753418, "lr": 0.0002, "elapsed_sec": 52809.08883857727, "step_time_sec": 8.240191304008476, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.029367923736572, "lr": 0.0002, "elapsed_sec": 52817.32783985138, "step_time_sec": 8.238819470978342, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9951798915863037, "lr": 0.0002, "elapsed_sec": 52825.56635570526, "step_time_sec": 8.238308388972655, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.046257495880127, "lr": 0.0002, "elapsed_sec": 52833.80444908142, "step_time_sec": 8.23792314104503, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9705734252929688, "lr": 0.0002, "elapsed_sec": 52842.04324698448, "step_time_sec": 8.238723009009846, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9437496662139893, "lr": 0.0002, "elapsed_sec": 52850.283816337585, "step_time_sec": 8.240293163980823, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9020493030548096, "lr": 0.0002, "elapsed_sec": 52858.52195930481, "step_time_sec": 8.238030419975985, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9931418895721436, "lr": 0.0002, "elapsed_sec": 52866.76267290115, "step_time_sec": 8.24052562599536, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.962578773498535, "lr": 0.0002, "elapsed_sec": 52875.001170158386, "step_time_sec": 8.238312041037716, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.083710670471191, "lr": 0.0002, "elapsed_sec": 52883.23930430412, "step_time_sec": 8.237956344033591, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8565726280212402, "lr": 0.0002, "elapsed_sec": 52891.47819495201, "step_time_sec": 8.238758553983644, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.991055727005005, "lr": 0.0002, "elapsed_sec": 52899.71698999405, "step_time_sec": 8.238601025019307, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.01342248916626, "lr": 0.0002, "elapsed_sec": 52907.95495939255, "step_time_sec": 8.237776836031117, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9190073013305664, "lr": 0.0002, "elapsed_sec": 52916.195054769516, "step_time_sec": 8.239989947993308, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.055221080780029, "lr": 0.0002, "elapsed_sec": 52924.43201470375, "step_time_sec": 8.236767872003838, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.986821174621582, "lr": 0.0002, "elapsed_sec": 52932.67004227638, "step_time_sec": 8.237895027035847, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.135401248931885, "lr": 0.0002, "elapsed_sec": 52940.908453941345, "step_time_sec": 8.238180938002188, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9906363487243652, "lr": 0.0002, "elapsed_sec": 52949.14779376984, "step_time_sec": 8.239205782010686, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9226794242858887, "lr": 0.0002, "elapsed_sec": 52957.386263370514, "step_time_sec": 8.238301808945835, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.073594570159912, "lr": 0.0002, "elapsed_sec": 52965.62333202362, "step_time_sec": 8.236942670948338, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9993772506713867, "lr": 0.0002, "elapsed_sec": 52973.862743377686, "step_time_sec": 8.23924194695428, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.013838291168213, "lr": 0.0002, "elapsed_sec": 52982.1018178463, "step_time_sec": 8.238862662983593, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.972028970718384, "lr": 0.0002, "elapsed_sec": 52990.34222865105, "step_time_sec": 8.240253853029571, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9887850284576416, "lr": 0.0002, "elapsed_sec": 52998.58253097534, "step_time_sec": 8.240138707973529, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.821018695831299, "lr": 0.0002, "elapsed_sec": 53006.82005596161, "step_time_sec": 8.23735258897068, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.936227798461914, "lr": 0.0002, "elapsed_sec": 53015.06038379669, "step_time_sec": 8.240162309026346, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.926345109939575, "lr": 0.0002, "elapsed_sec": 53023.29990887642, "step_time_sec": 8.239364170993213, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0978169441223145, "lr": 0.0002, "elapsed_sec": 53031.53948688507, "step_time_sec": 8.239425272971857, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.063263416290283, "lr": 0.0002, "elapsed_sec": 53039.77897334099, "step_time_sec": 8.239312710997183, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9559218883514404, "lr": 0.0002, "elapsed_sec": 53048.01631140709, "step_time_sec": 8.237148928048555, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9361939430236816, "lr": 0.0002, "elapsed_sec": 53056.257920742035, "step_time_sec": 8.241482624027412, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.031321048736572, "lr": 0.0002, "elapsed_sec": 53064.49714279175, "step_time_sec": 8.23904685798334, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8577394485473633, "lr": 0.0002, "elapsed_sec": 53072.73493695259, "step_time_sec": 8.23760505602695, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9656150341033936, "lr": 0.0002, "elapsed_sec": 53080.97522878647, "step_time_sec": 8.240146491036285, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.962374210357666, "lr": 0.0002, "elapsed_sec": 53089.21259832382, "step_time_sec": 8.237171510991175, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9710898399353027, "lr": 0.0002, "elapsed_sec": 53097.45297718048, "step_time_sec": 8.240216742968187, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.015817642211914, "lr": 0.0002, "elapsed_sec": 53105.69315004349, "step_time_sec": 8.240009660017677, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.024135589599609, "lr": 0.0002, "elapsed_sec": 53113.93111538887, "step_time_sec": 8.237837499997113, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8532776832580566, "lr": 0.0002, "elapsed_sec": 53122.170825481415, "step_time_sec": 8.239481701049954, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.130440711975098, "lr": 0.0002, "elapsed_sec": 53130.41034460068, "step_time_sec": 8.23932988604065, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.098199367523193, "lr": 0.0002, "elapsed_sec": 53138.64874958992, "step_time_sec": 8.23827493499266, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.00545597076416, "lr": 0.0002, "elapsed_sec": 53146.88858580589, "step_time_sec": 8.239691459981259, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0032758712768555, "lr": 0.0002, "elapsed_sec": 53155.12622880936, "step_time_sec": 8.23745992401382, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.038522720336914, "lr": 0.0002, "elapsed_sec": 53163.36566019058, "step_time_sec": 8.239285312010907, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9854393005371094, "lr": 0.0002, "elapsed_sec": 53171.60597848892, "step_time_sec": 8.240138703025877, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9210376739501953, "lr": 0.0002, "elapsed_sec": 53179.844859838486, "step_time_sec": 8.238731397024821, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.987408399581909, "lr": 0.0002, "elapsed_sec": 53188.084235429764, "step_time_sec": 8.239214856002945, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.017129898071289, "lr": 0.0002, "elapsed_sec": 53196.322419166565, "step_time_sec": 8.238064953009598, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.979620933532715, "lr": 0.0002, "elapsed_sec": 53204.563824653625, "step_time_sec": 8.241209651983809, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.045071601867676, "lr": 0.0002, "elapsed_sec": 53212.80340719223, "step_time_sec": 8.239414278010372, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9755120277404785, "lr": 0.0002, "elapsed_sec": 53221.04377055168, "step_time_sec": 8.24018983199494, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.037101745605469, "lr": 0.0002, "elapsed_sec": 53229.28362607956, "step_time_sec": 8.239709848014172, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.035794258117676, "lr": 0.0002, "elapsed_sec": 53237.52365422249, "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": 6406, "loss": 3.92087721824646, "lr": 0.0002, "elapsed_sec": 53245.76411938667, "step_time_sec": 8.240282826998737, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.028233528137207, "lr": 0.0002, "elapsed_sec": 53254.00256538391, "step_time_sec": 8.238305407983717, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9147722721099854, "lr": 0.0002, "elapsed_sec": 53262.23936152458, "step_time_sec": 8.236623766017146, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.004294395446777, "lr": 0.0002, "elapsed_sec": 53270.48053073883, "step_time_sec": 8.240938400034793, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.964216470718384, "lr": 0.0002, "elapsed_sec": 53278.719333171844, "step_time_sec": 8.238663759024348, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7771010398864746, "lr": 0.0002, "elapsed_sec": 53286.95793223381, "step_time_sec": 8.238452065968886, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.044083595275879, "lr": 0.0002, "elapsed_sec": 53295.19796395302, "step_time_sec": 8.239841961010825, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.888737201690674, "lr": 0.0002, "elapsed_sec": 53303.435849666595, "step_time_sec": 8.237736900046002, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.034656047821045, "lr": 0.0002, "elapsed_sec": 53311.67553448677, "step_time_sec": 8.239552001992706, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9988458156585693, "lr": 0.0002, "elapsed_sec": 53319.913709163666, "step_time_sec": 8.238019865995739, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.08994722366333, "lr": 0.0002, "elapsed_sec": 53328.15222287178, "step_time_sec": 8.238301984034479, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9397389888763428, "lr": 0.0002, "elapsed_sec": 53336.39287734032, "step_time_sec": 8.240509878029115, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.811572313308716, "lr": 0.0002, "elapsed_sec": 53344.640686273575, "step_time_sec": 8.247668200987391, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9205915927886963, "lr": 0.0002, "elapsed_sec": 53352.87823295593, "step_time_sec": 8.237378794990946, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.917470693588257, "lr": 0.0002, "elapsed_sec": 53361.11828374863, "step_time_sec": 8.239840286958497, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.01405143737793, "lr": 0.0002, "elapsed_sec": 53369.35609984398, "step_time_sec": 8.237655891978648, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.055323123931885, "lr": 0.0002, "elapsed_sec": 53377.598836660385, "step_time_sec": 8.242617228999734, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.910233974456787, "lr": 0.0002, "elapsed_sec": 53385.838305711746, "step_time_sec": 8.239300162007567, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.032903671264648, "lr": 0.0002, "elapsed_sec": 53394.07578611374, "step_time_sec": 8.237317468039691, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.009857177734375, "lr": 0.0002, "elapsed_sec": 53402.31717848778, "step_time_sec": 8.241180073993746, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.025378227233887, "lr": 0.0002, "elapsed_sec": 53410.55636239052, "step_time_sec": 8.239091366995126, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.048712730407715, "lr": 0.0002, "elapsed_sec": 53418.795798540115, "step_time_sec": 8.23919782001758, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.141977787017822, "lr": 0.0002, "elapsed_sec": 53427.034751176834, "step_time_sec": 8.238820433965884, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.062180519104004, "lr": 0.0002, "elapsed_sec": 53435.27227497101, "step_time_sec": 8.237386121007148, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8964107036590576, "lr": 0.0002, "elapsed_sec": 53443.51265215874, "step_time_sec": 8.240165119990706, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7419028282165527, "lr": 0.0002, "elapsed_sec": 53451.75265169144, "step_time_sec": 8.239860414993018, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9132814407348633, "lr": 0.0002, "elapsed_sec": 53459.991238832474, "step_time_sec": 8.238426559022628, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9870285987854004, "lr": 0.0002, "elapsed_sec": 53468.230820178986, "step_time_sec": 8.239436394011136, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.914407730102539, "lr": 0.0002, "elapsed_sec": 53476.46867990494, "step_time_sec": 8.237685492960736, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9230477809906006, "lr": 0.0002, "elapsed_sec": 53484.70803594589, "step_time_sec": 8.239172368019354, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.902092456817627, "lr": 0.0002, "elapsed_sec": 53492.94775533676, "step_time_sec": 8.239600571978372, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.078801155090332, "lr": 0.0002, "elapsed_sec": 53501.18785619736, "step_time_sec": 8.239940247032791, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.060160160064697, "lr": 0.0002, "elapsed_sec": 53509.42846131325, "step_time_sec": 8.24040317098843, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9739620685577393, "lr": 0.0002, "elapsed_sec": 53517.66669130325, "step_time_sec": 8.238129223987926, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.895505905151367, "lr": 0.0002, "elapsed_sec": 53525.906128168106, "step_time_sec": 8.239249352016486, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.991687297821045, "lr": 0.0002, "elapsed_sec": 53534.14613175392, "step_time_sec": 8.239883975009434, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.902270793914795, "lr": 0.0002, "elapsed_sec": 53542.38421082497, "step_time_sec": 8.23784048797097, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8501248359680176, "lr": 0.0002, "elapsed_sec": 53550.62431907654, "step_time_sec": 8.239981605962384, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8638148307800293, "lr": 0.0002, "elapsed_sec": 53558.86445856094, "step_time_sec": 8.239962083986029, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.039737224578857, "lr": 0.0002, "elapsed_sec": 53567.10284733772, "step_time_sec": 8.238220598024782, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.864349365234375, "lr": 0.0002, "elapsed_sec": 53575.34333920479, "step_time_sec": 8.24032957199961, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.6671924591064453, "lr": 0.0002, "elapsed_sec": 53583.58141207695, "step_time_sec": 8.237876921019051, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8318915367126465, "lr": 0.0002, "elapsed_sec": 53591.82196164131, "step_time_sec": 8.240424628020264, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8073620796203613, "lr": 0.0002, "elapsed_sec": 53600.061445236206, "step_time_sec": 8.239302277972456, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2728447914123535, "lr": 0.0002, "elapsed_sec": 53608.30032706261, "step_time_sec": 8.23872115497943, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.977107524871826, "lr": 0.0002, "elapsed_sec": 53616.540060043335, "step_time_sec": 8.239536284992937, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.907525062561035, "lr": 0.0002, "elapsed_sec": 53624.77953886986, "step_time_sec": 8.239318920008373, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.053891658782959, "lr": 0.0002, "elapsed_sec": 53633.019167900085, "step_time_sec": 8.239478921983391, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.906904458999634, "lr": 0.0002, "elapsed_sec": 53641.25752425194, "step_time_sec": 8.23816338001052, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.107548713684082, "lr": 0.0002, "elapsed_sec": 53649.49602842331, "step_time_sec": 8.238343652978074, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.005144119262695, "lr": 0.0002, "elapsed_sec": 53657.73742389679, "step_time_sec": 8.241247003956232, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9534380435943604, "lr": 0.0002, "elapsed_sec": 53665.97748494148, "step_time_sec": 8.239912143035326, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9149155616760254, "lr": 0.0002, "elapsed_sec": 53674.21481823921, "step_time_sec": 8.23719696898479, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.980017900466919, "lr": 0.0002, "elapsed_sec": 53682.454642534256, "step_time_sec": 8.239604717004113, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9133493900299072, "lr": 0.0002, "elapsed_sec": 53690.691850185394, "step_time_sec": 8.23703271400882, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.934905767440796, "lr": 0.0002, "elapsed_sec": 53698.93293905258, "step_time_sec": 8.240991186001338, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8261783123016357, "lr": 0.0002, "elapsed_sec": 53707.1731607914, "step_time_sec": 8.240023827005643, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.019887447357178, "lr": 0.0002, "elapsed_sec": 53715.40990829468, "step_time_sec": 8.23656381800538, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.829772472381592, "lr": 0.0002, "elapsed_sec": 53723.65067124367, "step_time_sec": 8.240571769012604, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9739344120025635, "lr": 0.0002, "elapsed_sec": 53731.8890273571, "step_time_sec": 8.238254320051055, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9027647972106934, "lr": 0.0002, "elapsed_sec": 53740.12702655792, "step_time_sec": 8.237795013003051, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.055357456207275, "lr": 0.0002, "elapsed_sec": 53748.367082834244, "step_time_sec": 8.239872080041096, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.020877838134766, "lr": 0.0002, "elapsed_sec": 53756.60573530197, "step_time_sec": 8.23849123402033, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.117302417755127, "lr": 0.0002, "elapsed_sec": 53764.84412026405, "step_time_sec": 8.238193690020125, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.023906707763672, "lr": 0.0002, "elapsed_sec": 53773.08304452896, "step_time_sec": 8.23880365199875, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9434707164764404, "lr": 0.0002, "elapsed_sec": 53781.322229623795, "step_time_sec": 8.238988347002305, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0530686378479, "lr": 0.0002, "elapsed_sec": 53789.56113195419, "step_time_sec": 8.238735852995887, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9436135292053223, "lr": 0.0002, "elapsed_sec": 53797.7991604805, "step_time_sec": 8.237881496024784, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.019955158233643, "lr": 0.0002, "elapsed_sec": 53806.038189888, "step_time_sec": 8.238830646965653, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0732221603393555, "lr": 0.0002, "elapsed_sec": 53814.27780652046, "step_time_sec": 8.239520456001628, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.962461471557617, "lr": 0.0002, "elapsed_sec": 53822.51808619499, "step_time_sec": 8.24004446901381, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.984222412109375, "lr": 0.0002, "elapsed_sec": 53830.757736206055, "step_time_sec": 8.239492458000313, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.982125759124756, "lr": 0.0002, "elapsed_sec": 53838.99586701393, "step_time_sec": 8.237967938010115, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.011144638061523, "lr": 0.0002, "elapsed_sec": 53847.2359187603, "step_time_sec": 8.239928301016334, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.798954486846924, "lr": 0.0002, "elapsed_sec": 53855.47484111786, "step_time_sec": 8.238709057040978, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9386801719665527, "lr": 0.0002, "elapsed_sec": 53863.713685035706, "step_time_sec": 8.238733801990747, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.892949342727661, "lr": 0.0002, "elapsed_sec": 53871.95312356949, "step_time_sec": 8.239268928999081, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.963604688644409, "lr": 0.0002, "elapsed_sec": 53880.19076418877, "step_time_sec": 8.237465562997386, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.013904094696045, "lr": 0.0002, "elapsed_sec": 53888.43124175072, "step_time_sec": 8.240316680981778, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9379825592041016, "lr": 0.0002, "elapsed_sec": 53896.67149710655, "step_time_sec": 8.240065863996278, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9836769104003906, "lr": 0.0002, "elapsed_sec": 53904.90946769714, "step_time_sec": 8.237825807009358, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.006441116333008, "lr": 0.0002, "elapsed_sec": 53913.14985203743, "step_time_sec": 8.240218346996699, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7379093170166016, "lr": 0.0002, "elapsed_sec": 53921.38911128044, "step_time_sec": 8.23908967198804, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8638501167297363, "lr": 0.0002, "elapsed_sec": 53941.14211678505, "step_time_sec": 19.752829934994224, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9178695678710938, "lr": 0.0002, "elapsed_sec": 53949.37989234924, "step_time_sec": 8.237616878002882, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.953542947769165, "lr": 0.0002, "elapsed_sec": 53957.61861896515, "step_time_sec": 8.23854620900238, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8760952949523926, "lr": 0.0002, "elapsed_sec": 53965.85805988312, "step_time_sec": 8.239277373999357, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8361318111419678, "lr": 0.0002, "elapsed_sec": 53974.09616923332, "step_time_sec": 8.237987319997046, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.978682518005371, "lr": 0.0002, "elapsed_sec": 53982.336624622345, "step_time_sec": 8.240255253971554, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0108232498168945, "lr": 0.0002, "elapsed_sec": 53990.5766479969, "step_time_sec": 8.239832694001961, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.848372220993042, "lr": 0.0002, "elapsed_sec": 53998.81502914429, "step_time_sec": 8.238266030035447, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8878061771392822, "lr": 0.0002, "elapsed_sec": 54007.05719661713, "step_time_sec": 8.241972516989335, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9866890907287598, "lr": 0.0002, "elapsed_sec": 54015.295130968094, "step_time_sec": 8.237742258992512, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.087180137634277, "lr": 0.0002, "elapsed_sec": 54023.53505110741, "step_time_sec": 8.239763582998421, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9711754322052, "lr": 0.0002, "elapsed_sec": 54031.775905132294, "step_time_sec": 30.632778770988807, "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": 3.796015739440918, "lr": 0.0002, "elapsed_sec": 54062.43616652489, "step_time_sec": 8.26764329400612, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8317034244537354, "lr": 0.0002, "elapsed_sec": 54070.67051935196, "step_time_sec": 8.234166683978401, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9288370609283447, "lr": 0.0002, "elapsed_sec": 54078.910021066666, "step_time_sec": 8.239297663036268, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.941249370574951, "lr": 0.0002, "elapsed_sec": 54087.14971804619, "step_time_sec": 8.239586499985307, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.926854133605957, "lr": 0.0002, "elapsed_sec": 54095.38611268997, "step_time_sec": 8.23622278001858, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.864217758178711, "lr": 0.0002, "elapsed_sec": 54103.62553834915, "step_time_sec": 8.239245018048678, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7831032276153564, "lr": 0.0002, "elapsed_sec": 54111.862971782684, "step_time_sec": 8.237265364965424, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.083745956420898, "lr": 0.0002, "elapsed_sec": 54120.10392665863, "step_time_sec": 8.240790419047698, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.940272092819214, "lr": 0.0002, "elapsed_sec": 54128.34287714958, "step_time_sec": 8.238834852993023, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7807469367980957, "lr": 0.0002, "elapsed_sec": 54136.58057165146, "step_time_sec": 8.237491495034192, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9271016120910645, "lr": 0.0002, "elapsed_sec": 54144.81992268562, "step_time_sec": 8.239167321997229, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.948462963104248, "lr": 0.0002, "elapsed_sec": 54153.058574438095, "step_time_sec": 8.238497040991206, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8264033794403076, "lr": 0.0002, "elapsed_sec": 54161.299729824066, "step_time_sec": 8.241012702055741, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.912482261657715, "lr": 0.0002, "elapsed_sec": 54169.53906369209, "step_time_sec": 8.239170729997568, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.949326276779175, "lr": 0.0002, "elapsed_sec": 54177.77736592293, "step_time_sec": 8.238132398982998, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.868852138519287, "lr": 0.0002, "elapsed_sec": 54186.01874470711, "step_time_sec": 8.241223261051346, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8397505283355713, "lr": 0.0002, "elapsed_sec": 54194.25858259201, "step_time_sec": 8.239727779000532, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.98117995262146, "lr": 0.0002, "elapsed_sec": 54202.49758148193, "step_time_sec": 8.238803030981217, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.077290058135986, "lr": 0.0002, "elapsed_sec": 54210.73658776283, "step_time_sec": 8.238829118025023, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.906355857849121, "lr": 0.0002, "elapsed_sec": 54218.97363257408, "step_time_sec": 8.236837969976477, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.036564350128174, "lr": 0.0002, "elapsed_sec": 54227.21649432182, "step_time_sec": 8.24278155300999, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.836592197418213, "lr": 0.0002, "elapsed_sec": 54235.4561150074, "step_time_sec": 8.239406378997955, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8520700931549072, "lr": 0.0002, "elapsed_sec": 54243.69492816925, "step_time_sec": 8.23864971002331, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.969554901123047, "lr": 0.0002, "elapsed_sec": 54251.93563437462, "step_time_sec": 8.240549473965075, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.070241928100586, "lr": 0.0002, "elapsed_sec": 54260.174753427505, "step_time_sec": 8.23893642501207, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.109667778015137, "lr": 0.0002, "elapsed_sec": 54268.414301633835, "step_time_sec": 8.239424670988228, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.082476615905762, "lr": 0.0002, "elapsed_sec": 54276.65360879898, "step_time_sec": 8.23913880001055, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.865201711654663, "lr": 0.0002, "elapsed_sec": 54284.89220690727, "step_time_sec": 8.238452589022927, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.973123073577881, "lr": 0.0002, "elapsed_sec": 54293.13270163536, "step_time_sec": 8.240287938970141, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.073785781860352, "lr": 0.0002, "elapsed_sec": 54301.372853040695, "step_time_sec": 8.239978803962003, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8549623489379883, "lr": 0.0002, "elapsed_sec": 54309.611929655075, "step_time_sec": 8.238907416991424, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.882114887237549, "lr": 0.0002, "elapsed_sec": 54317.85196733475, "step_time_sec": 8.239870174031239, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.834014892578125, "lr": 0.0002, "elapsed_sec": 54326.09133839607, "step_time_sec": 8.239216546004172, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8811326026916504, "lr": 0.0002, "elapsed_sec": 54334.33191990852, "step_time_sec": 8.240510092990007, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9379076957702637, "lr": 0.0002, "elapsed_sec": 54342.5719537735, "step_time_sec": 8.239814483036753, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9552314281463623, "lr": 0.0002, "elapsed_sec": 54350.81025862694, "step_time_sec": 8.238122905022465, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8708643913269043, "lr": 0.0002, "elapsed_sec": 54359.05123138428, "step_time_sec": 8.240822535997722, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.870706558227539, "lr": 0.0002, "elapsed_sec": 54367.28997135162, "step_time_sec": 8.23857050598599, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.104381561279297, "lr": 0.0002, "elapsed_sec": 54375.52937030792, "step_time_sec": 8.239197154005524, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.96932053565979, "lr": 0.0002, "elapsed_sec": 54383.77065682411, "step_time_sec": 8.24110768904211, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9808285236358643, "lr": 0.0002, "elapsed_sec": 54392.00981640816, "step_time_sec": 8.239015038998332, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8620498180389404, "lr": 0.0002, "elapsed_sec": 54400.2513794899, "step_time_sec": 8.241397753998172, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7857437133789062, "lr": 0.0002, "elapsed_sec": 54408.490961551666, "step_time_sec": 8.239454227965325, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.869361400604248, "lr": 0.0002, "elapsed_sec": 54416.73060297966, "step_time_sec": 8.239466851984616, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.932908535003662, "lr": 0.0002, "elapsed_sec": 54424.97075009346, "step_time_sec": 8.239998193981592, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8991260528564453, "lr": 0.0002, "elapsed_sec": 54433.21185517311, "step_time_sec": 8.240930583968293, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0705413818359375, "lr": 0.0002, "elapsed_sec": 54441.45004081726, "step_time_sec": 8.23800333600957, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9487247467041016, "lr": 0.0002, "elapsed_sec": 54449.689895153046, "step_time_sec": 8.239701526996214, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.877476930618286, "lr": 0.0002, "elapsed_sec": 54457.9291536808, "step_time_sec": 8.239071533025708, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.042028903961182, "lr": 0.0002, "elapsed_sec": 54466.16806483269, "step_time_sec": 8.238714713021182, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8559746742248535, "lr": 0.0002, "elapsed_sec": 54474.407838344574, "step_time_sec": 8.239665252971463, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.05300760269165, "lr": 0.0002, "elapsed_sec": 54482.645404577255, "step_time_sec": 8.237387854023837, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.935818672180176, "lr": 0.0002, "elapsed_sec": 54490.88648438454, "step_time_sec": 8.240898110030685, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9033899307250977, "lr": 0.0002, "elapsed_sec": 54499.126905202866, "step_time_sec": 8.24022443703143, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9109692573547363, "lr": 0.0002, "elapsed_sec": 54507.366540670395, "step_time_sec": 8.239499633957166, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.830523729324341, "lr": 0.0002, "elapsed_sec": 54515.60619163513, "step_time_sec": 8.239507496997248, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.96002459526062, "lr": 0.0002, "elapsed_sec": 54523.84365820885, "step_time_sec": 8.23730522498954, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.00355339050293, "lr": 0.0002, "elapsed_sec": 54532.08481502533, "step_time_sec": 8.240989861020353, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.824058771133423, "lr": 0.0002, "elapsed_sec": 54540.32586169243, "step_time_sec": 8.240896985051222, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.866405487060547, "lr": 0.0002, "elapsed_sec": 54548.56881356239, "step_time_sec": 8.242810135008767, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8738532066345215, "lr": 0.0002, "elapsed_sec": 54556.81414055824, "step_time_sec": 8.24508830398554, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7773468494415283, "lr": 0.0002, "elapsed_sec": 54565.05352091789, "step_time_sec": 8.239178278949112, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8432600498199463, "lr": 0.0002, "elapsed_sec": 54573.292627334595, "step_time_sec": 8.238961596041918, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.950626850128174, "lr": 0.0002, "elapsed_sec": 54581.53401350975, "step_time_sec": 8.24120326898992, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9641897678375244, "lr": 0.0002, "elapsed_sec": 54589.772830963135, "step_time_sec": 8.23870123695815, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.939210891723633, "lr": 0.0002, "elapsed_sec": 54598.012986421585, "step_time_sec": 8.239957184996456, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9705381393432617, "lr": 0.0002, "elapsed_sec": 54606.253363609314, "step_time_sec": 8.240231444011442, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.04776668548584, "lr": 0.0002, "elapsed_sec": 54614.4933013916, "step_time_sec": 8.23975975299254, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9412505626678467, "lr": 0.0002, "elapsed_sec": 54622.73306679726, "step_time_sec": 8.239627983013634, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8589625358581543, "lr": 0.0002, "elapsed_sec": 54630.973950624466, "step_time_sec": 8.240729004959576, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.816277503967285, "lr": 0.0002, "elapsed_sec": 54639.21314191818, "step_time_sec": 8.238995274004992, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9977521896362305, "lr": 0.0002, "elapsed_sec": 54647.45346951485, "step_time_sec": 8.24017676297808, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.930265426635742, "lr": 0.0002, "elapsed_sec": 54655.69362998009, "step_time_sec": 8.240017325966619, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.890045642852783, "lr": 0.0002, "elapsed_sec": 54663.93233180046, "step_time_sec": 8.238555821008049, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.871429204940796, "lr": 0.0002, "elapsed_sec": 54672.171901226044, "step_time_sec": 8.239348296017852, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8680191040039062, "lr": 0.0002, "elapsed_sec": 54680.409880399704, "step_time_sec": 8.237826374010183, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.889378070831299, "lr": 0.0002, "elapsed_sec": 54688.64981126785, "step_time_sec": 8.239756022987422, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.938924551010132, "lr": 0.0002, "elapsed_sec": 54696.88940382004, "step_time_sec": 8.23948956897948, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8814921379089355, "lr": 0.0002, "elapsed_sec": 54705.12677335739, "step_time_sec": 8.237140010984149, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7626984119415283, "lr": 0.0002, "elapsed_sec": 54713.36797618866, "step_time_sec": 8.241019507986493, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8004603385925293, "lr": 0.0002, "elapsed_sec": 54721.60731673241, "step_time_sec": 8.239176456991117, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8225643634796143, "lr": 0.0002, "elapsed_sec": 54729.846652030945, "step_time_sec": 8.239150488981977, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.951023578643799, "lr": 0.0002, "elapsed_sec": 54738.08542346954, "step_time_sec": 8.238645425997674, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.079883575439453, "lr": 0.0002, "elapsed_sec": 54746.323313713074, "step_time_sec": 8.237769740982912, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8765974044799805, "lr": 0.0002, "elapsed_sec": 54754.56532907486, "step_time_sec": 8.241804269026034, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.039687156677246, "lr": 0.0002, "elapsed_sec": 54762.805585861206, "step_time_sec": 8.24012032803148, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.921868324279785, "lr": 0.0002, "elapsed_sec": 54771.04424405098, "step_time_sec": 8.238498918013647, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.071284770965576, "lr": 0.0002, "elapsed_sec": 54779.28498220444, "step_time_sec": 8.240577467018738, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.862701892852783, "lr": 0.0002, "elapsed_sec": 54787.52408742905, "step_time_sec": 8.238902342040092, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.976705312728882, "lr": 0.0002, "elapsed_sec": 54795.76395344734, "step_time_sec": 8.239689244015608, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.985748052597046, "lr": 0.0002, "elapsed_sec": 54804.00347495079, "step_time_sec": 8.239426699990872, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7820546627044678, "lr": 0.0002, "elapsed_sec": 54812.24118089676, "step_time_sec": 8.237482646014541, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.011077404022217, "lr": 0.0002, "elapsed_sec": 54820.481189489365, "step_time_sec": 8.23981008096598, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.918595552444458, "lr": 0.0002, "elapsed_sec": 54828.72030425072, "step_time_sec": 8.238989111967385, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.060636043548584, "lr": 0.0002, "elapsed_sec": 54836.95935678482, "step_time_sec": 8.238901864970103, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.901395797729492, "lr": 0.0002, "elapsed_sec": 54845.19998073578, "step_time_sec": 8.240407592034899, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.054697513580322, "lr": 0.0002, "elapsed_sec": 54853.437264442444, "step_time_sec": 8.23716883501038, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.943406581878662, "lr": 0.0002, "elapsed_sec": 54861.67835068703, "step_time_sec": 8.240895747963805, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9076759815216064, "lr": 0.0002, "elapsed_sec": 54869.91928887367, "step_time_sec": 8.240772665012628, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8892953395843506, "lr": 0.0002, "elapsed_sec": 54878.15617656708, "step_time_sec": 8.236774019955192, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.891080856323242, "lr": 0.0002, "elapsed_sec": 54886.396782159805, "step_time_sec": 8.240452961996198, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.137224197387695, "lr": 0.0002, "elapsed_sec": 54894.63511323929, "step_time_sec": 8.238180865999311, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8568954467773438, "lr": 0.0002, "elapsed_sec": 54902.87511205673, "step_time_sec": 8.23984696000116, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.912449359893799, "lr": 0.0002, "elapsed_sec": 54911.11669540405, "step_time_sec": 8.241375988000073, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8891208171844482, "lr": 0.0002, "elapsed_sec": 54919.353585243225, "step_time_sec": 8.236744516005274, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.871131181716919, "lr": 0.0002, "elapsed_sec": 54927.592599630356, "step_time_sec": 8.238825137959793, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7615296840667725, "lr": 0.0002, "elapsed_sec": 54935.83185124397, "step_time_sec": 8.239068922004662, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.037793159484863, "lr": 0.0002, "elapsed_sec": 54944.07114481926, "step_time_sec": 8.239175565016922, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.906921148300171, "lr": 0.0002, "elapsed_sec": 54952.31184077263, "step_time_sec": 8.240472131001297, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9239213466644287, "lr": 0.0002, "elapsed_sec": 54960.55115199089, "step_time_sec": 8.239151018031407, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.07025671005249, "lr": 0.0002, "elapsed_sec": 54968.79214811325, "step_time_sec": 8.240850576024968, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.065872669219971, "lr": 0.0002, "elapsed_sec": 54977.0306699276, "step_time_sec": 8.238371519022621, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.920872688293457, "lr": 0.0002, "elapsed_sec": 54985.26925492287, "step_time_sec": 8.238378704001661, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9522571563720703, "lr": 0.0002, "elapsed_sec": 54993.50895667076, "step_time_sec": 8.239525540033355, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.141545295715332, "lr": 0.0002, "elapsed_sec": 55001.74871277809, "step_time_sec": 8.239609107025899, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8520991802215576, "lr": 0.0002, "elapsed_sec": 55009.98816037178, "step_time_sec": 8.239323596004397, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9869344234466553, "lr": 0.0002, "elapsed_sec": 55018.22794222832, "step_time_sec": 8.239563939976506, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.955918788909912, "lr": 0.0002, "elapsed_sec": 55026.46635532379, "step_time_sec": 8.238312658970244, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.010268211364746, "lr": 0.0002, "elapsed_sec": 55034.705909490585, "step_time_sec": 8.239333346951753, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.998732089996338, "lr": 0.0002, "elapsed_sec": 55042.94421362877, "step_time_sec": 8.238173498015385, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8296637535095215, "lr": 0.0002, "elapsed_sec": 55051.182559490204, "step_time_sec": 8.238150455988944, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7862331867218018, "lr": 0.0002, "elapsed_sec": 55059.421360731125, "step_time_sec": 8.238667306955904, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.20672082901001, "lr": 0.0002, "elapsed_sec": 55067.659108400345, "step_time_sec": 8.237579141976312, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.908282518386841, "lr": 0.0002, "elapsed_sec": 55075.89909410477, "step_time_sec": 8.239840985042974, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9793901443481445, "lr": 0.0002, "elapsed_sec": 55084.14058256149, "step_time_sec": 8.241326957999263, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9881575107574463, "lr": 0.0002, "elapsed_sec": 55092.37908744812, "step_time_sec": 8.238323397003114, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9293370246887207, "lr": 0.0002, "elapsed_sec": 55100.61949849129, "step_time_sec": 8.240219916973729, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.882486581802368, "lr": 0.0002, "elapsed_sec": 55108.85813546181, "step_time_sec": 8.23844998900313, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.113249778747559, "lr": 0.0002, "elapsed_sec": 55117.09842181206, "step_time_sec": 8.240173343976494, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.885348320007324, "lr": 0.0002, "elapsed_sec": 55125.33874678612, "step_time_sec": 8.240118417015765, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.058545112609863, "lr": 0.0002, "elapsed_sec": 55133.57669520378, "step_time_sec": 8.237788938975427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9794914722442627, "lr": 0.0002, "elapsed_sec": 55141.81714296341, "step_time_sec": 8.240283987019211, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9007437229156494, "lr": 0.0002, "elapsed_sec": 55150.0563147068, "step_time_sec": 8.239001684996765, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9597344398498535, "lr": 0.0002, "elapsed_sec": 55158.29544830322, "step_time_sec": 8.239024196052924, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8298299312591553, "lr": 0.0002, "elapsed_sec": 55166.53407120705, "step_time_sec": 8.238447284966242, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9284393787384033, "lr": 0.0002, "elapsed_sec": 55174.77206349373, "step_time_sec": 8.237795569992159, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.822780132293701, "lr": 0.0002, "elapsed_sec": 55183.012130498886, "step_time_sec": 8.239945889974479, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.723907232284546, "lr": 0.0002, "elapsed_sec": 55191.24998354912, "step_time_sec": 8.237648395996075, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8909623622894287, "lr": 0.0002, "elapsed_sec": 55199.489139318466, "step_time_sec": 8.239000372996088, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8858563899993896, "lr": 0.0002, "elapsed_sec": 55207.72994470596, "step_time_sec": 8.240649892017245, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.001953601837158, "lr": 0.0002, "elapsed_sec": 55215.968257427216, "step_time_sec": 8.238141731999349, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.961668014526367, "lr": 0.0002, "elapsed_sec": 55224.20895075798, "step_time_sec": 8.240565729967784, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9410529136657715, "lr": 0.0002, "elapsed_sec": 55232.44838142395, "step_time_sec": 8.239247065037489, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8809127807617188, "lr": 0.0002, "elapsed_sec": 55240.686128139496, "step_time_sec": 8.237559208006132, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8763742446899414, "lr": 0.0002, "elapsed_sec": 55248.925114393234, "step_time_sec": 8.238822531013284, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.989795446395874, "lr": 0.0002, "elapsed_sec": 55257.16310834885, "step_time_sec": 8.23787946102675, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.047483444213867, "lr": 0.0002, "elapsed_sec": 55265.402707099915, "step_time_sec": 8.239397345052566, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8385367393493652, "lr": 0.0002, "elapsed_sec": 55273.64170384407, "step_time_sec": 8.238845361978747, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9645843505859375, "lr": 0.0002, "elapsed_sec": 55281.879276275635, "step_time_sec": 8.237382830004208, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.869968891143799, "lr": 0.0002, "elapsed_sec": 55290.118465185165, "step_time_sec": 8.239028135023545, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.035573959350586, "lr": 0.0002, "elapsed_sec": 55298.356097221375, "step_time_sec": 8.237518537032884, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8956048488616943, "lr": 0.0002, "elapsed_sec": 55306.59522867203, "step_time_sec": 8.238893171015661, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9138569831848145, "lr": 0.0002, "elapsed_sec": 55314.836346149445, "step_time_sec": 8.240960447001271, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8852388858795166, "lr": 0.0002, "elapsed_sec": 55323.0734539032, "step_time_sec": 8.237000986002386, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.925466537475586, "lr": 0.0002, "elapsed_sec": 55331.31305575371, "step_time_sec": 8.239440231991466, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.004539966583252, "lr": 0.0002, "elapsed_sec": 55339.55174136162, "step_time_sec": 8.238491882046219, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7772016525268555, "lr": 0.0002, "elapsed_sec": 55347.790066957474, "step_time_sec": 8.238193964993116, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.972893238067627, "lr": 0.0002, "elapsed_sec": 55356.029557704926, "step_time_sec": 8.239296649000607, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9164979457855225, "lr": 0.0002, "elapsed_sec": 55364.26659679413, "step_time_sec": 8.236922020034399, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.932793378829956, "lr": 0.0002, "elapsed_sec": 55372.506048202515, "step_time_sec": 8.239243604009971, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9536893367767334, "lr": 0.0002, "elapsed_sec": 55380.7445704937, "step_time_sec": 8.23836537200259, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8766860961914062, "lr": 0.0002, "elapsed_sec": 55388.98443245888, "step_time_sec": 8.239715127972886, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.009826183319092, "lr": 0.0002, "elapsed_sec": 55397.22411298752, "step_time_sec": 8.239563757029828, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9762122631073, "lr": 0.0002, "elapsed_sec": 55405.46236228943, "step_time_sec": 8.238057052949443, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.081648349761963, "lr": 0.0002, "elapsed_sec": 55413.70159316063, "step_time_sec": 8.239103864005301, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.013877868652344, "lr": 0.0002, "elapsed_sec": 55421.94134807587, "step_time_sec": 8.239575724001043, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.205167293548584, "lr": 0.0002, "elapsed_sec": 55430.181391239166, "step_time_sec": 8.239845324016642, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.003230571746826, "lr": 0.0002, "elapsed_sec": 55438.4212038517, "step_time_sec": 8.239704499021173, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.973484992980957, "lr": 0.0002, "elapsed_sec": 55446.65935063362, "step_time_sec": 8.23800721700536, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.043490409851074, "lr": 0.0002, "elapsed_sec": 55454.900688648224, "step_time_sec": 8.241119179991074, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.99338436126709, "lr": 0.0002, "elapsed_sec": 55463.14021945, "step_time_sec": 8.23939129000064, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.081996440887451, "lr": 0.0002, "elapsed_sec": 55471.37762570381, "step_time_sec": 8.237187562976032, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.164555072784424, "lr": 0.0002, "elapsed_sec": 55479.616305828094, "step_time_sec": 8.238547398999799, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9913759231567383, "lr": 0.0002, "elapsed_sec": 55487.854077100754, "step_time_sec": 8.237606859998778, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9538462162017822, "lr": 0.0002, "elapsed_sec": 55496.093985557556, "step_time_sec": 8.239724799001124, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.799800157546997, "lr": 0.0002, "elapsed_sec": 55504.33307266235, "step_time_sec": 8.238987057993654, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.816662549972534, "lr": 0.0002, "elapsed_sec": 55512.570774555206, "step_time_sec": 8.237478490977082, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.875922679901123, "lr": 0.0002, "elapsed_sec": 55520.80954170227, "step_time_sec": 8.238621696014889, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.973761796951294, "lr": 0.0002, "elapsed_sec": 55529.04696440697, "step_time_sec": 8.237270855053794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.946556806564331, "lr": 0.0002, "elapsed_sec": 55537.28901553154, "step_time_sec": 8.24192583898548, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.981314182281494, "lr": 0.0002, "elapsed_sec": 55545.527639865875, "step_time_sec": 8.23841807997087, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.040491580963135, "lr": 0.0002, "elapsed_sec": 55553.76576280594, "step_time_sec": 8.237940674996935, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.856645107269287, "lr": 0.0002, "elapsed_sec": 55562.00473856926, "step_time_sec": 8.238833321025595, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.869938611984253, "lr": 0.0002, "elapsed_sec": 55570.24370265007, "step_time_sec": 8.23876348597696, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.17361307144165, "lr": 0.0002, "elapsed_sec": 55578.48264551163, "step_time_sec": 8.238851479021832, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.785615921020508, "lr": 0.0002, "elapsed_sec": 55586.721562862396, "step_time_sec": 8.238719307002611, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.898494243621826, "lr": 0.0002, "elapsed_sec": 55594.95949244499, "step_time_sec": 8.23777234199224, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.014037132263184, "lr": 0.0002, "elapsed_sec": 55603.199311733246, "step_time_sec": 8.239638708997518, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.012180328369141, "lr": 0.0002, "elapsed_sec": 55611.43595409393, "step_time_sec": 8.236472092976328, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.035093784332275, "lr": 0.0002, "elapsed_sec": 55619.67680597305, "step_time_sec": 8.24069706298178, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9446606636047363, "lr": 0.0002, "elapsed_sec": 55627.916719436646, "step_time_sec": 8.239762583980337, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.070714950561523, "lr": 0.0002, "elapsed_sec": 55636.15403389931, "step_time_sec": 8.237123837985564, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9052882194519043, "lr": 0.0002, "elapsed_sec": 55644.39205789566, "step_time_sec": 8.23787580197677, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.084108829498291, "lr": 0.0002, "elapsed_sec": 55652.63062429428, "step_time_sec": 8.238365412980784, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.037303924560547, "lr": 0.0002, "elapsed_sec": 55660.870067596436, "step_time_sec": 8.239284974057227, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.01933479309082, "lr": 0.0002, "elapsed_sec": 55669.10977268219, "step_time_sec": 8.239542733994313, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9465062618255615, "lr": 0.0002, "elapsed_sec": 55677.34729170799, "step_time_sec": 8.237392936018296, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9165613651275635, "lr": 0.0002, "elapsed_sec": 55685.58614587784, "step_time_sec": 8.238683048053645, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.179455757141113, "lr": 0.0002, "elapsed_sec": 55693.82433342934, "step_time_sec": 8.238025820988696, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9880266189575195, "lr": 0.0002, "elapsed_sec": 55702.06414556503, "step_time_sec": 8.239642448956147, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.056398391723633, "lr": 0.0002, "elapsed_sec": 55710.30401015282, "step_time_sec": 8.239663058018778, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0095086097717285, "lr": 0.0002, "elapsed_sec": 55718.5426363945, "step_time_sec": 8.23851942498004, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.988994598388672, "lr": 0.0002, "elapsed_sec": 55726.782898664474, "step_time_sec": 8.240070694009773, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.947988510131836, "lr": 0.0002, "elapsed_sec": 55735.02335071564, "step_time_sec": 8.240320469019935, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9439151287078857, "lr": 0.0002, "elapsed_sec": 55743.262878894806, "step_time_sec": 8.239317387982737, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.014496803283691, "lr": 0.0002, "elapsed_sec": 55751.52220106125, "step_time_sec": 8.25922643498052, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9795167446136475, "lr": 0.0002, "elapsed_sec": 55759.76253271103, "step_time_sec": 8.24009826296242, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8670670986175537, "lr": 0.0002, "elapsed_sec": 55768.00085926056, "step_time_sec": 8.23817412200151, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.931373119354248, "lr": 0.0002, "elapsed_sec": 55776.240978717804, "step_time_sec": 8.239951854047831, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.855891227722168, "lr": 0.0002, "elapsed_sec": 55784.48091650009, "step_time_sec": 8.23977261397522, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0146708488464355, "lr": 0.0002, "elapsed_sec": 55792.71947026253, "step_time_sec": 8.238388194004074, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.744743824005127, "lr": 0.0002, "elapsed_sec": 55800.9615790844, "step_time_sec": 8.241951317992061, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9218459129333496, "lr": 0.0002, "elapsed_sec": 55809.200919151306, "step_time_sec": 8.239207752980292, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9917664527893066, "lr": 0.0002, "elapsed_sec": 55817.439445734024, "step_time_sec": 8.238312505010981, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9598474502563477, "lr": 0.0002, "elapsed_sec": 55825.67822599411, "step_time_sec": 8.238636670983396, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.927927255630493, "lr": 0.0002, "elapsed_sec": 55833.915761232376, "step_time_sec": 8.237421389028896, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.00515079498291, "lr": 0.0002, "elapsed_sec": 55842.15691614151, "step_time_sec": 8.240983146009967, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.003079414367676, "lr": 0.0002, "elapsed_sec": 55850.39572644234, "step_time_sec": 8.238589500018861, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.942888021469116, "lr": 0.0002, "elapsed_sec": 55858.63410615921, "step_time_sec": 8.238269140012562, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.973400592803955, "lr": 0.0002, "elapsed_sec": 55866.873044490814, "step_time_sec": 8.23872865102021, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.994680166244507, "lr": 0.0002, "elapsed_sec": 55875.1103181839, "step_time_sec": 8.237128567998298, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.992034912109375, "lr": 0.0002, "elapsed_sec": 55883.35067677498, "step_time_sec": 8.24020824499894, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.954686403274536, "lr": 0.0002, "elapsed_sec": 55891.59148478508, "step_time_sec": 8.240644031960983, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.024511814117432, "lr": 0.0002, "elapsed_sec": 55899.82934546471, "step_time_sec": 8.2377366059809, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9707252979278564, "lr": 0.0002, "elapsed_sec": 55908.070026397705, "step_time_sec": 8.240471293975133, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9803307056427, "lr": 0.0002, "elapsed_sec": 55916.30813956261, "step_time_sec": 8.237998170952778, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9596755504608154, "lr": 0.0002, "elapsed_sec": 55924.54721570015, "step_time_sec": 8.23891241295496, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9716970920562744, "lr": 0.0002, "elapsed_sec": 55932.78601741791, "step_time_sec": 8.238634031033143, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.967466115951538, "lr": 0.0002, "elapsed_sec": 55941.02344918251, "step_time_sec": 8.23723259801045, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.903787851333618, "lr": 0.0002, "elapsed_sec": 55949.261986494064, "step_time_sec": 8.238385003001895, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9419100284576416, "lr": 0.0002, "elapsed_sec": 55957.50006389618, "step_time_sec": 8.237900239997543, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.079436779022217, "lr": 0.0002, "elapsed_sec": 55965.73890566826, "step_time_sec": 8.238697747001424, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.887969732284546, "lr": 0.0002, "elapsed_sec": 55973.97773647308, "step_time_sec": 8.238742007000837, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8769032955169678, "lr": 0.0002, "elapsed_sec": 55982.21638584137, "step_time_sec": 8.238440450048074, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.07723331451416, "lr": 0.0002, "elapsed_sec": 55990.456975221634, "step_time_sec": 8.240414279978722, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8977720737457275, "lr": 0.0002, "elapsed_sec": 55998.694536447525, "step_time_sec": 8.237411494017579, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9486420154571533, "lr": 0.0002, "elapsed_sec": 56006.93478155136, "step_time_sec": 8.240057358983904, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8547868728637695, "lr": 0.0002, "elapsed_sec": 56015.173320531845, "step_time_sec": 8.23840170702897, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9666295051574707, "lr": 0.0002, "elapsed_sec": 56023.41088914871, "step_time_sec": 8.237367483961862, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8635969161987305, "lr": 0.0002, "elapsed_sec": 56031.65132689476, "step_time_sec": 8.240272908995394, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.039173126220703, "lr": 0.0002, "elapsed_sec": 56039.890704870224, "step_time_sec": 8.239227860991377, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9419119358062744, "lr": 0.0002, "elapsed_sec": 56048.130685567856, "step_time_sec": 8.239843037968967, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9822402000427246, "lr": 0.0002, "elapsed_sec": 56056.3708012104, "step_time_sec": 8.239939431019593, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9332263469696045, "lr": 0.0002, "elapsed_sec": 56064.60821676254, "step_time_sec": 8.237260399968363, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8036000728607178, "lr": 0.0002, "elapsed_sec": 56072.8480155468, "step_time_sec": 8.239638706028927, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.998910427093506, "lr": 0.0002, "elapsed_sec": 56081.08713579178, "step_time_sec": 8.238933151995298, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9661877155303955, "lr": 0.0002, "elapsed_sec": 56089.325051784515, "step_time_sec": 8.237773569009732, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.993720054626465, "lr": 0.0002, "elapsed_sec": 56097.566969156265, "step_time_sec": 8.241735662973952, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.917814254760742, "lr": 0.0002, "elapsed_sec": 56105.80543422699, "step_time_sec": 8.238313795009162, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8920857906341553, "lr": 0.0002, "elapsed_sec": 56114.04526877403, "step_time_sec": 8.239716411044355, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.030815124511719, "lr": 0.0002, "elapsed_sec": 56122.28570985794, "step_time_sec": 8.240245449997019, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.066531658172607, "lr": 0.0002, "elapsed_sec": 56130.52324509621, "step_time_sec": 8.237377304991242, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.896127223968506, "lr": 0.0002, "elapsed_sec": 56138.76295661926, "step_time_sec": 8.23952934902627, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.928889036178589, "lr": 0.0002, "elapsed_sec": 56147.00177645683, "step_time_sec": 8.238620466028806, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.990788698196411, "lr": 0.0002, "elapsed_sec": 56155.241153001785, "step_time_sec": 8.239217087975703, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.085702896118164, "lr": 0.0002, "elapsed_sec": 56163.48030877113, "step_time_sec": 8.239004746021237, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.94232177734375, "lr": 0.0002, "elapsed_sec": 56171.7189116478, "step_time_sec": 8.238399577036034, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.921445369720459, "lr": 0.0002, "elapsed_sec": 56179.96017074585, "step_time_sec": 8.241106259985827, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.986551284790039, "lr": 0.0002, "elapsed_sec": 56188.19974184036, "step_time_sec": 8.239419199002441, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.92621111869812, "lr": 0.0002, "elapsed_sec": 56196.438086509705, "step_time_sec": 8.238180120009929, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.946894884109497, "lr": 0.0002, "elapsed_sec": 56204.67685055733, "step_time_sec": 8.238613009976689, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.900592088699341, "lr": 0.0002, "elapsed_sec": 56212.9144487381, "step_time_sec": 8.237423599988688, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.889585018157959, "lr": 0.0002, "elapsed_sec": 56221.154616594315, "step_time_sec": 8.24001719703665, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9718596935272217, "lr": 0.0002, "elapsed_sec": 56229.394807338715, "step_time_sec": 8.24006116896635, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.899707317352295, "lr": 0.0002, "elapsed_sec": 56237.63326764107, "step_time_sec": 8.23829075199319, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.763315439224243, "lr": 0.0002, "elapsed_sec": 56245.87402224541, "step_time_sec": 8.240588029031642, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.976727247238159, "lr": 0.0002, "elapsed_sec": 56254.11273312569, "step_time_sec": 8.238589336979203, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8863346576690674, "lr": 0.0002, "elapsed_sec": 56262.35200858116, "step_time_sec": 8.239070571027696, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9847359657287598, "lr": 0.0002, "elapsed_sec": 56270.592400074005, "step_time_sec": 8.24021055502817, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.093207359313965, "lr": 0.0002, "elapsed_sec": 56278.83023524284, "step_time_sec": 8.237728886015248, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9319655895233154, "lr": 0.0002, "elapsed_sec": 56287.070977687836, "step_time_sec": 8.240520610008389, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.973802089691162, "lr": 0.0002, "elapsed_sec": 56295.311052560806, "step_time_sec": 8.239899650972802, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.794856071472168, "lr": 0.0002, "elapsed_sec": 56303.550371170044, "step_time_sec": 8.239165725011844, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0356526374816895, "lr": 0.0002, "elapsed_sec": 56311.790239572525, "step_time_sec": 8.239687707042322, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.010056972503662, "lr": 0.0002, "elapsed_sec": 56320.02696156502, "step_time_sec": 8.236570114968345, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9358103275299072, "lr": 0.0002, "elapsed_sec": 56328.266479730606, "step_time_sec": 8.239365907968022, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.97593092918396, "lr": 0.0002, "elapsed_sec": 56336.50708580017, "step_time_sec": 8.240468994015828, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9392614364624023, "lr": 0.0002, "elapsed_sec": 56344.74467134476, "step_time_sec": 8.237362750980537, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.105329513549805, "lr": 0.0002, "elapsed_sec": 56352.98520350456, "step_time_sec": 8.24039420596091, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.015796661376953, "lr": 0.0002, "elapsed_sec": 56361.22301363945, "step_time_sec": 8.237598131992854, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.909001588821411, "lr": 0.0002, "elapsed_sec": 56369.46295285225, "step_time_sec": 8.239801505987998, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.91190767288208, "lr": 0.0002, "elapsed_sec": 56377.701780080795, "step_time_sec": 8.238670683000237, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8075037002563477, "lr": 0.0002, "elapsed_sec": 56385.93916344643, "step_time_sec": 8.237193920998834, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.85710072517395, "lr": 0.0002, "elapsed_sec": 56394.177903175354, "step_time_sec": 8.238668972975574, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.910494327545166, "lr": 0.0002, "elapsed_sec": 56402.416957855225, "step_time_sec": 8.238832184986677, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.096343517303467, "lr": 0.0002, "elapsed_sec": 56410.6579105854, "step_time_sec": 8.240781823988073, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.016355991363525, "lr": 0.0002, "elapsed_sec": 56418.89794397354, "step_time_sec": 8.239888014039025, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9024553298950195, "lr": 0.0002, "elapsed_sec": 56427.136471033096, "step_time_sec": 8.238335757981986, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.872807502746582, "lr": 0.0002, "elapsed_sec": 56435.37613773346, "step_time_sec": 8.239476386050228, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.091093063354492, "lr": 0.0002, "elapsed_sec": 56443.61640262604, "step_time_sec": 8.240128392004408, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9491922855377197, "lr": 0.0002, "elapsed_sec": 56451.85519838333, "step_time_sec": 8.238676698994823, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8736555576324463, "lr": 0.0002, "elapsed_sec": 56460.09609055519, "step_time_sec": 8.240678575006314, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9661223888397217, "lr": 0.0002, "elapsed_sec": 56468.333419799805, "step_time_sec": 8.237166897044517, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.025844573974609, "lr": 0.0002, "elapsed_sec": 56476.57356619835, "step_time_sec": 8.239981641992927, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.909283399581909, "lr": 0.0002, "elapsed_sec": 56484.813571453094, "step_time_sec": 8.23984613304492, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8701376914978027, "lr": 0.0002, "elapsed_sec": 56493.050288677216, "step_time_sec": 8.23658525798237, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8191683292388916, "lr": 0.0002, "elapsed_sec": 56501.289568424225, "step_time_sec": 8.23906440497376, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.041684150695801, "lr": 0.0002, "elapsed_sec": 56509.52657985687, "step_time_sec": 8.236857755982783, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8854496479034424, "lr": 0.0002, "elapsed_sec": 56517.76739168167, "step_time_sec": 8.24071995401755, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.93011736869812, "lr": 0.0002, "elapsed_sec": 56526.008130550385, "step_time_sec": 8.240525718021672, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9266467094421387, "lr": 0.0002, "elapsed_sec": 56534.246077775955, "step_time_sec": 8.237790628976654, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8518733978271484, "lr": 0.0002, "elapsed_sec": 56542.48697042465, "step_time_sec": 8.240673274965957, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8428900241851807, "lr": 0.0002, "elapsed_sec": 56550.72525238991, "step_time_sec": 8.23814038594719, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.982330799102783, "lr": 0.0002, "elapsed_sec": 56558.96440768242, "step_time_sec": 8.23903027601773, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.983916997909546, "lr": 0.0002, "elapsed_sec": 56567.20319032669, "step_time_sec": 8.238595460017677, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.881096601486206, "lr": 0.0002, "elapsed_sec": 56575.44075870514, "step_time_sec": 8.237392020004336, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.09675931930542, "lr": 0.0002, "elapsed_sec": 56583.68215084076, "step_time_sec": 8.241248387028463, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8521955013275146, "lr": 0.0002, "elapsed_sec": 56591.92013192177, "step_time_sec": 8.237829493999016, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.039164066314697, "lr": 0.0002, "elapsed_sec": 56600.1596827507, "step_time_sec": 8.2393762160209, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8329520225524902, "lr": 0.0002, "elapsed_sec": 56608.398897886276, "step_time_sec": 8.239103003987111, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9298102855682373, "lr": 0.0002, "elapsed_sec": 56616.63777065277, "step_time_sec": 8.238671417988371, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.841557025909424, "lr": 0.0002, "elapsed_sec": 56624.877135038376, "step_time_sec": 8.239202009979635, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.891873836517334, "lr": 0.0002, "elapsed_sec": 56633.11478829384, "step_time_sec": 8.237510738021228, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0517377853393555, "lr": 0.0002, "elapsed_sec": 56641.35444974899, "step_time_sec": 8.239450425026007, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9489660263061523, "lr": 0.0002, "elapsed_sec": 56649.594266176224, "step_time_sec": 8.239696168981027, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.136596202850342, "lr": 0.0002, "elapsed_sec": 56657.83299398422, "step_time_sec": 8.238555620948318, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.012908935546875, "lr": 0.0002, "elapsed_sec": 56666.074279785156, "step_time_sec": 8.24110019596992, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.790330410003662, "lr": 0.0002, "elapsed_sec": 56674.31451129913, "step_time_sec": 8.240100499999244, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.999500036239624, "lr": 0.0002, "elapsed_sec": 56682.55131649971, "step_time_sec": 8.236636947025545, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.106326580047607, "lr": 0.0002, "elapsed_sec": 56690.7908885479, "step_time_sec": 8.239423152001109, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.898899555206299, "lr": 0.0002, "elapsed_sec": 56699.02979040146, "step_time_sec": 8.238752976001706, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.836104154586792, "lr": 0.0002, "elapsed_sec": 56707.270516872406, "step_time_sec": 8.240508862014394, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9125936031341553, "lr": 0.0002, "elapsed_sec": 56715.50993824005, "step_time_sec": 8.239274306979496, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.896902084350586, "lr": 0.0002, "elapsed_sec": 56723.748036146164, "step_time_sec": 8.23797769198427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.917510747909546, "lr": 0.0002, "elapsed_sec": 56731.98674726486, "step_time_sec": 8.23851419397397, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9562082290649414, "lr": 0.0002, "elapsed_sec": 56740.22546887398, "step_time_sec": 8.238562287995592, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.882638931274414, "lr": 0.0002, "elapsed_sec": 56748.464327812195, "step_time_sec": 8.238716037012637, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.008416175842285, "lr": 0.0002, "elapsed_sec": 56756.70545601845, "step_time_sec": 8.240957982023247, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.83872652053833, "lr": 0.0002, "elapsed_sec": 56764.94311738014, "step_time_sec": 8.237494409026112, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.036539554595947, "lr": 0.0002, "elapsed_sec": 56773.183421850204, "step_time_sec": 8.240189784031827, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8651585578918457, "lr": 0.0002, "elapsed_sec": 56781.42271208763, "step_time_sec": 8.239080170053057, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0192155838012695, "lr": 0.0002, "elapsed_sec": 56789.66071295738, "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": 6833, "loss": 3.9969208240509033, "lr": 0.0002, "elapsed_sec": 56797.90085172653, "step_time_sec": 8.239979600999504, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.031489849090576, "lr": 0.0002, "elapsed_sec": 56806.137781858444, "step_time_sec": 8.236748063005507, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.864008665084839, "lr": 0.0002, "elapsed_sec": 56814.37811231613, "step_time_sec": 8.2402068930096, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.945443868637085, "lr": 0.0002, "elapsed_sec": 56822.61776638031, "step_time_sec": 8.239439167024102, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9369096755981445, "lr": 0.0002, "elapsed_sec": 56830.85643815994, "step_time_sec": 8.238501344982069, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.943070650100708, "lr": 0.0002, "elapsed_sec": 56839.095643758774, "step_time_sec": 8.239023384987377, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8881750106811523, "lr": 0.0002, "elapsed_sec": 56847.33575081825, "step_time_sec": 8.239964630978648, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8680596351623535, "lr": 0.0002, "elapsed_sec": 56855.575672626495, "step_time_sec": 8.239761573961005, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9824330806732178, "lr": 0.0002, "elapsed_sec": 56863.8145570755, "step_time_sec": 8.238739820022602, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7773520946502686, "lr": 0.0002, "elapsed_sec": 56872.05333828926, "step_time_sec": 8.238602999015711, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9039368629455566, "lr": 0.0002, "elapsed_sec": 56880.29304432869, "step_time_sec": 8.239531903003808, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8824872970581055, "lr": 0.0002, "elapsed_sec": 56888.53263068199, "step_time_sec": 8.239452989015263, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.807563066482544, "lr": 0.0002, "elapsed_sec": 56896.77222943306, "step_time_sec": 8.239426343003288, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.807795763015747, "lr": 0.0002, "elapsed_sec": 56905.080362558365, "step_time_sec": 8.245668504969217, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8669352531433105, "lr": 0.0002, "elapsed_sec": 56913.321593523026, "step_time_sec": 8.241047062969301, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9822731018066406, "lr": 0.0002, "elapsed_sec": 56921.55925321579, "step_time_sec": 8.237493471999187, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9998438358306885, "lr": 0.0002, "elapsed_sec": 56929.79935479164, "step_time_sec": 8.24000554898521, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9489963054656982, "lr": 0.0002, "elapsed_sec": 56938.04043030739, "step_time_sec": 8.240879060991574, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.973935604095459, "lr": 0.0002, "elapsed_sec": 56946.27912855148, "step_time_sec": 8.238543535990175, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.836230516433716, "lr": 0.0002, "elapsed_sec": 56954.520683288574, "step_time_sec": 8.241392390977126, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9131548404693604, "lr": 0.0002, "elapsed_sec": 56962.75904130936, "step_time_sec": 8.238140041008592, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.012600421905518, "lr": 0.0002, "elapsed_sec": 56970.99738693237, "step_time_sec": 8.238209607021417, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.022362232208252, "lr": 0.0002, "elapsed_sec": 56979.236842632294, "step_time_sec": 8.239267982018646, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.878511905670166, "lr": 0.0002, "elapsed_sec": 56987.47430038452, "step_time_sec": 8.237286083051004, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.6131937503814697, "lr": 0.0002, "elapsed_sec": 56995.714688539505, "step_time_sec": 8.240226087975316, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9730756282806396, "lr": 0.0002, "elapsed_sec": 57003.954157590866, "step_time_sec": 8.23937050101813, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9844603538513184, "lr": 0.0002, "elapsed_sec": 57012.192662000656, "step_time_sec": 8.238295315008145, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8529202938079834, "lr": 0.0002, "elapsed_sec": 57020.432990312576, "step_time_sec": 8.24015213700477, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7986371517181396, "lr": 0.0002, "elapsed_sec": 57028.67239522934, "step_time_sec": 8.239257624954917, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9247517585754395, "lr": 0.0002, "elapsed_sec": 57036.91205692291, "step_time_sec": 8.23951953498181, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8782434463500977, "lr": 0.0002, "elapsed_sec": 57045.150195360184, "step_time_sec": 8.237965048989281, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.005731105804443, "lr": 0.0002, "elapsed_sec": 57053.387271404266, "step_time_sec": 8.236910636012908, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.011485576629639, "lr": 0.0002, "elapsed_sec": 57061.62719607353, "step_time_sec": 8.239746447012294, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8755533695220947, "lr": 0.0002, "elapsed_sec": 57069.866263628006, "step_time_sec": 8.238886091974564, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8835434913635254, "lr": 0.0002, "elapsed_sec": 57078.10460305214, "step_time_sec": 8.238216158002615, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.822643995285034, "lr": 0.0002, "elapsed_sec": 57086.34533071518, "step_time_sec": 8.24058297497686, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7537057399749756, "lr": 0.0002, "elapsed_sec": 57094.58424592018, "step_time_sec": 8.238779793959111, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7956714630126953, "lr": 0.0002, "elapsed_sec": 57102.82480978966, "step_time_sec": 8.240338900999632, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.85737681388855, "lr": 0.0002, "elapsed_sec": 57111.06500029564, "step_time_sec": 8.240048281033523, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.046842098236084, "lr": 0.0002, "elapsed_sec": 57119.30538535118, "step_time_sec": 8.240273959003389, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.058998107910156, "lr": 0.0002, "elapsed_sec": 57127.54448771477, "step_time_sec": 8.238903046993073, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.90606951713562, "lr": 0.0002, "elapsed_sec": 57135.78488326073, "step_time_sec": 8.240210746997036, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.98116135597229, "lr": 0.0002, "elapsed_sec": 57144.02507066727, "step_time_sec": 8.240095916960854, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.971214532852173, "lr": 0.0002, "elapsed_sec": 57152.26526212692, "step_time_sec": 8.239988000015728, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8658666610717773, "lr": 0.0002, "elapsed_sec": 57160.504128456116, "step_time_sec": 8.238685054006055, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8849403858184814, "lr": 0.0002, "elapsed_sec": 57168.744550943375, "step_time_sec": 8.240281047997996, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.862447738647461, "lr": 0.0002, "elapsed_sec": 57176.984719753265, "step_time_sec": 8.239993422990665, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8596301078796387, "lr": 0.0002, "elapsed_sec": 57185.22239732742, "step_time_sec": 8.237512979016174, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8154022693634033, "lr": 0.0002, "elapsed_sec": 57193.462104320526, "step_time_sec": 8.239562497008592, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.747514486312866, "lr": 0.0002, "elapsed_sec": 57201.700770139694, "step_time_sec": 8.238452379009686, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.083434104919434, "lr": 0.0002, "elapsed_sec": 57209.94057607651, "step_time_sec": 8.239651184994727, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.060780048370361, "lr": 0.0002, "elapsed_sec": 57218.18017244339, "step_time_sec": 8.23948751098942, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.933271646499634, "lr": 0.0002, "elapsed_sec": 57226.41801524162, "step_time_sec": 8.23763443500502, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.911907434463501, "lr": 0.0002, "elapsed_sec": 57234.65846157074, "step_time_sec": 8.2403236280079, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.939422130584717, "lr": 0.0002, "elapsed_sec": 57242.897065877914, "step_time_sec": 8.238389478996396, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.970808744430542, "lr": 0.0002, "elapsed_sec": 57251.13701915741, "step_time_sec": 8.239846784970723, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9500222206115723, "lr": 0.0002, "elapsed_sec": 57259.3765835762, "step_time_sec": 8.239367712987587, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.944256544113159, "lr": 0.0002, "elapsed_sec": 57267.614751815796, "step_time_sec": 8.237991809961386, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.885279417037964, "lr": 0.0002, "elapsed_sec": 57275.854647397995, "step_time_sec": 8.239725047023967, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.971214771270752, "lr": 0.0002, "elapsed_sec": 57284.0936088562, "step_time_sec": 8.238851700967643, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.925567626953125, "lr": 0.0002, "elapsed_sec": 57292.3326728344, "step_time_sec": 8.238858942990191, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.901897668838501, "lr": 0.0002, "elapsed_sec": 57300.571613550186, "step_time_sec": 8.238780299026985, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.869126319885254, "lr": 0.0002, "elapsed_sec": 57308.80816459656, "step_time_sec": 8.236405107949395, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8951754570007324, "lr": 0.0002, "elapsed_sec": 57317.04940748215, "step_time_sec": 8.241079675033689, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9187066555023193, "lr": 0.0002, "elapsed_sec": 57325.28791260719, "step_time_sec": 8.238344948040321, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.826810836791992, "lr": 0.0002, "elapsed_sec": 57333.526223659515, "step_time_sec": 8.238143388007302, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.966205596923828, "lr": 0.0002, "elapsed_sec": 57341.76592516899, "step_time_sec": 8.23951725900406, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7242393493652344, "lr": 0.0002, "elapsed_sec": 57350.003787755966, "step_time_sec": 8.237729564018082, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.923696994781494, "lr": 0.0002, "elapsed_sec": 57358.24401926994, "step_time_sec": 8.240066787984688, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9346487522125244, "lr": 0.0002, "elapsed_sec": 57366.48324632645, "step_time_sec": 8.2390963640064, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.89151930809021, "lr": 0.0002, "elapsed_sec": 57374.723052978516, "step_time_sec": 8.239620183012448, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7606756687164307, "lr": 0.0002, "elapsed_sec": 57382.96280360222, "step_time_sec": 8.239563500974327, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7750964164733887, "lr": 0.0002, "elapsed_sec": 57391.202120780945, "step_time_sec": 8.2391635820386, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.695495367050171, "lr": 0.0002, "elapsed_sec": 57399.441673755646, "step_time_sec": 8.239396603952628, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9612579345703125, "lr": 0.0002, "elapsed_sec": 57407.681835889816, "step_time_sec": 8.239982973958831, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8070764541625977, "lr": 0.0002, "elapsed_sec": 57415.91994929314, "step_time_sec": 8.238003269943874, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9995174407958984, "lr": 0.0002, "elapsed_sec": 57424.16019320488, "step_time_sec": 8.240085251047276, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9197943210601807, "lr": 0.0002, "elapsed_sec": 57432.399228572845, "step_time_sec": 8.238823972991668, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.978628635406494, "lr": 0.0002, "elapsed_sec": 57440.63885593414, "step_time_sec": 8.239473684050608, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.814357042312622, "lr": 0.0002, "elapsed_sec": 57448.87882089615, "step_time_sec": 8.23979480098933, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9733171463012695, "lr": 0.0002, "elapsed_sec": 57457.11732530594, "step_time_sec": 8.238403528986964, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.904226779937744, "lr": 0.0002, "elapsed_sec": 57465.357716560364, "step_time_sec": 8.240181491011754, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.830960988998413, "lr": 0.0002, "elapsed_sec": 57473.59532046318, "step_time_sec": 8.237484150042292, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.041942596435547, "lr": 0.0002, "elapsed_sec": 57481.834965229034, "step_time_sec": 8.239447743981145, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8581554889678955, "lr": 0.0002, "elapsed_sec": 57490.07543039322, "step_time_sec": 8.240312943002209, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7033896446228027, "lr": 0.0002, "elapsed_sec": 57498.31416988373, "step_time_sec": 8.238558288023341, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.06596040725708, "lr": 0.0002, "elapsed_sec": 57506.554615974426, "step_time_sec": 8.240297726006247, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9612948894500732, "lr": 0.0002, "elapsed_sec": 57514.79292201996, "step_time_sec": 8.238153401995078, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.6778249740600586, "lr": 0.0002, "elapsed_sec": 57523.0307765007, "step_time_sec": 8.237667911977042, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8898096084594727, "lr": 0.0002, "elapsed_sec": 57531.271709918976, "step_time_sec": 8.240763101028278, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9388692378997803, "lr": 0.0002, "elapsed_sec": 57539.51136112213, "step_time_sec": 8.239512064028531, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.878091335296631, "lr": 0.0002, "elapsed_sec": 57547.75056695938, "step_time_sec": 8.239040540007409, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.244443893432617, "lr": 0.0002, "elapsed_sec": 57555.990421533585, "step_time_sec": 8.239724670012947, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.85357928276062, "lr": 0.0002, "elapsed_sec": 57564.227967739105, "step_time_sec": 8.237400721991435, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9327402114868164, "lr": 0.0002, "elapsed_sec": 57572.46764397621, "step_time_sec": 8.239450000983197, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8516459465026855, "lr": 0.0002, "elapsed_sec": 57580.70660614967, "step_time_sec": 8.238805453001987, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.799048900604248, "lr": 0.0002, "elapsed_sec": 57588.944513082504, "step_time_sec": 8.237762168981135, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9376301765441895, "lr": 0.0002, "elapsed_sec": 57597.18436574936, "step_time_sec": 8.239713720045984, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8949782848358154, "lr": 0.0002, "elapsed_sec": 57605.422411203384, "step_time_sec": 8.237854240986053, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.760523557662964, "lr": 0.0002, "elapsed_sec": 57613.66380786896, "step_time_sec": 8.241256480047014, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9850566387176514, "lr": 0.0002, "elapsed_sec": 57621.90220451355, "step_time_sec": 8.238201548054349, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.88750958442688, "lr": 0.0002, "elapsed_sec": 57630.140384197235, "step_time_sec": 8.238054955028929, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.007075309753418, "lr": 0.0002, "elapsed_sec": 57638.38013744354, "step_time_sec": 8.239568887045607, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9251151084899902, "lr": 0.0002, "elapsed_sec": 57646.618988990784, "step_time_sec": 8.238697954977397, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8034090995788574, "lr": 0.0002, "elapsed_sec": 57654.85852956772, "step_time_sec": 8.239406870037783, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.910463809967041, "lr": 0.0002, "elapsed_sec": 57663.098516225815, "step_time_sec": 8.239814164000563, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.968017339706421, "lr": 0.0002, "elapsed_sec": 57671.33670639992, "step_time_sec": 8.238048814993817, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.028711795806885, "lr": 0.0002, "elapsed_sec": 57679.57722854614, "step_time_sec": 8.240331545006484, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.742037534713745, "lr": 0.0002, "elapsed_sec": 57687.81687664986, "step_time_sec": 8.23948271095287, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.794152021408081, "lr": 0.0002, "elapsed_sec": 57696.0565366745, "step_time_sec": 8.239526360994205, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.820786952972412, "lr": 0.0002, "elapsed_sec": 57704.29664468765, "step_time_sec": 8.239959653990809, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.004482746124268, "lr": 0.0002, "elapsed_sec": 57712.53444004059, "step_time_sec": 8.237565636984073, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.891383409500122, "lr": 0.0002, "elapsed_sec": 57720.77431178093, "step_time_sec": 8.23973680799827, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9542696475982666, "lr": 0.0002, "elapsed_sec": 57729.01335167885, "step_time_sec": 8.238865241990425, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9053189754486084, "lr": 0.0002, "elapsed_sec": 57737.25229930878, "step_time_sec": 8.238847882952541, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.834171772003174, "lr": 0.0002, "elapsed_sec": 57745.49159026146, "step_time_sec": 8.239051262964495, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.897491455078125, "lr": 0.0002, "elapsed_sec": 57753.72881770134, "step_time_sec": 8.237097595003434, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0210862159729, "lr": 0.0002, "elapsed_sec": 57761.96890330315, "step_time_sec": 8.239909494994208, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9573519229888916, "lr": 0.0002, "elapsed_sec": 57770.20954322815, "step_time_sec": 8.240515855024569, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.667851448059082, "lr": 0.0002, "elapsed_sec": 57778.44772410393, "step_time_sec": 8.238008353975601, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.056049346923828, "lr": 0.0002, "elapsed_sec": 57786.68740582466, "step_time_sec": 8.239562798989937, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8758420944213867, "lr": 0.0002, "elapsed_sec": 57794.92686724663, "step_time_sec": 8.239288952026982, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.179079532623291, "lr": 0.0002, "elapsed_sec": 57803.16657066345, "step_time_sec": 8.239518314017914, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8144726753234863, "lr": 0.0002, "elapsed_sec": 57811.40608215332, "step_time_sec": 8.23931931098923, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.036978244781494, "lr": 0.0002, "elapsed_sec": 57819.643020391464, "step_time_sec": 8.236781822983176, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.084477424621582, "lr": 0.0002, "elapsed_sec": 57827.883200883865, "step_time_sec": 8.24005899601616, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.937988042831421, "lr": 0.0002, "elapsed_sec": 57836.12260532379, "step_time_sec": 8.239185563987121, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.894749879837036, "lr": 0.0002, "elapsed_sec": 57844.36178994179, "step_time_sec": 8.239075285964645, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.927476167678833, "lr": 0.0002, "elapsed_sec": 57852.60187625885, "step_time_sec": 8.239879994012881, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.039775371551514, "lr": 0.0002, "elapsed_sec": 57860.84054994583, "step_time_sec": 8.238536201009993, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9760794639587402, "lr": 0.0002, "elapsed_sec": 57869.080775260925, "step_time_sec": 8.24006299697794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.93058180809021, "lr": 0.0002, "elapsed_sec": 57877.32076525688, "step_time_sec": 8.239833572995849, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9856953620910645, "lr": 0.0002, "elapsed_sec": 57885.557896614075, "step_time_sec": 8.236953307990916, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.031898021697998, "lr": 0.0002, "elapsed_sec": 57893.798107624054, "step_time_sec": 8.240052392997313, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8499691486358643, "lr": 0.0002, "elapsed_sec": 57902.03760290146, "step_time_sec": 8.239303465001285, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.011518955230713, "lr": 0.0002, "elapsed_sec": 57910.277178525925, "step_time_sec": 8.239453324989881, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9183435440063477, "lr": 0.0002, "elapsed_sec": 57918.51834368706, "step_time_sec": 8.240982957009692, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9644217491149902, "lr": 0.0002, "elapsed_sec": 57926.75566196442, "step_time_sec": 8.237171148008201, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.919748067855835, "lr": 0.0002, "elapsed_sec": 57934.99581813812, "step_time_sec": 8.23994248197414, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8445777893066406, "lr": 0.0002, "elapsed_sec": 57943.2346804142, "step_time_sec": 8.238704752991907, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.975497007369995, "lr": 0.0002, "elapsed_sec": 57951.47434711456, "step_time_sec": 8.239505799952894, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.852952718734741, "lr": 0.0002, "elapsed_sec": 57959.71328020096, "step_time_sec": 8.238785484980326, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.057840347290039, "lr": 0.0002, "elapsed_sec": 57967.95065450668, "step_time_sec": 8.23720905103255, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8398008346557617, "lr": 0.0002, "elapsed_sec": 57976.191719293594, "step_time_sec": 8.240896484989207, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9472155570983887, "lr": 0.0002, "elapsed_sec": 57984.43025946617, "step_time_sec": 8.238385228964034, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.815385580062866, "lr": 0.0002, "elapsed_sec": 57992.668461084366, "step_time_sec": 8.238022692967206, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.016895771026611, "lr": 0.0002, "elapsed_sec": 58000.90761113167, "step_time_sec": 8.238985581032466, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.810255765914917, "lr": 0.0002, "elapsed_sec": 58009.14437866211, "step_time_sec": 8.236612461972982, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.778090000152588, "lr": 0.0002, "elapsed_sec": 58017.3844332695, "step_time_sec": 8.239940407976974, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.833340883255005, "lr": 0.0002, "elapsed_sec": 58025.62389779091, "step_time_sec": 8.239205827005208, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.6772844791412354, "lr": 0.0002, "elapsed_sec": 58033.862010240555, "step_time_sec": 8.238002156023867, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8733794689178467, "lr": 0.0002, "elapsed_sec": 58042.100491285324, "step_time_sec": 8.238276215037331, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.935462236404419, "lr": 0.0002, "elapsed_sec": 58050.34000778198, "step_time_sec": 8.239378356025554, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9190263748168945, "lr": 0.0002, "elapsed_sec": 58058.57852935791, "step_time_sec": 8.238326325954404, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8662447929382324, "lr": 0.0002, "elapsed_sec": 58066.81885242462, "step_time_sec": 8.240159877983388, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.830379009246826, "lr": 0.0002, "elapsed_sec": 58075.057054042816, "step_time_sec": 8.238055123016238, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9038023948669434, "lr": 0.0002, "elapsed_sec": 58083.29724740982, "step_time_sec": 8.240031973982695, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.950995683670044, "lr": 0.0002, "elapsed_sec": 58091.53556895256, "step_time_sec": 8.238135933992453, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8234288692474365, "lr": 0.0002, "elapsed_sec": 58099.77569460869, "step_time_sec": 8.239971348026302, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.013308048248291, "lr": 0.0002, "elapsed_sec": 58108.01406788826, "step_time_sec": 8.238245784013998, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.894068479537964, "lr": 0.0002, "elapsed_sec": 58116.25043082237, "step_time_sec": 8.236161956039723, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.019315719604492, "lr": 0.0002, "elapsed_sec": 58124.49009013176, "step_time_sec": 8.239467067993246, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.70851731300354, "lr": 0.0002, "elapsed_sec": 58132.72760796547, "step_time_sec": 8.237403465958778, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8826820850372314, "lr": 0.0002, "elapsed_sec": 58140.9710726738, "step_time_sec": 8.243292314000428, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2442216873168945, "lr": 0.0002, "elapsed_sec": 58149.233191251755, "step_time_sec": 8.26196054898901, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.892850875854492, "lr": 0.0002, "elapsed_sec": 58157.47357749939, "step_time_sec": 8.24022162903566, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.952038526535034, "lr": 0.0002, "elapsed_sec": 58165.711122751236, "step_time_sec": 8.237352882977575, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8024909496307373, "lr": 0.0002, "elapsed_sec": 58173.95099449158, "step_time_sec": 52.895026740967296, "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": 0.9997517460142262, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 7001, "loss": 3.9846394062042236, "lr": 0.0002, "elapsed_sec": 58226.855769872665, "step_time_sec": 8.249327622994315, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.922828197479248, "lr": 0.00019999999900581207, "elapsed_sec": 58235.09480857849, "step_time_sec": 8.238916789996438, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.022723197937012, "lr": 0.00019999999602324833, "elapsed_sec": 58243.3318901062, "step_time_sec": 8.236943210009485, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9655303955078125, "lr": 0.0001999999910523088, "elapsed_sec": 58251.57154035568, "step_time_sec": 8.239432457019575, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.042421817779541, "lr": 0.00019999998409299358, "elapsed_sec": 58259.81009578705, "step_time_sec": 8.238403260009363, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9388816356658936, "lr": 0.00019999997514530285, "elapsed_sec": 58268.04931640625, "step_time_sec": 8.239114989002701, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8499865531921387, "lr": 0.00019999996420923675, "elapsed_sec": 58276.28852176666, "step_time_sec": 8.239003703987692, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.911752462387085, "lr": 0.00019999995128479553, "elapsed_sec": 58284.5254547596, "step_time_sec": 8.236794112017378, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.941622257232666, "lr": 0.00019999993637197946, "elapsed_sec": 58292.7644777298, "step_time_sec": 8.23885991802672, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.990873336791992, "lr": 0.0001999999194707888, "elapsed_sec": 58301.00090551376, "step_time_sec": 8.236262538004667, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7927796840667725, "lr": 0.00019999990058122396, "elapsed_sec": 58309.24125099182, "step_time_sec": 8.240215971018188, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8936684131622314, "lr": 0.00019999987970328528, "elapsed_sec": 58317.4805123806, "step_time_sec": 8.23912682197988, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9123358726501465, "lr": 0.00019999985683697322, "elapsed_sec": 58325.71805238724, "step_time_sec": 8.237344822031446, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9090473651885986, "lr": 0.0001999998319822882, "elapsed_sec": 58333.95707988739, "step_time_sec": 8.238910095009487, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8370301723480225, "lr": 0.00019999980513923075, "elapsed_sec": 58342.196212768555, "step_time_sec": 8.238890192995314, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8558647632598877, "lr": 0.00019999977630780143, "elapsed_sec": 58350.43567276001, "step_time_sec": 8.23930651601404, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9882302284240723, "lr": 0.00019999974548800083, "elapsed_sec": 58358.67669034004, "step_time_sec": 8.240835951990448, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.919222831726074, "lr": 0.00019999971267982953, "elapsed_sec": 58366.9148516655, "step_time_sec": 8.237988465989474, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.991509199142456, "lr": 0.00019999967788328826, "elapsed_sec": 58375.15450716019, "step_time_sec": 8.239530910970643, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.898312568664551, "lr": 0.00019999964109837767, "elapsed_sec": 58383.39343237877, "step_time_sec": 8.238773229008075, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.052680492401123, "lr": 0.0001999996023250986, "elapsed_sec": 58391.63301372528, "step_time_sec": 8.23935986700235, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8945178985595703, "lr": 0.00019999956156345174, "elapsed_sec": 58399.8720035553, "step_time_sec": 8.238829638052266, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9292664527893066, "lr": 0.000199999518813438, "elapsed_sec": 58408.10960197449, "step_time_sec": 8.237467174010817, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.964738607406616, "lr": 0.00019999947407505818, "elapsed_sec": 58416.34943652153, "step_time_sec": 8.239624209993053, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.009016990661621, "lr": 0.00019999942734831325, "elapsed_sec": 58424.58953952789, "step_time_sec": 8.239960148988757, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.800933837890625, "lr": 0.00019999937863320413, "elapsed_sec": 58432.82720351219, "step_time_sec": 8.237471505010035, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8174822330474854, "lr": 0.00019999932792973184, "elapsed_sec": 58441.066934108734, "step_time_sec": 8.239562924951315, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.062860012054443, "lr": 0.0001999992752378974, "elapsed_sec": 58449.30284285545, "step_time_sec": 8.235786106029991, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.020078182220459, "lr": 0.00019999922055770187, "elapsed_sec": 58457.541699171066, "step_time_sec": 8.238692449056543, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.853832244873047, "lr": 0.0001999991638891464, "elapsed_sec": 58465.781166791916, "step_time_sec": 8.239335494989064, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8667964935302734, "lr": 0.00019999910523223213, "elapsed_sec": 58474.02007818222, "step_time_sec": 8.238774939032737, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9733901023864746, "lr": 0.00019999904458696023, "elapsed_sec": 58482.2607896328, "step_time_sec": 8.240479957021307, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.943859338760376, "lr": 0.00019999898195333197, "elapsed_sec": 58490.499210596085, "step_time_sec": 8.238265504012816, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9444706439971924, "lr": 0.0001999989173313486, "elapsed_sec": 58498.741039037704, "step_time_sec": 8.241672067029867, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8581526279449463, "lr": 0.0001999988507210115, "elapsed_sec": 58506.98038315773, "step_time_sec": 8.239175483002327, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.005946159362793, "lr": 0.00019999878212232196, "elapsed_sec": 58515.218482494354, "step_time_sec": 8.237912046024576, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9164202213287354, "lr": 0.00019999871153528138, "elapsed_sec": 58523.45824980736, "step_time_sec": 8.239624696027022, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9797887802124023, "lr": 0.00019999863895989121, "elapsed_sec": 58531.69814300537, "step_time_sec": 8.239709604007658, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.961235284805298, "lr": 0.00019999856439615297, "elapsed_sec": 58539.93706679344, "step_time_sec": 8.238747479976155, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7755956649780273, "lr": 0.00019999848784406815, "elapsed_sec": 58548.177522182465, "step_time_sec": 8.240337547962554, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.945769786834717, "lr": 0.00019999840930363832, "elapsed_sec": 58556.41649508476, "step_time_sec": 8.238834980002139, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8119404315948486, "lr": 0.00019999832877486506, "elapsed_sec": 58564.65688252449, "step_time_sec": 8.240206381015014, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8967316150665283, "lr": 0.00019999824625775004, "elapsed_sec": 58572.89575791359, "step_time_sec": 8.238674035004806, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8803908824920654, "lr": 0.0001999981617522949, "elapsed_sec": 58581.13532114029, "step_time_sec": 8.239405767002609, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7591280937194824, "lr": 0.00019999807525850142, "elapsed_sec": 58589.37571597099, "step_time_sec": 8.240291373978835, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.91580867767334, "lr": 0.00019999798677637132, "elapsed_sec": 58597.614342689514, "step_time_sec": 8.238408702018205, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9142563343048096, "lr": 0.00019999789630590644, "elapsed_sec": 58605.854325294495, "step_time_sec": 8.239859241002705, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9845235347747803, "lr": 0.00019999780384710863, "elapsed_sec": 58614.093153476715, "step_time_sec": 8.238672448031139, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.047027111053467, "lr": 0.0001999977093999797, "elapsed_sec": 58622.332020998, "step_time_sec": 8.238609623978846, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.208769798278809, "lr": 0.00019999761296452163, "elapsed_sec": 58630.57179522514, "step_time_sec": 8.239624943991657, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9334044456481934, "lr": 0.00019999751454073642, "elapsed_sec": 58638.80973696709, "step_time_sec": 8.23786160297459, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9105119705200195, "lr": 0.00019999741412862605, "elapsed_sec": 58647.04885697365, "step_time_sec": 8.238938970025629, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.870163917541504, "lr": 0.0001999973117281925, "elapsed_sec": 58655.28891682625, "step_time_sec": 8.239871015015524, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.061189651489258, "lr": 0.00019999720733943798, "elapsed_sec": 58663.526320934296, "step_time_sec": 8.237245861964766, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9331884384155273, "lr": 0.00019999710096236454, "elapsed_sec": 58671.76622772217, "step_time_sec": 8.239719895995222, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7817509174346924, "lr": 0.00019999699259697437, "elapsed_sec": 58680.00521969795, "step_time_sec": 8.238806648994796, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9967169761657715, "lr": 0.00019999688224326965, "elapsed_sec": 58688.24268960953, "step_time_sec": 8.237311109958682, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9451935291290283, "lr": 0.00019999676990125267, "elapsed_sec": 58696.48302435875, "step_time_sec": 8.240160354005639, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.025599479675293, "lr": 0.0001999966555709257, "elapsed_sec": 58704.720578193665, "step_time_sec": 8.23743049998302, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.925090789794922, "lr": 0.0001999965392522911, "elapsed_sec": 58712.9620950222, "step_time_sec": 8.241323264024686, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.04146146774292, "lr": 0.00019999642094535122, "elapsed_sec": 58721.20151305199, "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": 7062, "loss": 3.933095932006836, "lr": 0.00019999630065010846, "elapsed_sec": 58729.43942451477, "step_time_sec": 8.237741608987562, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8829517364501953, "lr": 0.00019999617836656528, "elapsed_sec": 58737.679231882095, "step_time_sec": 8.239659641985781, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.911328077316284, "lr": 0.0001999960540947242, "elapsed_sec": 58745.916254758835, "step_time_sec": 8.236854214046616, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8565804958343506, "lr": 0.00019999592783458774, "elapsed_sec": 58754.155537843704, "step_time_sec": 8.239121676015202, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.91752028465271, "lr": 0.00019999579958615845, "elapsed_sec": 58762.39514923096, "step_time_sec": 8.239397482015193, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8883821964263916, "lr": 0.00019999566934943898, "elapsed_sec": 58770.63431072235, "step_time_sec": 8.239019187050872, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.845201253890991, "lr": 0.00019999553712443197, "elapsed_sec": 58778.87295937538, "step_time_sec": 8.2384749869816, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.945277690887451, "lr": 0.0001999954029111401, "elapsed_sec": 58787.11161541939, "step_time_sec": 8.238542394014075, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.177441596984863, "lr": 0.00019999526670956617, "elapsed_sec": 58795.35173416138, "step_time_sec": 8.239910579985008, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.914821147918701, "lr": 0.00019999512851971287, "elapsed_sec": 58803.5912194252, "step_time_sec": 8.239321611006744, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.746396780014038, "lr": 0.00019999498834158308, "elapsed_sec": 58811.828708171844, "step_time_sec": 8.23735257104272, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8184313774108887, "lr": 0.00019999484617517964, "elapsed_sec": 58820.06921958923, "step_time_sec": 8.24035689898301, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.871594190597534, "lr": 0.00019999470202050544, "elapsed_sec": 58828.3081638813, "step_time_sec": 8.238764242036268, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.993866443634033, "lr": 0.00019999455587756343, "elapsed_sec": 58836.5482878685, "step_time_sec": 8.239936918020248, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.046595573425293, "lr": 0.00019999440774635659, "elapsed_sec": 58844.78745055199, "step_time_sec": 8.23903271195013, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.896282434463501, "lr": 0.00019999425762688797, "elapsed_sec": 58853.0240046978, "step_time_sec": 8.236355390981771, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9299564361572266, "lr": 0.00019999410551916057, "elapsed_sec": 58861.26409840584, "step_time_sec": 8.23995823302539, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8526477813720703, "lr": 0.00019999395142317755, "elapsed_sec": 58869.503522872925, "step_time_sec": 8.239282039983664, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8563640117645264, "lr": 0.000199993795338942, "elapsed_sec": 58877.74315595627, "step_time_sec": 8.239507737976965, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.863635778427124, "lr": 0.00019999363726645714, "elapsed_sec": 58885.98208642006, "step_time_sec": 8.238743975991383, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9198145866394043, "lr": 0.0001999934772057262, "elapsed_sec": 58894.21903991699, "step_time_sec": 8.2368107449729, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9243388175964355, "lr": 0.0001999933151567524, "elapsed_sec": 58902.459332466125, "step_time_sec": 8.24008006596705, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8760197162628174, "lr": 0.0001999931511195391, "elapsed_sec": 58910.696454286575, "step_time_sec": 8.236972483980935, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0395379066467285, "lr": 0.0001999929850940896, "elapsed_sec": 58918.935259103775, "step_time_sec": 8.238690496014897, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8824150562286377, "lr": 0.0001999928170804073, "elapsed_sec": 58927.175278663635, "step_time_sec": 8.239820110960864, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9221744537353516, "lr": 0.00019999264707849562, "elapsed_sec": 58935.41359829903, "step_time_sec": 8.238111465005204, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8920557498931885, "lr": 0.00019999247508835808, "elapsed_sec": 58943.65378832817, "step_time_sec": 8.240036910050549, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8673367500305176, "lr": 0.00019999230110999812, "elapsed_sec": 58951.89371371269, "step_time_sec": 8.239771530032158, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.979985475540161, "lr": 0.0001999921251434193, "elapsed_sec": 58960.13385081291, "step_time_sec": 8.239978336030617, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.15606164932251, "lr": 0.00019999194718862525, "elapsed_sec": 58968.37269759178, "step_time_sec": 8.238710038014688, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.925041675567627, "lr": 0.00019999176724561953, "elapsed_sec": 58976.61186027527, "step_time_sec": 8.23902050498873, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9336605072021484, "lr": 0.0001999915853144059, "elapsed_sec": 58984.85173988342, "step_time_sec": 8.239714108000044, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.006352424621582, "lr": 0.00019999140139498797, "elapsed_sec": 58993.091264009476, "step_time_sec": 8.239362106018234, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9781785011291504, "lr": 0.00019999121548736957, "elapsed_sec": 59001.330199718475, "step_time_sec": 8.23873621900566, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8859469890594482, "lr": 0.00019999102759155445, "elapsed_sec": 59009.57091712952, "step_time_sec": 8.24058631103253, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8432493209838867, "lr": 0.00019999083770754647, "elapsed_sec": 59017.811547756195, "step_time_sec": 8.240436395979486, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8208954334259033, "lr": 0.00019999064583534946, "elapsed_sec": 59026.05106115341, "step_time_sec": 8.239374473982025, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9152584075927734, "lr": 0.0001999904519749674, "elapsed_sec": 59034.28954958916, "step_time_sec": 8.23834195494419, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9194066524505615, "lr": 0.00019999025612640417, "elapsed_sec": 59042.5286359787, "step_time_sec": 8.238854695984628, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0102972984313965, "lr": 0.0001999900582896638, "elapsed_sec": 59050.77048945427, "step_time_sec": 8.241728927998338, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.864593744277954, "lr": 0.00019998985846475035, "elapsed_sec": 59059.00868368149, "step_time_sec": 8.238037878007162, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.872650623321533, "lr": 0.00019998965665166785, "elapsed_sec": 59067.24613714218, "step_time_sec": 8.237294460006524, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.105579853057861, "lr": 0.0001999894528504204, "elapsed_sec": 59075.48628616333, "step_time_sec": 8.239964004023932, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9018759727478027, "lr": 0.00019998924706101222, "elapsed_sec": 59083.72398471832, "step_time_sec": 8.237545086012688, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.812000274658203, "lr": 0.0001999890392834475, "elapsed_sec": 59091.96381735802, "step_time_sec": 8.239714303985238, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.874545097351074, "lr": 0.0001999888295177304, "elapsed_sec": 59100.20394515991, "step_time_sec": 8.2399410729995, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.252140045166016, "lr": 0.00019998861776386527, "elapsed_sec": 59108.44158029556, "step_time_sec": 8.237483372970019, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.935771942138672, "lr": 0.00019998840402185645, "elapsed_sec": 59116.682524204254, "step_time_sec": 8.240782606007997, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8435628414154053, "lr": 0.00019998818829170822, "elapsed_sec": 59124.92221736908, "step_time_sec": 8.239561184018385, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.900557279586792, "lr": 0.00019998797057342504, "elapsed_sec": 59133.16200757027, "step_time_sec": 8.239603624970187, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7728943824768066, "lr": 0.0001999877508670113, "elapsed_sec": 59141.40231823921, "step_time_sec": 8.24015515402425, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8816542625427246, "lr": 0.00019998752917247152, "elapsed_sec": 59149.6396176815, "step_time_sec": 8.237110046029557, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.866713285446167, "lr": 0.00019998730548981023, "elapsed_sec": 59157.88280057907, "step_time_sec": 8.243036010011565, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9669992923736572, "lr": 0.0001999870798190319, "elapsed_sec": 59166.12310051918, "step_time_sec": 8.240161955996882, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.026288032531738, "lr": 0.00019998685216014129, "elapsed_sec": 59174.360142707825, "step_time_sec": 8.23687952797627, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.646076202392578, "lr": 0.00019998662251314293, "elapsed_sec": 59182.600680828094, "step_time_sec": 8.240321538003627, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.971285820007324, "lr": 0.0001999863908780415, "elapsed_sec": 59190.83994555473, "step_time_sec": 8.239144221006427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.914503335952759, "lr": 0.00019998615725484178, "elapsed_sec": 59199.07982301712, "step_time_sec": 8.239708266977686, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.750370502471924, "lr": 0.0001999859216435485, "elapsed_sec": 59207.32042002678, "step_time_sec": 8.240485477028415, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.946413040161133, "lr": 0.00019998568404416646, "elapsed_sec": 59215.55819916725, "step_time_sec": 8.237591426004656, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.782824993133545, "lr": 0.00019998544445670055, "elapsed_sec": 59223.79922890663, "step_time_sec": 8.240814424003474, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8936831951141357, "lr": 0.0001999852028811556, "elapsed_sec": 59232.04008078575, "step_time_sec": 8.240725506970193, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.091573715209961, "lr": 0.0001999849593175366, "elapsed_sec": 59240.27730202675, "step_time_sec": 8.237112018978223, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9751312732696533, "lr": 0.00019998471376584843, "elapsed_sec": 59248.518377780914, "step_time_sec": 8.240866827953141, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8446767330169678, "lr": 0.00019998446622609617, "elapsed_sec": 59256.75612664223, "step_time_sec": 8.237548035976943, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.129807472229004, "lr": 0.00019998421669828483, "elapsed_sec": 59264.99626660347, "step_time_sec": 8.240030136017594, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8828964233398438, "lr": 0.00019998396518241952, "elapsed_sec": 59273.23556423187, "step_time_sec": 8.239119618956465, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.881936550140381, "lr": 0.00019998371167850537, "elapsed_sec": 59281.47447371483, "step_time_sec": 8.238760498003103, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.889039993286133, "lr": 0.0001999834561865475, "elapsed_sec": 59289.71500873566, "step_time_sec": 8.240339027019218, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9346182346343994, "lr": 0.0001999831987065512, "elapsed_sec": 59297.95398068428, "step_time_sec": 8.238798219012097, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.818347930908203, "lr": 0.0001999829392385217, "elapsed_sec": 59306.193156957626, "step_time_sec": 8.239049464988057, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7670977115631104, "lr": 0.00019998267778246427, "elapsed_sec": 59314.43395471573, "step_time_sec": 8.240646796009969, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9046030044555664, "lr": 0.0001999824143383842, "elapsed_sec": 59322.671698093414, "step_time_sec": 8.237541873997543, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9043288230895996, "lr": 0.00019998214890628696, "elapsed_sec": 59330.91424536705, "step_time_sec": 8.242444424016867, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9068710803985596, "lr": 0.0001999818814861779, "elapsed_sec": 59339.15601229668, "step_time_sec": 8.241580741014332, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.959334373474121, "lr": 0.00019998161207806248, "elapsed_sec": 59347.40658044815, "step_time_sec": 8.250469525053632, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.97086238861084, "lr": 0.0001999813406819462, "elapsed_sec": 59355.65078163147, "step_time_sec": 8.243927642004564, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.933826208114624, "lr": 0.00019998106729783458, "elapsed_sec": 59363.891132354736, "step_time_sec": 8.240150508994702, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9009361267089844, "lr": 0.00019998079192573323, "elapsed_sec": 59372.13141489029, "step_time_sec": 8.240137830027379, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9792275428771973, "lr": 0.00019998051456564776, "elapsed_sec": 59380.37073516846, "step_time_sec": 8.239158037991729, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9193854331970215, "lr": 0.0001999802352175838, "elapsed_sec": 59388.612251996994, "step_time_sec": 8.24135898501845, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8258073329925537, "lr": 0.00019997995388154705, "elapsed_sec": 59396.85131072998, "step_time_sec": 8.238889740954619, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.807612895965576, "lr": 0.00019997967055754325, "elapsed_sec": 59405.09265136719, "step_time_sec": 8.24119074497139, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0120625495910645, "lr": 0.0001999793852455782, "elapsed_sec": 59413.332518815994, "step_time_sec": 8.239727751992177, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0106730461120605, "lr": 0.00019997909794565772, "elapsed_sec": 59421.57106208801, "step_time_sec": 8.238360046001617, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.877129554748535, "lr": 0.0001999788086577876, "elapsed_sec": 59429.81109023094, "step_time_sec": 8.239858225977514, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.836390495300293, "lr": 0.00019997851738197384, "elapsed_sec": 59438.050639629364, "step_time_sec": 8.239392507006414, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.784987449645996, "lr": 0.00019997822411822231, "elapsed_sec": 59446.2901096344, "step_time_sec": 8.23932368698297, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0135178565979, "lr": 0.00019997792886653902, "elapsed_sec": 59454.53141641617, "step_time_sec": 8.241161921992898, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9373228549957275, "lr": 0.00019997763162693, "elapsed_sec": 59462.7700419426, "step_time_sec": 8.23840733495308, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.940656900405884, "lr": 0.00019997733239940127, "elapsed_sec": 59471.00892472267, "step_time_sec": 8.238765016023535, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.134439945220947, "lr": 0.00019997703118395898, "elapsed_sec": 59479.24797868729, "step_time_sec": 8.238879507989623, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.038015842437744, "lr": 0.00019997672798060926, "elapsed_sec": 59487.48830533028, "step_time_sec": 8.240132528997492, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.021806240081787, "lr": 0.00019997642278935825, "elapsed_sec": 59495.72715616226, "step_time_sec": 8.238684173033107, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.931635856628418, "lr": 0.00019997611561021225, "elapsed_sec": 59503.96658921242, "step_time_sec": 8.23933509294875, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.763801336288452, "lr": 0.00019997580644317745, "elapsed_sec": 59512.206079244614, "step_time_sec": 8.239272693987004, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.865941286087036, "lr": 0.0001999754952882602, "elapsed_sec": 59520.44877457619, "step_time_sec": 8.24252561898902, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.862478256225586, "lr": 0.00019997518214546684, "elapsed_sec": 59528.68706035614, "step_time_sec": 8.238130821031518, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.871570587158203, "lr": 0.0001999748670148037, "elapsed_sec": 59536.928087472916, "step_time_sec": 8.240859187033493, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.076233863830566, "lr": 0.0001999745498962773, "elapsed_sec": 59545.167016744614, "step_time_sec": 8.238798425009008, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9002959728240967, "lr": 0.0001999742307898941, "elapsed_sec": 59553.40557765961, "step_time_sec": 8.238385982986074, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.803551197052002, "lr": 0.0001999739096956605, "elapsed_sec": 59561.64604353905, "step_time_sec": 8.24030171794584, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.082396030426025, "lr": 0.00019997358661358317, "elapsed_sec": 59569.88383984566, "step_time_sec": 8.237600970023777, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.850252151489258, "lr": 0.00019997326154366863, "elapsed_sec": 59578.124474048615, "step_time_sec": 8.240482788998634, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.973081350326538, "lr": 0.00019997293448592354, "elapsed_sec": 59586.36419081688, "step_time_sec": 8.239593142003287, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7807774543762207, "lr": 0.00019997260544035456, "elapsed_sec": 59594.602957487106, "step_time_sec": 8.238565808045678, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.736100912094116, "lr": 0.00019997227440696837, "elapsed_sec": 59602.843379974365, "step_time_sec": 8.240238192956895, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.901413679122925, "lr": 0.00019997194138577179, "elapsed_sec": 59611.08386397362, "step_time_sec": 8.240359832008835, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9377975463867188, "lr": 0.00019997160637677154, "elapsed_sec": 59619.3224196434, "step_time_sec": 8.238356492947787, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.85109543800354, "lr": 0.00019997126937997448, "elapsed_sec": 59627.56301569939, "step_time_sec": 8.240478612016886, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9130144119262695, "lr": 0.00019997093039538747, "elapsed_sec": 59635.80186343193, "step_time_sec": 8.238638547016308, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9051766395568848, "lr": 0.00019997058942301747, "elapsed_sec": 59644.04096198082, "step_time_sec": 8.238983380957507, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.6940901279449463, "lr": 0.00019997024646287138, "elapsed_sec": 59652.280802965164, "step_time_sec": 8.239648601971567, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.045306205749512, "lr": 0.00019996990151495626, "elapsed_sec": 59660.51929593086, "step_time_sec": 8.23836537101306, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.802079677581787, "lr": 0.00019996955457927905, "elapsed_sec": 59668.76021242142, "step_time_sec": 8.240749671007507, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.786163091659546, "lr": 0.0001999692056558469, "elapsed_sec": 59677.00024318695, "step_time_sec": 8.239858223008923, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9436869621276855, "lr": 0.0001999688547446669, "elapsed_sec": 59685.23994898796, "step_time_sec": 8.239498209033627, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9765937328338623, "lr": 0.00019996850184574621, "elapsed_sec": 59693.478803396225, "step_time_sec": 8.238708232005592, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9384515285491943, "lr": 0.00019996814695909202, "elapsed_sec": 59701.717725753784, "step_time_sec": 8.238773379998747, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.963202476501465, "lr": 0.00019996779008471158, "elapsed_sec": 59709.959092378616, "step_time_sec": 8.241150918009225, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9115517139434814, "lr": 0.00019996743122261215, "elapsed_sec": 59718.19796252251, "step_time_sec": 8.23873982502846, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8501811027526855, "lr": 0.00019996707037280107, "elapsed_sec": 59726.43576025963, "step_time_sec": 8.23765247198753, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.949530839920044, "lr": 0.0001999667075352857, "elapsed_sec": 59734.67573070526, "step_time_sec": 8.239813571970444, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9300620555877686, "lr": 0.00019996634271007342, "elapsed_sec": 59742.91532111168, "step_time_sec": 8.23938923800597, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9105923175811768, "lr": 0.00019996597589717164, "elapsed_sec": 59751.155801057816, "step_time_sec": 8.240334784029983, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8758938312530518, "lr": 0.00019996560709658793, "elapsed_sec": 59759.39641523361, "step_time_sec": 8.24046634795377, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.737067461013794, "lr": 0.00019996523630832973, "elapsed_sec": 59767.63531804085, "step_time_sec": 8.23875478003174, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.906515121459961, "lr": 0.00019996486353240464, "elapsed_sec": 59775.87522602081, "step_time_sec": 8.239722506026737, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8470680713653564, "lr": 0.00019996448876882027, "elapsed_sec": 59784.11535477638, "step_time_sec": 8.240008170017973, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.768007278442383, "lr": 0.0001999641120175842, "elapsed_sec": 59792.353852272034, "step_time_sec": 8.238262765982654, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.895305633544922, "lr": 0.0001999637332787042, "elapsed_sec": 59800.59453678131, "step_time_sec": 8.240568329987582, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7739033699035645, "lr": 0.00019996335255218793, "elapsed_sec": 59808.83399224281, "step_time_sec": 8.239284083014354, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8134539127349854, "lr": 0.0001999629698380432, "elapsed_sec": 59817.07587814331, "step_time_sec": 8.24174689000938, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.80612850189209, "lr": 0.0001999625851362778, "elapsed_sec": 59825.31554436684, "step_time_sec": 8.239464870013762, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9325177669525146, "lr": 0.00019996219844689955, "elapsed_sec": 59833.55346250534, "step_time_sec": 8.237784372002352, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.944415807723999, "lr": 0.00019996180976991634, "elapsed_sec": 59841.79494333267, "step_time_sec": 8.241334675985854, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.937824010848999, "lr": 0.00019996141910533612, "elapsed_sec": 59850.03464508057, "step_time_sec": 8.239485992991831, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8470258712768555, "lr": 0.00019996102645316685, "elapsed_sec": 59858.27319145203, "step_time_sec": 8.238406227959786, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.014906883239746, "lr": 0.00019996063181341656, "elapsed_sec": 59866.51428794861, "step_time_sec": 8.24090937298024, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.805744171142578, "lr": 0.00019996023518609326, "elapsed_sec": 59874.7534930706, "step_time_sec": 8.23911402997328, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.01858377456665, "lr": 0.00019995983657120506, "elapsed_sec": 59882.994975566864, "step_time_sec": 8.24128658004338, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9013113975524902, "lr": 0.00019995943596876005, "elapsed_sec": 59891.23423576355, "step_time_sec": 8.239079567953013, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9496710300445557, "lr": 0.00019995903337876645, "elapsed_sec": 59899.47257089615, "step_time_sec": 8.23819923098199, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.775268077850342, "lr": 0.00019995862880123246, "elapsed_sec": 59907.71598958969, "step_time_sec": 8.243238773022313, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9787611961364746, "lr": 0.00019995822223616632, "elapsed_sec": 59915.95530819893, "step_time_sec": 8.23919048398966, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.904292345046997, "lr": 0.0001999578136835763, "elapsed_sec": 59924.19418883324, "step_time_sec": 8.238728477037512, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8567535877227783, "lr": 0.00019995740314347076, "elapsed_sec": 59932.434295892715, "step_time_sec": 8.239893447957002, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9897823333740234, "lr": 0.00019995699061585808, "elapsed_sec": 59940.67364239693, "step_time_sec": 8.239165535022039, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.06056022644043, "lr": 0.00019995657610074666, "elapsed_sec": 59948.91423249245, "step_time_sec": 8.240447400021367, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.971033811569214, "lr": 0.00019995615959814494, "elapsed_sec": 59957.1539747715, "step_time_sec": 8.23957331199199, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.815042495727539, "lr": 0.00019995574110806145, "elapsed_sec": 59965.39314842224, "step_time_sec": 8.238964930991642, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9583864212036133, "lr": 0.0001999553206305047, "elapsed_sec": 59973.63421869278, "step_time_sec": 8.240912490000483, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.949061870574951, "lr": 0.00019995489816548325, "elapsed_sec": 59981.87366890907, "step_time_sec": 8.239314414968248, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9281463623046875, "lr": 0.00019995447371300577, "elapsed_sec": 59990.11215567589, "step_time_sec": 8.238352972024586, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.046159267425537, "lr": 0.00019995404727308084, "elapsed_sec": 59998.352914094925, "step_time_sec": 8.240530808980111, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7994933128356934, "lr": 0.0001999536188457172, "elapsed_sec": 60006.591325998306, "step_time_sec": 8.238306913990527, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.6869733333587646, "lr": 0.00019995318843092362, "elapsed_sec": 60014.83251643181, "step_time_sec": 8.240970921004191, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.877101421356201, "lr": 0.0001999527560287088, "elapsed_sec": 60023.07292366028, "step_time_sec": 8.240241641993634, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.952345609664917, "lr": 0.00019995232163908162, "elapsed_sec": 60031.311168670654, "step_time_sec": 8.238085269986186, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9699974060058594, "lr": 0.00019995188526205092, "elapsed_sec": 60039.55245399475, "step_time_sec": 8.241131222981494, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7743871212005615, "lr": 0.0001999514468976256, "elapsed_sec": 60047.79374241829, "step_time_sec": 8.24110856000334, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.902017593383789, "lr": 0.00019995100654581456, "elapsed_sec": 60056.03214430809, "step_time_sec": 8.23824938701, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.844142436981201, "lr": 0.00019995056420662685, "elapsed_sec": 60064.27199292183, "step_time_sec": 8.239712919981685, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.939664125442505, "lr": 0.00019995011988007145, "elapsed_sec": 60072.512843847275, "step_time_sec": 8.24066157604102, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9678900241851807, "lr": 0.00019994967356615743, "elapsed_sec": 60080.75281214714, "step_time_sec": 8.239821650029626, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.748157262802124, "lr": 0.0001999492252648939, "elapsed_sec": 60088.99273324013, "step_time_sec": 8.239767176972236, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7250916957855225, "lr": 0.00019994877497629, "elapsed_sec": 60097.23123884201, "step_time_sec": 8.23835170903476, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.878065347671509, "lr": 0.00019994832270035489, "elapsed_sec": 60105.47327065468, "step_time_sec": 8.241849467973225, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9325172901153564, "lr": 0.0001999478684370978, "elapsed_sec": 60113.71332573891, "step_time_sec": 8.23985641199397, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7802646160125732, "lr": 0.00019994741218652803, "elapsed_sec": 60121.95250749588, "step_time_sec": 8.239059218030889, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.854168653488159, "lr": 0.00019994695394865485, "elapsed_sec": 60130.19331431389, "step_time_sec": 8.240597625961527, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.918154239654541, "lr": 0.0001999464937234876, "elapsed_sec": 60138.43178343773, "step_time_sec": 8.23834291199455, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.791266441345215, "lr": 0.00019994603151103572, "elapsed_sec": 60146.67006134987, "step_time_sec": 8.238115046988241, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9752390384674072, "lr": 0.00019994556731130856, "elapsed_sec": 60154.90943813324, "step_time_sec": 8.239211927982979, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.745255470275879, "lr": 0.00019994510112431564, "elapsed_sec": 60163.14739155769, "step_time_sec": 8.237795663008, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.010386943817139, "lr": 0.00019994463295006643, "elapsed_sec": 60171.38979244232, "step_time_sec": 8.242226447968278, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8632538318634033, "lr": 0.00019994416278857051, "elapsed_sec": 60179.62995624542, "step_time_sec": 8.240000760008115, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.980442523956299, "lr": 0.00019994369063983745, "elapsed_sec": 60187.86861562729, "step_time_sec": 8.238510768976994, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9319028854370117, "lr": 0.0001999432165038769, "elapsed_sec": 60196.10908102989, "step_time_sec": 8.240298936027102, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.836437463760376, "lr": 0.0001999427403806985, "elapsed_sec": 60204.347902059555, "step_time_sec": 8.238689678022638, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8248331546783447, "lr": 0.00019994226227031199, "elapsed_sec": 60212.58878993988, "step_time_sec": 8.240718664019369, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8473949432373047, "lr": 0.00019994178217272712, "elapsed_sec": 60220.8279030323, "step_time_sec": 8.238940447045024, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8138656616210938, "lr": 0.00019994130008795362, "elapsed_sec": 60229.06614637375, "step_time_sec": 8.23807417502394, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.991647720336914, "lr": 0.0001999408160160014, "elapsed_sec": 60237.30972266197, "step_time_sec": 8.243433399999049, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.900421142578125, "lr": 0.00019994032995688027, "elapsed_sec": 60245.54984116554, "step_time_sec": 8.239952004980296, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9770021438598633, "lr": 0.0001999398419106002, "elapsed_sec": 60253.788142204285, "step_time_sec": 8.23812568897847, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9112303256988525, "lr": 0.0001999393518771711, "elapsed_sec": 60262.02873277664, "step_time_sec": 8.240436745982151, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7827365398406982, "lr": 0.000199938859856603, "elapsed_sec": 60270.26804995537, "step_time_sec": 8.239109592046589, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8496174812316895, "lr": 0.00019993836584890587, "elapsed_sec": 60278.50859045982, "step_time_sec": 8.240403373958543, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.764373540878296, "lr": 0.00019993786985408987, "elapsed_sec": 60286.7473654747, "step_time_sec": 8.23859047004953, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.916581630706787, "lr": 0.00019993737187216505, "elapsed_sec": 60294.98492884636, "step_time_sec": 8.237413408001885, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8602113723754883, "lr": 0.00019993687190314157, "elapsed_sec": 60303.22632908821, "step_time_sec": 8.241276871005539, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8121681213378906, "lr": 0.0001999363699470297, "elapsed_sec": 60311.46477890015, "step_time_sec": 8.238267851003911, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9184682369232178, "lr": 0.00019993586600383956, "elapsed_sec": 60319.703505039215, "step_time_sec": 8.238573017995805, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9147226810455322, "lr": 0.0001999353600735815, "elapsed_sec": 60327.94422483444, "step_time_sec": 8.240585352003109, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7999117374420166, "lr": 0.00019993485215626585, "elapsed_sec": 60336.183482170105, "step_time_sec": 8.239058808016125, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9039382934570312, "lr": 0.00019993434225190295, "elapsed_sec": 60344.42397212982, "step_time_sec": 8.240343330951873, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8520522117614746, "lr": 0.00019993383036050316, "elapsed_sec": 60352.66418552399, "step_time_sec": 8.239979882957414, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8671727180480957, "lr": 0.00019993331648207694, "elapsed_sec": 60360.90287446976, "step_time_sec": 8.238551461021416, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8882861137390137, "lr": 0.00019993280061663483, "elapsed_sec": 60369.14381933212, "step_time_sec": 8.240770881006029, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9696052074432373, "lr": 0.00019993228276418728, "elapsed_sec": 60377.384004592896, "step_time_sec": 8.24004646600224, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.956449270248413, "lr": 0.00019993176292474487, "elapsed_sec": 60385.623431921005, "step_time_sec": 8.23925948998658, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9035329818725586, "lr": 0.0001999312410983182, "elapsed_sec": 60393.86323976517, "step_time_sec": 8.239632156968582, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9049549102783203, "lr": 0.00019993071728491792, "elapsed_sec": 60402.102507829666, "step_time_sec": 8.23913361504674, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.759434938430786, "lr": 0.00019993019148455469, "elapsed_sec": 60410.344559669495, "step_time_sec": 8.241844655014575, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.898050546646118, "lr": 0.0001999296636972393, "elapsed_sec": 60418.58483362198, "step_time_sec": 8.240116985980421, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8885951042175293, "lr": 0.00019992913392298243, "elapsed_sec": 60426.82302761078, "step_time_sec": 8.2380359970266, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9505984783172607, "lr": 0.00019992860216179492, "elapsed_sec": 60435.0640308857, "step_time_sec": 8.240859076962806, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9311318397521973, "lr": 0.00019992806841368763, "elapsed_sec": 60443.30332040787, "step_time_sec": 8.23917482997058, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9562630653381348, "lr": 0.0001999275326786714, "elapsed_sec": 60451.54198217392, "step_time_sec": 8.238456154009327, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.043307781219482, "lr": 0.0001999269949567572, "elapsed_sec": 60459.78323984146, "step_time_sec": 8.241097928956151, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7773332595825195, "lr": 0.00019992645524795596, "elapsed_sec": 60468.0226726532, "step_time_sec": 8.239242596027907, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.988309621810913, "lr": 0.00019992591355227874, "elapsed_sec": 60476.26336359978, "step_time_sec": 8.240587848995347, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.852682113647461, "lr": 0.00019992536986973656, "elapsed_sec": 60484.50360798836, "step_time_sec": 8.240027270978317, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.895556926727295, "lr": 0.00019992482420034047, "elapsed_sec": 60492.74166941643, "step_time_sec": 8.23789761599619, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.000462532043457, "lr": 0.00019992427654410167, "elapsed_sec": 60500.98289227486, "step_time_sec": 8.24108566797804, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9068353176116943, "lr": 0.00019992372690103125, "elapsed_sec": 60509.223266124725, "step_time_sec": 8.240233319986146, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.086538314819336, "lr": 0.00019992317527114045, "elapsed_sec": 60517.46214413643, "step_time_sec": 8.238686194992624, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7990097999572754, "lr": 0.00019992262165444055, "elapsed_sec": 60525.703120708466, "step_time_sec": 8.240798662998714, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.150505065917969, "lr": 0.00019992206605094282, "elapsed_sec": 60533.94224643707, "step_time_sec": 8.23902228398947, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9463112354278564, "lr": 0.00019992150846065857, "elapsed_sec": 60542.18231678009, "step_time_sec": 8.239862715010531, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.038095474243164, "lr": 0.0001999209488835992, "elapsed_sec": 60550.42238402367, "step_time_sec": 8.239864842966199, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.6911072731018066, "lr": 0.0001999203873197761, "elapsed_sec": 60558.66162753105, "step_time_sec": 8.239075733989011, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8661656379699707, "lr": 0.00019991982376920076, "elapsed_sec": 60566.9031252861, "step_time_sec": 8.241363808978349, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.6944282054901123, "lr": 0.00019991925823188464, "elapsed_sec": 60575.14294099808, "step_time_sec": 8.239669781003613, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.986647367477417, "lr": 0.00019991869070783928, "elapsed_sec": 60583.38146328926, "step_time_sec": 8.23834344104398, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.758873224258423, "lr": 0.00019991812119707624, "elapsed_sec": 60591.622467041016, "step_time_sec": 8.240878388984129, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8600738048553467, "lr": 0.00019991754969960715, "elapsed_sec": 60599.86202931404, "step_time_sec": 8.239350959949661, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.933868408203125, "lr": 0.00019991697621544365, "elapsed_sec": 60608.102132081985, "step_time_sec": 8.239948397967964, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.984386682510376, "lr": 0.00019991640074459746, "elapsed_sec": 60616.34139561653, "step_time_sec": 8.239101606013719, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8419888019561768, "lr": 0.00019991582328708033, "elapsed_sec": 60624.57996606827, "step_time_sec": 8.238381104019936, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.905548095703125, "lr": 0.000199915243842904, "elapsed_sec": 60632.81976914406, "step_time_sec": 8.239624109992292, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.83709979057312, "lr": 0.00019991466241208028, "elapsed_sec": 60641.05955576897, "step_time_sec": 8.239640348998364, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.991265058517456, "lr": 0.00019991407899462105, "elapsed_sec": 60649.298827171326, "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": 7296, "loss": 3.9601006507873535, "lr": 0.0001999134935905382, "elapsed_sec": 60657.538422584534, "step_time_sec": 8.239480789983645, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.768784761428833, "lr": 0.00019991290619984369, "elapsed_sec": 60665.77614426613, "step_time_sec": 8.237513415981084, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.834402561187744, "lr": 0.00019991231682254946, "elapsed_sec": 60674.017442941666, "step_time_sec": 8.241170194000006, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8766746520996094, "lr": 0.00019991172545866755, "elapsed_sec": 60682.255772829056, "step_time_sec": 8.238167363975663, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8616511821746826, "lr": 0.00019991113210821005, "elapsed_sec": 60701.97134780884, "step_time_sec": 19.71537529298803, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.808931827545166, "lr": 0.000199910536771189, "elapsed_sec": 60710.210453510284, "step_time_sec": 8.238939467992168, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.90185809135437, "lr": 0.00019990993944761658, "elapsed_sec": 60718.448543787, "step_time_sec": 8.237925214983989, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.754633665084839, "lr": 0.00019990934013750498, "elapsed_sec": 60726.687926054, "step_time_sec": 8.2392707199906, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8896899223327637, "lr": 0.00019990873884086636, "elapsed_sec": 60734.9272582531, "step_time_sec": 8.239111686008982, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9411561489105225, "lr": 0.00019990813555771308, "elapsed_sec": 60743.16604757309, "step_time_sec": 8.23867007798981, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9123892784118652, "lr": 0.00019990753028805737, "elapsed_sec": 60751.406913757324, "step_time_sec": 8.2406816809671, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.813396692276001, "lr": 0.00019990692303191158, "elapsed_sec": 60759.644565582275, "step_time_sec": 8.237489471968729, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.97078537940979, "lr": 0.00019990631378928814, "elapsed_sec": 60767.88225483894, "step_time_sec": 8.2375089499983, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9266598224639893, "lr": 0.00019990570256019942, "elapsed_sec": 60776.122211933136, "step_time_sec": 8.239797891990747, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.931255340576172, "lr": 0.0001999050893446579, "elapsed_sec": 60784.360877513885, "step_time_sec": 8.238534658041317, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0124921798706055, "lr": 0.00019990447414267612, "elapsed_sec": 60792.60168790817, "step_time_sec": 8.240635703026783, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.962611675262451, "lr": 0.00019990385695426658, "elapsed_sec": 60800.842035770416, "step_time_sec": 8.240154005994555, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.882237672805786, "lr": 0.0001999032377794419, "elapsed_sec": 60809.0813164711, "step_time_sec": 8.23916726699099, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7327966690063477, "lr": 0.0001999026166182147, "elapsed_sec": 60817.32180714607, "step_time_sec": 8.240315740986262, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8497493267059326, "lr": 0.0001999019934705976, "elapsed_sec": 60825.55939030647, "step_time_sec": 8.237389168993104, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.865628957748413, "lr": 0.0001999013683366034, "elapsed_sec": 60833.798640966415, "step_time_sec": 8.239086138026323, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8853864669799805, "lr": 0.00019990074121624477, "elapsed_sec": 60842.03880858421, "step_time_sec": 8.240024201979395, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9094369411468506, "lr": 0.00019990011210953455, "elapsed_sec": 60850.27782535553, "step_time_sec": 8.238838358956855, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9213783740997314, "lr": 0.00019989948101648556, "elapsed_sec": 60858.51830077171, "step_time_sec": 8.240317917021457, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9177451133728027, "lr": 0.0001998988479371106, "elapsed_sec": 60866.75787067413, "step_time_sec": 8.239423229009844, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.950397253036499, "lr": 0.0001998982128714227, "elapsed_sec": 60874.99711108208, "step_time_sec": 8.239066796028055, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.915530204772949, "lr": 0.0001998975758194347, "elapsed_sec": 60883.236862659454, "step_time_sec": 8.239546841999982, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.880584478378296, "lr": 0.0001998969367811597, "elapsed_sec": 60891.474600315094, "step_time_sec": 8.237597675004508, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8879873752593994, "lr": 0.00019989629575661063, "elapsed_sec": 60899.718782663345, "step_time_sec": 8.244005560001824, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9403698444366455, "lr": 0.00019989565274580062, "elapsed_sec": 60907.9595208168, "step_time_sec": 8.240593971975613, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.828874111175537, "lr": 0.00019989500774874277, "elapsed_sec": 60916.19787859917, "step_time_sec": 8.238183371024206, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.749549388885498, "lr": 0.00019989436076545025, "elapsed_sec": 60924.4391245842, "step_time_sec": 8.24114117299905, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9475362300872803, "lr": 0.00019989371179593622, "elapsed_sec": 60932.679867744446, "step_time_sec": 8.240571180009283, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7676572799682617, "lr": 0.00019989306084021395, "elapsed_sec": 60940.91951107979, "step_time_sec": 8.239443510014098, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.930677890777588, "lr": 0.00019989240789829667, "elapsed_sec": 60949.15913629532, "step_time_sec": 8.239463953999802, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9866790771484375, "lr": 0.00019989175297019775, "elapsed_sec": 60957.39981651306, "step_time_sec": 8.240507308975793, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.987128734588623, "lr": 0.0001998910960559305, "elapsed_sec": 60965.63887500763, "step_time_sec": 8.238912326982245, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8365731239318848, "lr": 0.00019989043715550837, "elapsed_sec": 60973.87930202484, "step_time_sec": 8.240248926973436, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.051498889923096, "lr": 0.00019988977626894474, "elapsed_sec": 60982.11699819565, "step_time_sec": 8.237525812000968, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.928306818008423, "lr": 0.0001998891133962531, "elapsed_sec": 60990.35787200928, "step_time_sec": 8.240732314996421, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.697174072265625, "lr": 0.00019988844853744702, "elapsed_sec": 60998.59706115723, "step_time_sec": 8.239028366981074, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8625106811523438, "lr": 0.00019988778169254, "elapsed_sec": 61006.837161540985, "step_time_sec": 8.239970349997748, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.933711528778076, "lr": 0.00019988711286154565, "elapsed_sec": 61015.07608747482, "step_time_sec": 8.238720979017671, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8765034675598145, "lr": 0.00019988644204447763, "elapsed_sec": 61023.31450533867, "step_time_sec": 8.238286464998964, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8940491676330566, "lr": 0.00019988576924134962, "elapsed_sec": 61031.5550365448, "step_time_sec": 8.240359250979964, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.938823938369751, "lr": 0.00019988509445217534, "elapsed_sec": 61039.79414892197, "step_time_sec": 8.238930516003165, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.88063645362854, "lr": 0.00019988441767696848, "elapsed_sec": 61048.032168626785, "step_time_sec": 8.237845091032796, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.870164632797241, "lr": 0.00019988373891574297, "elapsed_sec": 61056.27243971825, "step_time_sec": 8.240139845991507, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7094316482543945, "lr": 0.00019988305816851255, "elapsed_sec": 61064.51187300682, "step_time_sec": 8.23926423501689, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.053141117095947, "lr": 0.00019988237543529117, "elapsed_sec": 61072.75237226486, "step_time_sec": 8.240345546975732, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.895636558532715, "lr": 0.00019988169071609273, "elapsed_sec": 61080.9913213253, "step_time_sec": 8.23877055401681, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7973310947418213, "lr": 0.00019988100401093113, "elapsed_sec": 61089.22950553894, "step_time_sec": 8.238022408040706, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9204227924346924, "lr": 0.00019988031531982046, "elapsed_sec": 61097.47106003761, "step_time_sec": 8.241377326019574, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.80930495262146, "lr": 0.00019987962464277474, "elapsed_sec": 61105.70946216583, "step_time_sec": 8.238234796968754, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.964570999145508, "lr": 0.00019987893197980807, "elapsed_sec": 61113.94917011261, "step_time_sec": 8.239600636006799, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.011087894439697, "lr": 0.00019987823733093454, "elapsed_sec": 61122.18888306618, "step_time_sec": 8.239486085018143, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9587488174438477, "lr": 0.00019987754069616834, "elapsed_sec": 61130.42662405968, "step_time_sec": 8.237593433004804, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.001885414123535, "lr": 0.00019987684207552366, "elapsed_sec": 61138.66547179222, "step_time_sec": 8.238685143995099, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8908047676086426, "lr": 0.00019987614146901473, "elapsed_sec": 61146.905015945435, "step_time_sec": 8.239384437038098, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9264211654663086, "lr": 0.0001998754388766559, "elapsed_sec": 61155.142788887024, "step_time_sec": 8.237617987964768, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.6725778579711914, "lr": 0.00019987473429846145, "elapsed_sec": 61163.383274555206, "step_time_sec": 8.240369086968713, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9832077026367188, "lr": 0.0001998740277344458, "elapsed_sec": 61171.621619939804, "step_time_sec": 8.238112409017049, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8912851810455322, "lr": 0.00019987331918462327, "elapsed_sec": 61179.862100601196, "step_time_sec": 8.240295215975493, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9675538539886475, "lr": 0.0001998726086490084, "elapsed_sec": 61188.10061454773, "step_time_sec": 8.238376533961855, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8846991062164307, "lr": 0.00019987189612761565, "elapsed_sec": 61196.33907532692, "step_time_sec": 8.238331370986998, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9702510833740234, "lr": 0.0001998711816204595, "elapsed_sec": 61204.57885169983, "step_time_sec": 8.239574841980357, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9073433876037598, "lr": 0.00019987046512755458, "elapsed_sec": 61212.81792449951, "step_time_sec": 8.238926413992886, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9925649166107178, "lr": 0.0001998697466489155, "elapsed_sec": 61221.05720663071, "step_time_sec": 8.239077645004727, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.773456335067749, "lr": 0.0001998690261845569, "elapsed_sec": 61229.29699110985, "step_time_sec": 8.239651374984533, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.921989917755127, "lr": 0.0001998683037344935, "elapsed_sec": 61237.535581588745, "step_time_sec": 8.238391969003715, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.915062665939331, "lr": 0.00019986757929873992, "elapsed_sec": 61245.77462077141, "step_time_sec": 8.238884905993473, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.026802062988281, "lr": 0.00019986685287731108, "elapsed_sec": 61254.01400303841, "step_time_sec": 8.239276833017357, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8693108558654785, "lr": 0.00019986612447022174, "elapsed_sec": 61262.25393629074, "step_time_sec": 8.239702822989784, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.000009059906006, "lr": 0.0001998653940774867, "elapsed_sec": 61270.49255466461, "step_time_sec": 8.23848626500694, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8512585163116455, "lr": 0.00019986466169912096, "elapsed_sec": 61278.72997188568, "step_time_sec": 8.237285783980042, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.947557210922241, "lr": 0.0001998639273351394, "elapsed_sec": 61286.9695212841, "step_time_sec": 8.239392492047045, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.979034900665283, "lr": 0.00019986319098555695, "elapsed_sec": 61295.20868849754, "step_time_sec": 8.23899674002314, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.011239051818848, "lr": 0.0001998624526503887, "elapsed_sec": 61303.447303295135, "step_time_sec": 8.238446030009072, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9303362369537354, "lr": 0.00019986171232964966, "elapsed_sec": 61311.686686992645, "step_time_sec": 8.239187038969249, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.989004373550415, "lr": 0.00019986097002335496, "elapsed_sec": 61319.92412829399, "step_time_sec": 8.237284917966463, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9084832668304443, "lr": 0.00019986022573151977, "elapsed_sec": 61328.16565871239, "step_time_sec": 8.241401337028947, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9537158012390137, "lr": 0.00019985947945415914, "elapsed_sec": 61336.40670514107, "step_time_sec": 8.240819090977311, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9320061206817627, "lr": 0.00019985873119128845, "elapsed_sec": 61344.643669605255, "step_time_sec": 8.236846757004969, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8734235763549805, "lr": 0.00019985798094292285, "elapsed_sec": 61352.88338971138, "step_time_sec": 8.239559017994907, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9351770877838135, "lr": 0.00019985722870907767, "elapsed_sec": 61361.121537685394, "step_time_sec": 8.237958964018617, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9230661392211914, "lr": 0.00019985647448976824, "elapsed_sec": 61369.36071705818, "step_time_sec": 8.239013172045816, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.029635906219482, "lr": 0.00019985571828500998, "elapsed_sec": 61377.599645614624, "step_time_sec": 8.23877924104454, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.023353099822998, "lr": 0.00019985496009481828, "elapsed_sec": 61385.837653160095, "step_time_sec": 8.2378896799637, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.933006525039673, "lr": 0.00019985419991920862, "elapsed_sec": 61394.07716393471, "step_time_sec": 8.239336983999237, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9373035430908203, "lr": 0.0001998534377581965, "elapsed_sec": 61402.31589436531, "step_time_sec": 8.238504818989895, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.004594802856445, "lr": 0.00019985267361179743, "elapsed_sec": 61410.55507898331, "step_time_sec": 8.239024906011764, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.882838249206543, "lr": 0.00019985190748002704, "elapsed_sec": 61418.79354548454, "step_time_sec": 8.23834236501716, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9362878799438477, "lr": 0.00019985113936290093, "elapsed_sec": 61427.03211688995, "step_time_sec": 8.238398023007903, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8009493350982666, "lr": 0.0001998503692604348, "elapsed_sec": 61435.27303504944, "step_time_sec": 8.240731019002851, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7907955646514893, "lr": 0.00019984959717264425, "elapsed_sec": 61443.51299715042, "step_time_sec": 8.23980629799189, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9210493564605713, "lr": 0.00019984882309954515, "elapsed_sec": 61451.75174951553, "step_time_sec": 8.23856668395456, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7991411685943604, "lr": 0.00019984804704115323, "elapsed_sec": 61459.99083161354, "step_time_sec": 8.238948303973302, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9206020832061768, "lr": 0.00019984726899748432, "elapsed_sec": 61468.227939367294, "step_time_sec": 8.236974002968054, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8584413528442383, "lr": 0.00019984648896855428, "elapsed_sec": 61476.46760749817, "step_time_sec": 8.239459214964882, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.870539665222168, "lr": 0.00019984570695437902, "elapsed_sec": 61484.70727467537, "step_time_sec": 8.239505413977895, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.765882968902588, "lr": 0.0001998449229549745, "elapsed_sec": 61492.94467687607, "step_time_sec": 8.237248832010664, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8723459243774414, "lr": 0.0001998441369703567, "elapsed_sec": 61501.185384750366, "step_time_sec": 8.240518815990072, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.951950788497925, "lr": 0.0001998433490005417, "elapsed_sec": 61509.42240381241, "step_time_sec": 8.236874163965695, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.861231803894043, "lr": 0.0001998425590455455, "elapsed_sec": 61517.661786556244, "step_time_sec": 8.239210225001443, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8473660945892334, "lr": 0.0001998417671053842, "elapsed_sec": 61525.9006357193, "step_time_sec": 8.238726345007308, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7671897411346436, "lr": 0.000199840973180074, "elapsed_sec": 61534.13851070404, "step_time_sec": 8.23772477102466, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8360989093780518, "lr": 0.00019984017726963106, "elapsed_sec": 61542.377878665924, "step_time_sec": 8.239177428011317, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9687161445617676, "lr": 0.00019983937937407165, "elapsed_sec": 61550.61601066589, "step_time_sec": 8.237917024001945, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8910059928894043, "lr": 0.00019983857949341203, "elapsed_sec": 61558.856179475784, "step_time_sec": 8.24005348799983, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9743432998657227, "lr": 0.00019983777762766847, "elapsed_sec": 61567.09480667114, "step_time_sec": 8.23840927699348, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.987071990966797, "lr": 0.00019983697377685736, "elapsed_sec": 61575.33405613899, "step_time_sec": 8.239069001981989, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9057204723358154, "lr": 0.0001998361679409951, "elapsed_sec": 61583.57265496254, "step_time_sec": 8.238431980949827, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9649224281311035, "lr": 0.0001998353601200981, "elapsed_sec": 61591.8092520237, "step_time_sec": 8.236419531982392, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9056169986724854, "lr": 0.00019983455031418284, "elapsed_sec": 61600.04901766777, "step_time_sec": 8.239629692980088, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.996124267578125, "lr": 0.00019983373852326585, "elapsed_sec": 61608.28652572632, "step_time_sec": 8.23735840502195, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9377498626708984, "lr": 0.00019983292474736364, "elapsed_sec": 61616.526493787766, "step_time_sec": 8.239812056010123, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0341386795043945, "lr": 0.00019983210898649287, "elapsed_sec": 61624.767107486725, "step_time_sec": 8.24041234498145, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8927345275878906, "lr": 0.00019983129124067012, "elapsed_sec": 61633.003826618195, "step_time_sec": 8.23658126196824, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9528162479400635, "lr": 0.00019983047150991213, "elapsed_sec": 61641.24245405197, "step_time_sec": 8.23841505200835, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8994455337524414, "lr": 0.00019982964979423556, "elapsed_sec": 61649.48012089729, "step_time_sec": 8.237524761003442, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.929244041442871, "lr": 0.0001998288260936572, "elapsed_sec": 61657.71863627434, "step_time_sec": 8.238318368035834, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.002297878265381, "lr": 0.00019982800040819384, "elapsed_sec": 61665.958799123764, "step_time_sec": 8.2400183439604, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8687825202941895, "lr": 0.00019982717273786229, "elapsed_sec": 61674.19715833664, "step_time_sec": 8.238250915019307, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7988502979278564, "lr": 0.00019982634308267946, "elapsed_sec": 61682.435819625854, "step_time_sec": 8.238448648015037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8301475048065186, "lr": 0.00019982551144266227, "elapsed_sec": 61690.67388558388, "step_time_sec": 8.237904068024363, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.902782440185547, "lr": 0.0001998246778178277, "elapsed_sec": 61698.91246461868, "step_time_sec": 8.238448936957866, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9615118503570557, "lr": 0.00019982384220819269, "elapsed_sec": 61707.15175795555, "step_time_sec": 8.239158936019521, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7809488773345947, "lr": 0.00019982300461377433, "elapsed_sec": 61715.388370752335, "step_time_sec": 8.236412908008788, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8345842361450195, "lr": 0.00019982216503458967, "elapsed_sec": 61723.627378702164, "step_time_sec": 8.238871170964558, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9559476375579834, "lr": 0.0001998213234706559, "elapsed_sec": 61731.86749744415, "step_time_sec": 8.240021673962474, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9324686527252197, "lr": 0.00019982047992199006, "elapsed_sec": 61740.10799407959, "step_time_sec": 8.240251047012862, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.066919803619385, "lr": 0.00019981963438860947, "elapsed_sec": 61748.35057902336, "step_time_sec": 8.242474878032226, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7006852626800537, "lr": 0.00019981878687053132, "elapsed_sec": 61756.59040951729, "step_time_sec": 8.239624376001302, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.907270908355713, "lr": 0.00019981793736777292, "elapsed_sec": 61764.8302423954, "step_time_sec": 8.239618782012258, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.046222686767578, "lr": 0.00019981708588035155, "elapsed_sec": 61773.0684735775, "step_time_sec": 8.238126320997253, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.106822490692139, "lr": 0.0001998162324082846, "elapsed_sec": 61781.30647587776, "step_time_sec": 8.237792261992581, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9630837440490723, "lr": 0.0001998153769515895, "elapsed_sec": 61789.54567670822, "step_time_sec": 8.239018319000024, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9717297554016113, "lr": 0.00019981451951028365, "elapsed_sec": 61797.7852396965, "step_time_sec": 8.239398189005442, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8632383346557617, "lr": 0.00019981366008438458, "elapsed_sec": 61806.02408337593, "step_time_sec": 8.238694507046603, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.92897891998291, "lr": 0.00019981279867390978, "elapsed_sec": 61814.26380586624, "step_time_sec": 8.23960209597135, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.839299201965332, "lr": 0.00019981193527887685, "elapsed_sec": 61822.50059723854, "step_time_sec": 8.23658812197391, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.744436025619507, "lr": 0.00019981106989930337, "elapsed_sec": 61830.74351096153, "step_time_sec": 8.242740388028324, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.853924036026001, "lr": 0.00019981020253520698, "elapsed_sec": 61838.98304963112, "step_time_sec": 8.23943583900109, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8424582481384277, "lr": 0.0001998093331866054, "elapsed_sec": 61847.22066450119, "step_time_sec": 8.237430631997995, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8455617427825928, "lr": 0.0001998084618535164, "elapsed_sec": 61855.459224939346, "step_time_sec": 8.238370164996013, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.028924465179443, "lr": 0.00019980758853595763, "elapsed_sec": 61863.69728851318, "step_time_sec": 8.237886012007948, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.937045097351074, "lr": 0.00019980671323394696, "elapsed_sec": 61871.93631839752, "step_time_sec": 8.23889172700001, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7532570362091064, "lr": 0.0001998058359475023, "elapsed_sec": 61880.17529821396, "step_time_sec": 8.238862981030252, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8274786472320557, "lr": 0.00019980495667664148, "elapsed_sec": 61888.41362977028, "step_time_sec": 8.238148910051677, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.933509349822998, "lr": 0.0001998040754213824, "elapsed_sec": 61896.65198516846, "step_time_sec": 8.238189115014393, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.899492025375366, "lr": 0.0001998031921817431, "elapsed_sec": 61904.88897228241, "step_time_sec": 8.23679194797296, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8832547664642334, "lr": 0.00019980230695774154, "elapsed_sec": 61913.1292424202, "step_time_sec": 8.24014247203013, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7429006099700928, "lr": 0.00019980141974939582, "elapsed_sec": 61921.367211818695, "step_time_sec": 8.237772250024136, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.833735466003418, "lr": 0.00019980053055672405, "elapsed_sec": 61929.60515713692, "step_time_sec": 8.237801506998949, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9483258724212646, "lr": 0.00019979963937974426, "elapsed_sec": 61937.8445827961, "step_time_sec": 8.239228890044615, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7192602157592773, "lr": 0.0001997987462184747, "elapsed_sec": 61946.08263015747, "step_time_sec": 8.237923172011506, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7665789127349854, "lr": 0.00019979785107293357, "elapsed_sec": 61954.32146573067, "step_time_sec": 8.238718701002654, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8306922912597656, "lr": 0.00019979695394313916, "elapsed_sec": 61962.55970644951, "step_time_sec": 8.238042556971777, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8752100467681885, "lr": 0.0001997960548291097, "elapsed_sec": 61970.797085523605, "step_time_sec": 8.237215736007784, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.001779556274414, "lr": 0.0001997951537308636, "elapsed_sec": 61979.03670144081, "step_time_sec": 8.23949922400061, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.234477519989014, "lr": 0.00019979425064841917, "elapsed_sec": 61987.27465844154, "step_time_sec": 8.237789373029955, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.827665328979492, "lr": 0.00019979334558179485, "elapsed_sec": 61995.51447534561, "step_time_sec": 8.239624587004073, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.917962074279785, "lr": 0.0001997924385310091, "elapsed_sec": 62003.754222631454, "step_time_sec": 8.239555143984035, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8366050720214844, "lr": 0.00019979152949608044, "elapsed_sec": 62011.99335408211, "step_time_sec": 8.238988475000951, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.925773859024048, "lr": 0.0001997906184770274, "elapsed_sec": 62020.232474565506, "step_time_sec": 8.238946244004183, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.934936046600342, "lr": 0.0001997897054738685, "elapsed_sec": 62028.46947526932, "step_time_sec": 8.236859243013896, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9750161170959473, "lr": 0.00019978879048662245, "elapsed_sec": 62036.710065841675, "step_time_sec": 8.240474843012635, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8072023391723633, "lr": 0.00019978787351530785, "elapsed_sec": 62044.94875693321, "step_time_sec": 8.23843956599012, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.79707670211792, "lr": 0.0001997869545599434, "elapsed_sec": 62053.186474084854, "step_time_sec": 8.237604480993468, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.833981513977051, "lr": 0.0001997860336205479, "elapsed_sec": 62061.42559552193, "step_time_sec": 8.23892541101668, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.887166738510132, "lr": 0.00019978511069714005, "elapsed_sec": 62069.73864006996, "step_time_sec": 8.251684745016973, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.837022304534912, "lr": 0.00019978418578973873, "elapsed_sec": 62077.977105379105, "step_time_sec": 8.238317656039726, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7980103492736816, "lr": 0.00019978325889836274, "elapsed_sec": 62086.21522855759, "step_time_sec": 8.237942036997993, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.6937975883483887, "lr": 0.00019978233002303109, "elapsed_sec": 62094.454704761505, "step_time_sec": 8.23935219401028, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7561874389648438, "lr": 0.00019978139916376265, "elapsed_sec": 62102.692341804504, "step_time_sec": 8.23741921200417, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9035792350769043, "lr": 0.00019978046632057638, "elapsed_sec": 62110.93090963364, "step_time_sec": 8.238387869030703, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.039820671081543, "lr": 0.00019977953149349133, "elapsed_sec": 62119.16985797882, "step_time_sec": 8.23879593698075, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.109999656677246, "lr": 0.0001997785946825266, "elapsed_sec": 62127.408106803894, "step_time_sec": 8.238119738001842, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.848515033721924, "lr": 0.00019977765588770127, "elapsed_sec": 62135.64839529991, "step_time_sec": 8.240107998019084, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8530313968658447, "lr": 0.00019977671510903447, "elapsed_sec": 62143.88611149788, "step_time_sec": 8.237529837002512, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9488301277160645, "lr": 0.00019977577234654535, "elapsed_sec": 62152.125910282135, "step_time_sec": 8.239655595039949, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.994986057281494, "lr": 0.00019977482760025325, "elapsed_sec": 62160.3660595417, "step_time_sec": 8.239977219956927, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7781450748443604, "lr": 0.00019977388087017735, "elapsed_sec": 62168.60508465767, "step_time_sec": 8.238900222990196, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7027909755706787, "lr": 0.00019977293215633698, "elapsed_sec": 62176.844197034836, "step_time_sec": 8.238958931993693, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8628342151641846, "lr": 0.00019977198145875146, "elapsed_sec": 62185.083091259, "step_time_sec": 8.238699935027398, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.880840539932251, "lr": 0.00019977102877744024, "elapsed_sec": 62193.322432756424, "step_time_sec": 8.239200157986488, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.901362180709839, "lr": 0.00019977007411242268, "elapsed_sec": 62201.56000351906, "step_time_sec": 8.23738292499911, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8750767707824707, "lr": 0.0001997691174637183, "elapsed_sec": 62209.79813790321, "step_time_sec": 8.237965237989556, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9725911617279053, "lr": 0.00019976815883134658, "elapsed_sec": 62218.03751373291, "step_time_sec": 8.239287607022561, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.944908380508423, "lr": 0.0001997671982153271, "elapsed_sec": 62226.27528810501, "step_time_sec": 8.237560755980667, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9294936656951904, "lr": 0.0001997662356156794, "elapsed_sec": 62234.515375852585, "step_time_sec": 8.239971749018878, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.918998956680298, "lr": 0.00019976527103242317, "elapsed_sec": 62242.75379443169, "step_time_sec": 8.238228367990814, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.914259672164917, "lr": 0.00019976430446557805, "elapsed_sec": 62250.99159741402, "step_time_sec": 8.237627808004618, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9338197708129883, "lr": 0.00019976333591516373, "elapsed_sec": 62259.2317006588, "step_time_sec": 8.239991655980702, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.658738136291504, "lr": 0.00019976236538120002, "elapsed_sec": 62267.46931099892, "step_time_sec": 8.237471136962995, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8638973236083984, "lr": 0.00019976139286370667, "elapsed_sec": 62275.70965886116, "step_time_sec": 8.240151904989034, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9426984786987305, "lr": 0.0001997604183627035, "elapsed_sec": 62283.94876885414, "step_time_sec": 8.238909704959951, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8803677558898926, "lr": 0.0001997594418782104, "elapsed_sec": 62292.187751054764, "step_time_sec": 8.238875647017267, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.82511830329895, "lr": 0.00019975846341024734, "elapsed_sec": 62300.42649149895, "step_time_sec": 8.238542595994659, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7483878135681152, "lr": 0.00019975748295883417, "elapsed_sec": 62308.665516614914, "step_time_sec": 8.238911330990959, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.865069627761841, "lr": 0.00019975650052399098, "elapsed_sec": 62316.904470682144, "step_time_sec": 8.238745079026558, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8842384815216064, "lr": 0.00019975551610573776, "elapsed_sec": 62325.14321422577, "step_time_sec": 8.238577704993077, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9015517234802246, "lr": 0.00019975452970409456, "elapsed_sec": 62333.38044834137, "step_time_sec": 8.237068148038816, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8972058296203613, "lr": 0.00019975354131908157, "elapsed_sec": 62341.62054681778, "step_time_sec": 8.239940514031332, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.972536325454712, "lr": 0.00019975255095071887, "elapsed_sec": 62349.858726263046, "step_time_sec": 30.678432862041518, "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": 3.818413734436035, "lr": 0.0001997515585990267, "elapsed_sec": 62380.55089139938, "step_time_sec": 8.251241618010681, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7798373699188232, "lr": 0.00019975056426402532, "elapsed_sec": 62388.78835296631, "step_time_sec": 8.237337516970001, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8110244274139404, "lr": 0.00019974956794573494, "elapsed_sec": 62397.026451826096, "step_time_sec": 8.237937104015145, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.857407331466675, "lr": 0.0001997485696441759, "elapsed_sec": 62405.264191150665, "step_time_sec": 8.237530074024107, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9711573123931885, "lr": 0.0001997475693593686, "elapsed_sec": 62413.502207517624, "step_time_sec": 8.237875058955979, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9049854278564453, "lr": 0.00019974656709133344, "elapsed_sec": 62421.739446401596, "step_time_sec": 8.237055592995603, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8837904930114746, "lr": 0.0001997455628400908, "elapsed_sec": 62429.977830171585, "step_time_sec": 8.238196571008302, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.770113229751587, "lr": 0.0001997445566056612, "elapsed_sec": 62438.21406841278, "step_time_sec": 8.236115979962051, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.855475902557373, "lr": 0.00019974354838806516, "elapsed_sec": 62446.452952861786, "step_time_sec": 8.23868903098628, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8957812786102295, "lr": 0.00019974253818732325, "elapsed_sec": 62454.69249248505, "step_time_sec": 8.239437498967163, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7375216484069824, "lr": 0.00019974152600345604, "elapsed_sec": 62462.93142962456, "step_time_sec": 8.238743944035377, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9878180027008057, "lr": 0.00019974051183648417, "elapsed_sec": 62471.1710190773, "step_time_sec": 8.239389599009883, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.841080665588379, "lr": 0.00019973949568642837, "elapsed_sec": 62479.407895326614, "step_time_sec": 8.236730945005547, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.953040838241577, "lr": 0.00019973847755330932, "elapsed_sec": 62487.648186683655, "step_time_sec": 8.240165609982796, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.971088409423828, "lr": 0.0001997374574371478, "elapsed_sec": 62495.88737797737, "step_time_sec": 8.239008392032702, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9909303188323975, "lr": 0.00019973643533796462, "elapsed_sec": 62504.123681783676, "step_time_sec": 8.23617596999975, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8294589519500732, "lr": 0.0001997354112557806, "elapsed_sec": 62512.36233377457, "step_time_sec": 8.238426527997945, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.793224811553955, "lr": 0.00019973438519061668, "elapsed_sec": 62520.59993648529, "step_time_sec": 8.237422801030334, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.948045015335083, "lr": 0.0001997333571424937, "elapsed_sec": 62528.840017557144, "step_time_sec": 8.239958588965237, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.000608921051025, "lr": 0.0001997323271114327, "elapsed_sec": 62537.0790746212, "step_time_sec": 8.238885201979429, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8216919898986816, "lr": 0.00019973129509745464, "elapsed_sec": 62545.316575050354, "step_time_sec": 8.237305259041023, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8059542179107666, "lr": 0.0001997302611005806, "elapsed_sec": 62553.55608654022, "step_time_sec": 8.2393380519934, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.776517868041992, "lr": 0.00019972922512083167, "elapsed_sec": 62561.795013427734, "step_time_sec": 8.238811605027877, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.80631422996521, "lr": 0.00019972818715822892, "elapsed_sec": 62570.03565001488, "step_time_sec": 8.240460336965043, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8555052280426025, "lr": 0.00019972714721279357, "elapsed_sec": 62578.27432227135, "step_time_sec": 8.238531479029916, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.767500638961792, "lr": 0.00019972610528454682, "elapsed_sec": 62586.512457847595, "step_time_sec": 8.238006266998127, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.029902458190918, "lr": 0.00019972506137350991, "elapsed_sec": 62594.75236797333, "step_time_sec": 8.239681385981385, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.05493688583374, "lr": 0.00019972401547970413, "elapsed_sec": 62602.991666555405, "step_time_sec": 8.239150592999067, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.879362106323242, "lr": 0.0001997229676031508, "elapsed_sec": 62611.22969818115, "step_time_sec": 8.237849312950857, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.989467144012451, "lr": 0.0001997219177438713, "elapsed_sec": 62619.46880078316, "step_time_sec": 8.238925173995085, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7956180572509766, "lr": 0.0001997208659018871, "elapsed_sec": 62627.70628595352, "step_time_sec": 8.237378255988006, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.010842800140381, "lr": 0.00019971981207721952, "elapsed_sec": 62635.94653606415, "step_time_sec": 8.240070076019038, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7235634326934814, "lr": 0.00019971875626989018, "elapsed_sec": 62644.18669581413, "step_time_sec": 8.240007164014969, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8274412155151367, "lr": 0.00019971769847992052, "elapsed_sec": 62652.42612743378, "step_time_sec": 8.239205160993151, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8568313121795654, "lr": 0.00019971663870733218, "elapsed_sec": 62660.666149139404, "step_time_sec": 8.239864439994562, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7673144340515137, "lr": 0.00019971557695214672, "elapsed_sec": 62668.90377116203, "step_time_sec": 8.237491794046946, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7276759147644043, "lr": 0.00019971451321438582, "elapsed_sec": 62677.14445352554, "step_time_sec": 8.240497097023763, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8710639476776123, "lr": 0.00019971344749407117, "elapsed_sec": 62685.38335418701, "step_time_sec": 8.238688079989515, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8675427436828613, "lr": 0.00019971237979122446, "elapsed_sec": 62693.62151551247, "step_time_sec": 8.237992547976319, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.854727268218994, "lr": 0.00019971131010586757, "elapsed_sec": 62701.861728191376, "step_time_sec": 8.240065285004675, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.812962055206299, "lr": 0.00019971023843802223, "elapsed_sec": 62710.09908103943, "step_time_sec": 8.237180326017551, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8836374282836914, "lr": 0.0001997091647877103, "elapsed_sec": 62718.339081048965, "step_time_sec": 8.239867875992786, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.082110404968262, "lr": 0.0001997080891549537, "elapsed_sec": 62726.5786921978, "step_time_sec": 8.239466550003272, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.962279796600342, "lr": 0.00019970701153977435, "elapsed_sec": 62734.816257715225, "step_time_sec": 8.237346799985971, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.740494728088379, "lr": 0.00019970593194219423, "elapsed_sec": 62743.056787252426, "step_time_sec": 8.240416448039468, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.5900349617004395, "lr": 0.0001997048503622354, "elapsed_sec": 62751.294604063034, "step_time_sec": 8.237645847955719, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8309943675994873, "lr": 0.00019970376679991983, "elapsed_sec": 62759.534294605255, "step_time_sec": 8.239545862015802, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.907351493835449, "lr": 0.00019970268125526968, "elapsed_sec": 62767.7729074955, "step_time_sec": 8.238426887022797, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.921191930770874, "lr": 0.00019970159372830707, "elapsed_sec": 62776.01078724861, "step_time_sec": 8.23775280098198, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.865849733352661, "lr": 0.00019970050421905418, "elapsed_sec": 62784.24877977371, "step_time_sec": 8.23785506398417, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.952324867248535, "lr": 0.00019969941272753323, "elapsed_sec": 62792.48656129837, "step_time_sec": 8.237564439012203, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8448145389556885, "lr": 0.00019969831925376651, "elapsed_sec": 62800.7251496315, "step_time_sec": 8.238440948014613, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7970938682556152, "lr": 0.00019969722379777627, "elapsed_sec": 62808.964890003204, "step_time_sec": 8.2396064089844, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8249640464782715, "lr": 0.0001996961263595849, "elapsed_sec": 62817.20338988304, "step_time_sec": 8.238326455000788, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7416648864746094, "lr": 0.0001996950269392147, "elapsed_sec": 62825.44230890274, "step_time_sec": 8.238742609973997, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.828321933746338, "lr": 0.00019969392553668817, "elapsed_sec": 62833.67932963371, "step_time_sec": 8.236826613021549, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8482961654663086, "lr": 0.00019969282215202776, "elapsed_sec": 62841.920800447464, "step_time_sec": 8.24128647299949, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7725281715393066, "lr": 0.00019969171678525593, "elapsed_sec": 62850.16067028046, "step_time_sec": 8.239691433962435, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.907113790512085, "lr": 0.00019969060943639524, "elapsed_sec": 62858.39827609062, "step_time_sec": 8.237482781987637, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.624188184738159, "lr": 0.00019968950010546831, "elapsed_sec": 62866.63709616661, "step_time_sec": 8.238629892992321, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.746643304824829, "lr": 0.0001996883887924977, "elapsed_sec": 62874.87508058548, "step_time_sec": 8.237823686969932, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.907130002975464, "lr": 0.00019968727549750614, "elapsed_sec": 62883.11502432823, "step_time_sec": 8.239794455002993, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9263179302215576, "lr": 0.0001996861602205163, "elapsed_sec": 62891.354467868805, "step_time_sec": 8.239306545990985, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.70469331741333, "lr": 0.0001996850429615509, "elapsed_sec": 62899.59231352806, "step_time_sec": 8.237670473987237, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9424941539764404, "lr": 0.00019968392372063279, "elapsed_sec": 62907.83183145523, "step_time_sec": 8.239333203993738, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9111666679382324, "lr": 0.00019968280249778474, "elapsed_sec": 62916.07132601738, "step_time_sec": 8.23934479203308, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9097342491149902, "lr": 0.00019968167929302962, "elapsed_sec": 62924.31094765663, "step_time_sec": 8.23947611200856, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8553686141967773, "lr": 0.00019968055410639037, "elapsed_sec": 62932.55085873604, "step_time_sec": 8.239744951948524, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8394129276275635, "lr": 0.0001996794269378899, "elapsed_sec": 62940.790410757065, "step_time_sec": 8.239365326007828, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9593660831451416, "lr": 0.00019967829778755124, "elapsed_sec": 62949.03757405281, "step_time_sec": 8.246987171994988, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8485519886016846, "lr": 0.00019967716665539737, "elapsed_sec": 62957.27798438072, "step_time_sec": 8.240229438000824, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9821829795837402, "lr": 0.00019967603354145136, "elapsed_sec": 62965.51529741287, "step_time_sec": 8.237174478999805, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9038102626800537, "lr": 0.00019967489844573637, "elapsed_sec": 62973.75553035736, "step_time_sec": 8.240059967036359, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9409191608428955, "lr": 0.0001996737613682755, "elapsed_sec": 62981.99332165718, "step_time_sec": 8.237627288966905, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.927703380584717, "lr": 0.00019967262230909198, "elapsed_sec": 62990.23139190674, "step_time_sec": 8.237900596985128, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.875760793685913, "lr": 0.00019967148126820897, "elapsed_sec": 62998.47253894806, "step_time_sec": 8.241011031961534, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.838440179824829, "lr": 0.0001996703382456498, "elapsed_sec": 63006.71076107025, "step_time_sec": 8.238078580994625, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8107004165649414, "lr": 0.00019966919324143777, "elapsed_sec": 63014.95156407356, "step_time_sec": 8.240627065009903, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.908740997314453, "lr": 0.0001996680462555962, "elapsed_sec": 63023.19106245041, "step_time_sec": 8.239330106007401, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.892535448074341, "lr": 0.00019966689728814855, "elapsed_sec": 63031.42888665199, "step_time_sec": 8.237628823029809, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.81854248046875, "lr": 0.00019966574633911819, "elapsed_sec": 63039.66653418541, "step_time_sec": 8.237549035984557, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.047605991363525, "lr": 0.0001996645934085286, "elapsed_sec": 63047.90402054787, "step_time_sec": 8.237291502999142, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9272079467773438, "lr": 0.0001996634384964033, "elapsed_sec": 63056.14367580414, "step_time_sec": 8.239537420042325, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.964463949203491, "lr": 0.00019966228160276585, "elapsed_sec": 63064.38189411163, "step_time_sec": 8.238060990988743, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9235129356384277, "lr": 0.00019966112272763983, "elapsed_sec": 63072.62086081505, "step_time_sec": 8.238750714983325, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.743704319000244, "lr": 0.00019965996187104888, "elapsed_sec": 63080.859660863876, "step_time_sec": 8.23864015401341, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9915685653686523, "lr": 0.00019965879903301672, "elapsed_sec": 63089.096506118774, "step_time_sec": 8.236715117003769, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8919053077697754, "lr": 0.00019965763421356697, "elapsed_sec": 63097.33449292183, "step_time_sec": 8.237780558993109, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.770610809326172, "lr": 0.00019965646741272345, "elapsed_sec": 63105.573862314224, "step_time_sec": 8.239197721006349, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9293134212493896, "lr": 0.00019965529863050994, "elapsed_sec": 63113.813131809235, "step_time_sec": 8.239098502031993, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9014387130737305, "lr": 0.00019965412786695027, "elapsed_sec": 63122.05307865143, "step_time_sec": 8.239766295999289, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.00613260269165, "lr": 0.00019965295512206833, "elapsed_sec": 63130.29034328461, "step_time_sec": 8.237129301007371, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9388058185577393, "lr": 0.00019965178039588804, "elapsed_sec": 63138.529794216156, "step_time_sec": 8.239291200006846, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7919223308563232, "lr": 0.00019965060368843334, "elapsed_sec": 63146.76886463165, "step_time_sec": 8.238910886982922, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7523300647735596, "lr": 0.0001996494249997282, "elapsed_sec": 63155.007531404495, "step_time_sec": 8.238529348978773, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8404388427734375, "lr": 0.00019964824432979673, "elapsed_sec": 63163.24623274803, "step_time_sec": 8.238504673005082, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.936333417892456, "lr": 0.00019964706167866297, "elapsed_sec": 63171.48453736305, "step_time_sec": 8.238112305989489, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.855710983276367, "lr": 0.000199645877046351, "elapsed_sec": 63179.72552704811, "step_time_sec": 8.240834519034252, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.93524169921875, "lr": 0.00019964469043288506, "elapsed_sec": 63187.964874982834, "step_time_sec": 8.239218623028137, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.976714849472046, "lr": 0.0001996435018382893, "elapsed_sec": 63196.20244002342, "step_time_sec": 8.237447192019317, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.861191987991333, "lr": 0.00019964231126258793, "elapsed_sec": 63204.4423160553, "step_time_sec": 8.23968318401603, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.968980312347412, "lr": 0.0001996411187058053, "elapsed_sec": 63212.68097424507, "step_time_sec": 8.238453781988937, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9662256240844727, "lr": 0.00019963992416796567, "elapsed_sec": 63220.92168927193, "step_time_sec": 8.240558715013321, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8428895473480225, "lr": 0.00019963872764909343, "elapsed_sec": 63229.1609582901, "step_time_sec": 8.239127011969686, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9858431816101074, "lr": 0.00019963752914921299, "elapsed_sec": 63237.398854255676, "step_time_sec": 8.237717241980135, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9320127964019775, "lr": 0.00019963632866834877, "elapsed_sec": 63245.63982415199, "step_time_sec": 8.240804706001654, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9171996116638184, "lr": 0.00019963512620652525, "elapsed_sec": 63253.88017177582, "step_time_sec": 8.240259283978958, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7951700687408447, "lr": 0.000199633921763767, "elapsed_sec": 63262.11889052391, "step_time_sec": 8.238498377031647, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.983189582824707, "lr": 0.0001996327153400985, "elapsed_sec": 63270.35916018486, "step_time_sec": 8.240138983994257, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9120795726776123, "lr": 0.00019963150693554446, "elapsed_sec": 63278.59680104256, "step_time_sec": 8.237458530988079, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8591196537017822, "lr": 0.0001996302965501294, "elapsed_sec": 63286.8372862339, "step_time_sec": 8.240319611039013, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9882004261016846, "lr": 0.0001996290841838781, "elapsed_sec": 63295.07626032829, "step_time_sec": 8.23883200797718, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9601194858551025, "lr": 0.00019962786983681528, "elapsed_sec": 63303.314207315445, "step_time_sec": 8.237748494022526, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9057559967041016, "lr": 0.00019962665350896564, "elapsed_sec": 63311.55427622795, "step_time_sec": 8.239915413025301, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.945010185241699, "lr": 0.000199625435200354, "elapsed_sec": 63319.79250431061, "step_time_sec": 8.238065556972288, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.000380516052246, "lr": 0.0001996242149110053, "elapsed_sec": 63328.032191991806, "step_time_sec": 8.239521552983206, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9396491050720215, "lr": 0.0001996229926409443, "elapsed_sec": 63336.271247148514, "step_time_sec": 8.238900710013695, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9475409984588623, "lr": 0.000199621768390196, "elapsed_sec": 63344.510046482086, "step_time_sec": 8.23867519700434, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9630091190338135, "lr": 0.00019962054215878536, "elapsed_sec": 63352.749865055084, "step_time_sec": 8.239625549002085, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.049633026123047, "lr": 0.00019961931394673737, "elapsed_sec": 63360.989443063736, "step_time_sec": 8.239413685980253, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9009499549865723, "lr": 0.00019961808375407704, "elapsed_sec": 63369.22842025757, "step_time_sec": 8.23883384099463, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8552086353302, "lr": 0.00019961685158082957, "elapsed_sec": 63377.46711540222, "step_time_sec": 8.23853525595041, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9112374782562256, "lr": 0.00019961561742701998, "elapsed_sec": 63385.70528244972, "step_time_sec": 8.237986885011196, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.910386085510254, "lr": 0.00019961438129267346, "elapsed_sec": 63393.94421458244, "step_time_sec": 8.238797867030371, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.863760232925415, "lr": 0.0001996131431778153, "elapsed_sec": 63402.182399988174, "step_time_sec": 8.237984987033997, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7684454917907715, "lr": 0.00019961190308247063, "elapsed_sec": 63410.42236161232, "step_time_sec": 8.239858758985065, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.905127763748169, "lr": 0.00019961066100666482, "elapsed_sec": 63418.66232943535, "step_time_sec": 8.239775537978858, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8391921520233154, "lr": 0.00019960941695042317, "elapsed_sec": 63426.90054678917, "step_time_sec": 8.238044745987281, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8450264930725098, "lr": 0.00019960817091377105, "elapsed_sec": 63435.13965988159, "step_time_sec": 8.238980136986356, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8344321250915527, "lr": 0.0001996069228967339, "elapsed_sec": 63443.37868690491, "step_time_sec": 8.238824300002307, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.862192153930664, "lr": 0.00019960567289933713, "elapsed_sec": 63451.61740064621, "step_time_sec": 8.238575211027637, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9942879676818848, "lr": 0.00019960442092160628, "elapsed_sec": 63459.85803103447, "step_time_sec": 8.240455135994125, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.907728672027588, "lr": 0.00019960316696356685, "elapsed_sec": 63468.09528660774, "step_time_sec": 8.23710510903038, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.942929744720459, "lr": 0.0001996019110252444, "elapsed_sec": 63476.33564424515, "step_time_sec": 8.240238934988156, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8327395915985107, "lr": 0.00019960065310666456, "elapsed_sec": 63484.5751953125, "step_time_sec": 8.239394812029786, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8962063789367676, "lr": 0.000199599393207853, "elapsed_sec": 63492.812771081924, "step_time_sec": 8.237364401982632, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8848564624786377, "lr": 0.00019959813132883542, "elapsed_sec": 63501.05339074135, "step_time_sec": 8.240447321964893, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8640992641448975, "lr": 0.0001995968674696375, "elapsed_sec": 63509.29053735733, "step_time_sec": 8.236991708981805, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.835745334625244, "lr": 0.00019959560163028504, "elapsed_sec": 63517.52770447731, "step_time_sec": 8.236988780961838, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9360623359680176, "lr": 0.00019959433381080388, "elapsed_sec": 63525.76611447334, "step_time_sec": 8.23827082302887, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.178130149841309, "lr": 0.00019959306401121985, "elapsed_sec": 63534.00325632095, "step_time_sec": 8.236994980019517, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.875919818878174, "lr": 0.00019959179223155886, "elapsed_sec": 63542.24381542206, "step_time_sec": 8.240326417027973, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.987387180328369, "lr": 0.00019959051847184684, "elapsed_sec": 63550.480899095535, "step_time_sec": 8.236921502975747, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7620737552642822, "lr": 0.00019958924273210976, "elapsed_sec": 63558.72167801857, "step_time_sec": 8.240638518007472, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.829206943511963, "lr": 0.00019958796501237368, "elapsed_sec": 63566.960763692856, "step_time_sec": 8.238914735033177, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9477012157440186, "lr": 0.00019958668531266458, "elapsed_sec": 63575.198424100876, "step_time_sec": 8.23751641297713, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8526508808135986, "lr": 0.0001995854036330086, "elapsed_sec": 63583.436864852905, "step_time_sec": 8.238257682009134, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7675747871398926, "lr": 0.0001995841199734319, "elapsed_sec": 63591.673593997955, "step_time_sec": 8.236567802960053, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9311859607696533, "lr": 0.0001995828343339606, "elapsed_sec": 63599.91384601593, "step_time_sec": 8.24012092204066, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.947852849960327, "lr": 0.00019958154671462097, "elapsed_sec": 63608.15316057205, "step_time_sec": 8.239156280993484, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8317911624908447, "lr": 0.00019958025711543927, "elapsed_sec": 63616.39098381996, "step_time_sec": 8.237598157022148, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.72811222076416, "lr": 0.0001995789655364418, "elapsed_sec": 63624.630769729614, "step_time_sec": 8.239680441038217, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9526383876800537, "lr": 0.0001995776719776548, "elapsed_sec": 63632.868027210236, "step_time_sec": 8.237067207985092, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.771569013595581, "lr": 0.0001995763764391048, "elapsed_sec": 63641.107660770416, "step_time_sec": 8.239509392005857, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.804774045944214, "lr": 0.00019957507892081814, "elapsed_sec": 63649.347019672394, "step_time_sec": 8.239159363030922, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0076446533203125, "lr": 0.0001995737794228213, "elapsed_sec": 63657.58645939827, "step_time_sec": 8.23928096902091, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9344899654388428, "lr": 0.00019957247794514076, "elapsed_sec": 63665.82671737671, "step_time_sec": 8.24010154599091, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8547120094299316, "lr": 0.0001995711744878031, "elapsed_sec": 63674.066507577896, "step_time_sec": 8.239689361013006, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.914125442504883, "lr": 0.00019956986905083485, "elapsed_sec": 63682.307723522186, "step_time_sec": 8.240995281026699, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9613566398620605, "lr": 0.00019956856163426268, "elapsed_sec": 63690.54771280289, "step_time_sec": 8.239840613969136, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8536410331726074, "lr": 0.00019956725223811323, "elapsed_sec": 63698.78671169281, "step_time_sec": 8.238855161995161, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.886934757232666, "lr": 0.0001995659408624132, "elapsed_sec": 63707.02717733383, "step_time_sec": 8.240301871963311, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8989548683166504, "lr": 0.0001995646275071894, "elapsed_sec": 63715.266397476196, "step_time_sec": 8.239085317007266, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9091415405273438, "lr": 0.00019956331217246847, "elapsed_sec": 63723.5060904026, "step_time_sec": 8.239476067014039, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0371809005737305, "lr": 0.0001995619948582774, "elapsed_sec": 63731.746306180954, "step_time_sec": 8.24007439299021, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.011555194854736, "lr": 0.0001995606755646429, "elapsed_sec": 63739.98505568504, "step_time_sec": 8.238609894993715, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.869006872177124, "lr": 0.00019955935429159197, "elapsed_sec": 63748.22459411621, "step_time_sec": 8.239342605986167, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8313050270080566, "lr": 0.00019955803103915157, "elapsed_sec": 63756.46331143379, "step_time_sec": 8.23857017501723, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.801539659500122, "lr": 0.00019955670580734864, "elapsed_sec": 63764.70160126686, "step_time_sec": 8.238149474957027, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8468031883239746, "lr": 0.00019955537859621023, "elapsed_sec": 63772.94200849533, "step_time_sec": 8.240228187991306, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.032352447509766, "lr": 0.0001995540494057634, "elapsed_sec": 63781.18183040619, "step_time_sec": 8.239635823003482, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.743687629699707, "lr": 0.0001995527182360352, "elapsed_sec": 63789.42224764824, "step_time_sec": 8.240268452034798, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8903863430023193, "lr": 0.00019955138508705286, "elapsed_sec": 63797.66182136536, "step_time_sec": 8.23945019004168, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.986567974090576, "lr": 0.00019955004995884354, "elapsed_sec": 63805.8994808197, "step_time_sec": 8.237469009996857, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7788944244384766, "lr": 0.00019954871285143448, "elapsed_sec": 63814.13977217674, "step_time_sec": 8.240181664994452, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9812405109405518, "lr": 0.00019954737376485292, "elapsed_sec": 63822.37922692299, "step_time_sec": 8.239213949011173, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.981991767883301, "lr": 0.00019954603269912617, "elapsed_sec": 63830.61847996712, "step_time_sec": 8.239095052995253, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.919036865234375, "lr": 0.00019954468965428158, "elapsed_sec": 63838.858187675476, "step_time_sec": 8.239582215959672, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.895082473754883, "lr": 0.0001995433446303466, "elapsed_sec": 63847.09529590607, "step_time_sec": 8.236864400969353, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8401715755462646, "lr": 0.00019954199762734856, "elapsed_sec": 63855.335169792175, "step_time_sec": 8.239707754983101, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.063614368438721, "lr": 0.000199540648645315, "elapsed_sec": 63863.574556827545, "step_time_sec": 8.239203840028495, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.853280544281006, "lr": 0.00019953929768427343, "elapsed_sec": 63871.811170339584, "step_time_sec": 8.236486147972755, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.110435962677002, "lr": 0.00019953794474425137, "elapsed_sec": 63880.05155849457, "step_time_sec": 8.240222736028954, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8440353870391846, "lr": 0.00019953658982527642, "elapsed_sec": 63888.28995513916, "step_time_sec": 8.23821251100162, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7180631160736084, "lr": 0.0001995352329273762, "elapsed_sec": 63896.530165433884, "step_time_sec": 8.24008416198194, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9460906982421875, "lr": 0.00019953387405057844, "elapsed_sec": 63904.77026057243, "step_time_sec": 8.23992535599973, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.894566774368286, "lr": 0.0001995325131949108, "elapsed_sec": 63913.007742881775, "step_time_sec": 8.237281995010562, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.947288990020752, "lr": 0.00019953115036040102, "elapsed_sec": 63921.24774861336, "step_time_sec": 8.239837904984597, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9038376808166504, "lr": 0.00019952978554707692, "elapsed_sec": 63929.486599206924, "step_time_sec": 8.238666676043067, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.993041515350342, "lr": 0.0001995284187549663, "elapsed_sec": 63937.72527170181, "step_time_sec": 8.238507300964557, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.81885027885437, "lr": 0.0001995270499840971, "elapsed_sec": 63945.96529650688, "step_time_sec": 8.23984501796076, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.905566692352295, "lr": 0.00019952567923449722, "elapsed_sec": 63954.20265626907, "step_time_sec": 8.237255415995605, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9430794715881348, "lr": 0.00019952430650619453, "elapsed_sec": 63962.443343639374, "step_time_sec": 8.240475700993557, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.720228672027588, "lr": 0.00019952293179921715, "elapsed_sec": 63970.682493925095, "step_time_sec": 8.238992175029125, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0229082107543945, "lr": 0.00019952155511359303, "elapsed_sec": 63978.92147397995, "step_time_sec": 8.23885818803683, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8562707901000977, "lr": 0.00019952017644935026, "elapsed_sec": 63987.159348249435, "step_time_sec": 8.237690031004604, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8176403045654297, "lr": 0.00019951879580651697, "elapsed_sec": 63995.39712524414, "step_time_sec": 8.237627792987041, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9500982761383057, "lr": 0.00019951741318512128, "elapsed_sec": 64003.638035297394, "step_time_sec": 8.240769814001396, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9876346588134766, "lr": 0.00019951602858519146, "elapsed_sec": 64011.877345085144, "step_time_sec": 8.239147942047566, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.017998218536377, "lr": 0.00019951464200675567, "elapsed_sec": 64020.11729335785, "step_time_sec": 8.23977255995851, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8094429969787598, "lr": 0.00019951325344984226, "elapsed_sec": 64028.35677909851, "step_time_sec": 8.239293503051158, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.95987868309021, "lr": 0.00019951186291447946, "elapsed_sec": 64036.59449481964, "step_time_sec": 8.237540339992847, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.984440565109253, "lr": 0.0001995104704006957, "elapsed_sec": 64044.834284067154, "step_time_sec": 8.239652852003928, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.039976119995117, "lr": 0.00019950907590851936, "elapsed_sec": 64053.07317852974, "step_time_sec": 8.238733317994047, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.053621768951416, "lr": 0.00019950767943797886, "elapsed_sec": 64061.311052560806, "step_time_sec": 8.237696379015688, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.963435411453247, "lr": 0.00019950628098910268, "elapsed_sec": 64069.55155110359, "step_time_sec": 8.240317157004029, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9614367485046387, "lr": 0.0001995048805619194, "elapsed_sec": 64077.78896164894, "step_time_sec": 8.237250361009501, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9471123218536377, "lr": 0.0001995034781564575, "elapsed_sec": 64086.02836227417, "step_time_sec": 8.23925570998108, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.85294508934021, "lr": 0.00019950207377274562, "elapsed_sec": 64094.26877498627, "step_time_sec": 8.240270412003156, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.5730667114257812, "lr": 0.00019950066741081238, "elapsed_sec": 64102.50685405731, "step_time_sec": 8.23788994498318, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7048120498657227, "lr": 0.0001994992590706865, "elapsed_sec": 64110.74564862251, "step_time_sec": 8.238576267031021, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.065502643585205, "lr": 0.00019949784875239663, "elapsed_sec": 64118.982697963715, "step_time_sec": 8.236921189003624, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.989321231842041, "lr": 0.0001994964364559716, "elapsed_sec": 64127.22158789635, "step_time_sec": 8.238709805009421, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9148828983306885, "lr": 0.0001994950221814402, "elapsed_sec": 64135.460943460464, "step_time_sec": 8.239248322031926, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9399101734161377, "lr": 0.00019949360592883125, "elapsed_sec": 64143.70373558998, "step_time_sec": 8.242637916002423, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.967136859893799, "lr": 0.00019949218769817363, "elapsed_sec": 64151.961508750916, "step_time_sec": 8.257578048971482, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.855346202850342, "lr": 0.00019949076748949628, "elapsed_sec": 64160.202149391174, "step_time_sec": 8.240462800022215, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9685184955596924, "lr": 0.00019948934530282817, "elapsed_sec": 64168.441846609116, "step_time_sec": 8.239488507970236, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.027318954467773, "lr": 0.0001994879211381983, "elapsed_sec": 64176.6805832386, "step_time_sec": 8.238594369031489, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.943931818008423, "lr": 0.00019948649499563567, "elapsed_sec": 64184.91976881027, "step_time_sec": 8.239001145993825, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9908158779144287, "lr": 0.00019948506687516942, "elapsed_sec": 64193.157990932465, "step_time_sec": 8.238076812995132, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9159340858459473, "lr": 0.00019948363677682867, "elapsed_sec": 64201.39826464653, "step_time_sec": 8.240111541002989, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.904783248901367, "lr": 0.00019948220470064254, "elapsed_sec": 64209.63548254967, "step_time_sec": 8.23706472699996, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.006414413452148, "lr": 0.00019948077064664026, "elapsed_sec": 64217.87299966812, "step_time_sec": 8.237332200980745, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9802374839782715, "lr": 0.00019947933461485112, "elapsed_sec": 64226.112246751785, "step_time_sec": 8.239077189005911, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.906931161880493, "lr": 0.0001994778966053043, "elapsed_sec": 64234.349836587906, "step_time_sec": 8.237486692029051, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.91933012008667, "lr": 0.00019947645661802925, "elapsed_sec": 64242.58856797218, "step_time_sec": 8.23856520204572, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.073919296264648, "lr": 0.00019947501465305526, "elapsed_sec": 64250.82892012596, "step_time_sec": 8.240135697997175, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.078364372253418, "lr": 0.00019947357071041174, "elapsed_sec": 64259.066286087036, "step_time_sec": 8.237279012973886, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.974771738052368, "lr": 0.0001994721247901282, "elapsed_sec": 64267.30613565445, "step_time_sec": 8.239633416000288, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.088892936706543, "lr": 0.000199470676892234, "elapsed_sec": 64275.54389476776, "step_time_sec": 8.237577011052053, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.998133420944214, "lr": 0.00019946922701675882, "elapsed_sec": 64283.78473806381, "step_time_sec": 8.240652684995439, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9789764881134033, "lr": 0.00019946777516373213, "elapsed_sec": 64292.02390170097, "step_time_sec": 8.239022187015507, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9246625900268555, "lr": 0.00019946632133318356, "elapsed_sec": 64300.26182627678, "step_time_sec": 8.237786184996367, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.023693084716797, "lr": 0.0001994648655251428, "elapsed_sec": 64308.50186753273, "step_time_sec": 8.23985824099509, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.075754165649414, "lr": 0.00019946340773963943, "elapsed_sec": 64316.74014377594, "step_time_sec": 8.238093591004144, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0325703620910645, "lr": 0.0001994619479767033, "elapsed_sec": 64324.97889661789, "step_time_sec": 8.23859220201848, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.957479238510132, "lr": 0.00019946048623636412, "elapsed_sec": 64333.21784687042, "step_time_sec": 8.238809476024471, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.124735355377197, "lr": 0.0001994590225186517, "elapsed_sec": 64341.45681285858, "step_time_sec": 8.238788150018081, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9845259189605713, "lr": 0.00019945755682359594, "elapsed_sec": 64349.697607278824, "step_time_sec": 8.24063382402528, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.028678894042969, "lr": 0.00019945608915122667, "elapsed_sec": 64357.93623447418, "step_time_sec": 8.238462262030225, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9765989780426025, "lr": 0.00019945461950157382, "elapsed_sec": 64366.17637634277, "step_time_sec": 8.239920102991164, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.006913661956787, "lr": 0.00019945314787466744, "elapsed_sec": 64374.41504716873, "step_time_sec": 8.238569820998237, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.902982473373413, "lr": 0.00019945167427053745, "elapsed_sec": 64382.65352654457, "step_time_sec": 8.238250161986798, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8836276531219482, "lr": 0.00019945019868921392, "elapsed_sec": 64390.89444231987, "step_time_sec": 8.240758180967532, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.099615573883057, "lr": 0.000199448721130727, "elapsed_sec": 64399.133724451065, "step_time_sec": 8.239175866008736, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.007644176483154, "lr": 0.00019944724159510677, "elapsed_sec": 64407.37268972397, "step_time_sec": 8.238779597973917, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.091281414031982, "lr": 0.0001994457600823834, "elapsed_sec": 64415.61345529556, "step_time_sec": 8.240594737988431, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.079905033111572, "lr": 0.00019944427659258717, "elapsed_sec": 64423.851871967316, "step_time_sec": 8.238241085025948, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.220509052276611, "lr": 0.00019944279112574825, "elapsed_sec": 64432.09135246277, "step_time_sec": 8.239394807023928, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9616997241973877, "lr": 0.00019944130368189695, "elapsed_sec": 64440.33078479767, "step_time_sec": 8.239262371032964, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.140963077545166, "lr": 0.00019943981426106367, "elapsed_sec": 64448.56788468361, "step_time_sec": 8.23691273899749, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.012660503387451, "lr": 0.0001994383228632787, "elapsed_sec": 64456.808191776276, "step_time_sec": 8.240156448970083, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.038161277770996, "lr": 0.0001994368294885725, "elapsed_sec": 64465.0469493866, "step_time_sec": 8.238556072989013, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0348591804504395, "lr": 0.00019943533413697555, "elapsed_sec": 64473.28528523445, "step_time_sec": 8.238229988026433, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.03300666809082, "lr": 0.00019943383680851828, "elapsed_sec": 64481.52431678772, "step_time_sec": 8.238846942957025, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.003787517547607, "lr": 0.00019943233750323127, "elapsed_sec": 64489.7619535923, "step_time_sec": 8.237506682984531, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.045996189117432, "lr": 0.00019943083622114507, "elapsed_sec": 64498.00107431412, "step_time_sec": 8.238907307968475, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.065670490264893, "lr": 0.00019942933296229033, "elapsed_sec": 64506.23947763443, "step_time_sec": 8.238237169978675, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.083779335021973, "lr": 0.0001994278277266977, "elapsed_sec": 64514.47879910469, "step_time_sec": 8.239200078009162, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.088117599487305, "lr": 0.00019942632051439784, "elapsed_sec": 64522.71715545654, "step_time_sec": 8.238139920053072, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.285034656524658, "lr": 0.00019942481132542153, "elapsed_sec": 64530.95517373085, "step_time_sec": 8.237860077002551, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.147132873535156, "lr": 0.00019942330015979953, "elapsed_sec": 64539.19461774826, "step_time_sec": 8.239288543001749, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.948352813720703, "lr": 0.00019942178701756264, "elapsed_sec": 64547.433483839035, "step_time_sec": 8.238671895989683, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.153876304626465, "lr": 0.00019942027189874176, "elapsed_sec": 64555.671375989914, "step_time_sec": 8.237767081009224, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.230422496795654, "lr": 0.00019941875480336777, "elapsed_sec": 64563.91133975983, "step_time_sec": 8.239777211972978, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.985670804977417, "lr": 0.00019941723573147158, "elapsed_sec": 64572.14887046814, "step_time_sec": 8.23736061702948, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.994072675704956, "lr": 0.00019941571468308424, "elapsed_sec": 64580.38725566864, "step_time_sec": 8.238287393993232, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.965574264526367, "lr": 0.00019941419165823667, "elapsed_sec": 64588.62494516373, "step_time_sec": 8.237468758015893, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.142126083374023, "lr": 0.00019941266665696002, "elapsed_sec": 64596.863871097565, "step_time_sec": 8.23877206700854, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7817001342773438, "lr": 0.00019941113967928534, "elapsed_sec": 64605.10321855545, "step_time_sec": 8.239174611051567, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.009719371795654, "lr": 0.0001994096107252438, "elapsed_sec": 64613.34129023552, "step_time_sec": 8.23792077897815, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7901110649108887, "lr": 0.00019940807979486655, "elapsed_sec": 64621.57920384407, "step_time_sec": 8.237742388970219, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9877309799194336, "lr": 0.0001994065468881848, "elapsed_sec": 64629.817677259445, "step_time_sec": 8.23831769201206, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.260919570922852, "lr": 0.00019940501200522992, "elapsed_sec": 64638.056958436966, "step_time_sec": 8.239099883008748, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.160409927368164, "lr": 0.00019940347514603308, "elapsed_sec": 64646.29541492462, "step_time_sec": 8.23830851202365, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.197762966156006, "lr": 0.00019940193631062567, "elapsed_sec": 64654.53218746185, "step_time_sec": 8.23661829403136, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9264743328094482, "lr": 0.00019940039549903909, "elapsed_sec": 64662.771280527115, "step_time_sec": 8.238870528992265, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.880002021789551, "lr": 0.00019939885271130474, "elapsed_sec": 64671.00836730003, "step_time_sec": 8.23694691498531, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.058879375457764, "lr": 0.00019939730794745408, "elapsed_sec": 64679.24925541878, "step_time_sec": 8.240743997041136, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.022386074066162, "lr": 0.00019939576120751865, "elapsed_sec": 64687.48936891556, "step_time_sec": 8.239902046974748, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.960939645767212, "lr": 0.00019939421249152994, "elapsed_sec": 64695.72634482384, "step_time_sec": 8.23685480694985, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9766194820404053, "lr": 0.0001993926617995196, "elapsed_sec": 64703.9664914608, "step_time_sec": 8.239983392995782, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0239410400390625, "lr": 0.00019939110913151919, "elapsed_sec": 64712.20406603813, "step_time_sec": 8.237374873016961, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.01305627822876, "lr": 0.00019938955448756037, "elapsed_sec": 64720.4431040287, "step_time_sec": 8.238908927014563, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9870762825012207, "lr": 0.00019938799786767492, "elapsed_sec": 64728.68340539932, "step_time_sec": 8.240114798012655, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.121431827545166, "lr": 0.00019938643927189454, "elapsed_sec": 64736.92328810692, "step_time_sec": 8.23970624001231, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.097031116485596, "lr": 0.00019938487870025098, "elapsed_sec": 64745.1621594429, "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": 7789, "loss": 4.027121543884277, "lr": 0.00019938331615277613, "elapsed_sec": 64753.401616573334, "step_time_sec": 8.239332985016517, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.134923458099365, "lr": 0.00019938175162950185, "elapsed_sec": 64761.64009261131, "step_time_sec": 8.238306280982215, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.988961696624756, "lr": 0.00019938018513046, "elapsed_sec": 64769.87914800644, "step_time_sec": 8.238908689003438, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9894988536834717, "lr": 0.00019937861665568252, "elapsed_sec": 64778.11691093445, "step_time_sec": 8.237573037971742, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9767074584960938, "lr": 0.00019937704620520146, "elapsed_sec": 64786.356120824814, "step_time_sec": 8.239044197020121, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.142277717590332, "lr": 0.00019937547377904882, "elapsed_sec": 64794.59438729286, "step_time_sec": 8.238151424040552, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9889631271362305, "lr": 0.00019937389937725665, "elapsed_sec": 64802.83355426788, "step_time_sec": 8.238932662003208, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.98148775100708, "lr": 0.00019937232299985707, "elapsed_sec": 64811.07336139679, "step_time_sec": 8.23965321900323, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.005698204040527, "lr": 0.00019937074464688223, "elapsed_sec": 64819.31035733223, "step_time_sec": 8.236822898034006, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8461155891418457, "lr": 0.0001993691643183643, "elapsed_sec": 64827.54993200302, "step_time_sec": 8.23940818203846, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.130406856536865, "lr": 0.00019936758201433556, "elapsed_sec": 64835.78736400604, "step_time_sec": 8.237304266018327, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1305012702941895, "lr": 0.0001993659977348282, "elapsed_sec": 64844.02545118332, "step_time_sec": 8.237941660976503, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.129250526428223, "lr": 0.00019936441147987459, "elapsed_sec": 64852.265197753906, "step_time_sec": 8.239490239007864, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9789931774139404, "lr": 0.00019936282324950705, "elapsed_sec": 64860.5025510788, "step_time_sec": 8.237231614009943, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.028469085693359, "lr": 0.000199361233043758, "elapsed_sec": 64868.74085903168, "step_time_sec": 8.238127097021788, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.072717666625977, "lr": 0.00019935964086265984, "elapsed_sec": 64876.97921156883, "step_time_sec": 8.238215024000965, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.048421859741211, "lr": 0.00019935804670624508, "elapsed_sec": 64885.2187628746, "step_time_sec": 8.23938205599552, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0574212074279785, "lr": 0.00019935645057454616, "elapsed_sec": 64893.457738637924, "step_time_sec": 8.238792891032062, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.03955078125, "lr": 0.0001993548524675957, "elapsed_sec": 64901.69555473328, "step_time_sec": 8.237647541973274, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.059859752655029, "lr": 0.00019935325238542627, "elapsed_sec": 64909.9346241951, "step_time_sec": 8.23891732597258, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.093785285949707, "lr": 0.00019935165032807048, "elapsed_sec": 64918.17343091965, "step_time_sec": 8.238633583008777, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.245738506317139, "lr": 0.000199350046295561, "elapsed_sec": 64926.41206550598, "step_time_sec": 8.238522797997575, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.066725730895996, "lr": 0.0001993484402879306, "elapsed_sec": 64934.651185035706, "step_time_sec": 8.238900306983851, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.237240314483643, "lr": 0.00019934683230521196, "elapsed_sec": 64942.88818335533, "step_time_sec": 8.23684496601345, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.039348125457764, "lr": 0.00019934522234743793, "elapsed_sec": 64951.12770152092, "step_time_sec": 8.239380175014958, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0417799949646, "lr": 0.0001993436104146413, "elapsed_sec": 64959.365538835526, "step_time_sec": 8.23767812101869, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.133535861968994, "lr": 0.00019934199650685497, "elapsed_sec": 64967.60465359688, "step_time_sec": 8.238936901965644, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.040310382843018, "lr": 0.00019934038062411184, "elapsed_sec": 64975.84409117699, "step_time_sec": 8.239300476037897, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.232883453369141, "lr": 0.00019933876276644486, "elapsed_sec": 64984.08241271973, "step_time_sec": 8.238168304960709, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.038001537322998, "lr": 0.00019933714293388704, "elapsed_sec": 64992.32111310959, "step_time_sec": 8.238481872016564, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.099754333496094, "lr": 0.0001993355211264714, "elapsed_sec": 65000.55981349945, "step_time_sec": 8.238525983993895, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.972301959991455, "lr": 0.000199333897344231, "elapsed_sec": 65008.79884433746, "step_time_sec": 8.23892682499718, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.067893981933594, "lr": 0.000199332271587199, "elapsed_sec": 65017.03803706169, "step_time_sec": 8.239005866984371, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.122348308563232, "lr": 0.00019933064385540853, "elapsed_sec": 65025.275866508484, "step_time_sec": 8.237661062041298, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.114357948303223, "lr": 0.0001993290141488928, "elapsed_sec": 65033.5140080452, "step_time_sec": 8.23794815299334, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.07794713973999, "lr": 0.000199327382467685, "elapsed_sec": 65041.753336668015, "step_time_sec": 8.23916526202811, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.134124279022217, "lr": 0.00019932574881181844, "elapsed_sec": 65049.99257802963, "step_time_sec": 8.23911428899737, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.012925624847412, "lr": 0.00019932411318132644, "elapsed_sec": 65058.2320728302, "step_time_sec": 8.239326638053171, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9134504795074463, "lr": 0.00019932247557624235, "elapsed_sec": 65066.46933555603, "step_time_sec": 8.237117605982348, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.007635116577148, "lr": 0.00019932083599659957, "elapsed_sec": 65074.706856012344, "step_time_sec": 8.237351203977596, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9634182453155518, "lr": 0.00019931919444243152, "elapsed_sec": 65082.94426035881, "step_time_sec": 8.237205345008988, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1933112144470215, "lr": 0.0001993175509137717, "elapsed_sec": 65091.183495521545, "step_time_sec": 8.23908179096179, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.325395107269287, "lr": 0.00019931590541065364, "elapsed_sec": 65099.42293047905, "step_time_sec": 8.239266892021988, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2041850090026855, "lr": 0.00019931425793311082, "elapsed_sec": 65107.660792827606, "step_time_sec": 8.237705078034196, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.313807010650635, "lr": 0.00019931260848117692, "elapsed_sec": 65115.89987659454, "step_time_sec": 8.23891710001044, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9776153564453125, "lr": 0.00019931095705488558, "elapsed_sec": 65124.13760638237, "step_time_sec": 8.237580457003787, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.031420707702637, "lr": 0.0001993093036542704, "elapsed_sec": 65132.378702163696, "step_time_sec": 8.240874411014374, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8350236415863037, "lr": 0.00019930764827936522, "elapsed_sec": 65140.61758065224, "step_time_sec": 8.238692350045312, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9049997329711914, "lr": 0.00019930599093020366, "elapsed_sec": 65148.85517215729, "step_time_sec": 8.237463255994953, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.067839622497559, "lr": 0.00019930433160681961, "elapsed_sec": 65157.094760894775, "step_time_sec": 8.239403631014284, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9166154861450195, "lr": 0.00019930267030924688, "elapsed_sec": 65165.33178281784, "step_time_sec": 8.236862321966328, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.763345956802368, "lr": 0.00019930100703751934, "elapsed_sec": 65173.57161974907, "step_time_sec": 8.23966811603168, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9969143867492676, "lr": 0.00019929934179167095, "elapsed_sec": 65181.812144994736, "step_time_sec": 8.240330037020613, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.914391040802002, "lr": 0.00019929767457173566, "elapsed_sec": 65190.04910826683, "step_time_sec": 8.236815320968162, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.086727619171143, "lr": 0.00019929600537774744, "elapsed_sec": 65198.28831624985, "step_time_sec": 8.239061900007073, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.056152820587158, "lr": 0.00019929433420974035, "elapsed_sec": 65206.5266122818, "step_time_sec": 8.238079210976139, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9608917236328125, "lr": 0.00019929266106774845, "elapsed_sec": 65214.76626372337, "step_time_sec": 8.2395502589643, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8906500339508057, "lr": 0.0001992909859518059, "elapsed_sec": 65223.00665163994, "step_time_sec": 8.240176783991046, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.105356693267822, "lr": 0.00019928930886194685, "elapsed_sec": 65231.24424242973, "step_time_sec": 8.23740484699374, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.078134536743164, "lr": 0.00019928762979820548, "elapsed_sec": 65239.482714891434, "step_time_sec": 8.23829630902037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.229314804077148, "lr": 0.00019928594876061608, "elapsed_sec": 65247.72040295601, "step_time_sec": 8.237540389993228, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0569257736206055, "lr": 0.00019928426574921288, "elapsed_sec": 65255.95939230919, "step_time_sec": 8.23887195304269, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.009685039520264, "lr": 0.0001992825807640302, "elapsed_sec": 65264.19845199585, "step_time_sec": 8.238916101981886, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.113417148590088, "lr": 0.00019928089380510242, "elapsed_sec": 65272.435591220856, "step_time_sec": 8.23691586102359, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9658267498016357, "lr": 0.00019927920487246395, "elapsed_sec": 65280.67551088333, "step_time_sec": 8.239746648992877, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.32028341293335, "lr": 0.00019927751396614925, "elapsed_sec": 65288.91415286064, "step_time_sec": 8.238498403981794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0450663566589355, "lr": 0.00019927582108619278, "elapsed_sec": 65297.153512477875, "step_time_sec": 8.239174147020094, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.300934791564941, "lr": 0.00019927412623262908, "elapsed_sec": 65305.39239859581, "step_time_sec": 8.23876720899716, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8912060260772705, "lr": 0.0001992724294054927, "elapsed_sec": 65313.62939929962, "step_time_sec": 8.236783581029158, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9703125953674316, "lr": 0.00019927073060481824, "elapsed_sec": 65321.86940383911, "step_time_sec": 8.239823551964946, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8792929649353027, "lr": 0.00019926902983064034, "elapsed_sec": 65330.10744404793, "step_time_sec": 8.23801758303307, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.127682209014893, "lr": 0.0001992673270829937, "elapsed_sec": 65338.34783768654, "step_time_sec": 8.240114418964367, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.022460460662842, "lr": 0.00019926562236191307, "elapsed_sec": 65346.58768701553, "step_time_sec": 8.239720198034775, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.933616876602173, "lr": 0.00019926391566743316, "elapsed_sec": 65354.82532310486, "step_time_sec": 8.237400061974768, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.082702159881592, "lr": 0.00019926220699958882, "elapsed_sec": 65363.06355881691, "step_time_sec": 8.238097306981217, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9402124881744385, "lr": 0.00019926049635841486, "elapsed_sec": 65371.30194425583, "step_time_sec": 8.238217430014629, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.023147106170654, "lr": 0.0001992587837439462, "elapsed_sec": 65379.54123926163, "step_time_sec": 8.239134829025716, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.168360233306885, "lr": 0.00019925706915621775, "elapsed_sec": 65387.78065609932, "step_time_sec": 8.239304349990562, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9492604732513428, "lr": 0.00019925535259526447, "elapsed_sec": 65396.01828289032, "step_time_sec": 8.237413459981326, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0959343910217285, "lr": 0.00019925363406112135, "elapsed_sec": 65404.258509874344, "step_time_sec": 8.240072234999388, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.150475025177002, "lr": 0.0001992519135538235, "elapsed_sec": 65412.49814391136, "step_time_sec": 8.239431769005023, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.043162822723389, "lr": 0.00019925019107340595, "elapsed_sec": 65420.73668599129, "step_time_sec": 8.238372044987045, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9523110389709473, "lr": 0.00019924846661990386, "elapsed_sec": 65428.97560405731, "step_time_sec": 8.238771266012918, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.943387031555176, "lr": 0.0001992467401933524, "elapsed_sec": 65437.212928533554, "step_time_sec": 8.237114764982834, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.221593379974365, "lr": 0.00019924501179378673, "elapsed_sec": 65445.45205616951, "step_time_sec": 8.2389890680206, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.189465522766113, "lr": 0.00019924328142124215, "elapsed_sec": 65453.690012693405, "step_time_sec": 8.237776751047932, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.235128402709961, "lr": 0.00019924154907575392, "elapsed_sec": 65461.928520441055, "step_time_sec": 8.238326469960157, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.108043193817139, "lr": 0.0001992398147573574, "elapsed_sec": 65470.167709350586, "step_time_sec": 8.239031556004193, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.203024387359619, "lr": 0.0001992380784660879, "elapsed_sec": 65478.40500879288, "step_time_sec": 8.23715177597478, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9377005100250244, "lr": 0.0001992363402019809, "elapsed_sec": 65486.644553899765, "step_time_sec": 8.239401836995967, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9061708450317383, "lr": 0.00019923459996507177, "elapsed_sec": 65494.883051633835, "step_time_sec": 8.238287704996765, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9893648624420166, "lr": 0.0001992328577553961, "elapsed_sec": 65503.121935367584, "step_time_sec": 8.238698392000515, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.054037570953369, "lr": 0.00019923111357298933, "elapsed_sec": 65511.359914541245, "step_time_sec": 8.237842646019999, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.034422397613525, "lr": 0.0001992293674178871, "elapsed_sec": 65519.59845018387, "step_time_sec": 8.238356801040936, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9616310596466064, "lr": 0.00019922761929012497, "elapsed_sec": 65527.83753013611, "step_time_sec": 8.238897207949776, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.082688331604004, "lr": 0.00019922586918973858, "elapsed_sec": 65536.075268507, "step_time_sec": 8.23763033299474, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.02161169052124, "lr": 0.00019922411711676368, "elapsed_sec": 65544.3149857521, "step_time_sec": 8.239538897003513, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.117706298828125, "lr": 0.00019922236307123596, "elapsed_sec": 65552.55444550514, "step_time_sec": 8.23927363799885, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.076817989349365, "lr": 0.00019922060705319117, "elapsed_sec": 65560.79079985619, "step_time_sec": 8.236154942016583, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.086601257324219, "lr": 0.0001992188490626652, "elapsed_sec": 65569.02942156792, "step_time_sec": 8.238464521011338, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.921757698059082, "lr": 0.0001992170890996938, "elapsed_sec": 65577.26819014549, "step_time_sec": 8.238605696999002, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.021707057952881, "lr": 0.00019921532716431298, "elapsed_sec": 65585.506783247, "step_time_sec": 8.23842665401753, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.183347702026367, "lr": 0.00019921356325655856, "elapsed_sec": 65593.74565958977, "step_time_sec": 8.238708329969086, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9929404258728027, "lr": 0.00019921179737646659, "elapsed_sec": 65601.98307132721, "step_time_sec": 8.2372168879956, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.220789432525635, "lr": 0.00019921002952407303, "elapsed_sec": 65610.22161412239, "step_time_sec": 8.238327488012146, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.00883674621582, "lr": 0.00019920825969941397, "elapsed_sec": 65618.45882701874, "step_time_sec": 8.237073680036701, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.047542095184326, "lr": 0.00019920648790252546, "elapsed_sec": 65626.69929456711, "step_time_sec": 8.240302366029937, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9725899696350098, "lr": 0.00019920471413344368, "elapsed_sec": 65634.93669843674, "step_time_sec": 8.237263478979003, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.896693468093872, "lr": 0.00019920293839220478, "elapsed_sec": 65643.1752102375, "step_time_sec": 8.238310583983548, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.109404563903809, "lr": 0.00019920116067884502, "elapsed_sec": 65651.4136455059, "step_time_sec": 8.23827606899431, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.110105991363525, "lr": 0.00019919938099340057, "elapsed_sec": 65659.65107584, "step_time_sec": 8.237247109005693, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.081783771514893, "lr": 0.00019919759933590778, "elapsed_sec": 65667.8900001049, "step_time_sec": 8.238763932953589, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.155076026916504, "lr": 0.00019919581570640297, "elapsed_sec": 65676.12996792793, "step_time_sec": 8.239854978979565, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.325325965881348, "lr": 0.0001991940301049225, "elapsed_sec": 65684.36726593971, "step_time_sec": 8.237160776974633, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.060366630554199, "lr": 0.0001991922425315028, "elapsed_sec": 65692.60732316971, "step_time_sec": 8.239845051022712, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0857977867126465, "lr": 0.00019919045298618035, "elapsed_sec": 65700.84569311142, "step_time_sec": 8.2382023659884, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.189126014709473, "lr": 0.0001991886614689916, "elapsed_sec": 65709.08629155159, "step_time_sec": 8.240464340953622, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.110423564910889, "lr": 0.00019918686797997312, "elapsed_sec": 65717.32488894463, "step_time_sec": 8.238381989009213, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.224759578704834, "lr": 0.0001991850725191615, "elapsed_sec": 65725.56464624405, "step_time_sec": 8.23960975499358, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.135647296905518, "lr": 0.00019918327508659328, "elapsed_sec": 65733.80295276642, "step_time_sec": 8.238124419003725, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.15856409072876, "lr": 0.0001991814756823052, "elapsed_sec": 65742.04086828232, "step_time_sec": 8.237742799974512, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.163080215454102, "lr": 0.0001991796743063339, "elapsed_sec": 65750.28081274033, "step_time_sec": 8.239768980012741, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.130040645599365, "lr": 0.00019917787095871615, "elapsed_sec": 65758.520352602, "step_time_sec": 8.239381170016713, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.185481548309326, "lr": 0.00019917606563948868, "elapsed_sec": 65766.75719070435, "step_time_sec": 8.236740401014686, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1363630294799805, "lr": 0.00019917425834868838, "elapsed_sec": 65774.99640798569, "step_time_sec": 8.23900765599683, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.120447635650635, "lr": 0.00019917244908635207, "elapsed_sec": 65783.23403120041, "step_time_sec": 8.23745755595155, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.243378639221191, "lr": 0.00019917063785251663, "elapsed_sec": 65791.47435212135, "step_time_sec": 8.24015700101154, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.097774028778076, "lr": 0.00019916882464721903, "elapsed_sec": 65799.71421146393, "step_time_sec": 8.239685498992912, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.055596828460693, "lr": 0.00019916700947049618, "elapsed_sec": 65807.9517569542, "step_time_sec": 8.237358432961628, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.227053642272949, "lr": 0.00019916519232238518, "elapsed_sec": 65816.19169831276, "step_time_sec": 8.23980948497774, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.093580722808838, "lr": 0.00019916337320292308, "elapsed_sec": 65824.42975187302, "step_time_sec": 8.237868540047202, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.011078834533691, "lr": 0.00019916155211214696, "elapsed_sec": 65832.6693212986, "step_time_sec": 8.239368517999537, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.109439373016357, "lr": 0.00019915972905009392, "elapsed_sec": 65840.90853881836, "step_time_sec": 8.239049566967878, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.136724948883057, "lr": 0.0001991579040168012, "elapsed_sec": 65849.14512443542, "step_time_sec": 8.236438594001811, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.015157222747803, "lr": 0.00019915607701230596, "elapsed_sec": 65857.38475465775, "step_time_sec": 8.239414484996814, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.141786098480225, "lr": 0.0001991542480366455, "elapsed_sec": 65865.62265658379, "step_time_sec": 8.237745593010914, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.135159969329834, "lr": 0.00019915241708985714, "elapsed_sec": 65873.86167407036, "step_time_sec": 8.238855525967665, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.086280822753906, "lr": 0.00019915058417197814, "elapsed_sec": 65882.10056257248, "step_time_sec": 8.238701977999881, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.157222747802734, "lr": 0.00019914874928304597, "elapsed_sec": 65890.33863925934, "step_time_sec": 8.237939311016817, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.927227258682251, "lr": 0.000199146912423098, "elapsed_sec": 65898.57864689827, "step_time_sec": 8.239868501026649, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.119502067565918, "lr": 0.0001991450735921717, "elapsed_sec": 65906.81664299965, "step_time_sec": 8.23783568298677, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.225969314575195, "lr": 0.00019914323279030457, "elapsed_sec": 65915.0561466217, "step_time_sec": 8.239333886944223, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.3254828453063965, "lr": 0.00019914139001753415, "elapsed_sec": 65923.29581332207, "step_time_sec": 8.239464387006592, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1345391273498535, "lr": 0.00019913954527389803, "elapsed_sec": 65931.53420877457, "step_time_sec": 8.238254348980263, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.012420654296875, "lr": 0.00019913769855943382, "elapsed_sec": 65939.77493214607, "step_time_sec": 8.24052720097825, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.011897563934326, "lr": 0.00019913584987417917, "elapsed_sec": 65948.01376056671, "step_time_sec": 8.23869215301238, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.155936241149902, "lr": 0.0001991339992181718, "elapsed_sec": 65956.25277042389, "step_time_sec": 8.238811207003891, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.186992645263672, "lr": 0.00019913214659144946, "elapsed_sec": 65964.49191641808, "step_time_sec": 8.239046497968957, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.045071125030518, "lr": 0.0001991302919940499, "elapsed_sec": 65972.72927117348, "step_time_sec": 8.237127862987109, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.155885696411133, "lr": 0.00019912843542601097, "elapsed_sec": 65980.96970915794, "step_time_sec": 8.240321796038188, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.070905685424805, "lr": 0.00019912657688737055, "elapsed_sec": 65989.21035075188, "step_time_sec": 8.240443847957067, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.083942890167236, "lr": 0.00019912471637816646, "elapsed_sec": 65997.44910287857, "step_time_sec": 8.238589875982143, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2502241134643555, "lr": 0.00019912285389843675, "elapsed_sec": 66005.68936347961, "step_time_sec": 8.240066706028301, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9666261672973633, "lr": 0.00019912098944821933, "elapsed_sec": 66013.92766213417, "step_time_sec": 8.238133912032936, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.155055999755859, "lr": 0.00019911912302755225, "elapsed_sec": 66022.16769170761, "step_time_sec": 8.23985754000023, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.175886631011963, "lr": 0.00019911725463647354, "elapsed_sec": 66030.40700745583, "step_time_sec": 8.239145472995006, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.186875343322754, "lr": 0.00019911538427502134, "elapsed_sec": 66038.6458041668, "step_time_sec": 8.238635193032678, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.245680332183838, "lr": 0.00019911351194323377, "elapsed_sec": 66046.8857114315, "step_time_sec": 8.239732910995372, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.120739459991455, "lr": 0.00019911163764114905, "elapsed_sec": 66055.12481355667, "step_time_sec": 8.239006723975763, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.049898624420166, "lr": 0.0001991097613688054, "elapsed_sec": 66063.3652176857, "step_time_sec": 8.240150226978585, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.099194526672363, "lr": 0.00019910788312624102, "elapsed_sec": 66071.60568022728, "step_time_sec": 8.240326725004707, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.124925136566162, "lr": 0.00019910600291349426, "elapsed_sec": 66079.84458470345, "step_time_sec": 8.23871143499855, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.951512336730957, "lr": 0.0001991041207306035, "elapsed_sec": 66088.08467626572, "step_time_sec": 8.23994507198222, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.13266134262085, "lr": 0.00019910223657760705, "elapsed_sec": 66096.32427072525, "step_time_sec": 8.239405290980358, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0850653648376465, "lr": 0.00019910035045454338, "elapsed_sec": 66104.56275844574, "step_time_sec": 8.238360397983342, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0571417808532715, "lr": 0.00019909846236145098, "elapsed_sec": 66112.80202674866, "step_time_sec": 8.239066388981882, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.208737850189209, "lr": 0.00019909657229836826, "elapsed_sec": 66121.04114484787, "step_time_sec": 8.238988563010935, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.104941368103027, "lr": 0.00019909468026533387, "elapsed_sec": 66129.28108048439, "step_time_sec": 8.239736296003684, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1930413246154785, "lr": 0.00019909278626238633, "elapsed_sec": 66137.5207836628, "step_time_sec": 8.239517385023646, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.035356044769287, "lr": 0.0001990908902895643, "elapsed_sec": 66145.76007080078, "step_time_sec": 8.239140225981828, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.229515075683594, "lr": 0.00019908899234690643, "elapsed_sec": 66154.00148916245, "step_time_sec": 8.241262487019412, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4303107261657715, "lr": 0.00019908709243445142, "elapsed_sec": 66162.24121880531, "step_time_sec": 8.239558054017834, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.261460781097412, "lr": 0.00019908519055223805, "elapsed_sec": 66170.47963476181, "step_time_sec": 8.238240938982926, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.252439022064209, "lr": 0.00019908328670030506, "elapsed_sec": 66178.71914172173, "step_time_sec": 8.239350500050932, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.262598514556885, "lr": 0.0001990813808786913, "elapsed_sec": 66186.95715928078, "step_time_sec": 8.237865899980534, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.219221115112305, "lr": 0.00019907947308743564, "elapsed_sec": 66195.197327137, "step_time_sec": 8.24002730898792, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.831897020339966, "lr": 0.00019907756332657696, "elapsed_sec": 66203.4355711937, "step_time_sec": 8.238049742998555, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.268676280975342, "lr": 0.00019907565159615426, "elapsed_sec": 66211.67477440834, "step_time_sec": 8.239066158013884, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.095577239990234, "lr": 0.00019907373789620648, "elapsed_sec": 66219.91415333748, "step_time_sec": 8.23916360095609, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0538835525512695, "lr": 0.00019907182222677266, "elapsed_sec": 66228.15179467201, "step_time_sec": 8.237528686993755, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.004269599914551, "lr": 0.00019906990458789188, "elapsed_sec": 66236.3919968605, "step_time_sec": 8.24006536102388, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.052064895629883, "lr": 0.0001990679849796032, "elapsed_sec": 66244.63280701637, "step_time_sec": 8.240562302002218, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.144400596618652, "lr": 0.00019906606340194585, "elapsed_sec": 66252.86985206604, "step_time_sec": 8.236886642989703, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.23734188079834, "lr": 0.00019906413985495898, "elapsed_sec": 66261.10966706276, "step_time_sec": 8.23966369300615, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.161663055419922, "lr": 0.00019906221433868177, "elapsed_sec": 66269.34877085686, "step_time_sec": 8.238964365969878, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.147761821746826, "lr": 0.00019906028685315357, "elapsed_sec": 66277.58781504631, "step_time_sec": 8.238836876000278, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.046501636505127, "lr": 0.0001990583573984136, "elapsed_sec": 66285.82622790337, "step_time_sec": 8.238260722020641, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.087961673736572, "lr": 0.00019905642597450128, "elapsed_sec": 66294.06366062164, "step_time_sec": 8.237324683985207, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.168543815612793, "lr": 0.00019905449258145598, "elapsed_sec": 66302.30424952507, "step_time_sec": 8.240372032974847, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.126809120178223, "lr": 0.00019905255721931712, "elapsed_sec": 66310.54192137718, "step_time_sec": 8.237510106002446, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.162965774536133, "lr": 0.00019905061988812414, "elapsed_sec": 66318.7809536457, "step_time_sec": 8.238890833978076, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.090425968170166, "lr": 0.00019904868058791663, "elapsed_sec": 66327.02096486092, "step_time_sec": 8.239871218975168, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.122734069824219, "lr": 0.00019904673931873406, "elapsed_sec": 66335.25886702538, "step_time_sec": 8.237743425008375, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.288582801818848, "lr": 0.0001990447960806161, "elapsed_sec": 66343.49804592133, "step_time_sec": 8.238990746962372, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.119004726409912, "lr": 0.00019904285087360226, "elapsed_sec": 66351.73743343353, "step_time_sec": 8.2392831860343, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.149263381958008, "lr": 0.0001990409036977323, "elapsed_sec": 66359.97741985321, "step_time_sec": 8.239775003981777, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1718974113464355, "lr": 0.00019903895455304594, "elapsed_sec": 66368.21691536903, "step_time_sec": 8.239345666021109, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.113807201385498, "lr": 0.0001990370034395829, "elapsed_sec": 66376.45389175415, "step_time_sec": 8.236814099014737, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.986405849456787, "lr": 0.00019903505035738292, "elapsed_sec": 66384.69351530075, "step_time_sec": 8.239438418997452, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.155343532562256, "lr": 0.00019903309530648592, "elapsed_sec": 66392.9336013794, "step_time_sec": 8.239935312012676, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.006659507751465, "lr": 0.00019903113828693172, "elapsed_sec": 66401.17206788063, "step_time_sec": 8.238314749964047, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.089522361755371, "lr": 0.00019902917929876023, "elapsed_sec": 66409.41142106056, "step_time_sec": 8.239187219005544, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.101324081420898, "lr": 0.00019902721834201144, "elapsed_sec": 66417.64921140671, "step_time_sec": 8.237639667990152, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.124236583709717, "lr": 0.00019902525541672529, "elapsed_sec": 66425.88766288757, "step_time_sec": 8.238236283010338, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0989227294921875, "lr": 0.00019902329052294183, "elapsed_sec": 66434.1268351078, "step_time_sec": 8.239081509993412, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.047393798828125, "lr": 0.00019902132366070115, "elapsed_sec": 66442.36533355713, "step_time_sec": 8.238268750952557, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.153347969055176, "lr": 0.00019901935483004337, "elapsed_sec": 66450.6057741642, "step_time_sec": 8.240269089001231, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.077648162841797, "lr": 0.00019901738403100858, "elapsed_sec": 66458.84337854385, "step_time_sec": 8.237500319024548, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.073858737945557, "lr": 0.00019901541126363703, "elapsed_sec": 66467.08479571342, "step_time_sec": 8.241226550017018, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.020814418792725, "lr": 0.0001990134365279689, "elapsed_sec": 66475.32524776459, "step_time_sec": 8.24030236096587, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.063669204711914, "lr": 0.00019901145982404453, "elapsed_sec": 66483.56328749657, "step_time_sec": 8.23783778201323, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.102062225341797, "lr": 0.00019900948115190417, "elapsed_sec": 66491.80310583115, "step_time_sec": 52.3720329470234, "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": 1.0701706149848178, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 8001, "loss": 4.056682586669922, "lr": 0.00019900750051158822, "elapsed_sec": 66544.18442511559, "step_time_sec": 8.248907758039422, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.192892551422119, "lr": 0.00019900551790313705, "elapsed_sec": 66552.42295479774, "step_time_sec": 8.238312870031223, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.18613862991333, "lr": 0.0001990035333265911, "elapsed_sec": 66560.66231679916, "step_time_sec": 8.23922060901532, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1832499504089355, "lr": 0.0001990015467819908, "elapsed_sec": 66568.90103411674, "step_time_sec": 8.238534344010986, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1813530921936035, "lr": 0.00019899955826937674, "elapsed_sec": 66577.14206337929, "step_time_sec": 8.24087264901027, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.156681537628174, "lr": 0.00019899756778878942, "elapsed_sec": 66585.38125801086, "step_time_sec": 8.238989809993654, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.175780773162842, "lr": 0.00019899557534026942, "elapsed_sec": 66593.61887168884, "step_time_sec": 8.237438795971684, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.295270919799805, "lr": 0.00019899358092385742, "elapsed_sec": 66601.8582854271, "step_time_sec": 8.23926466802368, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.180217742919922, "lr": 0.00019899158453959406, "elapsed_sec": 66610.09720158577, "step_time_sec": 8.238705390016548, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.12326192855835, "lr": 0.00019898958618752005, "elapsed_sec": 66618.33592319489, "step_time_sec": 8.238610039988998, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.142792224884033, "lr": 0.0001989875858676762, "elapsed_sec": 66626.57448220253, "step_time_sec": 8.238392898987513, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1724724769592285, "lr": 0.00019898558358010326, "elapsed_sec": 66634.81254458427, "step_time_sec": 8.237852026009932, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.142791271209717, "lr": 0.00019898357932484205, "elapsed_sec": 66643.05323934555, "step_time_sec": 8.2405990610132, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0866289138793945, "lr": 0.00019898157310193343, "elapsed_sec": 66651.29195451736, "step_time_sec": 8.238490920048207, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.026249885559082, "lr": 0.00019897956491141838, "elapsed_sec": 66659.53198027611, "step_time_sec": 8.239895528007764, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.99882173538208, "lr": 0.0001989775547533378, "elapsed_sec": 66667.77159452438, "step_time_sec": 8.239427770022303, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.075664520263672, "lr": 0.0001989755426277327, "elapsed_sec": 66676.01059508324, "step_time_sec": 8.23886136704823, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.020655155181885, "lr": 0.00019897352853464413, "elapsed_sec": 66684.25124692917, "step_time_sec": 8.240508033020888, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.105766296386719, "lr": 0.00019897151247411316, "elapsed_sec": 66692.4905538559, "step_time_sec": 8.239081582985818, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.189101696014404, "lr": 0.0001989694944461809, "elapsed_sec": 66700.72787976265, "step_time_sec": 8.23717483796645, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0871758460998535, "lr": 0.00019896747445088847, "elapsed_sec": 66708.96786856651, "step_time_sec": 8.239863666007295, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.006349563598633, "lr": 0.00019896545248827713, "elapsed_sec": 66717.20945692062, "step_time_sec": 8.241480519995093, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.045265197753906, "lr": 0.00019896342855838805, "elapsed_sec": 66725.4489634037, "step_time_sec": 8.239287200965919, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.927088499069214, "lr": 0.00019896140266126256, "elapsed_sec": 66733.6891553402, "step_time_sec": 8.24004549102392, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.16167688369751, "lr": 0.00019895937479694193, "elapsed_sec": 66741.92749595642, "step_time_sec": 8.238183940004092, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.045387268066406, "lr": 0.00019895734496546755, "elapsed_sec": 66750.16820812225, "step_time_sec": 8.240448807016946, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.15035343170166, "lr": 0.0001989553131668808, "elapsed_sec": 66758.40950131416, "step_time_sec": 8.24111287103733, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.253774642944336, "lr": 0.0001989532794012231, "elapsed_sec": 66766.64693593979, "step_time_sec": 8.23731050704373, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.214136123657227, "lr": 0.000198951243668536, "elapsed_sec": 66774.88716244698, "step_time_sec": 8.240036825998686, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9534060955047607, "lr": 0.0001989492059688609, "elapsed_sec": 66783.12541484833, "step_time_sec": 8.238094968022779, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.033912658691406, "lr": 0.00019894716630223947, "elapsed_sec": 66791.36533284187, "step_time_sec": 8.23976229300024, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.095094203948975, "lr": 0.00019894512466871322, "elapsed_sec": 66799.60576152802, "step_time_sec": 8.240281172038522, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.014812469482422, "lr": 0.00019894308106832383, "elapsed_sec": 66807.8433496952, "step_time_sec": 8.237407125998288, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.05134916305542, "lr": 0.00019894103550111293, "elapsed_sec": 66816.08314204216, "step_time_sec": 8.239613488025498, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.081177711486816, "lr": 0.0001989389879671223, "elapsed_sec": 66824.32263660431, "step_time_sec": 8.239371486008167, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.142907619476318, "lr": 0.0001989369384663937, "elapsed_sec": 66832.56109070778, "step_time_sec": 8.23820295301266, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.019073486328125, "lr": 0.00019893488699896885, "elapsed_sec": 66840.80062055588, "step_time_sec": 8.239373668970075, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.109544277191162, "lr": 0.00019893283356488964, "elapsed_sec": 66849.03897166252, "step_time_sec": 8.238225277978927, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.035060405731201, "lr": 0.00019893077816419795, "elapsed_sec": 66857.27969431877, "step_time_sec": 8.240596113959327, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.249361515045166, "lr": 0.0001989287207969357, "elapsed_sec": 66865.52007818222, "step_time_sec": 8.240174168953672, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.213990211486816, "lr": 0.0001989266614631448, "elapsed_sec": 66873.75807595253, "step_time_sec": 8.237842126982287, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1855292320251465, "lr": 0.00019892460016286727, "elapsed_sec": 66881.99791193008, "step_time_sec": 8.239682318002451, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.982708215713501, "lr": 0.00019892253689614517, "elapsed_sec": 66890.23521995544, "step_time_sec": 8.23713492898969, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.114240646362305, "lr": 0.00019892047166302057, "elapsed_sec": 66898.47587752342, "step_time_sec": 8.24046471400652, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.147469997406006, "lr": 0.00019891840446353558, "elapsed_sec": 66906.71572899818, "step_time_sec": 8.239758523006458, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9803929328918457, "lr": 0.0001989163352977324, "elapsed_sec": 66914.95365786552, "step_time_sec": 8.23773103201529, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.078464508056641, "lr": 0.00019891426416565312, "elapsed_sec": 66923.19464612007, "step_time_sec": 8.240844632964581, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.144571781158447, "lr": 0.00019891219106734007, "elapsed_sec": 66931.43484973907, "step_time_sec": 8.240010962006636, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.312936782836914, "lr": 0.00019891011600283552, "elapsed_sec": 66939.67435574532, "step_time_sec": 8.239367877016775, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.121036529541016, "lr": 0.00019890803897218176, "elapsed_sec": 66947.9135210514, "step_time_sec": 8.239006649004295, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.981231451034546, "lr": 0.00019890595997542116, "elapsed_sec": 66956.15224194527, "step_time_sec": 8.23859911202453, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9819180965423584, "lr": 0.0001989038790125961, "elapsed_sec": 66964.39258503914, "step_time_sec": 8.240189325995743, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.024613857269287, "lr": 0.00019890179608374905, "elapsed_sec": 66972.63129615784, "step_time_sec": 8.238458344014362, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0336174964904785, "lr": 0.0001988997111889225, "elapsed_sec": 66980.8698644638, "step_time_sec": 8.238396384986117, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0807695388793945, "lr": 0.0001988976243281589, "elapsed_sec": 66989.11013388634, "step_time_sec": 8.240150451019872, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.03574800491333, "lr": 0.00019889553550150088, "elapsed_sec": 66997.34844946861, "step_time_sec": 8.238148030010052, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9333088397979736, "lr": 0.000198893444708991, "elapsed_sec": 67005.5881857872, "step_time_sec": 8.239535362983588, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8404581546783447, "lr": 0.00019889135195067192, "elapsed_sec": 67013.8295853138, "step_time_sec": 8.241214847948868, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9289324283599854, "lr": 0.00019888925722658628, "elapsed_sec": 67022.06857538223, "step_time_sec": 8.238862817001063, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9978106021881104, "lr": 0.00019888716053677687, "elapsed_sec": 67030.31006717682, "step_time_sec": 8.241353605000768, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.920567750930786, "lr": 0.00019888506188128635, "elapsed_sec": 67038.55048799515, "step_time_sec": 8.240235057019163, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.044269561767578, "lr": 0.00019888296126015762, "elapsed_sec": 67046.78788256645, "step_time_sec": 8.237194134970196, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.973794937133789, "lr": 0.00019888085867343346, "elapsed_sec": 67055.02803516388, "step_time_sec": 8.23997293697903, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.944727897644043, "lr": 0.00019887875412115676, "elapsed_sec": 67063.26686429977, "step_time_sec": 8.238678574038204, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9947962760925293, "lr": 0.00019887664760337046, "elapsed_sec": 67071.50762104988, "step_time_sec": 8.240586869011167, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9441092014312744, "lr": 0.00019887453912011748, "elapsed_sec": 67079.74763560295, "step_time_sec": 8.239866394957062, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.092377662658691, "lr": 0.00019887242867144085, "elapsed_sec": 67087.98444390297, "step_time_sec": 8.236676938016899, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.055724143981934, "lr": 0.00019887031625738362, "elapsed_sec": 67096.22458982468, "step_time_sec": 8.239966761961114, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.037166595458984, "lr": 0.00019886820187798882, "elapsed_sec": 67104.46448683739, "step_time_sec": 8.239777927985415, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.060751438140869, "lr": 0.00019886608553329962, "elapsed_sec": 67112.70331025124, "step_time_sec": 8.238597869989462, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.118757247924805, "lr": 0.00019886396722335913, "elapsed_sec": 67120.942111969, "step_time_sec": 8.238688774988987, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8684420585632324, "lr": 0.00019886184694821064, "elapsed_sec": 67129.17976880074, "step_time_sec": 8.237492583983112, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9350738525390625, "lr": 0.00019885972470789728, "elapsed_sec": 67137.4187514782, "step_time_sec": 8.238810387032572, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.267295837402344, "lr": 0.0001988576005024624, "elapsed_sec": 67145.6561806202, "step_time_sec": 8.237261781003326, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9713993072509766, "lr": 0.0001988554743319493, "elapsed_sec": 67153.89482355118, "step_time_sec": 8.238446814008057, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9753966331481934, "lr": 0.00019885334619640133, "elapsed_sec": 67162.13467383385, "step_time_sec": 8.239735634007957, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9468464851379395, "lr": 0.00019885121609586193, "elapsed_sec": 67170.37304019928, "step_time_sec": 8.238198524981271, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.072659969329834, "lr": 0.00019884908403037448, "elapsed_sec": 67178.61170315742, "step_time_sec": 8.238470035954379, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.090907096862793, "lr": 0.00019884694999998254, "elapsed_sec": 67186.85098838806, "step_time_sec": 8.23914111603517, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.175449371337891, "lr": 0.00019884481400472953, "elapsed_sec": 67195.08999609947, "step_time_sec": 8.238832275033928, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.077579498291016, "lr": 0.00019884267604465907, "elapsed_sec": 67203.3295109272, "step_time_sec": 8.239331394026522, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.006518363952637, "lr": 0.00019884053611981476, "elapsed_sec": 67211.56728935242, "step_time_sec": 8.237633131968323, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.026620388031006, "lr": 0.00019883839423024026, "elapsed_sec": 67219.80863904953, "step_time_sec": 8.241206675011199, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9120371341705322, "lr": 0.00019883625037597923, "elapsed_sec": 67228.048163414, "step_time_sec": 8.23937865201151, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.223368167877197, "lr": 0.00019883410455707535, "elapsed_sec": 67236.28720140457, "step_time_sec": 8.23881050903583, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.086198806762695, "lr": 0.00019883195677357246, "elapsed_sec": 67244.53569698334, "step_time_sec": 8.240981354028918, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.049865245819092, "lr": 0.0001988298070255143, "elapsed_sec": 67252.77635264397, "step_time_sec": 8.240458295040298, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.167847156524658, "lr": 0.00019882765531294472, "elapsed_sec": 67261.01503038406, "step_time_sec": 8.238536618999206, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.133911609649658, "lr": 0.0001988255016359076, "elapsed_sec": 67269.25498485565, "step_time_sec": 8.239783905970398, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9769997596740723, "lr": 0.0001988233459944469, "elapsed_sec": 67277.49391579628, "step_time_sec": 8.238799195038155, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.977856159210205, "lr": 0.00019882118838860654, "elapsed_sec": 67285.73322606087, "step_time_sec": 8.239100534992758, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.072641849517822, "lr": 0.00019881902881843053, "elapsed_sec": 67293.97191429138, "step_time_sec": 8.238528167945333, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.155237674713135, "lr": 0.00019881686728396295, "elapsed_sec": 67302.21077299118, "step_time_sec": 8.238714361039456, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.986772298812866, "lr": 0.0001988147037852478, "elapsed_sec": 67310.45059108734, "step_time_sec": 8.239620852982625, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.10488224029541, "lr": 0.0001988125383223293, "elapsed_sec": 67318.69129490852, "step_time_sec": 8.240550076006912, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.070102214813232, "lr": 0.00019881037089525153, "elapsed_sec": 67326.92999958992, "step_time_sec": 8.238560562953353, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7690203189849854, "lr": 0.00019880820150405874, "elapsed_sec": 67335.16963839531, "step_time_sec": 8.239513509033713, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.201648712158203, "lr": 0.00019880603014879513, "elapsed_sec": 67343.40741753578, "step_time_sec": 8.237560470995959, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.011394500732422, "lr": 0.000198803856829505, "elapsed_sec": 67351.64871072769, "step_time_sec": 8.24116617301479, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.109903335571289, "lr": 0.00019880168154623272, "elapsed_sec": 67359.88708686829, "step_time_sec": 8.238159945001826, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.018343448638916, "lr": 0.00019879950429902262, "elapsed_sec": 67368.12508153915, "step_time_sec": 8.237869047967251, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9679276943206787, "lr": 0.00019879732508791903, "elapsed_sec": 67376.36496376991, "step_time_sec": 8.239732080022804, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.966372489929199, "lr": 0.0001987951439129665, "elapsed_sec": 67384.6045100689, "step_time_sec": 8.239379499049392, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9529716968536377, "lr": 0.00019879296077420945, "elapsed_sec": 67392.84355926514, "step_time_sec": 8.238847755012102, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.219984531402588, "lr": 0.00019879077567169242, "elapsed_sec": 67401.08466815948, "step_time_sec": 8.240932978980709, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.996079683303833, "lr": 0.00019878858860545995, "elapsed_sec": 67409.32243394852, "step_time_sec": 8.237613869947381, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9476752281188965, "lr": 0.00019878639957555667, "elapsed_sec": 67417.56159615517, "step_time_sec": 8.239016206003726, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.066385269165039, "lr": 0.00019878420858202725, "elapsed_sec": 67425.80137491226, "step_time_sec": 8.239608103001956, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.041362762451172, "lr": 0.0001987820156249163, "elapsed_sec": 67434.03980660439, "step_time_sec": 8.238224501023069, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.066946983337402, "lr": 0.0001987798207042686, "elapsed_sec": 67442.2805275917, "step_time_sec": 8.240586358006112, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.034797191619873, "lr": 0.00019877762382012885, "elapsed_sec": 67461.99999380112, "step_time_sec": 19.719275947019923, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0133891105651855, "lr": 0.00019877542497254196, "elapsed_sec": 67470.2382555008, "step_time_sec": 8.238106578006409, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.973844528198242, "lr": 0.00019877322416155267, "elapsed_sec": 67478.47770404816, "step_time_sec": 8.23925251699984, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.122371196746826, "lr": 0.0001987710213872059, "elapsed_sec": 67486.71516108513, "step_time_sec": 8.237287044001278, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.051402568817139, "lr": 0.00019876881664954656, "elapsed_sec": 67494.95392465591, "step_time_sec": 8.238570733985398, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9449639320373535, "lr": 0.00019876660994861962, "elapsed_sec": 67503.1916334629, "step_time_sec": 8.237578216998372, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.077117443084717, "lr": 0.00019876440128447013, "elapsed_sec": 67511.43060326576, "step_time_sec": 8.238803046057, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.015913486480713, "lr": 0.00019876219065714305, "elapsed_sec": 67519.66937208176, "step_time_sec": 8.238619134994224, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8522231578826904, "lr": 0.0001987599780666835, "elapsed_sec": 67527.90673279762, "step_time_sec": 8.237214960972778, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.061702728271484, "lr": 0.00019875776351313662, "elapsed_sec": 67536.14590907097, "step_time_sec": 8.238997489970643, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.208593845367432, "lr": 0.00019875554699654756, "elapsed_sec": 67544.38487672806, "step_time_sec": 8.238841757993214, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.00676965713501, "lr": 0.00019875332851696152, "elapsed_sec": 67552.62526082993, "step_time_sec": 8.240227442991454, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.022021770477295, "lr": 0.00019875110807442373, "elapsed_sec": 67560.86474227905, "step_time_sec": 8.23925937502645, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.123520374298096, "lr": 0.00019874888566897946, "elapsed_sec": 67569.10323643684, "step_time_sec": 8.238366108969785, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9566731452941895, "lr": 0.0001987466613006741, "elapsed_sec": 67577.346596241, "step_time_sec": 8.243172597954981, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.037121295928955, "lr": 0.00019874443496955294, "elapsed_sec": 67585.58679986, "step_time_sec": 8.240010314970277, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0270867347717285, "lr": 0.00019874220667566146, "elapsed_sec": 67593.82554101944, "step_time_sec": 8.238576044037472, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.018221378326416, "lr": 0.000198739976419045, "elapsed_sec": 67602.06445622444, "step_time_sec": 8.238744851958472, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.015388011932373, "lr": 0.00019873774419974915, "elapsed_sec": 67610.30329322815, "step_time_sec": 8.238686081953347, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.184455394744873, "lr": 0.00019873551001781936, "elapsed_sec": 67618.54409384727, "step_time_sec": 8.240617130009923, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9809799194335938, "lr": 0.00019873327387330122, "elapsed_sec": 67626.78459739685, "step_time_sec": 8.240355286980048, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.02618408203125, "lr": 0.00019873103576624032, "elapsed_sec": 67635.02317810059, "step_time_sec": 8.238401374022942, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.926539897918701, "lr": 0.0001987287956966823, "elapsed_sec": 67643.26309275627, "step_time_sec": 8.239821780007333, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.083510875701904, "lr": 0.00019872655366467287, "elapsed_sec": 67651.50347232819, "step_time_sec": 8.24011100397911, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.998924493789673, "lr": 0.00019872430967025776, "elapsed_sec": 67659.74245095253, "step_time_sec": 8.238812310039066, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.933992385864258, "lr": 0.00019872206371348268, "elapsed_sec": 67667.98221635818, "step_time_sec": 8.239656160993036, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.071163654327393, "lr": 0.00019871981579439347, "elapsed_sec": 67676.2207608223, "step_time_sec": 8.23836667399155, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.919327735900879, "lr": 0.00019871756591303597, "elapsed_sec": 67684.46120643616, "step_time_sec": 8.24028684705263, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.02769660949707, "lr": 0.00019871531406945607, "elapsed_sec": 67692.69942903519, "step_time_sec": 8.238040835014544, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.107384204864502, "lr": 0.00019871306026369968, "elapsed_sec": 67700.93795609474, "step_time_sec": 8.238306270039175, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.14330530166626, "lr": 0.00019871080449581277, "elapsed_sec": 67709.17744231224, "step_time_sec": 8.239343627996277, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.099793910980225, "lr": 0.0001987085467658413, "elapsed_sec": 67717.41632127762, "step_time_sec": 8.238700061978307, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0453009605407715, "lr": 0.0001987062870738314, "elapsed_sec": 67725.65660095215, "step_time_sec": 8.240116889006458, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8258578777313232, "lr": 0.00019870402541982907, "elapsed_sec": 67733.89772367477, "step_time_sec": 8.24101671599783, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9963467121124268, "lr": 0.0001987017618038805, "elapsed_sec": 67742.14576554298, "step_time_sec": 8.247878241993021, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.137617588043213, "lr": 0.00019869949622603183, "elapsed_sec": 67750.39306783676, "step_time_sec": 8.247122544038575, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9441521167755127, "lr": 0.00019869722868632924, "elapsed_sec": 67758.6335105896, "step_time_sec": 8.240230468974914, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.097433567047119, "lr": 0.000198694959184819, "elapsed_sec": 67766.87302565575, "step_time_sec": 8.239292951009702, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.137248992919922, "lr": 0.00019869268772154734, "elapsed_sec": 67775.11258745193, "step_time_sec": 8.239435580966529, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.779921293258667, "lr": 0.0001986904142965607, "elapsed_sec": 67783.35194730759, "step_time_sec": 8.239186522958335, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.160998821258545, "lr": 0.0001986881389099053, "elapsed_sec": 67791.58958649635, "step_time_sec": 8.237436460971367, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.984435558319092, "lr": 0.00019868586156162764, "elapsed_sec": 67799.8307132721, "step_time_sec": 8.240966468991246, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9902868270874023, "lr": 0.00019868358225177415, "elapsed_sec": 67808.07065153122, "step_time_sec": 8.23982147400966, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.03211784362793, "lr": 0.00019868130098039128, "elapsed_sec": 67816.30894732475, "step_time_sec": 8.238091261999216, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.116573810577393, "lr": 0.0001986790177475256, "elapsed_sec": 67824.54809331894, "step_time_sec": 8.239010966964997, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.823762893676758, "lr": 0.0001986767325532236, "elapsed_sec": 67832.78644013405, "step_time_sec": 8.238151855999604, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.104883670806885, "lr": 0.00019867444539753198, "elapsed_sec": 67841.02596759796, "step_time_sec": 8.239313385973219, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.016132354736328, "lr": 0.0001986721562804973, "elapsed_sec": 67849.2666079998, "step_time_sec": 8.24052800796926, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.832104206085205, "lr": 0.0001986698652021663, "elapsed_sec": 67857.50526714325, "step_time_sec": 8.238464972004294, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8185975551605225, "lr": 0.00019866757216258565, "elapsed_sec": 67865.74525427818, "step_time_sec": 8.23981390701374, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0063676834106445, "lr": 0.00019866527716180219, "elapsed_sec": 67873.98496341705, "step_time_sec": 8.2395393900224, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7874696254730225, "lr": 0.00019866298019986263, "elapsed_sec": 67882.22378110886, "step_time_sec": 8.238604963989928, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9841277599334717, "lr": 0.00019866068127681388, "elapsed_sec": 67890.46403574944, "step_time_sec": 8.240108280966524, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9461240768432617, "lr": 0.0001986583803927028, "elapsed_sec": 67898.70258188248, "step_time_sec": 8.238366223988123, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.15695858001709, "lr": 0.00019865607754757631, "elapsed_sec": 67906.94305062294, "step_time_sec": 8.24033340398455, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1577301025390625, "lr": 0.0001986537727414814, "elapsed_sec": 67915.18280434608, "step_time_sec": 8.239599982043728, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0156707763671875, "lr": 0.00019865146597446503, "elapsed_sec": 67923.42177700996, "step_time_sec": 8.238740290980786, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7575180530548096, "lr": 0.00019864915724657426, "elapsed_sec": 67931.6619796753, "step_time_sec": 8.24006956804078, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.946812152862549, "lr": 0.0001986468465578562, "elapsed_sec": 67939.9000787735, "step_time_sec": 8.237914723984431, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.035053730010986, "lr": 0.00019864453390835792, "elapsed_sec": 67948.13998508453, "step_time_sec": 8.239784069010057, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.07174825668335, "lr": 0.00019864221929812665, "elapsed_sec": 67956.37966799736, "step_time_sec": 8.239501480013132, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8791375160217285, "lr": 0.00019863990272720957, "elapsed_sec": 67964.61705684662, "step_time_sec": 8.23724885500269, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.113656997680664, "lr": 0.0001986375841956539, "elapsed_sec": 67972.85916256905, "step_time_sec": 8.241894572973251, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.062368392944336, "lr": 0.0001986352637035069, "elapsed_sec": 67981.09815502167, "step_time_sec": 8.238803357991856, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.064084529876709, "lr": 0.00019863294125081597, "elapsed_sec": 67989.33854293823, "step_time_sec": 8.240227022964973, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.000715255737305, "lr": 0.00019863061683762843, "elapsed_sec": 67997.57827877998, "step_time_sec": 8.239570160978474, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.932481527328491, "lr": 0.00019862829046399165, "elapsed_sec": 68005.8175137043, "step_time_sec": 8.239059227984399, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.023656368255615, "lr": 0.00019862596212995314, "elapsed_sec": 68014.0565237999, "step_time_sec": 8.238815758028068, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.21256160736084, "lr": 0.00019862363183556034, "elapsed_sec": 68022.29479146004, "step_time_sec": 8.238107471959665, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.148900032043457, "lr": 0.0001986212995808608, "elapsed_sec": 68030.53255581856, "step_time_sec": 8.237640391977038, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.011319637298584, "lr": 0.00019861896536590203, "elapsed_sec": 68038.77272677422, "step_time_sec": 8.240083906974178, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9785215854644775, "lr": 0.0001986166291907317, "elapsed_sec": 68047.0108332634, "step_time_sec": 8.237841727037448, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.03267240524292, "lr": 0.0001986142910553974, "elapsed_sec": 68055.2497754097, "step_time_sec": 8.238790874020196, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.162441730499268, "lr": 0.00019861195095994687, "elapsed_sec": 68063.49018549919, "step_time_sec": 8.24024932301836, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.09072208404541, "lr": 0.00019860960890442778, "elapsed_sec": 68071.7279305458, "step_time_sec": 8.237602510023862, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8663196563720703, "lr": 0.00019860726488888793, "elapsed_sec": 68079.96832990646, "step_time_sec": 8.240216391975991, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.984553813934326, "lr": 0.0001986049189133751, "elapsed_sec": 68088.20709443092, "step_time_sec": 8.238650726969354, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9360296726226807, "lr": 0.00019860257097793714, "elapsed_sec": 68096.44648885727, "step_time_sec": 8.239209171035327, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9853384494781494, "lr": 0.00019860022108262188, "elapsed_sec": 68104.6861038208, "step_time_sec": 8.239453558984678, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8590245246887207, "lr": 0.00019859786922747736, "elapsed_sec": 68112.92323637009, "step_time_sec": 8.236924850032665, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.912370204925537, "lr": 0.00019859551541255143, "elapsed_sec": 68121.16371393204, "step_time_sec": 8.240303622966167, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.025242328643799, "lr": 0.00019859315963789218, "elapsed_sec": 68129.40408086777, "step_time_sec": 8.2401835169876, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9082741737365723, "lr": 0.0001985908019035476, "elapsed_sec": 68137.64247703552, "step_time_sec": 8.238246840017382, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.008313179016113, "lr": 0.00019858844220956576, "elapsed_sec": 68145.88255810738, "step_time_sec": 8.239886787952855, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.039829254150391, "lr": 0.00019858608055599484, "elapsed_sec": 68154.12029314041, "step_time_sec": 8.237632392032538, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.00634765625, "lr": 0.00019858371694288294, "elapsed_sec": 68162.36092686653, "step_time_sec": 8.24038232897874, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9970972537994385, "lr": 0.0001985813513702783, "elapsed_sec": 68170.60141777992, "step_time_sec": 8.240331044013146, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.889460325241089, "lr": 0.00019857898383822918, "elapsed_sec": 68178.8391160965, "step_time_sec": 8.23756636295002, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.027390003204346, "lr": 0.00019857661434678384, "elapsed_sec": 68187.07872629166, "step_time_sec": 8.239430293033365, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9588866233825684, "lr": 0.00019857424289599057, "elapsed_sec": 68195.31666874886, "step_time_sec": 8.237744276004378, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.007119178771973, "lr": 0.00019857186948589778, "elapsed_sec": 68203.55588841438, "step_time_sec": 8.23902337695472, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.03296422958374, "lr": 0.00019856949411655383, "elapsed_sec": 68211.79478025436, "step_time_sec": 8.238762384979054, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.06384801864624, "lr": 0.00019856711678800724, "elapsed_sec": 68220.0338947773, "step_time_sec": 8.238939497969113, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.959534168243408, "lr": 0.00019856473750030638, "elapsed_sec": 68228.27505373955, "step_time_sec": 8.241035480983555, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.010073184967041, "lr": 0.00019856235625349987, "elapsed_sec": 68236.51437473297, "step_time_sec": 8.23911271902034, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.041409969329834, "lr": 0.0001985599730476362, "elapsed_sec": 68244.75279307365, "step_time_sec": 8.238276396004949, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0884881019592285, "lr": 0.00019855758788276404, "elapsed_sec": 68252.99201583862, "step_time_sec": 8.239068496972322, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.064872741699219, "lr": 0.000198555200758932, "elapsed_sec": 68261.23003149033, "step_time_sec": 8.237846493953839, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.021731376647949, "lr": 0.00019855281167618872, "elapsed_sec": 68269.47146344185, "step_time_sec": 8.24125698802527, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.000600337982178, "lr": 0.000198550420634583, "elapsed_sec": 68277.71067142487, "step_time_sec": 8.239034351019654, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.012820720672607, "lr": 0.00019854802763416356, "elapsed_sec": 68285.94871044159, "step_time_sec": 8.237868298019748, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9344077110290527, "lr": 0.0001985456326749792, "elapsed_sec": 68294.1873562336, "step_time_sec": 8.238535918993875, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.983242988586426, "lr": 0.00019854323575707876, "elapsed_sec": 68302.42494773865, "step_time_sec": 8.23735690698959, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8919806480407715, "lr": 0.00019854083688051113, "elapsed_sec": 68310.6640021801, "step_time_sec": 8.238917971029878, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.976940870285034, "lr": 0.00019853843604532523, "elapsed_sec": 68318.90385103226, "step_time_sec": 8.23966143396683, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9189534187316895, "lr": 0.00019853603325157003, "elapsed_sec": 68327.14239692688, "step_time_sec": 8.238369193044491, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.965428352355957, "lr": 0.00019853362849929452, "elapsed_sec": 68335.3820681572, "step_time_sec": 8.239550043013878, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8308277130126953, "lr": 0.00019853122178854776, "elapsed_sec": 68343.62311196327, "step_time_sec": 8.240832624025643, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.025537490844727, "lr": 0.00019852881311937883, "elapsed_sec": 68351.86202812195, "step_time_sec": 8.23872585297795, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.000360488891602, "lr": 0.0001985264024918368, "elapsed_sec": 68360.10116457939, "step_time_sec": 8.238995184015948, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9100699424743652, "lr": 0.0001985239899059709, "elapsed_sec": 68368.33962607384, "step_time_sec": 8.23828853899613, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.086151123046875, "lr": 0.00019852157536183026, "elapsed_sec": 68376.57925081253, "step_time_sec": 8.239425589039456, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9444003105163574, "lr": 0.0001985191588594642, "elapsed_sec": 68384.81848478317, "step_time_sec": 8.239071760966908, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.070782661437988, "lr": 0.00019851674039892198, "elapsed_sec": 68393.05805397034, "step_time_sec": 8.239410342997871, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1162495613098145, "lr": 0.00019851431998025286, "elapsed_sec": 68401.29852938652, "step_time_sec": 8.24029181903461, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8684582710266113, "lr": 0.00019851189760350626, "elapsed_sec": 68409.5370414257, "step_time_sec": 8.238355781999417, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.090465545654297, "lr": 0.00019850947326873158, "elapsed_sec": 68417.77682423592, "step_time_sec": 8.23961111297831, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8833606243133545, "lr": 0.00019850704697597826, "elapsed_sec": 68426.01662230492, "step_time_sec": 8.239647445036098, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9721739292144775, "lr": 0.00019850461872529575, "elapsed_sec": 68434.25520563126, "step_time_sec": 8.238445013994351, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9002201557159424, "lr": 0.0001985021885167336, "elapsed_sec": 68442.4943716526, "step_time_sec": 8.238979423011187, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.041241645812988, "lr": 0.00019849975635034136, "elapsed_sec": 68450.73349308968, "step_time_sec": 8.238918462011497, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9226012229919434, "lr": 0.00019849732222616863, "elapsed_sec": 68458.97486186028, "step_time_sec": 8.241263606003486, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.057671070098877, "lr": 0.00019849488614426505, "elapsed_sec": 68467.21517944336, "step_time_sec": 8.24010504502803, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.027329444885254, "lr": 0.0001984924481046803, "elapsed_sec": 68475.45459747314, "step_time_sec": 8.239266169024631, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.07737398147583, "lr": 0.0001984900081074641, "elapsed_sec": 68483.69573187828, "step_time_sec": 8.241003389004618, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.86499285697937, "lr": 0.0001984875661526662, "elapsed_sec": 68491.93596076965, "step_time_sec": 8.240037340961862, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9876768589019775, "lr": 0.00019848512224033646, "elapsed_sec": 68500.17409515381, "step_time_sec": 8.23796310700709, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.135261535644531, "lr": 0.00019848267637052463, "elapsed_sec": 68508.41279554367, "step_time_sec": 8.23859194398392, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.189062118530273, "lr": 0.00019848022854328064, "elapsed_sec": 68516.65137863159, "step_time_sec": 8.238396242959425, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.821190118789673, "lr": 0.00019847777875865442, "elapsed_sec": 68524.8911883831, "step_time_sec": 8.239611093013082, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.144472599029541, "lr": 0.0001984753270166959, "elapsed_sec": 68533.13069915771, "step_time_sec": 8.23938902997179, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.061832904815674, "lr": 0.0001984728733174551, "elapsed_sec": 68541.3678495884, "step_time_sec": 8.236962689028587, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1823248863220215, "lr": 0.00019847041766098204, "elapsed_sec": 68549.60659599304, "step_time_sec": 8.238596904964652, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.003362655639648, "lr": 0.00019846796004732684, "elapsed_sec": 68557.84409427643, "step_time_sec": 8.237355465011206, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.06093692779541, "lr": 0.00019846550047653954, "elapsed_sec": 68566.08510828018, "step_time_sec": 8.24083167500794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.02784538269043, "lr": 0.0001984630389486704, "elapsed_sec": 68574.32375359535, "step_time_sec": 8.23850986897014, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.032571315765381, "lr": 0.00019846057546376957, "elapsed_sec": 68582.56109046936, "step_time_sec": 8.23711208900204, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2485270500183105, "lr": 0.00019845811002188726, "elapsed_sec": 68590.80131673813, "step_time_sec": 8.240018839016557, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.203496932983398, "lr": 0.0001984556426230738, "elapsed_sec": 68599.04203915596, "step_time_sec": 8.240607508982066, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0623040199279785, "lr": 0.0001984531732673795, "elapsed_sec": 68607.28022885323, "step_time_sec": 8.237956181983463, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.212924957275391, "lr": 0.0001984507019548547, "elapsed_sec": 68615.51985907555, "step_time_sec": 8.239463286008686, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.118709564208984, "lr": 0.0001984482286855498, "elapsed_sec": 68623.75632095337, "step_time_sec": 8.236339349998161, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0981550216674805, "lr": 0.00019844575345951524, "elapsed_sec": 68631.9944319725, "step_time_sec": 8.2379308140371, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.21776008605957, "lr": 0.0001984432762768015, "elapsed_sec": 68640.23409795761, "step_time_sec": 8.239488470018841, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.014316558837891, "lr": 0.0001984407971374591, "elapsed_sec": 68648.47284650803, "step_time_sec": 8.238638644979801, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9311912059783936, "lr": 0.00019843831604153862, "elapsed_sec": 68656.71178102493, "step_time_sec": 8.238757535000332, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9737117290496826, "lr": 0.00019843583298909066, "elapsed_sec": 68664.94822025299, "step_time_sec": 8.236259181983769, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.086938858032227, "lr": 0.00019843334798016582, "elapsed_sec": 68673.1876437664, "step_time_sec": 8.23925473599229, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1034016609191895, "lr": 0.0001984308610148148, "elapsed_sec": 68681.42626047134, "step_time_sec": 8.238404173986055, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.115342140197754, "lr": 0.00019842837209308828, "elapsed_sec": 68689.66574048996, "step_time_sec": 8.239328378986102, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.045120716094971, "lr": 0.0001984258812150371, "elapsed_sec": 68697.90731024742, "step_time_sec": 8.24142048502108, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.155984401702881, "lr": 0.00019842338838071198, "elapsed_sec": 68706.14487314224, "step_time_sec": 8.237377437006216, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.193171501159668, "lr": 0.00019842089359016382, "elapsed_sec": 68714.3858332634, "step_time_sec": 8.240852075978182, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9838640689849854, "lr": 0.00019841839684344342, "elapsed_sec": 68722.62602114677, "step_time_sec": 8.239966413995717, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.062353610992432, "lr": 0.00019841589814060178, "elapsed_sec": 68730.86348199844, "step_time_sec": 8.237287447031122, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.151947021484375, "lr": 0.00019841339748168983, "elapsed_sec": 68739.10320830345, "step_time_sec": 8.239605339011177, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.854443311691284, "lr": 0.00019841089486675854, "elapsed_sec": 68747.34132313728, "step_time_sec": 8.237903607019689, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9750561714172363, "lr": 0.000198408390295859, "elapsed_sec": 68755.58080863953, "step_time_sec": 8.239325549046043, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.012582778930664, "lr": 0.00019840588376904218, "elapsed_sec": 68763.82080030441, "step_time_sec": 8.239859021035954, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.216809272766113, "lr": 0.00019840337528635933, "elapsed_sec": 68772.05955910683, "step_time_sec": 8.238554268027656, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.10590934753418, "lr": 0.0001984008648478615, "elapsed_sec": 68780.30008530617, "step_time_sec": 8.240399719972629, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.042895793914795, "lr": 0.00019839835245360002, "elapsed_sec": 68788.53781723976, "step_time_sec": 8.23754675197415, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.99155330657959, "lr": 0.00019839583810362597, "elapsed_sec": 68796.77735805511, "step_time_sec": 8.239412513037678, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.974729061126709, "lr": 0.0001983933217979907, "elapsed_sec": 68805.01767349243, "step_time_sec": 8.240106716984883, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.073658466339111, "lr": 0.00019839080353674555, "elapsed_sec": 68813.25510978699, "step_time_sec": 8.23728767898865, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9959495067596436, "lr": 0.00019838828331994186, "elapsed_sec": 68821.49562191963, "step_time_sec": 8.240278294018935, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.048308849334717, "lr": 0.00019838576114763099, "elapsed_sec": 68829.7347176075, "step_time_sec": 8.238927209982648, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9590518474578857, "lr": 0.0001983832370198644, "elapsed_sec": 68837.9736316204, "step_time_sec": 8.238791713025421, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.064263343811035, "lr": 0.00019838071093669358, "elapsed_sec": 68846.21368312836, "step_time_sec": 8.239837843051646, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.125349998474121, "lr": 0.00019837818289817006, "elapsed_sec": 68854.45126605034, "step_time_sec": 8.237471019965596, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9330434799194336, "lr": 0.00019837565290434535, "elapsed_sec": 68862.69194245338, "step_time_sec": 8.240455187042244, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.011566162109375, "lr": 0.00019837312095527106, "elapsed_sec": 68870.93120002747, "step_time_sec": 8.239109878020827, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.074172496795654, "lr": 0.00019837058705099886, "elapsed_sec": 68879.1707007885, "step_time_sec": 8.239400594960898, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.107644081115723, "lr": 0.00019836805119158036, "elapsed_sec": 68887.40996599197, "step_time_sec": 8.23907834495185, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2162981033325195, "lr": 0.00019836551337706735, "elapsed_sec": 68895.64843845367, "step_time_sec": 8.23825414199382, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.046988010406494, "lr": 0.0001983629736075115, "elapsed_sec": 68903.88764762878, "step_time_sec": 8.239069743955042, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.096335411071777, "lr": 0.0001983604318829647, "elapsed_sec": 68912.12784671783, "step_time_sec": 8.240015333984047, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9622650146484375, "lr": 0.00019835788820347873, "elapsed_sec": 68920.36646914482, "step_time_sec": 8.238461349974386, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.952540874481201, "lr": 0.00019835534256910548, "elapsed_sec": 68928.60618686676, "step_time_sec": 8.239593637990765, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.227589130401611, "lr": 0.00019835279497989685, "elapsed_sec": 68936.84658885002, "step_time_sec": 8.240176457038615, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.963644504547119, "lr": 0.00019835024543590478, "elapsed_sec": 68945.0872554779, "step_time_sec": 8.240504688001238, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.087680339813232, "lr": 0.00019834769393718132, "elapsed_sec": 68953.32646632195, "step_time_sec": 8.239079892984591, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.220900535583496, "lr": 0.00019834514048377845, "elapsed_sec": 68961.56484174728, "step_time_sec": 8.23818481498165, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.994558811187744, "lr": 0.00019834258507574827, "elapsed_sec": 68969.80696058273, "step_time_sec": 8.241953033953905, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9870550632476807, "lr": 0.00019834002771314287, "elapsed_sec": 68978.04481101036, "step_time_sec": 8.237672802992165, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.958730459213257, "lr": 0.00019833746839601445, "elapsed_sec": 68986.28374052048, "step_time_sec": 8.238754348014481, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.027617931365967, "lr": 0.00019833490712441518, "elapsed_sec": 68994.52314710617, "step_time_sec": 8.239294596016407, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.946584939956665, "lr": 0.00019833234389839727, "elapsed_sec": 69002.76096725464, "step_time_sec": 8.237596841005143, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.070052623748779, "lr": 0.00019832977871801302, "elapsed_sec": 69011.00063443184, "step_time_sec": 8.23951844300609, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8954193592071533, "lr": 0.00019832721158331474, "elapsed_sec": 69019.2407681942, "step_time_sec": 8.240005547995679, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.007557392120361, "lr": 0.00019832464249435478, "elapsed_sec": 69027.47856259346, "step_time_sec": 8.237628731003497, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9913392066955566, "lr": 0.00019832207145118553, "elapsed_sec": 69035.7188501358, "step_time_sec": 8.240084695047699, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9150338172912598, "lr": 0.00019831949845385946, "elapsed_sec": 69043.95536780357, "step_time_sec": 8.236399112036452, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9381232261657715, "lr": 0.00019831692350242894, "elapsed_sec": 69052.1952893734, "step_time_sec": 8.23969965800643, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9649853706359863, "lr": 0.0001983143465969466, "elapsed_sec": 69060.43651199341, "step_time_sec": 8.241078351973556, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9057085514068604, "lr": 0.00019831176773746492, "elapsed_sec": 69068.6744658947, "step_time_sec": 8.237800324044656, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.184070110321045, "lr": 0.00019830918692403653, "elapsed_sec": 69076.91284251213, "step_time_sec": 8.238179208012298, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.057801723480225, "lr": 0.00019830660415671406, "elapsed_sec": 69085.1527967453, "step_time_sec": 8.239814294036478, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8749656677246094, "lr": 0.00019830401943555012, "elapsed_sec": 69093.39143490791, "step_time_sec": 8.238440018030815, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7997725009918213, "lr": 0.00019830143276059752, "elapsed_sec": 69101.63089227676, "step_time_sec": 8.239288763026707, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.045384883880615, "lr": 0.00019829884413190895, "elapsed_sec": 69109.8689045906, "step_time_sec": 8.237833458988462, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.852389335632324, "lr": 0.00019829625354953723, "elapsed_sec": 69118.10974311829, "step_time_sec": 8.240720148955006, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.08962869644165, "lr": 0.00019829366101353513, "elapsed_sec": 69126.34731078148, "step_time_sec": 8.237413181981537, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.950108051300049, "lr": 0.00019829106652395558, "elapsed_sec": 69134.5867292881, "step_time_sec": 8.239259233989287, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.97416615486145, "lr": 0.00019828847008085152, "elapsed_sec": 69142.82549381256, "step_time_sec": 8.238558971963357, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.999972105026245, "lr": 0.00019828587168427582, "elapsed_sec": 69151.06353855133, "step_time_sec": 8.237870421027765, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.908198118209839, "lr": 0.00019828327133428154, "elapsed_sec": 69159.30315327644, "step_time_sec": 8.239461946999654, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.045266151428223, "lr": 0.00019828066903092165, "elapsed_sec": 69167.54230332375, "step_time_sec": 8.23896434402559, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.982449769973755, "lr": 0.00019827806477424925, "elapsed_sec": 69175.78080439568, "step_time_sec": 8.238296114024706, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.922518014907837, "lr": 0.00019827545856431747, "elapsed_sec": 69184.02005410194, "step_time_sec": 8.23914566199528, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.108544826507568, "lr": 0.00019827285040117942, "elapsed_sec": 69192.25737810135, "step_time_sec": 8.23710549599491, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.892063856124878, "lr": 0.00019827024028488832, "elapsed_sec": 69200.49680948257, "step_time_sec": 8.239260154019576, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.868999719619751, "lr": 0.0001982676282154974, "elapsed_sec": 69208.73660731316, "step_time_sec": 8.239693853014614, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.095437526702881, "lr": 0.0001982650141930599, "elapsed_sec": 69216.97362661362, "step_time_sec": 8.236809164984152, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.015335559844971, "lr": 0.00019826239821762917, "elapsed_sec": 69225.21210122108, "step_time_sec": 8.238313608977478, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.956280469894409, "lr": 0.00019825978028925853, "elapsed_sec": 69233.44945907593, "step_time_sec": 8.237198273011018, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.988445281982422, "lr": 0.00019825716040800136, "elapsed_sec": 69241.68837785721, "step_time_sec": 8.238773835008033, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9519217014312744, "lr": 0.0001982545385739111, "elapsed_sec": 69249.92828702927, "step_time_sec": 8.239719339006115, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.015009880065918, "lr": 0.00019825191478704127, "elapsed_sec": 69258.16727423668, "step_time_sec": 8.238892683992162, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.04771089553833, "lr": 0.0001982492890474453, "elapsed_sec": 69266.40690374374, "step_time_sec": 8.239439853990916, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.95335054397583, "lr": 0.00019824666135517677, "elapsed_sec": 69274.64483761787, "step_time_sec": 8.237803479016293, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9888038635253906, "lr": 0.00019824403171028927, "elapsed_sec": 69282.88466477394, "step_time_sec": 8.239601950976066, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.123377323150635, "lr": 0.00019824140011283643, "elapsed_sec": 69291.1242287159, "step_time_sec": 8.239380949002225, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.082393169403076, "lr": 0.0001982387665628719, "elapsed_sec": 69299.36376237869, "step_time_sec": 8.239423754974268, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.042642593383789, "lr": 0.00019823613106044946, "elapsed_sec": 69307.60341143608, "step_time_sec": 8.239450552966446, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.034353256225586, "lr": 0.00019823349360562274, "elapsed_sec": 69315.84083080292, "step_time_sec": 8.23723400599556, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.243001937866211, "lr": 0.0001982308541984456, "elapsed_sec": 69324.0799446106, "step_time_sec": 8.238995359977707, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9127109050750732, "lr": 0.00019822821283897187, "elapsed_sec": 69332.31943678856, "step_time_sec": 8.239264284959063, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0609893798828125, "lr": 0.0001982255695272554, "elapsed_sec": 69340.5571463108, "step_time_sec": 8.237552814010996, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.031339168548584, "lr": 0.00019822292426335006, "elapsed_sec": 69348.79812812805, "step_time_sec": 8.240857800992671, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.058644771575928, "lr": 0.00019822027704730988, "elapsed_sec": 69357.03688192368, "step_time_sec": 8.238538105972111, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.054409980773926, "lr": 0.00019821762787918876, "elapsed_sec": 69365.27521419525, "step_time_sec": 8.238157265004702, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0643157958984375, "lr": 0.00019821497675904078, "elapsed_sec": 69373.51318144798, "step_time_sec": 8.237812239967752, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.969248056411743, "lr": 0.00019821232368692, "elapsed_sec": 69381.75122213364, "step_time_sec": 8.237863451009616, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.056159496307373, "lr": 0.00019820966866288054, "elapsed_sec": 69389.99000144005, "step_time_sec": 8.238600386015605, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.003138065338135, "lr": 0.0001982070116869765, "elapsed_sec": 69398.22744774818, "step_time_sec": 8.237291930010542, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9327714443206787, "lr": 0.00019820435275926207, "elapsed_sec": 69406.46650981903, "step_time_sec": 8.23886046296684, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.068239212036133, "lr": 0.00019820169187979153, "elapsed_sec": 69414.70613598824, "step_time_sec": 8.239520325965714, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.15150785446167, "lr": 0.00019819902904861906, "elapsed_sec": 69422.94434952736, "step_time_sec": 8.238008654036094, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.030246257781982, "lr": 0.00019819636426579907, "elapsed_sec": 69431.18455171585, "step_time_sec": 8.240015303017572, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1804680824279785, "lr": 0.0001981936975313858, "elapsed_sec": 69439.42299365997, "step_time_sec": 8.238321114971768, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.02236795425415, "lr": 0.0001981910288454337, "elapsed_sec": 69447.66356492043, "step_time_sec": 8.240362422016915, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1787943840026855, "lr": 0.0001981883582079972, "elapsed_sec": 69455.90389323235, "step_time_sec": 8.240156555024441, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.028029441833496, "lr": 0.00019818568561913072, "elapsed_sec": 69464.14048910141, "step_time_sec": 8.236500885046553, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.020724773406982, "lr": 0.00019818301107888875, "elapsed_sec": 69472.37996649742, "step_time_sec": 8.239223352982663, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1084747314453125, "lr": 0.0001981803345873259, "elapsed_sec": 69480.61903882027, "step_time_sec": 8.238908612984233, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.093842029571533, "lr": 0.0001981776561444967, "elapsed_sec": 69488.85846161842, "step_time_sec": 8.239300098037347, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.021267890930176, "lr": 0.00019817497575045585, "elapsed_sec": 69497.09868192673, "step_time_sec": 8.24002276698593, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.018041133880615, "lr": 0.00019817229340525792, "elapsed_sec": 69505.33619379997, "step_time_sec": 8.237371986964718, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.012214183807373, "lr": 0.00019816960910895766, "elapsed_sec": 69513.57601571083, "step_time_sec": 8.239583414979279, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9361071586608887, "lr": 0.00019816692286160978, "elapsed_sec": 69521.81605505943, "step_time_sec": 8.239883504982572, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.105990409851074, "lr": 0.0001981642346632691, "elapsed_sec": 69530.05421996117, "step_time_sec": 8.237988070002757, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.939262628555298, "lr": 0.00019816154451399043, "elapsed_sec": 69538.29311561584, "step_time_sec": 8.238709608966019, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9963183403015137, "lr": 0.0001981588524138286, "elapsed_sec": 69546.53078579903, "step_time_sec": 8.237562123977114, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.021909236907959, "lr": 0.00019815615836283855, "elapsed_sec": 69554.77028989792, "step_time_sec": 8.239286910975352, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.140045642852783, "lr": 0.0001981534623610752, "elapsed_sec": 69563.00913929939, "step_time_sec": 8.238703425973654, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.6359920501708984, "lr": 0.00019815076440859357, "elapsed_sec": 69571.24685502052, "step_time_sec": 8.237525160016958, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9888429641723633, "lr": 0.00019814806450544863, "elapsed_sec": 69579.48601126671, "step_time_sec": 8.239043648005463, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0422797203063965, "lr": 0.0001981453626516955, "elapsed_sec": 69587.72369933128, "step_time_sec": 8.237466282967944, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.064445495605469, "lr": 0.0001981426588473892, "elapsed_sec": 69595.96321892738, "step_time_sec": 8.239418226003181, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9967775344848633, "lr": 0.00019813995309258492, "elapsed_sec": 69604.20140266418, "step_time_sec": 8.238016696006525, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9697999954223633, "lr": 0.00019813724538733787, "elapsed_sec": 69612.43984484673, "step_time_sec": 8.238192435004748, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2496137619018555, "lr": 0.0001981345357317032, "elapsed_sec": 69620.67945933342, "step_time_sec": 8.239488559949677, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.006503582000732, "lr": 0.00019813182412573623, "elapsed_sec": 69628.91678977013, "step_time_sec": 8.237152817018796, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.05308198928833, "lr": 0.00019812911056949224, "elapsed_sec": 69637.15675163269, "step_time_sec": 8.239799522969406, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.09701681137085, "lr": 0.00019812639506302653, "elapsed_sec": 69645.39646840096, "step_time_sec": 8.23954015399795, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.995798349380493, "lr": 0.00019812367760639452, "elapsed_sec": 69653.63411355019, "step_time_sec": 8.237469213025179, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.018124103546143, "lr": 0.00019812095819965163, "elapsed_sec": 69661.87452816963, "step_time_sec": 8.240289087989368, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.97200870513916, "lr": 0.0001981182368428533, "elapsed_sec": 69670.11332678795, "step_time_sec": 8.238648600003216, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9560487270355225, "lr": 0.000198115513536055, "elapsed_sec": 69678.35275363922, "step_time_sec": 8.239240640948992, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.172845363616943, "lr": 0.00019811278827931237, "elapsed_sec": 69686.59209728241, "step_time_sec": 8.239232317020651, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9756219387054443, "lr": 0.00019811006107268091, "elapsed_sec": 69694.8327448368, "step_time_sec": 8.240412543003913, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.01151180267334, "lr": 0.00019810733191621624, "elapsed_sec": 69703.0733640194, "step_time_sec": 8.24048825999489, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.11593770980835, "lr": 0.000198104600809974, "elapsed_sec": 69711.31155395508, "step_time_sec": 8.238015184004325, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.017248153686523, "lr": 0.00019810186775400996, "elapsed_sec": 69719.55228447914, "step_time_sec": 8.240584651997779, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8760712146759033, "lr": 0.00019809913274837978, "elapsed_sec": 69727.7928109169, "step_time_sec": 8.240317568008322, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.948821544647217, "lr": 0.00019809639579313927, "elapsed_sec": 69736.03076076508, "step_time_sec": 8.23780307098059, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.003030300140381, "lr": 0.00019809365688834425, "elapsed_sec": 69744.2709467411, "step_time_sec": 8.240067655046005, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.842041015625, "lr": 0.00019809091603405058, "elapsed_sec": 69752.51083445549, "step_time_sec": 8.239661875995807, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.978259563446045, "lr": 0.00019808817323031417, "elapsed_sec": 69760.74956083298, "step_time_sec": 8.238573316019028, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.999769926071167, "lr": 0.00019808542847719087, "elapsed_sec": 69768.98896408081, "step_time_sec": 8.239303093985654, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.119112014770508, "lr": 0.00019808268177473674, "elapsed_sec": 69777.22795152664, "step_time_sec": 8.238789690949488, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.188812732696533, "lr": 0.0001980799331230078, "elapsed_sec": 69785.46732568741, "step_time_sec": 8.23919987899717, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.007892608642578, "lr": 0.00019807718252206006, "elapsed_sec": 69793.70570731163, "step_time_sec": 8.238252217008267, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.022631645202637, "lr": 0.00019807442997194963, "elapsed_sec": 69801.94358825684, "step_time_sec": 8.237695135001559, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9546315670013428, "lr": 0.00019807167547273267, "elapsed_sec": 69810.18273472786, "step_time_sec": 8.239019117027055, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.063533782958984, "lr": 0.0001980689190244653, "elapsed_sec": 69818.42096948624, "step_time_sec": 8.238036073977128, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.892184019088745, "lr": 0.00019806616062720376, "elapsed_sec": 69826.65871310234, "step_time_sec": 8.237555142026395, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9851040840148926, "lr": 0.0001980634002810043, "elapsed_sec": 69834.89734888077, "step_time_sec": 8.238495601981413, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.869943380355835, "lr": 0.0001980606379859233, "elapsed_sec": 69843.13486433029, "step_time_sec": 8.237362803018186, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.265614032745361, "lr": 0.00019805787374201693, "elapsed_sec": 69851.37302350998, "step_time_sec": 8.237979542987887, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.12556791305542, "lr": 0.00019805510754934168, "elapsed_sec": 69859.61142802238, "step_time_sec": 8.238251906994265, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.950434923171997, "lr": 0.00019805233940795394, "elapsed_sec": 69867.85142230988, "step_time_sec": 8.239801131014246, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.964054822921753, "lr": 0.00019804956931791014, "elapsed_sec": 69876.09045743942, "step_time_sec": 8.238871831970755, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.014566898345947, "lr": 0.00019804679727926678, "elapsed_sec": 69884.32857346535, "step_time_sec": 8.237988025997765, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9983508586883545, "lr": 0.0001980440232920804, "elapsed_sec": 69892.56729626656, "step_time_sec": 8.238508234033361, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.094244956970215, "lr": 0.00019804124735640758, "elapsed_sec": 69900.80541348457, "step_time_sec": 8.23793195397593, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9335010051727295, "lr": 0.0001980384694723049, "elapsed_sec": 69909.04505991936, "step_time_sec": 8.239515547989868, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.021105766296387, "lr": 0.00019803568963982905, "elapsed_sec": 69917.28267002106, "step_time_sec": 8.237419879995286, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9828126430511475, "lr": 0.00019803290785903668, "elapsed_sec": 69925.52091431618, "step_time_sec": 8.238069398968946, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.990337610244751, "lr": 0.00019803012412998456, "elapsed_sec": 69933.75975203514, "step_time_sec": 8.238650904037058, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9386637210845947, "lr": 0.00019802733845272945, "elapsed_sec": 69941.99658179283, "step_time_sec": 8.236728270014282, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.890089273452759, "lr": 0.00019802455082732813, "elapsed_sec": 69950.23660874367, "step_time_sec": 8.239788060018327, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.196176528930664, "lr": 0.00019802176125383747, "elapsed_sec": 69958.4775865078, "step_time_sec": 8.24081558996113, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.985527515411377, "lr": 0.0001980189697323144, "elapsed_sec": 69966.7157318592, "step_time_sec": 8.23803681699792, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.922712802886963, "lr": 0.00019801617626281578, "elapsed_sec": 69974.95716881752, "step_time_sec": 8.241251399973407, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.982189655303955, "lr": 0.0001980133808453986, "elapsed_sec": 69983.19564294815, "step_time_sec": 8.238263881008606, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9656035900115967, "lr": 0.0001980105834801199, "elapsed_sec": 69991.4350028038, "step_time_sec": 8.239177567011211, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9546353816986084, "lr": 0.00019800778416703666, "elapsed_sec": 69999.67506432533, "step_time_sec": 8.239956968987826, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.933586597442627, "lr": 0.00019800498290620605, "elapsed_sec": 70007.91252565384, "step_time_sec": 8.237253845029045, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.891446113586426, "lr": 0.0001980021796976852, "elapsed_sec": 70016.1525361538, "step_time_sec": 8.239859527966473, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7342031002044678, "lr": 0.0001979993745415312, "elapsed_sec": 70024.39215755463, "step_time_sec": 8.239496105001308, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9240520000457764, "lr": 0.0001979965674378013, "elapsed_sec": 70032.63199520111, "step_time_sec": 8.239594596030656, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.016793251037598, "lr": 0.00019799375838655274, "elapsed_sec": 70040.8719842434, "step_time_sec": 8.239827122015413, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9535934925079346, "lr": 0.0001979909473878428, "elapsed_sec": 70049.10956549644, "step_time_sec": 8.237443391990382, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.874455213546753, "lr": 0.00019798813444172884, "elapsed_sec": 70057.35004758835, "step_time_sec": 8.240269714035094, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.965100049972534, "lr": 0.00019798531954826819, "elapsed_sec": 70065.58862662315, "step_time_sec": 8.23841451300541, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.048595905303955, "lr": 0.00019798250270751828, "elapsed_sec": 70073.827521801, "step_time_sec": 8.23872051096987, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.022965431213379, "lr": 0.00019797968391953652, "elapsed_sec": 70082.0678369999, "step_time_sec": 8.240156813000795, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.915031909942627, "lr": 0.00019797686318438046, "elapsed_sec": 70090.30752754211, "step_time_sec": 8.23952876997646, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.050239086151123, "lr": 0.00019797404050210757, "elapsed_sec": 70098.54759168625, "step_time_sec": 8.239912759978324, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.124481678009033, "lr": 0.0001979712158727754, "elapsed_sec": 70106.78687429428, "step_time_sec": 8.239107138011605, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9625511169433594, "lr": 0.0001979683892964416, "elapsed_sec": 70115.02547097206, "step_time_sec": 8.23842580098426, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.063180446624756, "lr": 0.0001979655607731638, "elapsed_sec": 70123.26484251022, "step_time_sec": 8.239265165990219, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.880282402038574, "lr": 0.0001979627303029997, "elapsed_sec": 70131.50383591652, "step_time_sec": 8.23882866895292, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9026339054107666, "lr": 0.000197959897886007, "elapsed_sec": 70139.7442536354, "step_time_sec": 8.240254745003767, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.113536357879639, "lr": 0.00019795706352224344, "elapsed_sec": 70147.98252940178, "step_time_sec": 8.238112525024917, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.149618625640869, "lr": 0.00019795422721176687, "elapsed_sec": 70156.21988797188, "step_time_sec": 8.237132104986813, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.023253917694092, "lr": 0.0001979513889546351, "elapsed_sec": 70164.45962762833, "step_time_sec": 8.23964442597935, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.055795669555664, "lr": 0.00019794854875090605, "elapsed_sec": 70172.69953632355, "step_time_sec": 8.239711728005204, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7751052379608154, "lr": 0.00019794570660063758, "elapsed_sec": 70180.93797278404, "step_time_sec": 8.238266064028721, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9510533809661865, "lr": 0.00019794286250388772, "elapsed_sec": 70189.17776226997, "step_time_sec": 8.239661074010655, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9902446269989014, "lr": 0.00019794001646071446, "elapsed_sec": 70197.41653871536, "step_time_sec": 8.238575341005344, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.003978252410889, "lr": 0.00019793716847117578, "elapsed_sec": 70205.65595817566, "step_time_sec": 8.239239844027907, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.953986167907715, "lr": 0.00019793431853532982, "elapsed_sec": 70213.89541387558, "step_time_sec": 8.23934560600901, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.960223913192749, "lr": 0.00019793146665323468, "elapsed_sec": 70222.13263368607, "step_time_sec": 8.237031102005858, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9483730792999268, "lr": 0.00019792861282494852, "elapsed_sec": 70230.3721241951, "step_time_sec": 8.239313522994053, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9511005878448486, "lr": 0.00019792575705052955, "elapsed_sec": 70238.60970211029, "step_time_sec": 8.237402086961083, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.007688999176025, "lr": 0.000197922899330036, "elapsed_sec": 70246.84973549843, "step_time_sec": 8.239875216037035, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.830763101577759, "lr": 0.00019792003966352616, "elapsed_sec": 70255.08941245079, "step_time_sec": 8.239549082994927, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.113415241241455, "lr": 0.00019791717805105834, "elapsed_sec": 70263.3273806572, "step_time_sec": 8.237759571988136, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.110217094421387, "lr": 0.00019791431449269086, "elapsed_sec": 70271.56693077087, "step_time_sec": 8.23942979396088, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.977034568786621, "lr": 0.0001979114489884822, "elapsed_sec": 70279.80566978455, "step_time_sec": 8.23857892700471, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9169414043426514, "lr": 0.00019790858153849073, "elapsed_sec": 70288.04561424255, "step_time_sec": 8.239758495998103, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.998943328857422, "lr": 0.000197905712142775, "elapsed_sec": 70296.28452897072, "step_time_sec": 8.238721772038843, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9592113494873047, "lr": 0.00019790284080139342, "elapsed_sec": 70304.52280068398, "step_time_sec": 8.238166905997787, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.060206413269043, "lr": 0.00019789996751440464, "elapsed_sec": 70312.76239657402, "step_time_sec": 8.2393752039643, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.035740852355957, "lr": 0.00019789709228186723, "elapsed_sec": 70321.00166583061, "step_time_sec": 8.239066070993431, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.220952987670898, "lr": 0.0001978942151038398, "elapsed_sec": 70329.24049258232, "step_time_sec": 8.238694672996644, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.966356039047241, "lr": 0.00019789133598038106, "elapsed_sec": 70337.479824543, "step_time_sec": 8.239144281949848, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9651033878326416, "lr": 0.00019788845491154971, "elapsed_sec": 70345.71701169014, "step_time_sec": 8.23702390800463, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.015133380889893, "lr": 0.0001978855718974045, "elapsed_sec": 70353.95631623268, "step_time_sec": 8.239168443949893, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.974846601486206, "lr": 0.00019788268693800423, "elapsed_sec": 70362.1939997673, "step_time_sec": 8.2375388739747, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.091501712799072, "lr": 0.00019787980003340775, "elapsed_sec": 70370.43357157707, "step_time_sec": 8.239356920996215, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.065727233886719, "lr": 0.00019787691118367394, "elapsed_sec": 70378.67291235924, "step_time_sec": 8.239185044018086, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.991811752319336, "lr": 0.00019787402038886164, "elapsed_sec": 70386.91142988205, "step_time_sec": 8.238385967037175, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9639315605163574, "lr": 0.00019787112764902993, "elapsed_sec": 70395.15116143227, "step_time_sec": 8.239591838035267, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.098399639129639, "lr": 0.0001978682329642377, "elapsed_sec": 70403.39028191566, "step_time_sec": 8.238926559977699, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9214019775390625, "lr": 0.000197865336334544, "elapsed_sec": 70411.6291384697, "step_time_sec": 8.23867141304072, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.099039554595947, "lr": 0.00019786243776000791, "elapsed_sec": 70419.8690803051, "step_time_sec": 8.239806380996015, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.091133117675781, "lr": 0.00019785953724068856, "elapsed_sec": 70428.10607171059, "step_time_sec": 8.23680658097146, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.061649799346924, "lr": 0.00019785663477664514, "elapsed_sec": 70436.34731292725, "step_time_sec": 8.24110014399048, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.120619297027588, "lr": 0.00019785373036793673, "elapsed_sec": 70444.58720231056, "step_time_sec": 8.239725167979486, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.021332263946533, "lr": 0.00019785082401462266, "elapsed_sec": 70452.82491207123, "step_time_sec": 8.237521545030177, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.987074613571167, "lr": 0.00019784791571676215, "elapsed_sec": 70461.06349182129, "step_time_sec": 8.238401454989798, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.057887077331543, "lr": 0.00019784500547441456, "elapsed_sec": 70469.30192875862, "step_time_sec": 8.238305160950404, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.275471210479736, "lr": 0.00019784209328763916, "elapsed_sec": 70477.54174137115, "step_time_sec": 8.239627943956293, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.168691635131836, "lr": 0.0001978391791564954, "elapsed_sec": 70485.78209066391, "step_time_sec": 8.240163241047412, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.909820795059204, "lr": 0.00019783626308104272, "elapsed_sec": 70494.02045536041, "step_time_sec": 8.238209555973299, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.164149284362793, "lr": 0.00019783334506134057, "elapsed_sec": 70502.25980234146, "step_time_sec": 8.239218890026677, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.029312610626221, "lr": 0.00019783042509744842, "elapsed_sec": 70510.49703764915, "step_time_sec": 8.237044890003745, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.102151870727539, "lr": 0.0001978275031894259, "elapsed_sec": 70518.73736214638, "step_time_sec": 8.240165817027446, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.013274669647217, "lr": 0.0001978245793373325, "elapsed_sec": 70526.97842907906, "step_time_sec": 8.240987462049816, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.04205846786499, "lr": 0.00019782165354122793, "elapsed_sec": 70535.21536803246, "step_time_sec": 8.23675863200333, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.779843807220459, "lr": 0.00019781872580117184, "elapsed_sec": 70543.45470094681, "step_time_sec": 8.239128076005727, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.998295307159424, "lr": 0.0001978157961172239, "elapsed_sec": 70551.69265723228, "step_time_sec": 8.237786508980207, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.043628215789795, "lr": 0.00019781286448944388, "elapsed_sec": 70559.932451725, "step_time_sec": 8.23967697401531, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.046354293823242, "lr": 0.00019780993091789162, "elapsed_sec": 70568.17098045349, "step_time_sec": 8.238336071022786, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.020158767700195, "lr": 0.00019780699540262683, "elapsed_sec": 70576.41069173813, "step_time_sec": 8.239503181015607, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.853602886199951, "lr": 0.00019780405794370947, "elapsed_sec": 70584.65197181702, "step_time_sec": 8.241200169024523, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9610819816589355, "lr": 0.00019780111854119942, "elapsed_sec": 70592.89023256302, "step_time_sec": 8.238077777961735, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.01049280166626, "lr": 0.0001977981771951566, "elapsed_sec": 70601.12896847725, "step_time_sec": 8.238531073962804, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0296173095703125, "lr": 0.00019779523390564102, "elapsed_sec": 70609.36896133423, "step_time_sec": 8.239836173015647, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.964823007583618, "lr": 0.00019779228867271268, "elapsed_sec": 70617.60733509064, "step_time_sec": 8.238250456983224, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9414641857147217, "lr": 0.0001977893414964317, "elapsed_sec": 70625.84796357155, "step_time_sec": 8.240422616014257, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.76320219039917, "lr": 0.00019778639237685814, "elapsed_sec": 70634.08686876297, "step_time_sec": 8.23872218595352, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.08675479888916, "lr": 0.00019778344131405216, "elapsed_sec": 70642.32630634308, "step_time_sec": 8.239262824994512, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1360859870910645, "lr": 0.0001977804883080739, "elapsed_sec": 70650.56610035896, "step_time_sec": 8.239646917965729, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.958098888397217, "lr": 0.00019777753335898363, "elapsed_sec": 70658.8051276207, "step_time_sec": 8.238877205003519, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9916012287139893, "lr": 0.0001977745764668416, "elapsed_sec": 70667.0451540947, "step_time_sec": 30.906441764032934, "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": 4.132518768310547, "lr": 0.00019777161763170808, "elapsed_sec": 70697.96860861778, "step_time_sec": 8.256375917990226, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.005486488342285, "lr": 0.0001977686568536435, "elapsed_sec": 70706.20454525948, "step_time_sec": 8.235751083004288, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.071681976318359, "lr": 0.00019776569413270813, "elapsed_sec": 70714.44298768044, "step_time_sec": 8.238354306959081, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.030925273895264, "lr": 0.00019776272946896246, "elapsed_sec": 70722.68182015419, "step_time_sec": 8.238606523023918, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9675791263580322, "lr": 0.00019775976286246692, "elapsed_sec": 70730.91910910606, "step_time_sec": 8.237161585013382, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.957789182662964, "lr": 0.00019775679431328203, "elapsed_sec": 70739.15957450867, "step_time_sec": 8.240319124015514, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9730236530303955, "lr": 0.00019775382382146832, "elapsed_sec": 70747.39904117584, "step_time_sec": 8.239274722000118, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.135871887207031, "lr": 0.00019775085138708638, "elapsed_sec": 70755.63813138008, "step_time_sec": 8.23892919399077, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.195960521697998, "lr": 0.0001977478770101968, "elapsed_sec": 70763.87763381004, "step_time_sec": 8.239332994970027, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.122639179229736, "lr": 0.0001977449006908603, "elapsed_sec": 70772.11383652687, "step_time_sec": 8.236044935008977, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.032736301422119, "lr": 0.00019774192242913747, "elapsed_sec": 70780.35359287262, "step_time_sec": 8.23962606297573, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8934993743896484, "lr": 0.00019773894222508917, "elapsed_sec": 70788.59266424179, "step_time_sec": 8.23893951100763, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9688682556152344, "lr": 0.00019773596007877607, "elapsed_sec": 70796.83267235756, "step_time_sec": 8.23978415800957, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.036070346832275, "lr": 0.00019773297599025907, "elapsed_sec": 70805.07294678688, "step_time_sec": 8.240119222027715, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8859729766845703, "lr": 0.00019772998995959897, "elapsed_sec": 70813.30983614922, "step_time_sec": 8.236744714959059, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8509528636932373, "lr": 0.0001977270019868567, "elapsed_sec": 70821.5498137474, "step_time_sec": 8.239782959979493, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9441721439361572, "lr": 0.00019772401207209317, "elapsed_sec": 70829.78801298141, "step_time_sec": 8.238039153045975, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9295740127563477, "lr": 0.0001977210202153694, "elapsed_sec": 70838.02632164955, "step_time_sec": 8.238160111010075, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.094825744628906, "lr": 0.00019771802641674636, "elapsed_sec": 70846.2658522129, "step_time_sec": 8.239361991989426, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0524373054504395, "lr": 0.00019771503067628508, "elapsed_sec": 70854.50371980667, "step_time_sec": 8.237744870013557, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.296297073364258, "lr": 0.0001977120329940467, "elapsed_sec": 70862.74432945251, "step_time_sec": 8.240388227044605, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.991168260574341, "lr": 0.0001977090333700924, "elapsed_sec": 70870.98377919197, "step_time_sec": 8.239296452025883, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0234222412109375, "lr": 0.00019770603180448328, "elapsed_sec": 70879.222209692, "step_time_sec": 8.238325999991503, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.109086990356445, "lr": 0.00019770302829728054, "elapsed_sec": 70887.46193599701, "step_time_sec": 8.239535942964721, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9535715579986572, "lr": 0.0001977000228485455, "elapsed_sec": 70895.70058226585, "step_time_sec": 8.238466595008504, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.978050708770752, "lr": 0.00019769701545833938, "elapsed_sec": 70903.9405040741, "step_time_sec": 8.239788676961325, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9765255451202393, "lr": 0.00019769400612672355, "elapsed_sec": 70912.18013095856, "step_time_sec": 8.239423409977462, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.942861318588257, "lr": 0.0001976909948537594, "elapsed_sec": 70920.41903686523, "step_time_sec": 8.23877547297161, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9709625244140625, "lr": 0.00019768798163950833, "elapsed_sec": 70928.65943169594, "step_time_sec": 8.240188238036353, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.915346384048462, "lr": 0.00019768496648403175, "elapsed_sec": 70936.89795041084, "step_time_sec": 8.23835571797099, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.939253091812134, "lr": 0.00019768194938739116, "elapsed_sec": 70945.13710141182, "step_time_sec": 8.23901369003579, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.953043222427368, "lr": 0.00019767893034964814, "elapsed_sec": 70953.37669634819, "step_time_sec": 8.23940721101826, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.894005537033081, "lr": 0.00019767590937086422, "elapsed_sec": 70961.61502146721, "step_time_sec": 8.238139376975596, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.924309253692627, "lr": 0.00019767288645110103, "elapsed_sec": 70969.85546684265, "step_time_sec": 8.24028555402765, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.891953468322754, "lr": 0.0001976698615904202, "elapsed_sec": 70978.09563565254, "step_time_sec": 8.240010485984385, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9992992877960205, "lr": 0.0001976668347888834, "elapsed_sec": 70986.3346915245, "step_time_sec": 8.238932280044537, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.978868007659912, "lr": 0.0001976638060465524, "elapsed_sec": 70994.57308959961, "step_time_sec": 8.238229992974084, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.845621109008789, "lr": 0.00019766077536348894, "elapsed_sec": 71002.81076669693, "step_time_sec": 8.237418311007787, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8721816539764404, "lr": 0.00019765774273975483, "elapsed_sec": 71011.0494992733, "step_time_sec": 8.238584194972645, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.993739604949951, "lr": 0.0001976547081754119, "elapsed_sec": 71019.28895640373, "step_time_sec": 8.239326826995239, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.986706256866455, "lr": 0.0001976516716705221, "elapsed_sec": 71027.52811980247, "step_time_sec": 8.238968764024321, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.004554748535156, "lr": 0.00019764863322514726, "elapsed_sec": 71035.76823401451, "step_time_sec": 8.239957201993093, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.893526792526245, "lr": 0.00019764559283934943, "elapsed_sec": 71044.00568151474, "step_time_sec": 8.237319382023998, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8631317615509033, "lr": 0.00019764255051319058, "elapsed_sec": 71052.24560809135, "step_time_sec": 8.239758233015891, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9445037841796875, "lr": 0.0001976395062467327, "elapsed_sec": 71060.48529720306, "step_time_sec": 8.239535900007468, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.038633346557617, "lr": 0.000197636460040038, "elapsed_sec": 71068.7229321003, "step_time_sec": 8.237437516974751, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.938842535018921, "lr": 0.00019763341189316848, "elapsed_sec": 71076.96316075325, "step_time_sec": 8.240050970984157, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8827054500579834, "lr": 0.00019763036180618637, "elapsed_sec": 71085.20267343521, "step_time_sec": 8.239346153044607, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9513726234436035, "lr": 0.00019762730977915388, "elapsed_sec": 71093.44247531891, "step_time_sec": 8.239652586984448, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.124321937561035, "lr": 0.0001976242558121332, "elapsed_sec": 71101.68327021599, "step_time_sec": 8.240622967015952, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.146805763244629, "lr": 0.0001976211999051866, "elapsed_sec": 71109.92087125778, "step_time_sec": 8.237414631992579, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.989283800125122, "lr": 0.0001976181420583765, "elapsed_sec": 71118.16079902649, "step_time_sec": 8.23978426400572, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9167070388793945, "lr": 0.00019761508227176516, "elapsed_sec": 71126.3991997242, "step_time_sec": 8.238256544980686, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.971914052963257, "lr": 0.00019761202054541506, "elapsed_sec": 71134.63787078857, "step_time_sec": 8.238516282988712, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9809839725494385, "lr": 0.00019760895687938855, "elapsed_sec": 71142.87737035751, "step_time_sec": 8.239351635042112, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.858712673187256, "lr": 0.0001976058912737482, "elapsed_sec": 71151.1156103611, "step_time_sec": 8.23801813396858, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.010191917419434, "lr": 0.00019760282372855646, "elapsed_sec": 71159.35532069206, "step_time_sec": 8.239621974993497, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9995691776275635, "lr": 0.0001975997542438759, "elapsed_sec": 71167.59453201294, "step_time_sec": 8.23896186298225, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.848912000656128, "lr": 0.00019759668281976915, "elapsed_sec": 71175.83307909966, "step_time_sec": 8.23839383298764, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9237544536590576, "lr": 0.00019759360945629885, "elapsed_sec": 71184.07245087624, "step_time_sec": 8.239152994006872, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.112552642822266, "lr": 0.00019759053415352765, "elapsed_sec": 71192.31060361862, "step_time_sec": 8.238004806044046, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.993830919265747, "lr": 0.00019758745691151827, "elapsed_sec": 71200.55049777031, "step_time_sec": 8.23978751100367, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9507434368133545, "lr": 0.00019758437773033347, "elapsed_sec": 71208.79025959969, "step_time_sec": 8.239592147001531, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.027410507202148, "lr": 0.00019758129661003604, "elapsed_sec": 71217.0285410881, "step_time_sec": 8.238094807020389, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.022548198699951, "lr": 0.00019757821355068884, "elapsed_sec": 71225.26772284508, "step_time_sec": 8.239021293993574, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.027929782867432, "lr": 0.00019757512855235468, "elapsed_sec": 71233.50468349457, "step_time_sec": 8.236820757039823, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.033995628356934, "lr": 0.00019757204161509654, "elapsed_sec": 71241.74488067627, "step_time_sec": 8.24001536500873, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9906811714172363, "lr": 0.00019756895273897738, "elapsed_sec": 71249.98469567299, "step_time_sec": 8.239637537044473, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.188595771789551, "lr": 0.00019756586192406014, "elapsed_sec": 71258.22266864777, "step_time_sec": 8.237824288022239, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.02266788482666, "lr": 0.0001975627691704079, "elapsed_sec": 71266.46189880371, "step_time_sec": 8.239040791988373, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.967125415802002, "lr": 0.0001975596744780837, "elapsed_sec": 71274.6986951828, "step_time_sec": 8.236711612029467, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.087804794311523, "lr": 0.00019755657784715065, "elapsed_sec": 71282.93823361397, "step_time_sec": 8.239312220015563, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.17816686630249, "lr": 0.00019755347927767194, "elapsed_sec": 71291.17716908455, "step_time_sec": 8.238766207010485, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0349626541137695, "lr": 0.0001975503787697107, "elapsed_sec": 71299.41549396515, "step_time_sec": 8.238192792981863, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9728779792785645, "lr": 0.00019754727632333022, "elapsed_sec": 71307.6559138298, "step_time_sec": 8.240277100994717, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0074944496154785, "lr": 0.00019754417193859372, "elapsed_sec": 71315.89396309853, "step_time_sec": 8.237901766959112, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8759913444519043, "lr": 0.00019754106561556457, "elapsed_sec": 71324.1337378025, "step_time_sec": 8.23962870601099, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.08576774597168, "lr": 0.00019753795735430612, "elapsed_sec": 71332.3732213974, "step_time_sec": 8.239294892002363, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1500725746154785, "lr": 0.00019753484715488164, "elapsed_sec": 71340.61248469353, "step_time_sec": 8.239141477970406, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.128258228302002, "lr": 0.00019753173501735468, "elapsed_sec": 71348.85244965553, "step_time_sec": 8.239696557982825, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.009453296661377, "lr": 0.0001975286209417887, "elapsed_sec": 71357.09159040451, "step_time_sec": 8.23900565499207, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.125479698181152, "lr": 0.00019752550492824713, "elapsed_sec": 71365.3313934803, "step_time_sec": 8.239655257028062, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0280256271362305, "lr": 0.0001975223869767936, "elapsed_sec": 71373.57095742226, "step_time_sec": 8.239371116971597, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.113317012786865, "lr": 0.0001975192670874916, "elapsed_sec": 71381.80885696411, "step_time_sec": 8.237739063042682, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.047945499420166, "lr": 0.0001975161452604049, "elapsed_sec": 71390.04964280128, "step_time_sec": 8.240610224020202, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.051123142242432, "lr": 0.00019751302149559704, "elapsed_sec": 71398.2884414196, "step_time_sec": 8.23863530397648, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1101179122924805, "lr": 0.00019750989579313174, "elapsed_sec": 71406.52667808533, "step_time_sec": 8.238122698967345, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9929654598236084, "lr": 0.0001975067681530728, "elapsed_sec": 71414.76721358299, "step_time_sec": 8.240347099024802, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.16916036605835, "lr": 0.00019750363857548396, "elapsed_sec": 71423.00458192825, "step_time_sec": 8.237156916991808, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.994952917098999, "lr": 0.00019750050706042908, "elapsed_sec": 71431.24314141273, "step_time_sec": 8.238387079036329, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.081553936004639, "lr": 0.000197497373607972, "elapsed_sec": 71439.4825565815, "step_time_sec": 8.239257887995336, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.270846843719482, "lr": 0.00019749423821817657, "elapsed_sec": 71447.71996355057, "step_time_sec": 8.237274963990785, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.855515718460083, "lr": 0.00019749110089110685, "elapsed_sec": 71455.95920538902, "step_time_sec": 8.239073646022007, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.967067003250122, "lr": 0.0001974879616268267, "elapsed_sec": 71464.19757246971, "step_time_sec": 8.238195214013103, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9614593982696533, "lr": 0.00019748482042540025, "elapsed_sec": 71472.43629407883, "step_time_sec": 8.238506602006964, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.06135368347168, "lr": 0.00019748167728689148, "elapsed_sec": 71480.67535924911, "step_time_sec": 8.238956660963595, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.111574172973633, "lr": 0.0001974785322113645, "elapsed_sec": 71488.91401124, "step_time_sec": 8.238443984009791, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9651706218719482, "lr": 0.00019747538519888352, "elapsed_sec": 71497.15241789818, "step_time_sec": 8.238237304030918, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9699161052703857, "lr": 0.0001974722362495126, "elapsed_sec": 71505.3899178505, "step_time_sec": 8.237372117990162, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.983936071395874, "lr": 0.00019746908536331608, "elapsed_sec": 71513.62881803513, "step_time_sec": 8.238699754991103, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9330549240112305, "lr": 0.00019746593254035813, "elapsed_sec": 71521.86798596382, "step_time_sec": 8.239013247017283, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9523208141326904, "lr": 0.0001974627777807031, "elapsed_sec": 71530.1062567234, "step_time_sec": 8.238105345983058, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.064650058746338, "lr": 0.00019745962108441527, "elapsed_sec": 71538.34609651566, "step_time_sec": 8.239667246001773, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9218008518218994, "lr": 0.00019745646245155908, "elapsed_sec": 71546.58350872993, "step_time_sec": 8.237275078019593, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.090191841125488, "lr": 0.0001974533018821989, "elapsed_sec": 71554.82358121872, "step_time_sec": 8.239885606977623, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.020819187164307, "lr": 0.0001974501393763992, "elapsed_sec": 71563.06283569336, "step_time_sec": 8.239107228000648, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.066112041473389, "lr": 0.0001974469749342245, "elapsed_sec": 71571.30038762093, "step_time_sec": 8.237355141027365, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.948715925216675, "lr": 0.00019744380855573932, "elapsed_sec": 71579.53957438469, "step_time_sec": 8.239040312997531, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.984174966812134, "lr": 0.0001974406402410082, "elapsed_sec": 71587.77776575089, "step_time_sec": 8.2380103119649, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.053863048553467, "lr": 0.0001974374699900958, "elapsed_sec": 71596.01789522171, "step_time_sec": 8.240016928000841, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8175415992736816, "lr": 0.00019743429780306673, "elapsed_sec": 71604.25774478912, "step_time_sec": 8.239694549003616, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.025447368621826, "lr": 0.00019743112367998574, "elapsed_sec": 71612.49653124809, "step_time_sec": 8.238542610954028, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.981618881225586, "lr": 0.0001974279476209175, "elapsed_sec": 71620.73547101021, "step_time_sec": 8.23883578600362, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9274747371673584, "lr": 0.0001974247696259268, "elapsed_sec": 71628.97545337677, "step_time_sec": 8.239832144987304, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9325907230377197, "lr": 0.0001974215896950785, "elapsed_sec": 71637.21524572372, "step_time_sec": 8.23963124200236, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.873703718185425, "lr": 0.00019741840782843738, "elapsed_sec": 71645.45361924171, "step_time_sec": 8.23816505796276, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9257214069366455, "lr": 0.00019741522402606836, "elapsed_sec": 71653.69252085686, "step_time_sec": 8.238763470028061, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.185261249542236, "lr": 0.00019741203828803637, "elapsed_sec": 71661.93209552765, "step_time_sec": 8.239412063965574, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.063347816467285, "lr": 0.00019740885061440638, "elapsed_sec": 71670.17191934586, "step_time_sec": 8.239620071020909, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8728721141815186, "lr": 0.00019740566100524338, "elapsed_sec": 71678.40983939171, "step_time_sec": 8.237757291004527, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.980116605758667, "lr": 0.00019740246946061243, "elapsed_sec": 71686.64900875092, "step_time_sec": 8.238976850989275, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8889012336730957, "lr": 0.00019739927598057865, "elapsed_sec": 71694.88686418533, "step_time_sec": 8.237741092976648, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.98715877532959, "lr": 0.0001973960805652071, "elapsed_sec": 71703.1268594265, "step_time_sec": 8.239797105954494, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8687422275543213, "lr": 0.00019739288321456302, "elapsed_sec": 71711.36563205719, "step_time_sec": 8.238571713038255, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.018735408782959, "lr": 0.00019738968392871156, "elapsed_sec": 71719.6041560173, "step_time_sec": 8.238352967018727, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.922302484512329, "lr": 0.00019738648270771798, "elapsed_sec": 71727.84289693832, "step_time_sec": 8.238606721977703, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9565463066101074, "lr": 0.00019738327955164757, "elapsed_sec": 71736.0803000927, "step_time_sec": 8.237276106956415, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9983620643615723, "lr": 0.00019738007446056565, "elapsed_sec": 71744.31937289238, "step_time_sec": 8.23889663699083, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.015885829925537, "lr": 0.0001973768674345376, "elapsed_sec": 71752.5582036972, "step_time_sec": 8.238594627007842, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.096930980682373, "lr": 0.00019737365847362877, "elapsed_sec": 71760.79679989815, "step_time_sec": 8.238489055016544, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.116652488708496, "lr": 0.00019737044757790466, "elapsed_sec": 71769.036506176, "step_time_sec": 8.239528578007594, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.915846586227417, "lr": 0.00019736723474743075, "elapsed_sec": 71777.27409434319, "step_time_sec": 8.237447805004194, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.01417350769043, "lr": 0.0001973640199822725, "elapsed_sec": 71785.51206922531, "step_time_sec": 8.237760023970623, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.194796562194824, "lr": 0.00019736080328249555, "elapsed_sec": 71793.74963617325, "step_time_sec": 8.237413797003683, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9929232597351074, "lr": 0.00019735758464816547, "elapsed_sec": 71801.98948121071, "step_time_sec": 8.239750126027502, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9575953483581543, "lr": 0.00019735436407934784, "elapsed_sec": 71810.22831106186, "step_time_sec": 8.2386182390037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9605774879455566, "lr": 0.00019735114157610844, "elapsed_sec": 71818.46529364586, "step_time_sec": 8.236808698042296, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9715659618377686, "lr": 0.0001973479171385129, "elapsed_sec": 71826.70430898666, "step_time_sec": 8.238908224971965, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0498809814453125, "lr": 0.00019734469076662704, "elapsed_sec": 71834.94246125221, "step_time_sec": 8.237940989027265, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.948580026626587, "lr": 0.00019734146246051665, "elapsed_sec": 71843.18110871315, "step_time_sec": 8.23852001799969, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.095165252685547, "lr": 0.00019733823222024753, "elapsed_sec": 71851.41999292374, "step_time_sec": 8.23868875700282, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.040938377380371, "lr": 0.00019733500004588558, "elapsed_sec": 71859.65679502487, "step_time_sec": 8.236630180035718, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.045351505279541, "lr": 0.00019733176593749675, "elapsed_sec": 71867.89620614052, "step_time_sec": 8.239269388956018, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.961190700531006, "lr": 0.00019732852989514692, "elapsed_sec": 71876.13481712341, "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": 8645, "loss": 4.081014156341553, "lr": 0.00019732529191890212, "elapsed_sec": 71884.37435078621, "step_time_sec": 8.23936736700125, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.983388662338257, "lr": 0.00019732205200882842, "elapsed_sec": 71892.61232233047, "step_time_sec": 8.237796865985729, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9766016006469727, "lr": 0.00019731881016499183, "elapsed_sec": 71900.8506000042, "step_time_sec": 8.238076962996274, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.966855525970459, "lr": 0.0001973155663874585, "elapsed_sec": 71909.08996415138, "step_time_sec": 8.239234426000621, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.632573127746582, "lr": 0.00019731232067629457, "elapsed_sec": 71917.32870745659, "step_time_sec": 8.238583433965687, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.095155239105225, "lr": 0.00019730907303156626, "elapsed_sec": 71925.56749153137, "step_time_sec": 8.238602027005982, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.154163360595703, "lr": 0.00019730582345333975, "elapsed_sec": 71933.80713367462, "step_time_sec": 8.23946121003246, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1209893226623535, "lr": 0.00019730257194168136, "elapsed_sec": 71942.04429721832, "step_time_sec": 8.23701489000814, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.849937915802002, "lr": 0.00019729931849665736, "elapsed_sec": 71950.28408527374, "step_time_sec": 8.239642653963529, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.127583026885986, "lr": 0.00019729606311833414, "elapsed_sec": 71958.52346992493, "step_time_sec": 8.239217332971748, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.094069957733154, "lr": 0.00019729280580677806, "elapsed_sec": 71966.76316380501, "step_time_sec": 8.23952843999723, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.210899829864502, "lr": 0.00019728954656205556, "elapsed_sec": 71975.00204229355, "step_time_sec": 8.238678405992687, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.073883056640625, "lr": 0.00019728628538423309, "elapsed_sec": 71983.23910236359, "step_time_sec": 8.236948625009973, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9862735271453857, "lr": 0.00019728302227337718, "elapsed_sec": 71991.47881221771, "step_time_sec": 8.239520768984221, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.967865228652954, "lr": 0.0001972797572295544, "elapsed_sec": 71999.71818566322, "step_time_sec": 8.23918190097902, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.099422931671143, "lr": 0.00019727649025283125, "elapsed_sec": 72007.95770215988, "step_time_sec": 8.23937104397919, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.958043098449707, "lr": 0.00019727322134327442, "elapsed_sec": 72016.19730067253, "step_time_sec": 8.239456214010715, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8363213539123535, "lr": 0.00019726995050095057, "elapsed_sec": 72024.4348499775, "step_time_sec": 8.237397688033525, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0094122886657715, "lr": 0.00019726667772592644, "elapsed_sec": 72032.67409014702, "step_time_sec": 8.239003843977116, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.941166639328003, "lr": 0.00019726340301826867, "elapsed_sec": 72040.91170096397, "step_time_sec": 8.237465142970905, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9643239974975586, "lr": 0.00019726012637804414, "elapsed_sec": 72049.1503329277, "step_time_sec": 8.23852347698994, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8964836597442627, "lr": 0.00019725684780531964, "elapsed_sec": 72057.38968634605, "step_time_sec": 8.239169580978341, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9948854446411133, "lr": 0.000197253567300162, "elapsed_sec": 72065.62670516968, "step_time_sec": 8.236838752985932, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7654130458831787, "lr": 0.00019725028486263817, "elapsed_sec": 72073.86627864838, "step_time_sec": 8.239436010015197, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.982416868209839, "lr": 0.00019724700049281505, "elapsed_sec": 72082.10363197327, "step_time_sec": 8.237186138983816, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.947575092315674, "lr": 0.00019724371419075967, "elapsed_sec": 72090.34260773659, "step_time_sec": 8.238787384994794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.886683464050293, "lr": 0.000197240425956539, "elapsed_sec": 72098.58229088783, "step_time_sec": 8.23951493500499, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.868267059326172, "lr": 0.00019723713579022014, "elapsed_sec": 72106.81942176819, "step_time_sec": 8.236962440016214, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9915876388549805, "lr": 0.00019723384369187017, "elapsed_sec": 72115.05886888504, "step_time_sec": 8.239289969031233, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.049576282501221, "lr": 0.0001972305496615562, "elapsed_sec": 72123.29855680466, "step_time_sec": 8.239532643987332, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9963252544403076, "lr": 0.00019722725369934545, "elapsed_sec": 72131.53825688362, "step_time_sec": 8.239563189039472, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.987229585647583, "lr": 0.0001972239558053051, "elapsed_sec": 72139.77747273445, "step_time_sec": 8.238988571974915, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9910080432891846, "lr": 0.00019722065597950245, "elapsed_sec": 72148.01524114609, "step_time_sec": 8.237648701993749, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.897193670272827, "lr": 0.00019721735422200475, "elapsed_sec": 72156.25564813614, "step_time_sec": 8.24026811501244, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9380903244018555, "lr": 0.00019721405053287937, "elapsed_sec": 72164.49455356598, "step_time_sec": 8.238741516019218, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.859483003616333, "lr": 0.00019721074491219366, "elapsed_sec": 72172.73398566246, "step_time_sec": 8.239226214995142, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9434945583343506, "lr": 0.00019720743736001506, "elapsed_sec": 72180.97451066971, "step_time_sec": 8.240349236992188, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9564826488494873, "lr": 0.000197204127876411, "elapsed_sec": 72189.21299958229, "step_time_sec": 8.238356983056292, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8252267837524414, "lr": 0.00019720081646144898, "elapsed_sec": 72197.45255684853, "step_time_sec": 8.23941893700976, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7183799743652344, "lr": 0.0001971975031151965, "elapsed_sec": 72205.69119668007, "step_time_sec": 8.238442272995599, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.014428615570068, "lr": 0.0001971941878377212, "elapsed_sec": 72213.92969202995, "step_time_sec": 8.238329635001719, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0291290283203125, "lr": 0.00019719087062909063, "elapsed_sec": 72222.16938257217, "step_time_sec": 8.239572354010306, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.201815128326416, "lr": 0.00019718755148937246, "elapsed_sec": 72230.40738224983, "step_time_sec": 8.2378095969907, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.014715194702148, "lr": 0.00019718423041863438, "elapsed_sec": 72238.64855885506, "step_time_sec": 8.240977476991247, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9499669075012207, "lr": 0.00019718090741694414, "elapsed_sec": 72246.88806843758, "step_time_sec": 8.239344881032594, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.950338125228882, "lr": 0.00019717758248436948, "elapsed_sec": 72255.12554574013, "step_time_sec": 8.237373513984494, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8841068744659424, "lr": 0.0001971742556209782, "elapsed_sec": 72263.3659684658, "step_time_sec": 8.240207750000991, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8698229789733887, "lr": 0.0001971709268268382, "elapsed_sec": 72271.60510063171, "step_time_sec": 8.238988104974851, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.948779582977295, "lr": 0.00019716759610201727, "elapsed_sec": 72279.84440755844, "step_time_sec": 8.239089899987448, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.032359600067139, "lr": 0.00019716426344658342, "elapsed_sec": 72288.0830206871, "step_time_sec": 8.238464010006282, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9601361751556396, "lr": 0.0001971609288606046, "elapsed_sec": 72296.32090497017, "step_time_sec": 8.237765476980712, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.989097833633423, "lr": 0.00019715759234414877, "elapsed_sec": 72304.5594136715, "step_time_sec": 8.238333610002883, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9506332874298096, "lr": 0.000197154253897284, "elapsed_sec": 72312.7989525795, "step_time_sec": 8.239324543974362, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.861182928085327, "lr": 0.00019715091352007842, "elapsed_sec": 72321.03722500801, "step_time_sec": 8.238117408007383, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.186324596405029, "lr": 0.0001971475712126001, "elapsed_sec": 72329.27616643906, "step_time_sec": 8.238813780015334, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8543522357940674, "lr": 0.0001971442269749172, "elapsed_sec": 72337.51249098778, "step_time_sec": 8.236177842016332, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8260419368743896, "lr": 0.00019714088080709794, "elapsed_sec": 72345.75370407104, "step_time_sec": 8.240996963984799, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9995365142822266, "lr": 0.00019713753270921052, "elapsed_sec": 72353.99323964119, "step_time_sec": 8.239367470960133, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9889450073242188, "lr": 0.0001971341826813233, "elapsed_sec": 72362.2325387001, "step_time_sec": 8.239122403028887, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.995671272277832, "lr": 0.00019713083072350453, "elapsed_sec": 72370.47224521637, "step_time_sec": 8.239531634026207, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.121635437011719, "lr": 0.0001971274768358226, "elapsed_sec": 72378.71018767357, "step_time_sec": 8.237767816986889, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7934763431549072, "lr": 0.00019712412101834592, "elapsed_sec": 72386.95013928413, "step_time_sec": 8.239783316035755, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9769818782806396, "lr": 0.00019712076327114283, "elapsed_sec": 72395.18963122368, "step_time_sec": 8.239323390997015, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8802969455718994, "lr": 0.00019711740359428195, "elapsed_sec": 72403.42754936218, "step_time_sec": 8.237753241963219, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.332512378692627, "lr": 0.00019711404198783172, "elapsed_sec": 72411.66609358788, "step_time_sec": 8.23844959505368, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.990304470062256, "lr": 0.0001971106784518607, "elapsed_sec": 72419.90400147438, "step_time_sec": 8.237662753963377, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9631502628326416, "lr": 0.00019710731298643746, "elapsed_sec": 72428.21286916733, "step_time_sec": 8.246693629014771, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.006702899932861, "lr": 0.0001971039455916307, "elapsed_sec": 72436.45124292374, "step_time_sec": 8.23822573700454, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.016692161560059, "lr": 0.00019710057626750902, "elapsed_sec": 72444.68954730034, "step_time_sec": 8.238147104042582, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9368512630462646, "lr": 0.00019709720501414122, "elapsed_sec": 72452.92751193047, "step_time_sec": 8.237774982990231, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.001521587371826, "lr": 0.00019709383183159595, "elapsed_sec": 72461.16697335243, "step_time_sec": 8.239266368036624, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9839670658111572, "lr": 0.00019709045671994206, "elapsed_sec": 72469.40558099747, "step_time_sec": 8.238467395014595, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8677356243133545, "lr": 0.0001970870796792484, "elapsed_sec": 72477.6461801529, "step_time_sec": 8.240463143971283, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.102250099182129, "lr": 0.00019708370070958378, "elapsed_sec": 72485.88548779488, "step_time_sec": 8.239092306990642, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.978381872177124, "lr": 0.00019708031981101716, "elapsed_sec": 72494.1241850853, "step_time_sec": 8.23857525701169, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.002940654754639, "lr": 0.00019707693698361748, "elapsed_sec": 72502.36465334892, "step_time_sec": 8.240255094016902, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.112248420715332, "lr": 0.0001970735522274537, "elapsed_sec": 72510.60375642776, "step_time_sec": 8.23896176397102, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.03501558303833, "lr": 0.00019707016554259486, "elapsed_sec": 72518.84254336357, "step_time_sec": 8.238635187968612, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9285786151885986, "lr": 0.00019706677692911005, "elapsed_sec": 72527.08174848557, "step_time_sec": 8.239032951008994, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8148043155670166, "lr": 0.00019706338638706834, "elapsed_sec": 72535.31938123703, "step_time_sec": 8.237549036974087, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.996046781539917, "lr": 0.0001970599939165389, "elapsed_sec": 72543.5601670742, "step_time_sec": 8.24058755702572, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9978792667388916, "lr": 0.00019705659951759093, "elapsed_sec": 72551.80089139938, "step_time_sec": 8.240579908015206, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7756810188293457, "lr": 0.00019705320319029362, "elapsed_sec": 72560.04070162773, "step_time_sec": 8.239591062010732, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8554489612579346, "lr": 0.00019704980493471624, "elapsed_sec": 72568.28044199944, "step_time_sec": 8.239599939028267, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9306318759918213, "lr": 0.0001970464047509281, "elapsed_sec": 72576.51851606369, "step_time_sec": 8.237918096012436, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.901839017868042, "lr": 0.00019704300263899852, "elapsed_sec": 72584.75714993477, "step_time_sec": 8.23843728902284, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.007005214691162, "lr": 0.00019703959859899693, "elapsed_sec": 72592.99628663063, "step_time_sec": 8.23896909801988, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0627241134643555, "lr": 0.00019703619263099272, "elapsed_sec": 72601.2342607975, "step_time_sec": 8.237785473989788, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8659920692443848, "lr": 0.00019703278473505535, "elapsed_sec": 72609.47461748123, "step_time_sec": 8.240212055039592, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8564629554748535, "lr": 0.0001970293749112543, "elapsed_sec": 72617.7134642601, "step_time_sec": 8.238691134029068, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.899839162826538, "lr": 0.00019702596315965913, "elapsed_sec": 72625.95209169388, "step_time_sec": 8.238453860976733, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8823487758636475, "lr": 0.00019702254948033944, "elapsed_sec": 72634.19155597687, "step_time_sec": 8.239377210964449, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.907799482345581, "lr": 0.00019701913387336483, "elapsed_sec": 72642.42909884453, "step_time_sec": 8.237366169982124, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.017571926116943, "lr": 0.00019701571633880493, "elapsed_sec": 72650.66887259483, "step_time_sec": 8.239578250970226, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7309632301330566, "lr": 0.00019701229687672946, "elapsed_sec": 72658.90825605392, "step_time_sec": 8.239271664002445, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8258464336395264, "lr": 0.00019700887548720818, "elapsed_sec": 72667.14788413048, "step_time_sec": 8.239457625022624, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8885717391967773, "lr": 0.00019700545217031084, "elapsed_sec": 72675.38730740547, "step_time_sec": 8.23923945397837, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7901597023010254, "lr": 0.0001970020269261072, "elapsed_sec": 72683.625613451, "step_time_sec": 8.238113348023035, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.090856552124023, "lr": 0.0001969985997546672, "elapsed_sec": 72691.86442565918, "step_time_sec": 8.238700105983298, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9480185508728027, "lr": 0.00019699517065606067, "elapsed_sec": 72700.10374546051, "step_time_sec": 8.239134686999023, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7008655071258545, "lr": 0.0001969917396303576, "elapsed_sec": 72708.34166264534, "step_time_sec": 8.237758492992725, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.029067039489746, "lr": 0.00019698830667762792, "elapsed_sec": 72716.58114099503, "step_time_sec": 8.239329523989, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.084933757781982, "lr": 0.00019698487179794163, "elapsed_sec": 72724.81941366196, "step_time_sec": 8.238057087990455, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9814281463623047, "lr": 0.00019698143499136884, "elapsed_sec": 72733.05996394157, "step_time_sec": 8.240375577006489, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9833171367645264, "lr": 0.00019697799625797955, "elapsed_sec": 72741.29839515686, "step_time_sec": 8.238240128965117, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.926907777786255, "lr": 0.00019697455559784395, "elapsed_sec": 72749.5375623703, "step_time_sec": 8.238994088955224, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.034268379211426, "lr": 0.00019697111301103222, "elapsed_sec": 72757.77561306953, "step_time_sec": 8.237906966998708, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.918370485305786, "lr": 0.00019696766849761448, "elapsed_sec": 72766.01315760612, "step_time_sec": 8.2373767690151, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.897737741470337, "lr": 0.00019696422205766107, "elapsed_sec": 72774.25204062462, "step_time_sec": 8.23876741703134, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9706950187683105, "lr": 0.00019696077369124227, "elapsed_sec": 72782.49217057228, "step_time_sec": 8.239890786004253, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.014541149139404, "lr": 0.00019695732339842836, "elapsed_sec": 72790.73055958748, "step_time_sec": 8.23826364701381, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.919870615005493, "lr": 0.0001969538711792897, "elapsed_sec": 72798.96980404854, "step_time_sec": 8.239119834033772, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.827800750732422, "lr": 0.00019695041703389672, "elapsed_sec": 72807.20668029785, "step_time_sec": 8.236679791996721, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9118871688842773, "lr": 0.00019694696096231984, "elapsed_sec": 72815.44756436348, "step_time_sec": 8.240707176970318, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.796936273574829, "lr": 0.0001969435029646296, "elapsed_sec": 72823.68784451485, "step_time_sec": 8.240097284025978, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.632542848587036, "lr": 0.00019694004304089644, "elapsed_sec": 72831.92588973045, "step_time_sec": 8.237954184005503, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.011921405792236, "lr": 0.00019693658119119097, "elapsed_sec": 72840.16564941406, "step_time_sec": 8.23956423002528, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9697389602661133, "lr": 0.0001969331174155838, "elapsed_sec": 72848.40449953079, "step_time_sec": 8.238702548027504, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.928884506225586, "lr": 0.00019692965171414552, "elapsed_sec": 72856.64413166046, "step_time_sec": 8.239471947017591, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.806795597076416, "lr": 0.00019692618408694685, "elapsed_sec": 72864.88395166397, "step_time_sec": 8.239678669022396, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9446628093719482, "lr": 0.00019692271453405848, "elapsed_sec": 72873.12261509895, "step_time_sec": 8.238477666978724, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9963009357452393, "lr": 0.00019691924305555118, "elapsed_sec": 72881.36280632019, "step_time_sec": 8.240043675003108, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9669382572174072, "lr": 0.00019691576965149576, "elapsed_sec": 72889.60184931755, "step_time_sec": 8.238825116015505, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.899052619934082, "lr": 0.00019691229432196303, "elapsed_sec": 72897.8420214653, "step_time_sec": 8.240027759980876, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9158313274383545, "lr": 0.00019690881706702385, "elapsed_sec": 72906.08156466484, "step_time_sec": 8.239394470991101, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.031892776489258, "lr": 0.0001969053378867492, "elapsed_sec": 72914.31929326057, "step_time_sec": 8.237633207987528, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9843921661376953, "lr": 0.00019690185678120998, "elapsed_sec": 72922.55914664268, "step_time_sec": 8.239588016003836, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9767961502075195, "lr": 0.00019689837375047714, "elapsed_sec": 72930.79910564423, "step_time_sec": 8.239807747013401, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.048766613006592, "lr": 0.00019689488879462183, "elapsed_sec": 72939.03768849373, "step_time_sec": 8.238461984030437, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9109601974487305, "lr": 0.00019689140191371505, "elapsed_sec": 72947.27658605576, "step_time_sec": 8.238718527019955, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9589288234710693, "lr": 0.0001968879131078279, "elapsed_sec": 72955.51428771019, "step_time_sec": 8.237507092999294, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8701486587524414, "lr": 0.00019688442237703155, "elapsed_sec": 72963.75434231758, "step_time_sec": 8.239893599995412, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9125561714172363, "lr": 0.00019688092972139717, "elapsed_sec": 72971.99523949623, "step_time_sec": 8.240790976968128, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.952308416366577, "lr": 0.00019687743514099602, "elapsed_sec": 72980.23325967789, "step_time_sec": 8.237853653961793, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.902543783187866, "lr": 0.00019687393863589934, "elapsed_sec": 72988.47313952446, "step_time_sec": 8.239677312027197, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.030396461486816, "lr": 0.00019687044020617843, "elapsed_sec": 72996.7112197876, "step_time_sec": 8.237908290000632, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9447834491729736, "lr": 0.00019686693985190468, "elapsed_sec": 73004.95033359528, "step_time_sec": 8.238958783040289, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.003999710083008, "lr": 0.00019686343757314943, "elapsed_sec": 73013.19017124176, "step_time_sec": 8.239661262952723, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.003420829772949, "lr": 0.0001968599333699841, "elapsed_sec": 73021.42887306213, "step_time_sec": 8.238504935987294, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.975955009460449, "lr": 0.00019685642724248022, "elapsed_sec": 73029.6666791439, "step_time_sec": 8.237683895044029, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.896554470062256, "lr": 0.00019685291919070922, "elapsed_sec": 73037.90570259094, "step_time_sec": 8.238833270035684, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9322104454040527, "lr": 0.00019684940921474266, "elapsed_sec": 73046.14592385292, "step_time_sec": 8.240045254991855, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8305208683013916, "lr": 0.00019684589731465212, "elapsed_sec": 73054.38511824608, "step_time_sec": 8.23909986403305, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.007645130157471, "lr": 0.00019684238349050926, "elapsed_sec": 73062.62397408485, "step_time_sec": 8.238668380014133, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9377901554107666, "lr": 0.0001968388677423857, "elapsed_sec": 73070.86409497261, "step_time_sec": 8.239913057012018, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.933258533477783, "lr": 0.00019683535007035313, "elapsed_sec": 73079.103110075, "step_time_sec": 8.23891471803654, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.181575775146484, "lr": 0.0001968318304744833, "elapsed_sec": 73087.34095835686, "step_time_sec": 8.237659447011538, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.017903804779053, "lr": 0.00019682830895484803, "elapsed_sec": 73095.57956886292, "step_time_sec": 8.23838524200255, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.950995445251465, "lr": 0.00019682478551151904, "elapsed_sec": 73103.81784844398, "step_time_sec": 8.238131607999094, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9212119579315186, "lr": 0.0001968212601445683, "elapsed_sec": 73112.05744552612, "step_time_sec": 8.239485118014272, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9413533210754395, "lr": 0.00019681773285406762, "elapsed_sec": 73120.29726219177, "step_time_sec": 8.23962967901025, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.055429458618164, "lr": 0.00019681420364008897, "elapsed_sec": 73128.53547239304, "step_time_sec": 8.238057308015414, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.946355104446411, "lr": 0.00019681067250270433, "elapsed_sec": 73136.77525997162, "step_time_sec": 8.239621438959148, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.056571006774902, "lr": 0.00019680713944198567, "elapsed_sec": 73145.01270341873, "step_time_sec": 8.237297954969108, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8922109603881836, "lr": 0.00019680360445800509, "elapsed_sec": 73153.25177955627, "step_time_sec": 8.238883438985795, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.061653137207031, "lr": 0.00019680006755083465, "elapsed_sec": 73161.49072980881, "step_time_sec": 8.238773083023261, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.893509864807129, "lr": 0.0001967965287205465, "elapsed_sec": 73169.72894048691, "step_time_sec": 8.238098126021214, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.085799217224121, "lr": 0.0001967929879672128, "elapsed_sec": 73177.96789646149, "step_time_sec": 8.238789068011101, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.871300458908081, "lr": 0.00019678944529090577, "elapsed_sec": 73186.20377850533, "step_time_sec": 8.235660801990889, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.956660509109497, "lr": 0.00019678590069169765, "elapsed_sec": 73194.44209980965, "step_time_sec": 8.238249998015817, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9300827980041504, "lr": 0.00019678235416966072, "elapsed_sec": 73202.68152713776, "step_time_sec": 8.239171898982022, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.094172954559326, "lr": 0.0001967788057248673, "elapsed_sec": 73210.92074584961, "step_time_sec": 8.239056869002525, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9561822414398193, "lr": 0.00019677525535738975, "elapsed_sec": 73219.16057753563, "step_time_sec": 8.239694164018147, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8997209072113037, "lr": 0.00019677170306730052, "elapsed_sec": 73227.39897680283, "step_time_sec": 8.238242124964017, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.884782552719116, "lr": 0.00019676814885467205, "elapsed_sec": 73235.63792681694, "step_time_sec": 8.238721867965069, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9171526432037354, "lr": 0.00019676459271957678, "elapsed_sec": 73243.87736558914, "step_time_sec": 8.239280209993012, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8919677734375, "lr": 0.00019676103466208724, "elapsed_sec": 73252.1154537201, "step_time_sec": 8.237954160955269, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8477375507354736, "lr": 0.000196757474682276, "elapsed_sec": 73260.35329532623, "step_time_sec": 8.237684946041554, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.994257688522339, "lr": 0.00019675391278021567, "elapsed_sec": 73268.59065556526, "step_time_sec": 8.237166473991238, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8813540935516357, "lr": 0.0001967503489559789, "elapsed_sec": 73276.82826972008, "step_time_sec": 8.237464044010267, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0172648429870605, "lr": 0.00019674678320963831, "elapsed_sec": 73285.06808519363, "step_time_sec": 8.239654631004669, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8996150493621826, "lr": 0.0001967432155412667, "elapsed_sec": 73293.30655789375, "step_time_sec": 8.23829318099888, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.111760139465332, "lr": 0.00019673964595093678, "elapsed_sec": 73301.54547905922, "step_time_sec": 8.238776789978147, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9578258991241455, "lr": 0.00019673607443872136, "elapsed_sec": 73309.78324151039, "step_time_sec": 8.23755863297265, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.086950778961182, "lr": 0.00019673250100469327, "elapsed_sec": 73318.02262282372, "step_time_sec": 8.239247835997958, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8080670833587646, "lr": 0.0001967289256489254, "elapsed_sec": 73326.26171350479, "step_time_sec": 8.238934854976833, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.867974281311035, "lr": 0.00019672534837149062, "elapsed_sec": 73334.50103068352, "step_time_sec": 8.239194895024411, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9020466804504395, "lr": 0.00019672176917246193, "elapsed_sec": 73342.74088573456, "step_time_sec": 8.239662272040732, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.863764762878418, "lr": 0.0001967181880519123, "elapsed_sec": 73350.97799134254, "step_time_sec": 8.236906003032345, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8724775314331055, "lr": 0.0001967146050099148, "elapsed_sec": 73359.21702194214, "step_time_sec": 8.23891640000511, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.115745544433594, "lr": 0.00019671102004654244, "elapsed_sec": 73367.45614051819, "step_time_sec": 8.23895295400871, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.020416736602783, "lr": 0.0001967074331618684, "elapsed_sec": 73375.69618320465, "step_time_sec": 8.239828686986584, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8463799953460693, "lr": 0.00019670384435596574, "elapsed_sec": 73383.9356765747, "step_time_sec": 8.23939853097545, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9687957763671875, "lr": 0.0001967002536289077, "elapsed_sec": 73392.1724767685, "step_time_sec": 8.236602632969152, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8995723724365234, "lr": 0.00019669666098076757, "elapsed_sec": 73400.4117128849, "step_time_sec": 8.239075680961832, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.911104440689087, "lr": 0.00019669306641161848, "elapsed_sec": 73408.64961123466, "step_time_sec": 8.237727970001288, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.912273645401001, "lr": 0.00019668946992153387, "elapsed_sec": 73416.88803815842, "step_time_sec": 8.238196395977866, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9011056423187256, "lr": 0.000196685871510587, "elapsed_sec": 73425.12686181068, "step_time_sec": 8.238696497981437, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.097728252410889, "lr": 0.00019668227117885126, "elapsed_sec": 73433.36548566818, "step_time_sec": 8.238476808008272, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8322901725769043, "lr": 0.00019667866892640013, "elapsed_sec": 73441.60566806793, "step_time_sec": 8.24000106396852, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.027629375457764, "lr": 0.00019667506475330704, "elapsed_sec": 73449.84537529945, "step_time_sec": 8.239542292954866, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.134387969970703, "lr": 0.00019667145865964546, "elapsed_sec": 73458.08411169052, "step_time_sec": 8.2385641989531, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9476613998413086, "lr": 0.00019666785064548898, "elapsed_sec": 73466.32296395302, "step_time_sec": 8.238736715982668, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9523065090179443, "lr": 0.00019666424071091115, "elapsed_sec": 73474.56111741066, "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": 8839, "loss": 3.847774028778076, "lr": 0.0001966606288559856, "elapsed_sec": 73482.80016899109, "step_time_sec": 8.238859313016292, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.7028937339782715, "lr": 0.00019665701508078597, "elapsed_sec": 73491.03763842583, "step_time_sec": 8.237260932975914, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.935866594314575, "lr": 0.000196653399385386, "elapsed_sec": 73499.27717113495, "step_time_sec": 8.239425720006693, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.977365493774414, "lr": 0.00019664978176985943, "elapsed_sec": 73507.51618885994, "step_time_sec": 8.238840320962481, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.943040132522583, "lr": 0.00019664616223427997, "elapsed_sec": 73515.75352740288, "step_time_sec": 8.237123309983872, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.045994281768799, "lr": 0.0001966425407787215, "elapsed_sec": 73523.99283957481, "step_time_sec": 8.239151360990945, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.069547176361084, "lr": 0.0001966389174032578, "elapsed_sec": 73532.23251104355, "step_time_sec": 8.239547918026801, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.933945417404175, "lr": 0.0001966352921079629, "elapsed_sec": 73540.47042775154, "step_time_sec": 8.237735185015481, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.063468933105469, "lr": 0.00019663166489291057, "elapsed_sec": 73548.7094540596, "step_time_sec": 8.238871414971072, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9925522804260254, "lr": 0.0001966280357581749, "elapsed_sec": 73556.94806408882, "step_time_sec": 8.23841359099606, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8928940296173096, "lr": 0.00019662440470382984, "elapsed_sec": 73565.18816256523, "step_time_sec": 8.239991418027785, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.84287428855896, "lr": 0.00019662077172994947, "elapsed_sec": 73573.4273416996, "step_time_sec": 8.238989637990016, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.922490358352661, "lr": 0.00019661713683660786, "elapsed_sec": 73581.66499781609, "step_time_sec": 8.237473117012996, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.940826892852783, "lr": 0.00019661350002387915, "elapsed_sec": 73589.9044020176, "step_time_sec": 8.23921894596424, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.085476398468018, "lr": 0.0001966098612918375, "elapsed_sec": 73598.14215159416, "step_time_sec": 8.237604532041587, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.176437854766846, "lr": 0.00019660622064055714, "elapsed_sec": 73606.38080263138, "step_time_sec": 8.238460623950232, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9563674926757812, "lr": 0.00019660257807011227, "elapsed_sec": 73614.62060427666, "step_time_sec": 8.239697532029822, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9294774532318115, "lr": 0.00019659893358057724, "elapsed_sec": 73622.85910654068, "step_time_sec": 8.238309937005397, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7764859199523926, "lr": 0.00019659528717202632, "elapsed_sec": 73631.09866094589, "step_time_sec": 8.239390722999815, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.016053199768066, "lr": 0.00019659163884453392, "elapsed_sec": 73639.3358669281, "step_time_sec": 8.237006270966958, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.056880474090576, "lr": 0.00019658798859817434, "elapsed_sec": 73647.57464957237, "step_time_sec": 8.238658648042474, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.002335071563721, "lr": 0.00019658433643302217, "elapsed_sec": 73655.81436228752, "step_time_sec": 8.239539006957784, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.866833209991455, "lr": 0.00019658068234915178, "elapsed_sec": 73664.05195426941, "step_time_sec": 8.237427018990275, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9773404598236084, "lr": 0.00019657702634663773, "elapsed_sec": 73672.29194808006, "step_time_sec": 8.239784188976046, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.10667085647583, "lr": 0.00019657336842555457, "elapsed_sec": 73680.5293841362, "step_time_sec": 8.237295917002484, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.970280885696411, "lr": 0.0001965697085859769, "elapsed_sec": 73688.76769733429, "step_time_sec": 8.23816640698351, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.835294008255005, "lr": 0.00019656604682797936, "elapsed_sec": 73697.00627660751, "step_time_sec": 8.238431988982484, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.987184524536133, "lr": 0.00019656238315163667, "elapsed_sec": 73705.24479746819, "step_time_sec": 8.238315439026337, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.87929630279541, "lr": 0.00019655871755702345, "elapsed_sec": 73713.48330783844, "step_time_sec": 8.238343038014136, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.94614839553833, "lr": 0.00019655505004421455, "elapsed_sec": 73721.7207763195, "step_time_sec": 8.237324170011561, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.926199197769165, "lr": 0.0001965513806132847, "elapsed_sec": 73729.96017742157, "step_time_sec": 8.239268124976661, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.882805824279785, "lr": 0.00019654770926430873, "elapsed_sec": 73738.20005750656, "step_time_sec": 8.239658036036417, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.02074670791626, "lr": 0.00019654403599736158, "elapsed_sec": 73746.4698073864, "step_time_sec": 8.269633744028397, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9772183895111084, "lr": 0.00019654036081251809, "elapsed_sec": 73754.70816230774, "step_time_sec": 8.23819204402389, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9128875732421875, "lr": 0.00019653668370985325, "elapsed_sec": 73762.94716453552, "step_time_sec": 8.238814526004717, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.889209747314453, "lr": 0.00019653300468944201, "elapsed_sec": 73771.18745923042, "step_time_sec": 8.24009247700451, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8054981231689453, "lr": 0.00019652932375135944, "elapsed_sec": 73779.42467284203, "step_time_sec": 8.237055332981981, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9048986434936523, "lr": 0.0001965256408956806, "elapsed_sec": 73787.6635298729, "step_time_sec": 8.238721998990513, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8371996879577637, "lr": 0.0001965219561224806, "elapsed_sec": 73795.90274310112, "step_time_sec": 8.238998675951734, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8534657955169678, "lr": 0.00019651826943183455, "elapsed_sec": 73804.14253211021, "step_time_sec": 8.239661113999318, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.032876491546631, "lr": 0.00019651458082381768, "elapsed_sec": 73812.38082313538, "step_time_sec": 8.23809578397777, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8090877532958984, "lr": 0.00019651089029850517, "elapsed_sec": 73820.6190867424, "step_time_sec": 8.238087497011293, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9848039150238037, "lr": 0.00019650719785597234, "elapsed_sec": 73828.85763764381, "step_time_sec": 8.238404890988022, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.007301330566406, "lr": 0.00019650350349629443, "elapsed_sec": 73837.09640288353, "step_time_sec": 8.238589595945086, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.061511516571045, "lr": 0.0001964998072195468, "elapsed_sec": 73845.33625507355, "step_time_sec": 8.239698998979293, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7992265224456787, "lr": 0.00019649610902580487, "elapsed_sec": 73853.57443594933, "step_time_sec": 8.23799917002907, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.001128673553467, "lr": 0.00019649240891514403, "elapsed_sec": 73861.8117120266, "step_time_sec": 8.237151635985356, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9390981197357178, "lr": 0.00019648870688763974, "elapsed_sec": 73870.05057573318, "step_time_sec": 8.23869593202835, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.158516883850098, "lr": 0.0001964850029433675, "elapsed_sec": 73878.28892230988, "step_time_sec": 8.238145088020246, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9213476181030273, "lr": 0.0001964812970824028, "elapsed_sec": 73886.52788043022, "step_time_sec": 8.238815990975127, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.004551887512207, "lr": 0.00019647758930482127, "elapsed_sec": 73894.7663834095, "step_time_sec": 8.238359390990809, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9886248111724854, "lr": 0.00019647387961069856, "elapsed_sec": 73903.00386548042, "step_time_sec": 8.237299466971308, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9890263080596924, "lr": 0.00019647016800011024, "elapsed_sec": 73911.24287080765, "step_time_sec": 8.238843200029805, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9360556602478027, "lr": 0.00019646645447313207, "elapsed_sec": 73919.48190808296, "step_time_sec": 8.238863784994464, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8919990062713623, "lr": 0.00019646273902983974, "elapsed_sec": 73927.72181344032, "step_time_sec": 8.239776036993135, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9950664043426514, "lr": 0.00019645902167030903, "elapsed_sec": 73935.95984196663, "step_time_sec": 8.23780890397029, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.897597551345825, "lr": 0.00019645530239461575, "elapsed_sec": 73944.19755125046, "step_time_sec": 8.237543212017044, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.025477886199951, "lr": 0.00019645158120283577, "elapsed_sec": 73952.4345369339, "step_time_sec": 8.236817631055601, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.860584259033203, "lr": 0.00019644785809504496, "elapsed_sec": 73960.67270469666, "step_time_sec": 8.238053736975417, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9223673343658447, "lr": 0.00019644413307131923, "elapsed_sec": 73968.91217184067, "step_time_sec": 8.239304940972943, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.050828456878662, "lr": 0.00019644040613173456, "elapsed_sec": 73977.15246391296, "step_time_sec": 8.240090382983908, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9833970069885254, "lr": 0.00019643667727636702, "elapsed_sec": 73985.38996934891, "step_time_sec": 8.237317029968835, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.023550033569336, "lr": 0.00019643294650529254, "elapsed_sec": 73993.62931275368, "step_time_sec": 8.239216350018978, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9375078678131104, "lr": 0.0001964292138185873, "elapsed_sec": 74001.86593818665, "step_time_sec": 8.23647715599509, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.936061143875122, "lr": 0.00019642547921632736, "elapsed_sec": 74010.10578131676, "step_time_sec": 8.239673217991367, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.809943914413452, "lr": 0.00019642174269858892, "elapsed_sec": 74018.34518527985, "step_time_sec": 8.239244950993452, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9439215660095215, "lr": 0.00019641800426544815, "elapsed_sec": 74026.58329200745, "step_time_sec": 8.237920331011992, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9566259384155273, "lr": 0.00019641426391698132, "elapsed_sec": 74034.82214808464, "step_time_sec": 8.238759986998048, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.073773384094238, "lr": 0.0001964105216532647, "elapsed_sec": 74043.05948805809, "step_time_sec": 8.237157880037557, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9314167499542236, "lr": 0.00019640677747437458, "elapsed_sec": 74051.29924154282, "step_time_sec": 8.239573731960263, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.053800582885742, "lr": 0.0001964030313803874, "elapsed_sec": 74059.5383515358, "step_time_sec": 8.238900460011791, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.799722909927368, "lr": 0.00019639928337137946, "elapsed_sec": 74067.77672934532, "step_time_sec": 8.238247102010064, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.961670160293579, "lr": 0.00019639553344742721, "elapsed_sec": 74076.01588392258, "step_time_sec": 8.238984344003256, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9937894344329834, "lr": 0.00019639178160860718, "elapsed_sec": 74084.25358986855, "step_time_sec": 8.237520092981867, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.021622180938721, "lr": 0.00019638802785499585, "elapsed_sec": 74092.49274921417, "step_time_sec": 8.239005576004274, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.128122329711914, "lr": 0.00019638427218666976, "elapsed_sec": 74100.73149394989, "step_time_sec": 8.238599737000186, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.869234561920166, "lr": 0.00019638051460370555, "elapsed_sec": 74108.96966457367, "step_time_sec": 8.238004866987467, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.855001211166382, "lr": 0.0001963767551061798, "elapsed_sec": 74117.20865631104, "step_time_sec": 8.23884015600197, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.016526699066162, "lr": 0.00019637299369416915, "elapsed_sec": 74125.44627785683, "step_time_sec": 8.237432427995373, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.851140022277832, "lr": 0.0001963692303677504, "elapsed_sec": 74133.68478441238, "step_time_sec": 8.23828627099283, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.834512948989868, "lr": 0.00019636546512700025, "elapsed_sec": 74141.92386746407, "step_time_sec": 8.238927875005174, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.149495601654053, "lr": 0.0001963616979719955, "elapsed_sec": 74150.1618475914, "step_time_sec": 8.237841971975286, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.937788963317871, "lr": 0.00019635792890281296, "elapsed_sec": 74158.4008140564, "step_time_sec": 8.238791505049448, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.23712158203125, "lr": 0.0001963541579195295, "elapsed_sec": 74178.04310560226, "step_time_sec": 19.642144902958535, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9735701084136963, "lr": 0.00019635038502222198, "elapsed_sec": 74186.27553796768, "step_time_sec": 8.23227403103374, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.009598255157471, "lr": 0.00019634661021096744, "elapsed_sec": 74194.51359510422, "step_time_sec": 8.237913878983818, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9698567390441895, "lr": 0.00019634283348584277, "elapsed_sec": 74202.75207209587, "step_time_sec": 8.23828573198989, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.982482671737671, "lr": 0.00019633905484692506, "elapsed_sec": 74210.99172759056, "step_time_sec": 8.239453780988697, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.026020526885986, "lr": 0.00019633527429429134, "elapsed_sec": 74219.2319278717, "step_time_sec": 8.240080340998247, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.844299077987671, "lr": 0.00019633149182801872, "elapsed_sec": 74227.46971988678, "step_time_sec": 8.237572025042027, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.96846604347229, "lr": 0.00019632770744818428, "elapsed_sec": 74235.70888900757, "step_time_sec": 8.23899701697519, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.004842758178711, "lr": 0.0001963239211548653, "elapsed_sec": 74243.94811820984, "step_time_sec": 8.239111045957543, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9642951488494873, "lr": 0.00019632013294813893, "elapsed_sec": 74252.18683433533, "step_time_sec": 8.238515596021898, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9644088745117188, "lr": 0.0001963163428280824, "elapsed_sec": 74260.42671871185, "step_time_sec": 8.23973583400948, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.056100368499756, "lr": 0.00019631255079477305, "elapsed_sec": 74268.66395926476, "step_time_sec": 8.237111355003435, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.14504337310791, "lr": 0.00019630875684828822, "elapsed_sec": 74276.90392684937, "step_time_sec": 8.239757278002799, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9496591091156006, "lr": 0.00019630496098870526, "elapsed_sec": 74285.14328813553, "step_time_sec": 8.23920359800104, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.077210426330566, "lr": 0.0001963011632161016, "elapsed_sec": 74293.38148641586, "step_time_sec": 8.238053594017401, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.958280563354492, "lr": 0.00019629736353055465, "elapsed_sec": 74301.62091588974, "step_time_sec": 8.239227114012465, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.925912380218506, "lr": 0.00019629356193214193, "elapsed_sec": 74309.85839557648, "step_time_sec": 8.237349677016027, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.057387351989746, "lr": 0.00019628975842094098, "elapsed_sec": 74318.09855389595, "step_time_sec": 8.239996185002383, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.911409854888916, "lr": 0.00019628595299702936, "elapsed_sec": 74326.33770036697, "step_time_sec": 8.238981298985891, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.961225748062134, "lr": 0.00019628214566048464, "elapsed_sec": 74334.57566952705, "step_time_sec": 8.237818153051194, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.061281681060791, "lr": 0.00019627833641138451, "elapsed_sec": 74342.81373810768, "step_time_sec": 8.237887091003358, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.866569757461548, "lr": 0.00019627452524980662, "elapsed_sec": 74351.05025243759, "step_time_sec": 8.236316995986272, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.976654291152954, "lr": 0.00019627071217582872, "elapsed_sec": 74359.28811883926, "step_time_sec": 8.237735411035828, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9420390129089355, "lr": 0.0001962668971895286, "elapsed_sec": 74367.52701115608, "step_time_sec": 8.23871229897486, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.857124090194702, "lr": 0.00019626308029098397, "elapsed_sec": 74375.7661561966, "step_time_sec": 8.23895387595985, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.040016174316406, "lr": 0.00019625926148027274, "elapsed_sec": 74384.00604510307, "step_time_sec": 8.239740464021452, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9542438983917236, "lr": 0.0001962554407574728, "elapsed_sec": 74392.24230623245, "step_time_sec": 8.23610058502527, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9608473777770996, "lr": 0.000196251618122662, "elapsed_sec": 74400.48215913773, "step_time_sec": 8.239682609040756, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.910412549972534, "lr": 0.00019624779357591836, "elapsed_sec": 74408.72100186348, "step_time_sec": 8.238669949001633, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8449535369873047, "lr": 0.00019624396711731987, "elapsed_sec": 74416.96038913727, "step_time_sec": 8.239277360029519, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.847135305404663, "lr": 0.00019624013874694455, "elapsed_sec": 74425.19993925095, "step_time_sec": 8.239320575026795, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.038069248199463, "lr": 0.00019623630846487048, "elapsed_sec": 74433.43757104874, "step_time_sec": 8.237504801014438, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8616623878479004, "lr": 0.00019623247627117575, "elapsed_sec": 74441.67702746391, "step_time_sec": 8.239262871968094, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7949445247650146, "lr": 0.00019622864216593854, "elapsed_sec": 74449.91519451141, "step_time_sec": 8.237989851972088, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.121454238891602, "lr": 0.00019622480614923702, "elapsed_sec": 74458.15333008766, "step_time_sec": 8.238029990985524, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9877045154571533, "lr": 0.00019622096822114946, "elapsed_sec": 74466.39208960533, "step_time_sec": 8.23859803302912, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.235767364501953, "lr": 0.00019621712838175406, "elapsed_sec": 74474.62881422043, "step_time_sec": 8.236565453989897, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.956448793411255, "lr": 0.0001962132866311292, "elapsed_sec": 74482.86849999428, "step_time_sec": 8.239435433992185, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.052552223205566, "lr": 0.00019620944296935318, "elapsed_sec": 74491.10704183578, "step_time_sec": 8.23839750996558, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8819079399108887, "lr": 0.0001962055973965044, "elapsed_sec": 74499.34589099884, "step_time_sec": 8.238712508988101, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8905484676361084, "lr": 0.00019620174991266133, "elapsed_sec": 74507.58460021019, "step_time_sec": 8.23852218396496, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.947021961212158, "lr": 0.00019619790051790236, "elapsed_sec": 74515.8209593296, "step_time_sec": 8.236229011032265, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.064578533172607, "lr": 0.000196194049212306, "elapsed_sec": 74524.06088614464, "step_time_sec": 8.239783726981841, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.013185501098633, "lr": 0.00019619019599595084, "elapsed_sec": 74532.29960942268, "step_time_sec": 8.238577009004075, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.98726487159729, "lr": 0.00019618634086891545, "elapsed_sec": 74540.53862404823, "step_time_sec": 8.238805390021298, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9934656620025635, "lr": 0.00019618248383127844, "elapsed_sec": 74548.77666115761, "step_time_sec": 8.237891543016303, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9502737522125244, "lr": 0.00019617862488311845, "elapsed_sec": 74557.01462936401, "step_time_sec": 8.237797835026868, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9514071941375732, "lr": 0.0001961747640245142, "elapsed_sec": 74565.25320029259, "step_time_sec": 8.238377842993941, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.142460823059082, "lr": 0.0001961709012555444, "elapsed_sec": 74573.49082136154, "step_time_sec": 8.237497587047983, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.005944728851318, "lr": 0.0001961670365762879, "elapsed_sec": 74581.73051428795, "step_time_sec": 8.239486999984365, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0030951499938965, "lr": 0.0001961631699868234, "elapsed_sec": 74589.9693300724, "step_time_sec": 8.238653305044863, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8648266792297363, "lr": 0.00019615930148722988, "elapsed_sec": 74598.20707559586, "step_time_sec": 8.237603725981899, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.012717247009277, "lr": 0.0001961554310775861, "elapsed_sec": 74606.44555401802, "step_time_sec": 8.238325999001972, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.886756420135498, "lr": 0.0001961515587579711, "elapsed_sec": 74614.6830997467, "step_time_sec": 8.237396679003723, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.108677387237549, "lr": 0.0001961476845284638, "elapsed_sec": 74622.9224755764, "step_time_sec": 8.239197051036172, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.035627841949463, "lr": 0.00019614380838914325, "elapsed_sec": 74631.16130518913, "step_time_sec": 8.23862368700793, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.928786516189575, "lr": 0.00019613993034008843, "elapsed_sec": 74639.40014314651, "step_time_sec": 8.238673444022425, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9704577922821045, "lr": 0.00019613605038137848, "elapsed_sec": 74647.64001822472, "step_time_sec": 8.239697399025317, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.93615460395813, "lr": 0.00019613216851309252, "elapsed_sec": 74655.8782081604, "step_time_sec": 8.238004020997323, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9523138999938965, "lr": 0.00019612828473530966, "elapsed_sec": 74664.11771273613, "step_time_sec": 8.239348257018719, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9524824619293213, "lr": 0.0001961243990481092, "elapsed_sec": 74672.35669589043, "step_time_sec": 8.238825912005268, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9655966758728027, "lr": 0.00019612051145157032, "elapsed_sec": 74680.59477472305, "step_time_sec": 8.237914872996043, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.97217059135437, "lr": 0.0001961166219457723, "elapsed_sec": 74688.83311223984, "step_time_sec": 8.238201076979749, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.910724639892578, "lr": 0.00019611273053079452, "elapsed_sec": 74697.07138943672, "step_time_sec": 8.238112931023352, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.178840637207031, "lr": 0.00019610883720671627, "elapsed_sec": 74705.31109833717, "step_time_sec": 8.23953361302847, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.021416187286377, "lr": 0.00019610494197361695, "elapsed_sec": 74713.55046796799, "step_time_sec": 8.239235804008786, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9956207275390625, "lr": 0.00019610104483157607, "elapsed_sec": 74721.78834700584, "step_time_sec": 8.237660131999291, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.923987865447998, "lr": 0.00019609714578067305, "elapsed_sec": 74730.02707719803, "step_time_sec": 8.23858109000139, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9537179470062256, "lr": 0.00019609324482098742, "elapsed_sec": 74738.26639008522, "step_time_sec": 8.239172501023859, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.046153545379639, "lr": 0.0001960893419525987, "elapsed_sec": 74746.50522041321, "step_time_sec": 8.238636931986548, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9442336559295654, "lr": 0.00019608543717558653, "elapsed_sec": 74754.74552536011, "step_time_sec": 8.24016752402531, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.062425136566162, "lr": 0.00019608153049003056, "elapsed_sec": 74762.98204922676, "step_time_sec": 8.236322842014488, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.078917503356934, "lr": 0.00019607762189601042, "elapsed_sec": 74771.22212314606, "step_time_sec": 8.239917354017962, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.899070978164673, "lr": 0.00019607371139360578, "elapsed_sec": 74779.46104073524, "step_time_sec": 8.23875338898506, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.990565776824951, "lr": 0.00019606979898289647, "elapsed_sec": 74787.70055270195, "step_time_sec": 8.239345380978193, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.104745388031006, "lr": 0.00019606588466396227, "elapsed_sec": 74795.93941354752, "step_time_sec": 8.23870672204066, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.922900676727295, "lr": 0.00019606196843688294, "elapsed_sec": 74804.17660236359, "step_time_sec": 8.237060863000806, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.06795597076416, "lr": 0.0001960580503017384, "elapsed_sec": 74812.41405439377, "step_time_sec": 8.237254615989514, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9783966541290283, "lr": 0.00019605413025860856, "elapsed_sec": 74820.65304684639, "step_time_sec": 53.457662651955616, "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": 0.9841871590469964, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 9001, "loss": 4.1182427406311035, "lr": 0.00019605020830757333, "elapsed_sec": 74874.11941385269, "step_time_sec": 8.247456060955301, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.336606025695801, "lr": 0.00019604628444871273, "elapsed_sec": 74882.35828304291, "step_time_sec": 8.238668432983104, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.067700386047363, "lr": 0.00019604235868210678, "elapsed_sec": 74890.59793639183, "step_time_sec": 8.239462694036774, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.182339191436768, "lr": 0.00019603843100783548, "elapsed_sec": 74898.8360388279, "step_time_sec": 8.237970547983423, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.235947608947754, "lr": 0.00019603450142597899, "elapsed_sec": 74907.07582521439, "step_time_sec": 8.239607091003563, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.267392158508301, "lr": 0.0001960305699366174, "elapsed_sec": 74915.31269669533, "step_time_sec": 8.236707545001991, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.180914402008057, "lr": 0.00019602663653983096, "elapsed_sec": 74923.55064678192, "step_time_sec": 8.237816109030973, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.211843967437744, "lr": 0.0001960227012356998, "elapsed_sec": 74931.7902059555, "step_time_sec": 8.239430904970504, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.111215114593506, "lr": 0.0001960187640243043, "elapsed_sec": 74940.02883934975, "step_time_sec": 8.238443808979355, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.069055080413818, "lr": 0.00019601482490572462, "elapsed_sec": 74948.27757811546, "step_time_sec": 8.248613577976357, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.122313022613525, "lr": 0.00019601088388004114, "elapsed_sec": 74956.51915001869, "step_time_sec": 8.241357252001762, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.963381290435791, "lr": 0.00019600694094733422, "elapsed_sec": 74964.75758099556, "step_time_sec": 8.23822439898504, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.079054355621338, "lr": 0.00019600299610768432, "elapsed_sec": 74972.99610853195, "step_time_sec": 8.238352731976192, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.054175853729248, "lr": 0.00019599904936117184, "elapsed_sec": 74981.23572921753, "step_time_sec": 8.23943956597941, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.062033176422119, "lr": 0.0001959951007078773, "elapsed_sec": 74989.4734685421, "step_time_sec": 8.23756171495188, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9668986797332764, "lr": 0.0001959911501478812, "elapsed_sec": 74997.71272921562, "step_time_sec": 8.239091663970612, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9439995288848877, "lr": 0.00019598719768126413, "elapsed_sec": 75005.9508960247, "step_time_sec": 8.238009697990492, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.101912021636963, "lr": 0.00019598324330810668, "elapsed_sec": 75014.1891245842, "step_time_sec": 8.238096537999809, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.975205659866333, "lr": 0.0001959792870284895, "elapsed_sec": 75022.42668604851, "step_time_sec": 8.237374320975505, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9178428649902344, "lr": 0.00019597532884249328, "elapsed_sec": 75030.6647875309, "step_time_sec": 8.237940985010937, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.986279249191284, "lr": 0.0001959713687501987, "elapsed_sec": 75038.90437436104, "step_time_sec": 8.239428572996985, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.928454637527466, "lr": 0.00019596740675168657, "elapsed_sec": 75047.14086532593, "step_time_sec": 8.23631381103769, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.941535711288452, "lr": 0.00019596344284703767, "elapsed_sec": 75055.3803870678, "step_time_sec": 8.239371452014893, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.958235263824463, "lr": 0.00019595947703633288, "elapsed_sec": 75063.61830091476, "step_time_sec": 8.237767050042748, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9112088680267334, "lr": 0.00019595550931965298, "elapsed_sec": 75071.8562772274, "step_time_sec": 8.237763922021259, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8715624809265137, "lr": 0.00019595153969707897, "elapsed_sec": 75080.09540915489, "step_time_sec": 8.238991137011908, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.920408248901367, "lr": 0.00019594756816869178, "elapsed_sec": 75088.33305382729, "step_time_sec": 8.237468202016316, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9139022827148438, "lr": 0.0001959435947345724, "elapsed_sec": 75096.5710966587, "step_time_sec": 8.237875726015773, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.956714630126953, "lr": 0.00019593961939480187, "elapsed_sec": 75104.80944108963, "step_time_sec": 8.238247962028254, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.92684268951416, "lr": 0.00019593564214946124, "elapsed_sec": 75113.0493183136, "step_time_sec": 8.239694620016962, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.907581329345703, "lr": 0.00019593166299863165, "elapsed_sec": 75121.28875923157, "step_time_sec": 8.239260756003205, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9300215244293213, "lr": 0.00019592768194239425, "elapsed_sec": 75129.5253868103, "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": 9033, "loss": 3.9311087131500244, "lr": 0.0001959236989808302, "elapsed_sec": 75137.76479530334, "step_time_sec": 8.239236453024205, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0378947257995605, "lr": 0.00019591971411402073, "elapsed_sec": 75146.00308585167, "step_time_sec": 8.238149286946282, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.004055976867676, "lr": 0.00019591572734204712, "elapsed_sec": 75154.24139809608, "step_time_sec": 8.238153243029956, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.009671688079834, "lr": 0.00019591173866499067, "elapsed_sec": 75162.47929644585, "step_time_sec": 8.237732997047715, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.961395263671875, "lr": 0.00019590774808293272, "elapsed_sec": 75170.71807098389, "step_time_sec": 8.238595164031722, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.960265874862671, "lr": 0.00019590375559595469, "elapsed_sec": 75178.95739889145, "step_time_sec": 8.23923766601365, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.88692307472229, "lr": 0.00019589976120413797, "elapsed_sec": 75187.1956603527, "step_time_sec": 8.23807402397506, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8248848915100098, "lr": 0.00019589576490756396, "elapsed_sec": 75195.43422031403, "step_time_sec": 8.238415798987262, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.947761058807373, "lr": 0.00019589176670631428, "elapsed_sec": 75203.67218637466, "step_time_sec": 8.237794120970648, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8202097415924072, "lr": 0.00019588776660047036, "elapsed_sec": 75211.90897011757, "step_time_sec": 8.236551960988436, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.884861707687378, "lr": 0.00019588376459011382, "elapsed_sec": 75220.14790296555, "step_time_sec": 8.238817397970706, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9143855571746826, "lr": 0.00019587976067532632, "elapsed_sec": 75228.38588786125, "step_time_sec": 8.237806631019339, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.991361618041992, "lr": 0.00019587575485618944, "elapsed_sec": 75236.62702775002, "step_time_sec": 8.240983943047468, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9074366092681885, "lr": 0.00019587174713278492, "elapsed_sec": 75244.86713123322, "step_time_sec": 8.239955611992627, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.078254222869873, "lr": 0.00019586773750519446, "elapsed_sec": 75253.10424232483, "step_time_sec": 8.236894079949707, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.942944288253784, "lr": 0.00019586372597349985, "elapsed_sec": 75261.34434843063, "step_time_sec": 8.239951566967648, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.030629634857178, "lr": 0.00019585971253778292, "elapsed_sec": 75269.58304190636, "step_time_sec": 8.238534156000242, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.796104907989502, "lr": 0.00019585569719812547, "elapsed_sec": 75277.82290768623, "step_time_sec": 8.239669244969264, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.902493953704834, "lr": 0.00019585167995460941, "elapsed_sec": 75286.06156253815, "step_time_sec": 8.238512364972848, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.963130235671997, "lr": 0.00019584766080731669, "elapsed_sec": 75294.29906773567, "step_time_sec": 8.23734609503299, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9451591968536377, "lr": 0.00019584363975632925, "elapsed_sec": 75302.53672862053, "step_time_sec": 8.237475779955275, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.111509323120117, "lr": 0.00019583961680172906, "elapsed_sec": 75310.77514123917, "step_time_sec": 8.238246352993883, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9126393795013428, "lr": 0.00019583559194359822, "elapsed_sec": 75319.01370882988, "step_time_sec": 8.238443028007168, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.918891429901123, "lr": 0.0001958315651820188, "elapsed_sec": 75327.25292801857, "step_time_sec": 8.239026936993469, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9385476112365723, "lr": 0.0001958275365170729, "elapsed_sec": 75335.4907925129, "step_time_sec": 8.237739222007804, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.940096139907837, "lr": 0.00019582350594884272, "elapsed_sec": 75343.73025107384, "step_time_sec": 8.239288279030006, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8244333267211914, "lr": 0.00019581947347741037, "elapsed_sec": 75351.96843600273, "step_time_sec": 8.238009192980826, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.019630432128906, "lr": 0.0001958154391028582, "elapsed_sec": 75360.2082297802, "step_time_sec": 8.239627004950307, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8919167518615723, "lr": 0.0001958114028252684, "elapsed_sec": 75368.44869208336, "step_time_sec": 8.240333378023934, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9419915676116943, "lr": 0.00019580736464472332, "elapsed_sec": 75376.68663668633, "step_time_sec": 8.237702365964651, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.897451162338257, "lr": 0.0001958033245613053, "elapsed_sec": 75384.92512130737, "step_time_sec": 8.238380251976196, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.925755262374878, "lr": 0.00019579928257509676, "elapsed_sec": 75393.16359710693, "step_time_sec": 8.238261674006935, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9028897285461426, "lr": 0.0001957952386861801, "elapsed_sec": 75401.40229773521, "step_time_sec": 8.23854302399559, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.988042116165161, "lr": 0.0001957911928946378, "elapsed_sec": 75409.64124774933, "step_time_sec": 8.238792191026732, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.991497039794922, "lr": 0.00019578714520055235, "elapsed_sec": 75417.87804794312, "step_time_sec": 8.236665459000506, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.829228639602661, "lr": 0.00019578309560400634, "elapsed_sec": 75426.11719560623, "step_time_sec": 8.23895538598299, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.899782419204712, "lr": 0.0001957790441050823, "elapsed_sec": 75434.3549451828, "step_time_sec": 8.23758325300878, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9329605102539062, "lr": 0.0001957749907038629, "elapsed_sec": 75442.59364676476, "step_time_sec": 8.238535723008681, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.813714027404785, "lr": 0.00019577093540043078, "elapsed_sec": 75450.83291649818, "step_time_sec": 8.239079103979748, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9585764408111572, "lr": 0.00019576687819486868, "elapsed_sec": 75459.06988883018, "step_time_sec": 8.23685215698788, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.159763813018799, "lr": 0.00019576281908725927, "elapsed_sec": 75467.31000208855, "step_time_sec": 8.239939582010265, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.007811069488525, "lr": 0.0001957587580776854, "elapsed_sec": 75475.54908752441, "step_time_sec": 8.238895251997747, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9127109050750732, "lr": 0.00019575469516622986, "elapsed_sec": 75483.78852272034, "step_time_sec": 8.239305530034471, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9094903469085693, "lr": 0.00019575063035297548, "elapsed_sec": 75492.02676558495, "step_time_sec": 8.238037873001304, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9559597969055176, "lr": 0.00019574656363800517, "elapsed_sec": 75500.26414060593, "step_time_sec": 8.237213303975295, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.82585072517395, "lr": 0.00019574249502140193, "elapsed_sec": 75508.5030465126, "step_time_sec": 8.238793753029313, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.033276557922363, "lr": 0.00019573842450324865, "elapsed_sec": 75516.74044442177, "step_time_sec": 8.237220574985258, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.002468109130859, "lr": 0.0001957343520836284, "elapsed_sec": 75524.97956466675, "step_time_sec": 8.238938249996863, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8541271686553955, "lr": 0.00019573027776262418, "elapsed_sec": 75533.21817278862, "step_time_sec": 8.238434821018018, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.972357988357544, "lr": 0.00019572620154031912, "elapsed_sec": 75541.45622372627, "step_time_sec": 8.237878354964778, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9687905311584473, "lr": 0.00019572212341679635, "elapsed_sec": 75549.69586443901, "step_time_sec": 8.239444547973108, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.912482738494873, "lr": 0.00019571804339213897, "elapsed_sec": 75557.93347358704, "step_time_sec": 8.23743805399863, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7847869396209717, "lr": 0.0001957139614664303, "elapsed_sec": 75566.17343783379, "step_time_sec": 8.239799740025774, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8282880783081055, "lr": 0.00019570987763975348, "elapsed_sec": 75574.41287469864, "step_time_sec": 8.23933195299469, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8885080814361572, "lr": 0.00019570579191219184, "elapsed_sec": 75582.6513876915, "step_time_sec": 8.23830479796743, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9066362380981445, "lr": 0.0001957017042838287, "elapsed_sec": 75590.88977384567, "step_time_sec": 8.238205713045318, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0458903312683105, "lr": 0.00019569761475474745, "elapsed_sec": 75599.1284840107, "step_time_sec": 8.238609942025505, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8705990314483643, "lr": 0.00019569352332503143, "elapsed_sec": 75607.3675107956, "step_time_sec": 8.238823554012924, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9116790294647217, "lr": 0.00019568942999476414, "elapsed_sec": 75615.60734629631, "step_time_sec": 8.239666167006362, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9806265830993652, "lr": 0.00019568533476402902, "elapsed_sec": 75623.84372854233, "step_time_sec": 8.236213967029471, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.017192363739014, "lr": 0.00019568123763290956, "elapsed_sec": 75632.08358645439, "step_time_sec": 8.239661875995807, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.247344017028809, "lr": 0.0001956771386014894, "elapsed_sec": 75640.32124638557, "step_time_sec": 8.237508414953481, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8738150596618652, "lr": 0.00019567303766985207, "elapsed_sec": 75648.5607020855, "step_time_sec": 8.239302476053126, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.070064544677734, "lr": 0.0001956689348380812, "elapsed_sec": 75656.80033683777, "step_time_sec": 8.239497242961079, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0863213539123535, "lr": 0.0001956648301062605, "elapsed_sec": 75665.03814673424, "step_time_sec": 8.237621814012527, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9275152683258057, "lr": 0.00019566072347447361, "elapsed_sec": 75673.27587223053, "step_time_sec": 8.237565462011844, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9189889430999756, "lr": 0.0001956566149428044, "elapsed_sec": 75681.5147280693, "step_time_sec": 8.238689310033806, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.832955837249756, "lr": 0.00019565250451133653, "elapsed_sec": 75689.75467586517, "step_time_sec": 8.23978600895498, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8577346801757812, "lr": 0.00019564839218015392, "elapsed_sec": 75697.99368333817, "step_time_sec": 8.23884416196961, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8528082370758057, "lr": 0.00019564427794934038, "elapsed_sec": 75706.23060846329, "step_time_sec": 8.236768274044152, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.897136926651001, "lr": 0.00019564016181897986, "elapsed_sec": 75714.46973800659, "step_time_sec": 8.239010065968614, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8237833976745605, "lr": 0.00019563604378915623, "elapsed_sec": 75722.70745038986, "step_time_sec": 8.237492156040389, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9985055923461914, "lr": 0.00019563192385995353, "elapsed_sec": 75730.9468626976, "step_time_sec": 8.239249677979387, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8700852394104004, "lr": 0.00019562780203145576, "elapsed_sec": 75739.18639087677, "step_time_sec": 8.239391632960178, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9569015502929688, "lr": 0.000195623678303747, "elapsed_sec": 75747.42329406738, "step_time_sec": 8.236752768978477, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8715574741363525, "lr": 0.00019561955267691136, "elapsed_sec": 75755.66219878197, "step_time_sec": 8.23869199003093, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.808088779449463, "lr": 0.0001956154251510329, "elapsed_sec": 75763.8995540142, "step_time_sec": 8.237188856001012, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.081539154052734, "lr": 0.00019561129572619583, "elapsed_sec": 75772.13808321953, "step_time_sec": 8.238343823992182, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.942218065261841, "lr": 0.0001956071644024844, "elapsed_sec": 75780.37560844421, "step_time_sec": 8.237358053971548, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.039863109588623, "lr": 0.00019560303117998286, "elapsed_sec": 75788.61366295815, "step_time_sec": 8.237896713020746, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8940038681030273, "lr": 0.00019559889605877547, "elapsed_sec": 75796.85194396973, "step_time_sec": 8.23815824801568, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9598276615142822, "lr": 0.00019559475903894658, "elapsed_sec": 75805.08953404427, "step_time_sec": 8.237354237004183, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.042883396148682, "lr": 0.0001955906201205805, "elapsed_sec": 75813.32766819, "step_time_sec": 8.237975844996981, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.021080017089844, "lr": 0.00019558647930376175, "elapsed_sec": 75821.56578803062, "step_time_sec": 8.237946063978598, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.037897109985352, "lr": 0.0001955823365885747, "elapsed_sec": 75829.80393600464, "step_time_sec": 8.238033090019599, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.009278297424316, "lr": 0.00019557819197510383, "elapsed_sec": 75838.04228878021, "step_time_sec": 8.238162775000092, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.03026819229126, "lr": 0.0001955740454634337, "elapsed_sec": 75846.28008818626, "step_time_sec": 8.237623343011364, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.995053768157959, "lr": 0.00019556989705364885, "elapsed_sec": 75854.5180516243, "step_time_sec": 8.237864974013064, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.003589153289795, "lr": 0.00019556574674583387, "elapsed_sec": 75862.75570750237, "step_time_sec": 8.237481515971012, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8765134811401367, "lr": 0.00019556159454007342, "elapsed_sec": 75870.9942958355, "step_time_sec": 8.23843923700042, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.005810737609863, "lr": 0.00019555744043645219, "elapsed_sec": 75879.23254847527, "step_time_sec": 8.238098795991391, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9375882148742676, "lr": 0.00019555328443505486, "elapsed_sec": 75887.47064900398, "step_time_sec": 8.237926324014552, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.100334644317627, "lr": 0.00019554912653596622, "elapsed_sec": 75895.70897984505, "step_time_sec": 8.238139763008803, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8901174068450928, "lr": 0.00019554496673927104, "elapsed_sec": 75903.94611811638, "step_time_sec": 8.237024510046467, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.021087169647217, "lr": 0.00019554080504505418, "elapsed_sec": 75912.18433904648, "step_time_sec": 8.238008561020251, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0247955322265625, "lr": 0.00019553664145340047, "elapsed_sec": 75920.42226600647, "step_time_sec": 8.2377719610231, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8153464794158936, "lr": 0.00019553247596439487, "elapsed_sec": 75928.66069746017, "step_time_sec": 8.238271716050804, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9915904998779297, "lr": 0.0001955283085781223, "elapsed_sec": 75936.89852786064, "step_time_sec": 8.237679562007543, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.098775386810303, "lr": 0.00019552413929466774, "elapsed_sec": 75945.13603878021, "step_time_sec": 8.2373365489766, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.974508762359619, "lr": 0.00019551996811411625, "elapsed_sec": 75953.37432336807, "step_time_sec": 8.23813102801796, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.076490879058838, "lr": 0.00019551579503655287, "elapsed_sec": 75961.61381864548, "step_time_sec": 8.239305739989504, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.003586292266846, "lr": 0.00019551162006206275, "elapsed_sec": 75969.85311937332, "step_time_sec": 8.239162594021764, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.865837335586548, "lr": 0.00019550744319073095, "elapsed_sec": 75978.09198522568, "step_time_sec": 8.23867902602069, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.943939447402954, "lr": 0.00019550326442264266, "elapsed_sec": 75986.32851362228, "step_time_sec": 8.236400326015428, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.936917304992676, "lr": 0.0001954990837578832, "elapsed_sec": 75994.56729578972, "step_time_sec": 8.238645314006135, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.051975727081299, "lr": 0.00019549490119653775, "elapsed_sec": 76002.80452251434, "step_time_sec": 8.237002189038321, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.047999858856201, "lr": 0.00019549071673869158, "elapsed_sec": 76011.04381585121, "step_time_sec": 8.239129970024806, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9471843242645264, "lr": 0.00019548653038443012, "elapsed_sec": 76019.28204607964, "step_time_sec": 8.23810965998564, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9432101249694824, "lr": 0.00019548234213383864, "elapsed_sec": 76027.51973032951, "step_time_sec": 8.237477107963059, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.121108055114746, "lr": 0.00019547815198700264, "elapsed_sec": 76035.75648188591, "step_time_sec": 8.236586097977124, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.973578929901123, "lr": 0.00019547395994400755, "elapsed_sec": 76043.99388217926, "step_time_sec": 8.237238914996851, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9666664600372314, "lr": 0.00019546976600493882, "elapsed_sec": 76052.23505187035, "step_time_sec": 8.241002360999119, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2410688400268555, "lr": 0.000195465570169882, "elapsed_sec": 76060.4736135006, "step_time_sec": 8.238386026991066, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.960355281829834, "lr": 0.00019546137243892268, "elapsed_sec": 76068.71124768257, "step_time_sec": 8.237506875011604, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.114609718322754, "lr": 0.00019545717281214647, "elapsed_sec": 76076.95185613632, "step_time_sec": 8.240411946026143, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.122466087341309, "lr": 0.00019545297128963898, "elapsed_sec": 76085.18917512894, "step_time_sec": 8.237208259000909, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.99951434135437, "lr": 0.00019544876787148594, "elapsed_sec": 76093.42777299881, "step_time_sec": 8.2384182230453, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.048870086669922, "lr": 0.000195444562557773, "elapsed_sec": 76101.6661772728, "step_time_sec": 8.238219649996608, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.009411334991455, "lr": 0.00019544035534858602, "elapsed_sec": 76109.90521717072, "step_time_sec": 8.238874049042352, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.952637195587158, "lr": 0.0001954361462440107, "elapsed_sec": 76118.14440321922, "step_time_sec": 8.239075565012172, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.889164447784424, "lr": 0.00019543193524413293, "elapsed_sec": 76126.38158941269, "step_time_sec": 8.236955493979622, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9646897315979004, "lr": 0.0001954277223490386, "elapsed_sec": 76134.62190937996, "step_time_sec": 8.240254778007511, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.147776126861572, "lr": 0.0001954235075588136, "elapsed_sec": 76142.86150074005, "step_time_sec": 8.23934053000994, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2973737716674805, "lr": 0.0001954192908735439, "elapsed_sec": 76151.10000252724, "step_time_sec": 8.238359873997979, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8376893997192383, "lr": 0.00019541507229331548, "elapsed_sec": 76159.33902907372, "step_time_sec": 8.238858851022087, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.138631820678711, "lr": 0.0001954108518182144, "elapsed_sec": 76167.57715463638, "step_time_sec": 8.237932781979907, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9175055027008057, "lr": 0.00019540662944832668, "elapsed_sec": 76175.81618618965, "step_time_sec": 8.238871770969126, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.781923532485962, "lr": 0.0001954024051837385, "elapsed_sec": 76184.0535800457, "step_time_sec": 8.237212751992047, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.101552486419678, "lr": 0.00019539817902453595, "elapsed_sec": 76192.292375803, "step_time_sec": 8.238620310032275, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.110054969787598, "lr": 0.0001953939509708052, "elapsed_sec": 76200.53100728989, "step_time_sec": 8.238468457013369, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.026519298553467, "lr": 0.00019538972102263255, "elapsed_sec": 76208.7690422535, "step_time_sec": 8.237896562030073, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0109543800354, "lr": 0.0001953854891801042, "elapsed_sec": 76217.00872063637, "step_time_sec": 8.239551882026717, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.012553691864014, "lr": 0.00019538125544330649, "elapsed_sec": 76225.2477941513, "step_time_sec": 8.238899273972493, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0261149406433105, "lr": 0.0001953770198123257, "elapsed_sec": 76233.48520350456, "step_time_sec": 8.237206601013895, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.235623359680176, "lr": 0.00019537278228724833, "elapsed_sec": 76241.72412538528, "step_time_sec": 8.23877013096353, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9904439449310303, "lr": 0.00019536854286816068, "elapsed_sec": 76249.96141028404, "step_time_sec": 8.237147203006316, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.089064598083496, "lr": 0.00019536430155514927, "elapsed_sec": 76258.20219564438, "step_time_sec": 8.240592900954653, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9188179969787598, "lr": 0.00019536005834830056, "elapsed_sec": 76266.44091653824, "step_time_sec": 8.238567560969386, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.938154458999634, "lr": 0.00019535581324770112, "elapsed_sec": 76274.67982983589, "step_time_sec": 8.238756901002489, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8816001415252686, "lr": 0.0001953515662534375, "elapsed_sec": 76282.9179944992, "step_time_sec": 8.23793865297921, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.008584976196289, "lr": 0.00019534731736559633, "elapsed_sec": 76291.15589880943, "step_time_sec": 8.237749147985596, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.98372745513916, "lr": 0.00019534306658426423, "elapsed_sec": 76299.39423131943, "step_time_sec": 8.238165453018155, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.000899314880371, "lr": 0.00019533881390952792, "elapsed_sec": 76307.63240098953, "step_time_sec": 8.238027074025013, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.121679306030273, "lr": 0.00019533455934147407, "elapsed_sec": 76315.87045311928, "step_time_sec": 8.23790678998921, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.12558650970459, "lr": 0.0001953303028801895, "elapsed_sec": 76324.10954785347, "step_time_sec": 8.23891958402237, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.129696846008301, "lr": 0.00019532604452576104, "elapsed_sec": 76332.34738826752, "step_time_sec": 8.237635169003624, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0327935218811035, "lr": 0.00019532178427827547, "elapsed_sec": 76340.58632230759, "step_time_sec": 8.238767179020215, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.075285911560059, "lr": 0.00019531752213781972, "elapsed_sec": 76348.82474589348, "step_time_sec": 8.238286729960237, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.99216890335083, "lr": 0.00019531325810448065, "elapsed_sec": 76357.06305003166, "step_time_sec": 8.238156146020629, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1997809410095215, "lr": 0.00019530899217834528, "elapsed_sec": 76365.30242371559, "step_time_sec": 8.239239085989539, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.921315908432007, "lr": 0.0001953047243595006, "elapsed_sec": 76373.54037046432, "step_time_sec": 8.237737801973708, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.042794704437256, "lr": 0.00019530045464803358, "elapsed_sec": 76381.78016018867, "step_time_sec": 8.239624898007605, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.962080955505371, "lr": 0.0001952961830440314, "elapsed_sec": 76390.01755261421, "step_time_sec": 8.237257184984628, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.07108736038208, "lr": 0.00019529190954758108, "elapsed_sec": 76398.25790381432, "step_time_sec": 8.240156083018519, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.862708568572998, "lr": 0.0001952876341587698, "elapsed_sec": 76406.49621844292, "step_time_sec": 8.238182769040577, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.029727458953857, "lr": 0.00019528335687768483, "elapsed_sec": 76414.73444700241, "step_time_sec": 8.238019602023996, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9728033542633057, "lr": 0.0001952790777044133, "elapsed_sec": 76422.9742500782, "step_time_sec": 8.239580386027228, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8829867839813232, "lr": 0.0001952747966390425, "elapsed_sec": 76431.21268224716, "step_time_sec": 8.238255352014676, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.187767505645752, "lr": 0.00019527051368165973, "elapsed_sec": 76439.45099186897, "step_time_sec": 8.238138328015339, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.079963684082031, "lr": 0.00019526622883235238, "elapsed_sec": 76447.68897628784, "step_time_sec": 8.237831225036643, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.935908794403076, "lr": 0.0001952619420912078, "elapsed_sec": 76455.92573165894, "step_time_sec": 8.236569366999902, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8502426147460938, "lr": 0.0001952576534583134, "elapsed_sec": 76464.16391682625, "step_time_sec": 8.238035508955363, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.031631946563721, "lr": 0.00019525336293375664, "elapsed_sec": 76472.40232777596, "step_time_sec": 8.23824854195118, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9096577167510986, "lr": 0.00019524907051762506, "elapsed_sec": 76480.6417131424, "step_time_sec": 8.239194420981221, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9190938472747803, "lr": 0.00019524477621000617, "elapsed_sec": 76488.88073801994, "step_time_sec": 8.23893233097624, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.012990951538086, "lr": 0.00019524048001098754, "elapsed_sec": 76497.11828541756, "step_time_sec": 8.237379488011356, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.05958890914917, "lr": 0.0001952361819206568, "elapsed_sec": 76505.35718297958, "step_time_sec": 8.238718577020336, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.94757080078125, "lr": 0.00019523188193910158, "elapsed_sec": 76513.59569072723, "step_time_sec": 8.238311370019801, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.966355800628662, "lr": 0.0001952275800664096, "elapsed_sec": 76521.83417105675, "step_time_sec": 8.238299665041268, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.890021562576294, "lr": 0.00019522327630266864, "elapsed_sec": 76530.07379245758, "step_time_sec": 8.239453439018689, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9948232173919678, "lr": 0.00019521897064796634, "elapsed_sec": 76538.31054711342, "step_time_sec": 8.236603622033726, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.957589864730835, "lr": 0.0001952146631023906, "elapsed_sec": 76546.55067396164, "step_time_sec": 8.23995877901325, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.983574628829956, "lr": 0.00019521035366602927, "elapsed_sec": 76554.78803038597, "step_time_sec": 8.23725213605212, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.003323554992676, "lr": 0.00019520604233897015, "elapsed_sec": 76563.02660298347, "step_time_sec": 8.238392946019303, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.061043739318848, "lr": 0.00019520172912130125, "elapsed_sec": 76571.26558947563, "step_time_sec": 8.238780220970511, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9807214736938477, "lr": 0.00019519741401311053, "elapsed_sec": 76579.50249290466, "step_time_sec": 8.236778541002423, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.06028938293457, "lr": 0.0001951930970144859, "elapsed_sec": 76587.7410056591, "step_time_sec": 8.238311190973036, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.006115436553955, "lr": 0.00019518877812551555, "elapsed_sec": 76595.97805523872, "step_time_sec": 8.236891962995287, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.946608304977417, "lr": 0.0001951844573462874, "elapsed_sec": 76604.21777367592, "step_time_sec": 8.23958502302412, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.933872938156128, "lr": 0.0001951801346768897, "elapsed_sec": 76612.45681715012, "step_time_sec": 8.238850591995288, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8433239459991455, "lr": 0.00019517581011741048, "elapsed_sec": 76620.69566750526, "step_time_sec": 8.238677132001612, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.900782823562622, "lr": 0.00019517148366793804, "elapsed_sec": 76628.93441772461, "step_time_sec": 8.23867860599421, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.019572734832764, "lr": 0.00019516715532856053, "elapsed_sec": 76637.1715350151, "step_time_sec": 8.23690085299313, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.781595468521118, "lr": 0.0001951628250993663, "elapsed_sec": 76645.41085958481, "step_time_sec": 8.239145042025484, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8378517627716064, "lr": 0.0001951584929804436, "elapsed_sec": 76653.6505112648, "step_time_sec": 8.23945727798855, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.867974281311035, "lr": 0.00019515415897188078, "elapsed_sec": 76661.88902378082, "step_time_sec": 8.23834135197103, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.931142568588257, "lr": 0.00019514982307376627, "elapsed_sec": 76670.12803196907, "step_time_sec": 8.23887547897175, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8440005779266357, "lr": 0.00019514548528618843, "elapsed_sec": 76678.36615276337, "step_time_sec": 8.237936241959687, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.825967788696289, "lr": 0.00019514114560923582, "elapsed_sec": 76686.6054699421, "step_time_sec": 8.239159527001902, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.114511489868164, "lr": 0.00019513680404299684, "elapsed_sec": 76694.84485793114, "step_time_sec": 8.239215666020755, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.046640872955322, "lr": 0.00019513246058756005, "elapsed_sec": 76703.08325624466, "step_time_sec": 8.238165347022004, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.901256561279297, "lr": 0.00019512811524301409, "elapsed_sec": 76711.32300853729, "step_time_sec": 8.239614352991339, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.907233715057373, "lr": 0.0001951237680094475, "elapsed_sec": 76719.56025099754, "step_time_sec": 8.23712611902738, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.957731008529663, "lr": 0.00019511941888694897, "elapsed_sec": 76727.80028247833, "step_time_sec": 8.239808802958578, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9132397174835205, "lr": 0.00019511506787560723, "elapsed_sec": 76736.03860020638, "step_time_sec": 8.238123974995688, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.126821517944336, "lr": 0.00019511071497551097, "elapsed_sec": 76744.2762131691, "step_time_sec": 8.237462528981268, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8018014430999756, "lr": 0.00019510636018674896, "elapsed_sec": 76752.51515364647, "step_time_sec": 8.23877817200264, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9463062286376953, "lr": 0.00019510200350941002, "elapsed_sec": 76760.75221943855, "step_time_sec": 8.23687741800677, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.829258441925049, "lr": 0.00019509764494358303, "elapsed_sec": 76768.99085402489, "step_time_sec": 8.238521735009272, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8609373569488525, "lr": 0.0001950932844893568, "elapsed_sec": 76777.22862529755, "step_time_sec": 8.237594424979761, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.134371757507324, "lr": 0.0001950889221468203, "elapsed_sec": 76785.46670866013, "step_time_sec": 8.237888681003824, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9658217430114746, "lr": 0.00019508455791606252, "elapsed_sec": 76793.70552039146, "step_time_sec": 8.238669642014429, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.152176380157471, "lr": 0.00019508019179717245, "elapsed_sec": 76801.94354057312, "step_time_sec": 8.237816111999564, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9838366508483887, "lr": 0.00019507582379023908, "elapsed_sec": 76810.1827223301, "step_time_sec": 8.23906833701767, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.001331329345703, "lr": 0.00019507145389535154, "elapsed_sec": 76818.42176485062, "step_time_sec": 8.238823347957805, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9652628898620605, "lr": 0.00019506708211259894, "elapsed_sec": 76826.66100883484, "step_time_sec": 8.239090485032648, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.973140239715576, "lr": 0.00019506270844207044, "elapsed_sec": 76834.90018081665, "step_time_sec": 8.23903027101187, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.045483112335205, "lr": 0.00019505833288385523, "elapsed_sec": 76843.13791441917, "step_time_sec": 8.237520204973407, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.93977952003479, "lr": 0.00019505395543804253, "elapsed_sec": 76851.37713599205, "step_time_sec": 8.23908080701949, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.996561288833618, "lr": 0.0001950495761047216, "elapsed_sec": 76859.61670064926, "step_time_sec": 8.23943481297465, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9651496410369873, "lr": 0.0001950451948839818, "elapsed_sec": 76867.85504341125, "step_time_sec": 8.238117775996216, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.91634202003479, "lr": 0.00019504081177591244, "elapsed_sec": 76876.09404397011, "step_time_sec": 8.238848007982597, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8448660373687744, "lr": 0.00019503642678060293, "elapsed_sec": 76884.33065867424, "step_time_sec": 8.236445826012641, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9555747509002686, "lr": 0.00019503203989814266, "elapsed_sec": 76892.57062315941, "step_time_sec": 8.239816236018669, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7424633502960205, "lr": 0.00019502765112862114, "elapsed_sec": 76900.8090801239, "step_time_sec": 8.23830311599886, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.05700159072876, "lr": 0.00019502326047212784, "elapsed_sec": 76909.04826259613, "step_time_sec": 8.239000318979379, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0471391677856445, "lr": 0.00019501886792875234, "elapsed_sec": 76917.28850746155, "step_time_sec": 8.240102990996093, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9736759662628174, "lr": 0.00019501447349858415, "elapsed_sec": 76925.52651953697, "step_time_sec": 8.237820383976214, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8962879180908203, "lr": 0.00019501007718171294, "elapsed_sec": 76933.7653055191, "step_time_sec": 8.23863124003401, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.009008407592773, "lr": 0.00019500567897822837, "elapsed_sec": 76942.0027384758, "step_time_sec": 8.237239670008421, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.154933929443359, "lr": 0.00019500127888822012, "elapsed_sec": 76950.24097776413, "step_time_sec": 8.238120404013898, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.084139823913574, "lr": 0.00019499687691177793, "elapsed_sec": 76958.47887015343, "step_time_sec": 8.23769722902216, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.152382850646973, "lr": 0.00019499247304899159, "elapsed_sec": 76966.7161769867, "step_time_sec": 8.237130408990197, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.933112382888794, "lr": 0.00019498806729995085, "elapsed_sec": 76974.95598363876, "step_time_sec": 8.239684498985298, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.898538589477539, "lr": 0.0001949836596647456, "elapsed_sec": 76983.19533491135, "step_time_sec": 8.239144453022163, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.006677627563477, "lr": 0.00019497925014346575, "elapsed_sec": 76991.43424844742, "step_time_sec": 8.238727775053121, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.994880437850952, "lr": 0.00019497483873620117, "elapsed_sec": 76999.67313718796, "step_time_sec": 8.238733880978543, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.037013053894043, "lr": 0.0001949704254430419, "elapsed_sec": 77007.90954566002, "step_time_sec": 8.236233112053014, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.048264980316162, "lr": 0.00019496601026407787, "elapsed_sec": 77016.14783263206, "step_time_sec": 8.23811066703638, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.023288249969482, "lr": 0.0001949615931993991, "elapsed_sec": 77024.38453388214, "step_time_sec": 8.2365535950521, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.144123554229736, "lr": 0.0001949571742490958, "elapsed_sec": 77032.62445926666, "step_time_sec": 8.23974306799937, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.038559913635254, "lr": 0.000194952753413258, "elapsed_sec": 77040.86341190338, "step_time_sec": 8.23882244102424, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.984159231185913, "lr": 0.00019494833069197583, "elapsed_sec": 77049.10077047348, "step_time_sec": 8.237176463997457, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.144804954528809, "lr": 0.00019494390608533954, "elapsed_sec": 77057.34012627602, "step_time_sec": 8.23919311404461, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.127832889556885, "lr": 0.00019493947959343933, "elapsed_sec": 77065.57775402069, "step_time_sec": 8.237443582038395, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9905734062194824, "lr": 0.0001949350512163655, "elapsed_sec": 77073.81644392014, "step_time_sec": 8.238497257989366, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.097464084625244, "lr": 0.00019493062095420833, "elapsed_sec": 77082.05542826653, "step_time_sec": 8.238819630991202, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.006852149963379, "lr": 0.00019492618880705823, "elapsed_sec": 77090.29103016853, "step_time_sec": 8.235472076979931, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.001836776733398, "lr": 0.0001949217547750055, "elapsed_sec": 77098.5296959877, "step_time_sec": 8.238466216018423, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9847464561462402, "lr": 0.00019491731885814062, "elapsed_sec": 77106.76652526855, "step_time_sec": 8.236673180013895, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.914154529571533, "lr": 0.00019491288105655401, "elapsed_sec": 77115.00553512573, "step_time_sec": 8.238866069994401, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.006962299346924, "lr": 0.00019490844137033626, "elapsed_sec": 77123.24373364449, "step_time_sec": 8.238025164988358, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.041153907775879, "lr": 0.00019490399979957783, "elapsed_sec": 77131.48217988014, "step_time_sec": 8.23827854602132, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9713356494903564, "lr": 0.00019489955634436934, "elapsed_sec": 77139.7212677002, "step_time_sec": 8.23894561699126, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.029297351837158, "lr": 0.0001948951110048014, "elapsed_sec": 77147.9579026699, "step_time_sec": 8.236450871976558, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.106260776519775, "lr": 0.00019489066378096465, "elapsed_sec": 77156.19662475586, "step_time_sec": 8.238584732986055, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.118600845336914, "lr": 0.0001948862146729498, "elapsed_sec": 77164.43450713158, "step_time_sec": 8.23773039795924, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.174189567565918, "lr": 0.00019488176368084758, "elapsed_sec": 77172.67338871956, "step_time_sec": 8.238669925020076, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.407171249389648, "lr": 0.00019487731080474876, "elapsed_sec": 77180.91136455536, "step_time_sec": 8.237835604988504, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.43705940246582, "lr": 0.0001948728560447441, "elapsed_sec": 77189.14874601364, "step_time_sec": 8.237247286946513, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.226590156555176, "lr": 0.00019486839940092456, "elapsed_sec": 77197.38685846329, "step_time_sec": 8.237894743971992, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.174172878265381, "lr": 0.00019486394087338095, "elapsed_sec": 77205.62412691116, "step_time_sec": 8.237106066022534, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.171121120452881, "lr": 0.00019485948046220425, "elapsed_sec": 77213.86335349083, "step_time_sec": 8.239064167020842, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.118418216705322, "lr": 0.00019485501816748533, "elapsed_sec": 77222.10157227516, "step_time_sec": 8.238052606990095, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.034657955169678, "lr": 0.00019485055398931527, "elapsed_sec": 77230.3389954567, "step_time_sec": 8.237302726949565, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.019030570983887, "lr": 0.0001948460879277851, "elapsed_sec": 77238.5786049366, "step_time_sec": 8.239418169949204, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.390316963195801, "lr": 0.00019484161998298586, "elapsed_sec": 77246.81529259682, "step_time_sec": 8.236497688980307, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2331671714782715, "lr": 0.0001948371501550087, "elapsed_sec": 77255.05450034142, "step_time_sec": 8.239042354049161, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.112171649932861, "lr": 0.0001948326784439448, "elapsed_sec": 77263.2926454544, "step_time_sec": 8.237966191023588, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.087848663330078, "lr": 0.00019482820484988532, "elapsed_sec": 77271.5299873352, "step_time_sec": 8.237223940959666, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.292346477508545, "lr": 0.00019482372937292146, "elapsed_sec": 77279.76956868172, "step_time_sec": 8.239428893022705, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1985344886779785, "lr": 0.00019481925201314458, "elapsed_sec": 77288.00672745705, "step_time_sec": 8.236974766012281, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.177952766418457, "lr": 0.00019481477277064592, "elapsed_sec": 77296.24543213844, "step_time_sec": 8.238549617002718, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.10152530670166, "lr": 0.00019481029164551686, "elapsed_sec": 77304.48314881325, "step_time_sec": 8.237589866039343, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.089390277862549, "lr": 0.00019480580863784875, "elapsed_sec": 77312.72180056572, "step_time_sec": 8.238421222951729, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.007646560668945, "lr": 0.00019480132374773306, "elapsed_sec": 77320.96162605286, "step_time_sec": 8.239683202991728, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.061314105987549, "lr": 0.0001947968369752612, "elapsed_sec": 77329.19951295853, "step_time_sec": 8.237777665955946, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.118502616882324, "lr": 0.00019479234832052471, "elapsed_sec": 77337.44326400757, "step_time_sec": 8.243535306013655, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.177554130554199, "lr": 0.00019478785778361514, "elapsed_sec": 77345.68761181831, "step_time_sec": 8.244149125996046, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.109888076782227, "lr": 0.00019478336536462405, "elapsed_sec": 77353.92593884468, "step_time_sec": 8.238251515955199, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.11248254776001, "lr": 0.00019477887106364306, "elapsed_sec": 77362.16511845589, "step_time_sec": 8.238951230014209, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.232721328735352, "lr": 0.0001947743748807638, "elapsed_sec": 77370.40222549438, "step_time_sec": 8.236950683989562, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.214748859405518, "lr": 0.000194769876816078, "elapsed_sec": 77378.64128923416, "step_time_sec": 8.23886333196424, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.251294136047363, "lr": 0.00019476537686967738, "elapsed_sec": 77386.87959647179, "step_time_sec": 8.238148668955546, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.095942974090576, "lr": 0.00019476087504165368, "elapsed_sec": 77395.1163341999, "step_time_sec": 8.236634785018396, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9608421325683594, "lr": 0.00019475637133209875, "elapsed_sec": 77403.35588598251, "step_time_sec": 8.23934118199395, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.045053482055664, "lr": 0.00019475186574110444, "elapsed_sec": 77411.59405899048, "step_time_sec": 8.238028720021248, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.167144298553467, "lr": 0.00019474735826876258, "elapsed_sec": 77419.8333556652, "step_time_sec": 8.239094714983366, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.967867851257324, "lr": 0.0001947428489151652, "elapsed_sec": 77428.07189655304, "step_time_sec": 8.238415100960992, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8202123641967773, "lr": 0.00019473833768040414, "elapsed_sec": 77436.3097088337, "step_time_sec": 8.237685221014544, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.079310894012451, "lr": 0.00019473382456457146, "elapsed_sec": 77444.54814386368, "step_time_sec": 8.238275015959516, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.115165710449219, "lr": 0.0001947293095677592, "elapsed_sec": 77452.7855796814, "step_time_sec": 8.237193793000188, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.074765682220459, "lr": 0.0001947247926900594, "elapsed_sec": 77461.02476263046, "step_time_sec": 8.239038555009756, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.054830074310303, "lr": 0.00019472027393156423, "elapsed_sec": 77469.26342225075, "step_time_sec": 8.238505969988182, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.095990180969238, "lr": 0.00019471575329236583, "elapsed_sec": 77477.50078821182, "step_time_sec": 8.237180685973726, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.140671253204346, "lr": 0.00019471123077255636, "elapsed_sec": 77485.74057388306, "step_time_sec": 8.239648076007143, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.268886089324951, "lr": 0.0001947067063722281, "elapsed_sec": 77493.97761154175, "step_time_sec": 8.236854491988197, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.036461353302002, "lr": 0.00019470218009147328, "elapsed_sec": 77502.21662545204, "step_time_sec": 8.238870330969803, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9986307621002197, "lr": 0.00019469765193038418, "elapsed_sec": 77510.45577430725, "step_time_sec": 8.238990402023774, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.034311294555664, "lr": 0.0001946931218890532, "elapsed_sec": 77518.69380903244, "step_time_sec": 8.23784297396196, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.085796356201172, "lr": 0.00019468858996757272, "elapsed_sec": 77526.93259549141, "step_time_sec": 8.238621028023772, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8475186824798584, "lr": 0.00019468405616603516, "elapsed_sec": 77535.17054748535, "step_time_sec": 8.237838934001047, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1348419189453125, "lr": 0.00019467952048453293, "elapsed_sec": 77543.4098174572, "step_time_sec": 8.239066295034718, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9567813873291016, "lr": 0.0001946749829231586, "elapsed_sec": 77551.64773631096, "step_time_sec": 8.237749781052116, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.075841426849365, "lr": 0.00019467044348200468, "elapsed_sec": 77559.88658761978, "step_time_sec": 8.238702673988882, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.127989292144775, "lr": 0.0001946659021611637, "elapsed_sec": 77568.12481832504, "step_time_sec": 8.238060327013955, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9542911052703857, "lr": 0.00019466135896072835, "elapsed_sec": 77576.36253881454, "step_time_sec": 8.237575527979061, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.952486276626587, "lr": 0.00019465681388079124, "elapsed_sec": 77584.60105967522, "step_time_sec": 8.238345496996772, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.156352519989014, "lr": 0.00019465226692144506, "elapsed_sec": 77592.91355323792, "step_time_sec": 8.248570040042978, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.060172080993652, "lr": 0.00019464771808278255, "elapsed_sec": 77601.1524784565, "step_time_sec": 8.238770621013828, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.091479301452637, "lr": 0.00019464316736489648, "elapsed_sec": 77609.38946318626, "step_time_sec": 8.236829129047692, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.15731954574585, "lr": 0.00019463861476787963, "elapsed_sec": 77617.62742638588, "step_time_sec": 8.237763852986973, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.012267589569092, "lr": 0.00019463406029182487, "elapsed_sec": 77625.86450719833, "step_time_sec": 8.236954394029453, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.180094242095947, "lr": 0.00019462950393682506, "elapsed_sec": 77634.10261559486, "step_time_sec": 8.237899653962813, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.225440502166748, "lr": 0.00019462494570297317, "elapsed_sec": 77642.3411026001, "step_time_sec": 8.238365560013335, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.953155994415283, "lr": 0.0001946203855903621, "elapsed_sec": 77650.57962203026, "step_time_sec": 8.238334714027587, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.098118782043457, "lr": 0.00019461582359908488, "elapsed_sec": 77658.81869387627, "step_time_sec": 8.238891226996202, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.187959671020508, "lr": 0.00019461125972923454, "elapsed_sec": 77667.05583810806, "step_time_sec": 8.236988304008264, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9655601978302, "lr": 0.00019460669398090413, "elapsed_sec": 77675.29407143593, "step_time_sec": 8.238053776032757, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.945415735244751, "lr": 0.0001946021263541868, "elapsed_sec": 77683.53285956383, "step_time_sec": 8.238608932006173, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.024172306060791, "lr": 0.00019459755684917568, "elapsed_sec": 77691.77109408379, "step_time_sec": 8.238054374000058, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0712995529174805, "lr": 0.00019459298546596396, "elapsed_sec": 77700.0097231865, "step_time_sec": 8.238475653983187, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0390849113464355, "lr": 0.00019458841220464486, "elapsed_sec": 77708.24753975868, "step_time_sec": 8.237650005030446, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.946399211883545, "lr": 0.00019458383706531167, "elapsed_sec": 77716.48741459846, "step_time_sec": 8.239724206039682, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9507734775543213, "lr": 0.00019457926004805767, "elapsed_sec": 77724.72588276863, "step_time_sec": 8.23831791500561, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.035670280456543, "lr": 0.0001945746811529762, "elapsed_sec": 77732.96455383301, "step_time_sec": 8.2384403150063, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.094283103942871, "lr": 0.00019457010038016065, "elapsed_sec": 77741.20460605621, "step_time_sec": 8.239917763974518, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.185435771942139, "lr": 0.00019456551772970446, "elapsed_sec": 77749.44189023972, "step_time_sec": 8.237117409007624, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.082812786102295, "lr": 0.00019456093320170103, "elapsed_sec": 77757.68208837509, "step_time_sec": 8.24002676497912, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.085611820220947, "lr": 0.0001945563467962439, "elapsed_sec": 77765.92103815079, "step_time_sec": 8.238800357969012, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.035006046295166, "lr": 0.0001945517585134266, "elapsed_sec": 77774.15930891037, "step_time_sec": 8.238079175993335, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.969635248184204, "lr": 0.00019454716835334273, "elapsed_sec": 77782.39806175232, "step_time_sec": 8.238662997027859, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.913773775100708, "lr": 0.0001945425763160858, "elapsed_sec": 77790.63545155525, "step_time_sec": 8.237176724011078, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.012186527252197, "lr": 0.00019453798240174955, "elapsed_sec": 77798.8732676506, "step_time_sec": 8.23761625395855, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.876089096069336, "lr": 0.00019453338661042766, "elapsed_sec": 77807.11127066612, "step_time_sec": 8.23784939700272, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.034594535827637, "lr": 0.00019452878894221382, "elapsed_sec": 77815.34898614883, "step_time_sec": 8.237546899006702, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.025942802429199, "lr": 0.0001945241893972018, "elapsed_sec": 77823.58817195892, "step_time_sec": 8.23901839996688, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8705852031707764, "lr": 0.0001945195879754854, "elapsed_sec": 77831.82558107376, "step_time_sec": 8.23720991902519, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.035558700561523, "lr": 0.00019451498467715846, "elapsed_sec": 77840.06449627876, "step_time_sec": 8.23878140503075, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.150724411010742, "lr": 0.00019451037950231486, "elapsed_sec": 77848.30216860771, "step_time_sec": 8.237509374972433, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.987410306930542, "lr": 0.00019450577245104856, "elapsed_sec": 77856.54018688202, "step_time_sec": 8.237850923032966, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.71364164352417, "lr": 0.00019450116352345346, "elapsed_sec": 77864.77795386314, "step_time_sec": 8.237637748999987, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8978147506713867, "lr": 0.00019449655271962357, "elapsed_sec": 77873.01536154747, "step_time_sec": 8.237220097042155, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9451467990875244, "lr": 0.0001944919400396529, "elapsed_sec": 77881.25269460678, "step_time_sec": 8.237144852988422, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.016474723815918, "lr": 0.00019448732548363554, "elapsed_sec": 77889.48976898193, "step_time_sec": 8.236925231991336, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.073068618774414, "lr": 0.0001944827090516656, "elapsed_sec": 77897.7290380001, "step_time_sec": 8.23908540501725, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.986685276031494, "lr": 0.00019447809074383724, "elapsed_sec": 77905.96634602547, "step_time_sec": 8.23721250204835, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.158895015716553, "lr": 0.0001944734705602446, "elapsed_sec": 77914.2049293518, "step_time_sec": 8.23837768100202, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9891316890716553, "lr": 0.00019446884850098193, "elapsed_sec": 77922.44235992432, "step_time_sec": 8.23727649700595, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9465010166168213, "lr": 0.0001944642245661435, "elapsed_sec": 77930.67945384979, "step_time_sec": 8.236953095009085, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.136425971984863, "lr": 0.0001944595987558236, "elapsed_sec": 77938.9193713665, "step_time_sec": 8.239766545011662, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.91591477394104, "lr": 0.00019445497107011656, "elapsed_sec": 77947.15763354301, "step_time_sec": 8.238042119948659, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.071968078613281, "lr": 0.00019445034150911676, "elapsed_sec": 77955.39679551125, "step_time_sec": 8.239014631020837, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.030137538909912, "lr": 0.0001944457100729186, "elapsed_sec": 77963.63512778282, "step_time_sec": 8.238151694997214, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9352900981903076, "lr": 0.00019444107676161654, "elapsed_sec": 77971.87224411964, "step_time_sec": 8.236935253022239, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.11196231842041, "lr": 0.00019443644157530506, "elapsed_sec": 77980.11178278923, "step_time_sec": 8.239418530021794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.095266819000244, "lr": 0.00019443180451407874, "elapsed_sec": 77988.34949970245, "step_time_sec": 8.237548918987159, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.117903709411621, "lr": 0.00019442716557803208, "elapsed_sec": 77996.58992052078, "step_time_sec": 8.240268796042074, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.94673490524292, "lr": 0.00019442252476725973, "elapsed_sec": 78004.8286306858, "step_time_sec": 8.238554447016213, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.027241230010986, "lr": 0.00019441788208185632, "elapsed_sec": 78013.0658864975, "step_time_sec": 8.237021345004905, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8185055255889893, "lr": 0.00019441323752191652, "elapsed_sec": 78021.30493235588, "step_time_sec": 8.238886900013313, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9456183910369873, "lr": 0.00019440859108753504, "elapsed_sec": 78029.54276299477, "step_time_sec": 8.237688943045214, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.024649620056152, "lr": 0.00019440394277880666, "elapsed_sec": 78037.78210639954, "step_time_sec": 8.239157723030075, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8560824394226074, "lr": 0.0001943992925958262, "elapsed_sec": 78046.02109217644, "step_time_sec": 8.23885220097145, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0353899002075195, "lr": 0.0001943946405386884, "elapsed_sec": 78054.25853705406, "step_time_sec": 8.237248838995583, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.041299819946289, "lr": 0.00019438998660748825, "elapsed_sec": 78062.4971241951, "step_time_sec": 8.238418148015626, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.069791793823242, "lr": 0.00019438533080232056, "elapsed_sec": 78070.73377990723, "step_time_sec": 8.236460734973662, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.07273530960083, "lr": 0.00019438067312328036, "elapsed_sec": 78078.97329425812, "step_time_sec": 8.239398324047215, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.045269012451172, "lr": 0.0001943760135704626, "elapsed_sec": 78087.21300697327, "step_time_sec": 8.239565264026169, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9849984645843506, "lr": 0.0001943713521439623, "elapsed_sec": 78095.4491610527, "step_time_sec": 8.235944796004333, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9128315448760986, "lr": 0.00019436668884387454, "elapsed_sec": 78103.68760180473, "step_time_sec": 8.238286578969564, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.005390644073486, "lr": 0.00019436202367029438, "elapsed_sec": 78111.92395663261, "step_time_sec": 8.236164758971427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9954802989959717, "lr": 0.00019435735662331701, "elapsed_sec": 78120.16229867935, "step_time_sec": 8.238184326968621, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9775941371917725, "lr": 0.00019435268770303757, "elapsed_sec": 78128.39962625504, "step_time_sec": 8.237157607043628, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.958256721496582, "lr": 0.00019434801690955132, "elapsed_sec": 78136.6386988163, "step_time_sec": 8.23892255598912, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.964557647705078, "lr": 0.00019434334424295346, "elapsed_sec": 78144.87749814987, "step_time_sec": 8.238686116994359, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8645431995391846, "lr": 0.00019433866970333934, "elapsed_sec": 78153.11410808563, "step_time_sec": 8.236389130004682, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.898890733718872, "lr": 0.00019433399329080425, "elapsed_sec": 78161.35331106186, "step_time_sec": 8.23905511497287, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.037949562072754, "lr": 0.00019432931500544355, "elapsed_sec": 78169.59111166, "step_time_sec": 8.237644505978096, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.010610103607178, "lr": 0.0001943246348473527, "elapsed_sec": 78177.83077001572, "step_time_sec": 8.239486447011586, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.984623908996582, "lr": 0.0001943199528166271, "elapsed_sec": 78186.06953787804, "step_time_sec": 8.238607560051605, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.033960342407227, "lr": 0.00019431526891336225, "elapsed_sec": 78194.30703401566, "step_time_sec": 8.237299024011008, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9945311546325684, "lr": 0.00019431058313765367, "elapsed_sec": 78202.54600906372, "step_time_sec": 8.238814789045136, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7781248092651367, "lr": 0.00019430589548959692, "elapsed_sec": 78210.78373456001, "step_time_sec": 8.237541532027535, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.920118570327759, "lr": 0.00019430120596928757, "elapsed_sec": 78219.0220644474, "step_time_sec": 8.238192782970145, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.03650426864624, "lr": 0.00019429651457682134, "elapsed_sec": 78227.25932979584, "step_time_sec": 8.237087822053581, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.009178161621094, "lr": 0.00019429182131229378, "elapsed_sec": 78235.49674367905, "step_time_sec": 8.237269596022088, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.924252986907959, "lr": 0.0001942871261758007, "elapsed_sec": 78243.73693227768, "step_time_sec": 8.240011331974529, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.242064476013184, "lr": 0.00019428242916743783, "elapsed_sec": 78251.97336411476, "step_time_sec": 8.236321100965142, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.034132480621338, "lr": 0.00019427773028730092, "elapsed_sec": 78260.21234345436, "step_time_sec": 8.238761255983263, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.183067798614502, "lr": 0.00019427302953548586, "elapsed_sec": 78268.45103693008, "step_time_sec": 8.238542037957814, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.180225849151611, "lr": 0.00019426832691208845, "elapsed_sec": 78276.68915081024, "step_time_sec": 8.237939639016986, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9629690647125244, "lr": 0.00019426362241720463, "elapsed_sec": 78284.92777085304, "step_time_sec": 8.23847778403433, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.973005771636963, "lr": 0.00019425891605093034, "elapsed_sec": 78293.1654818058, "step_time_sec": 8.237540172005538, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.08891487121582, "lr": 0.00019425420781336156, "elapsed_sec": 78301.40371346474, "step_time_sec": 8.238079031987581, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.954044818878174, "lr": 0.0001942494977045943, "elapsed_sec": 78309.64226651192, "step_time_sec": 8.23837751802057, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9736101627349854, "lr": 0.00019424478572472461, "elapsed_sec": 78317.88148546219, "step_time_sec": 8.239077321020886, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9078469276428223, "lr": 0.00019424007187384863, "elapsed_sec": 78326.11976790428, "step_time_sec": 8.238146340008825, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.760610580444336, "lr": 0.00019423535615206242, "elapsed_sec": 78334.35700774193, "step_time_sec": 8.237081832019612, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.112363338470459, "lr": 0.00019423063855946219, "elapsed_sec": 78342.59581923485, "step_time_sec": 8.23859447799623, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0159173011779785, "lr": 0.00019422591909614418, "elapsed_sec": 78350.83339571953, "step_time_sec": 8.237400764017366, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8746509552001953, "lr": 0.00019422119776220457, "elapsed_sec": 78359.07314682007, "step_time_sec": 8.239644429006148, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9456872940063477, "lr": 0.0001942164745577397, "elapsed_sec": 78367.3119251728, "step_time_sec": 8.238543560029939, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9285600185394287, "lr": 0.00019421174948284584, "elapsed_sec": 78375.54963588715, "step_time_sec": 8.237559421046171, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.974612236022949, "lr": 0.00019420702253761937, "elapsed_sec": 78383.78828668594, "step_time_sec": 8.238513304037042, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8561513423919678, "lr": 0.00019420229372215674, "elapsed_sec": 78392.02668714523, "step_time_sec": 8.238242872001138, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.006877422332764, "lr": 0.00019419756303655435, "elapsed_sec": 78400.26550483704, "step_time_sec": 8.2386693249573, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.066925048828125, "lr": 0.00019419283048090864, "elapsed_sec": 78408.50340557098, "step_time_sec": 8.237729774962645, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8165595531463623, "lr": 0.0001941880960553162, "elapsed_sec": 78416.74097037315, "step_time_sec": 8.2374134939746, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.062892913818359, "lr": 0.0001941833597598735, "elapsed_sec": 78424.9795050621, "step_time_sec": 8.238390661950689, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8823628425598145, "lr": 0.00019417862159467722, "elapsed_sec": 78433.21592330933, "step_time_sec": 8.236194684985094, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.289017200469971, "lr": 0.00019417388155982388, "elapsed_sec": 78441.4542760849, "step_time_sec": 8.238243304018397, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9531033039093018, "lr": 0.00019416913965541023, "elapsed_sec": 78449.69284820557, "step_time_sec": 8.238403762981761, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9151031970977783, "lr": 0.00019416439588153294, "elapsed_sec": 78457.93124198914, "step_time_sec": 8.238239929021802, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9164466857910156, "lr": 0.00019415965023828879, "elapsed_sec": 78466.16970920563, "step_time_sec": 8.238320959964767, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.984097719192505, "lr": 0.00019415490272577451, "elapsed_sec": 78474.40724492073, "step_time_sec": 8.237385796033777, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.99869441986084, "lr": 0.00019415015334408694, "elapsed_sec": 78482.64669299126, "step_time_sec": 8.239231345010921, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9876961708068848, "lr": 0.00019414540209332294, "elapsed_sec": 78490.88484597206, "step_time_sec": 8.237988567969296, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9648053646087646, "lr": 0.00019414064897357942, "elapsed_sec": 78499.1235613823, "step_time_sec": 8.238555537012871, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.048493385314941, "lr": 0.0001941358939849533, "elapsed_sec": 78507.36198019981, "step_time_sec": 8.238293545029592, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8834218978881836, "lr": 0.00019413113712754153, "elapsed_sec": 78515.59956622124, "step_time_sec": 8.237395305011887, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.882219076156616, "lr": 0.00019412637840144115, "elapsed_sec": 78523.83857655525, "step_time_sec": 8.238870914967265, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.034607887268066, "lr": 0.0001941216178067492, "elapsed_sec": 78532.07807803154, "step_time_sec": 8.23939227097435, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.099276065826416, "lr": 0.00019411685534356275, "elapsed_sec": 78540.31783246994, "step_time_sec": 8.239530706021469, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.049160480499268, "lr": 0.0001941120910119789, "elapsed_sec": 78548.55861544609, "step_time_sec": 8.240622468001675, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0022501945495605, "lr": 0.00019410732481209492, "elapsed_sec": 78556.79569983482, "step_time_sec": 8.236885423015337, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.011683940887451, "lr": 0.0001941025567440079, "elapsed_sec": 78565.03526544571, "step_time_sec": 8.239504969969857, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.027884483337402, "lr": 0.00019409778680781512, "elapsed_sec": 78573.27273726463, "step_time_sec": 8.237226455996279, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.122255325317383, "lr": 0.00019409301500361383, "elapsed_sec": 78581.51090145111, "step_time_sec": 8.238047541002743, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1392645835876465, "lr": 0.00019408824133150136, "elapsed_sec": 78589.75028324127, "step_time_sec": 8.239201473013964, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9604945182800293, "lr": 0.0001940834657915751, "elapsed_sec": 78597.98775649071, "step_time_sec": 8.237274028011598, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9495651721954346, "lr": 0.0001940786883839324, "elapsed_sec": 78606.22722244263, "step_time_sec": 8.239328898023814, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9872312545776367, "lr": 0.00019407390910867068, "elapsed_sec": 78614.46506857872, "step_time_sec": 8.237672273011412, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.07373046875, "lr": 0.00019406912796588745, "elapsed_sec": 78622.70492601395, "step_time_sec": 8.239669233968016, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.023993015289307, "lr": 0.00019406434495568015, "elapsed_sec": 78630.94338560104, "step_time_sec": 8.238347773964051, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9267358779907227, "lr": 0.00019405956007814638, "elapsed_sec": 78639.18132996559, "step_time_sec": 8.237727394967806, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.176957607269287, "lr": 0.0001940547733333837, "elapsed_sec": 78647.42012095451, "step_time_sec": 8.23862682597246, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.137021064758301, "lr": 0.00019404998472148975, "elapsed_sec": 78655.65774083138, "step_time_sec": 8.237453808018472, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.04119873046875, "lr": 0.00019404519424256214, "elapsed_sec": 78663.8981089592, "step_time_sec": 8.240222161985002, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.060086250305176, "lr": 0.00019404040189669863, "elapsed_sec": 78672.13662028313, "step_time_sec": 8.23834334599087, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.043354511260986, "lr": 0.00019403560768399685, "elapsed_sec": 78680.3731687069, "step_time_sec": 8.236403912014794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.93481707572937, "lr": 0.0001940308116045547, "elapsed_sec": 78688.61077904701, "step_time_sec": 8.237407097010873, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.002753257751465, "lr": 0.00019402601365846992, "elapsed_sec": 78696.84972167015, "step_time_sec": 8.2387684059795, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.031991481781006, "lr": 0.00019402121384584032, "elapsed_sec": 78705.08843946457, "step_time_sec": 8.238596863986459, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.895904779434204, "lr": 0.00019401641216676386, "elapsed_sec": 78713.32694911957, "step_time_sec": 8.238300783967134, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.028195858001709, "lr": 0.0001940116086213384, "elapsed_sec": 78721.5647456646, "step_time_sec": 8.237670454022009, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0956711769104, "lr": 0.000194006803209662, "elapsed_sec": 78729.80382704735, "step_time_sec": 8.238890631007962, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.054383277893066, "lr": 0.00019400199593183254, "elapsed_sec": 78738.04065299034, "step_time_sec": 8.236666866985615, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8428962230682373, "lr": 0.0001939971867879481, "elapsed_sec": 78746.27911162376, "step_time_sec": 8.2383440049598, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.987243890762329, "lr": 0.00019399237577810683, "elapsed_sec": 78754.51731348038, "step_time_sec": 8.238045053032693, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.165281772613525, "lr": 0.00019398756290240675, "elapsed_sec": 78762.75528693199, "step_time_sec": 8.237769679981284, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.251726150512695, "lr": 0.00019398274816094602, "elapsed_sec": 78770.99389243126, "step_time_sec": 8.238416440028232, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.261185646057129, "lr": 0.00019397793155382287, "elapsed_sec": 78779.23114037514, "step_time_sec": 8.237111198017374, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.104146480560303, "lr": 0.0001939731130811355, "elapsed_sec": 78787.47002840042, "step_time_sec": 8.238700337009504, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.082162380218506, "lr": 0.0001939682927429822, "elapsed_sec": 78795.70745635033, "step_time_sec": 8.237308763025794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.072154998779297, "lr": 0.00019396347053946125, "elapsed_sec": 78803.94547653198, "step_time_sec": 8.237838014028966, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.079604625701904, "lr": 0.000193958646470671, "elapsed_sec": 78812.18482279778, "step_time_sec": 8.239161559031345, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.979358434677124, "lr": 0.00019395382053670986, "elapsed_sec": 78820.42109155655, "step_time_sec": 8.23607807396911, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.897441864013672, "lr": 0.0001939489927376762, "elapsed_sec": 78828.66112089157, "step_time_sec": 8.23990014300216, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0940117835998535, "lr": 0.00019394416307366854, "elapsed_sec": 78836.89940428734, "step_time_sec": 8.238110742997378, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.087461471557617, "lr": 0.00019393933154478527, "elapsed_sec": 78845.1381123066, "step_time_sec": 8.238507658999879, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9939475059509277, "lr": 0.00019393449815112502, "elapsed_sec": 78853.37678933144, "step_time_sec": 8.23853924596915, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.047393321990967, "lr": 0.0001939296628927863, "elapsed_sec": 78861.614538908, "step_time_sec": 8.237596464983653, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.028186321258545, "lr": 0.00019392482576986776, "elapsed_sec": 78869.85268616676, "step_time_sec": 8.23797266400652, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.044715404510498, "lr": 0.00019391998678246802, "elapsed_sec": 78878.08974909782, "step_time_sec": 8.236918834969401, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.12681770324707, "lr": 0.00019391514593068576, "elapsed_sec": 78886.32892894745, "step_time_sec": 8.238992413971573, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.195313453674316, "lr": 0.00019391030321461975, "elapsed_sec": 78894.5657966137, "step_time_sec": 8.236770364048425, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.119227409362793, "lr": 0.00019390545863436868, "elapsed_sec": 78902.80336070061, "step_time_sec": 8.237342667998746, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9322948455810547, "lr": 0.0001939006121900314, "elapsed_sec": 78911.04197454453, "step_time_sec": 8.23847946402384, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.105030536651611, "lr": 0.00019389576388170675, "elapsed_sec": 78919.27915525436, "step_time_sec": 8.23699054500321, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.141928672790527, "lr": 0.00019389091370949357, "elapsed_sec": 78927.51917791367, "step_time_sec": 8.239859635010362, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2000555992126465, "lr": 0.0001938860616734908, "elapsed_sec": 78935.75639867783, "step_time_sec": 8.237096319964621, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.12462854385376, "lr": 0.00019388120777379735, "elapsed_sec": 78943.9956202507, "step_time_sec": 8.23902163299499, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.130284309387207, "lr": 0.00019387635201051224, "elapsed_sec": 78952.2330391407, "step_time_sec": 8.237270102021284, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.159825801849365, "lr": 0.00019387149438373453, "elapsed_sec": 78960.4710097313, "step_time_sec": 8.237790515006054, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.030231475830078, "lr": 0.0001938666348935632, "elapsed_sec": 78968.71115326881, "step_time_sec": 8.240043610043358, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.144689559936523, "lr": 0.00019386177354009743, "elapsed_sec": 78976.94931840897, "step_time_sec": 8.238026135019027, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.991422176361084, "lr": 0.00019385691032343632, "elapsed_sec": 78985.18836712837, "step_time_sec": 31.526128610013984, "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": 4.084225177764893, "lr": 0.00019385204524367906, "elapsed_sec": 79016.72470736504, "step_time_sec": 8.24855308199767, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.180433750152588, "lr": 0.00019384717830092486, "elapsed_sec": 79024.95539307594, "step_time_sec": 8.230539238953497, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.164090156555176, "lr": 0.000193842309495273, "elapsed_sec": 79033.1868185997, "step_time_sec": 8.231224018963985, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.978128433227539, "lr": 0.00019383743882682273, "elapsed_sec": 79041.41809082031, "step_time_sec": 8.23108423902886, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.024938106536865, "lr": 0.00019383256629567342, "elapsed_sec": 79049.6504354477, "step_time_sec": 8.23220994998701, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.029533386230469, "lr": 0.0001938276919019244, "elapsed_sec": 79057.8814022541, "step_time_sec": 8.230773083050735, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.179406642913818, "lr": 0.00019382281564567512, "elapsed_sec": 79066.11482524872, "step_time_sec": 8.233282318979036, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.115835189819336, "lr": 0.00019381793752702497, "elapsed_sec": 79074.3515150547, "step_time_sec": 8.236506004992407, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1740546226501465, "lr": 0.00019381305754607352, "elapsed_sec": 79082.58999037743, "step_time_sec": 8.23832039401168, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.169172763824463, "lr": 0.0001938081757029202, "elapsed_sec": 79090.82481384277, "step_time_sec": 8.234654226980638, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.023182392120361, "lr": 0.00019380329199766461, "elapsed_sec": 79099.06308984756, "step_time_sec": 8.238102022965904, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.954077959060669, "lr": 0.00019379840643040633, "elapsed_sec": 79107.30056786537, "step_time_sec": 8.237344621040393, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.040851593017578, "lr": 0.00019379351900124502, "elapsed_sec": 79115.53729104996, "step_time_sec": 8.236555048963055, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.184023380279541, "lr": 0.00019378862971028033, "elapsed_sec": 79123.77544569969, "step_time_sec": 8.238011876994278, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9436655044555664, "lr": 0.00019378373855761196, "elapsed_sec": 79132.01192164421, "step_time_sec": 8.23628725099843, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.141924858093262, "lr": 0.0001937788455433397, "elapsed_sec": 79140.25171160698, "step_time_sec": 8.23959922499489, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0429816246032715, "lr": 0.0001937739506675633, "elapsed_sec": 79148.49119257927, "step_time_sec": 8.239327389048412, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.096783638000488, "lr": 0.00019376905393038256, "elapsed_sec": 79156.72953557968, "step_time_sec": 8.238195850979537, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9007723331451416, "lr": 0.00019376415533189743, "elapsed_sec": 79164.96836352348, "step_time_sec": 8.23867305595195, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9034154415130615, "lr": 0.00019375925487220776, "elapsed_sec": 79173.20542144775, "step_time_sec": 8.23685798002407, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.110035419464111, "lr": 0.00019375435255141347, "elapsed_sec": 79181.4450340271, "step_time_sec": 8.239436838019174, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.103397846221924, "lr": 0.0001937494483696145, "elapsed_sec": 79189.68341207504, "step_time_sec": 8.238232266972773, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9997050762176514, "lr": 0.00019374454232691098, "elapsed_sec": 79197.92188858986, "step_time_sec": 8.238296070019715, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.200720310211182, "lr": 0.00019373963442340287, "elapsed_sec": 79206.16041111946, "step_time_sec": 8.23837049404392, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.158773899078369, "lr": 0.0001937347246591903, "elapsed_sec": 79214.39659047127, "step_time_sec": 8.236021263990551, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.030430793762207, "lr": 0.00019372981303437337, "elapsed_sec": 79222.63510656357, "step_time_sec": 8.23837120400276, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.037315368652344, "lr": 0.0001937248995490523, "elapsed_sec": 79230.87314462662, "step_time_sec": 8.237853750004433, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.140174388885498, "lr": 0.0001937199842033272, "elapsed_sec": 79239.11192703247, "step_time_sec": 8.23862001200905, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.127808094024658, "lr": 0.0001937150669972984, "elapsed_sec": 79247.35045933723, "step_time_sec": 8.2384317459655, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0626630783081055, "lr": 0.00019371014793106612, "elapsed_sec": 79255.58830070496, "step_time_sec": 8.237628649978433, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.089317798614502, "lr": 0.0001937052270047307, "elapsed_sec": 79263.82808589935, "step_time_sec": 8.239620384993032, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.115924835205078, "lr": 0.00019370030421839252, "elapsed_sec": 79272.06459403038, "step_time_sec": 8.236412100028247, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.064099311828613, "lr": 0.00019369537957215194, "elapsed_sec": 79280.30454850197, "step_time_sec": 8.239781694952399, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1144185066223145, "lr": 0.00019369045306610942, "elapsed_sec": 79288.54165506363, "step_time_sec": 8.236917557020206, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.057675361633301, "lr": 0.0001936855247003654, "elapsed_sec": 79296.7793700695, "step_time_sec": 8.237598124018405, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.001878261566162, "lr": 0.0001936805944750204, "elapsed_sec": 79305.01871776581, "step_time_sec": 8.23916483100038, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.989158868789673, "lr": 0.00019367566239017494, "elapsed_sec": 79313.25616717339, "step_time_sec": 8.237233588006347, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9376227855682373, "lr": 0.00019367072844592964, "elapsed_sec": 79321.49690699577, "step_time_sec": 8.240569835994393, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.011443138122559, "lr": 0.00019366579264238511, "elapsed_sec": 79329.73678159714, "step_time_sec": 8.239729323948268, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.171151161193848, "lr": 0.00019366085497964206, "elapsed_sec": 79337.97452950478, "step_time_sec": 8.237582047993783, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0066609382629395, "lr": 0.00019365591545780105, "elapsed_sec": 79346.21297168732, "step_time_sec": 8.238305778999347, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.109755039215088, "lr": 0.00019365097407696293, "elapsed_sec": 79354.45108151436, "step_time_sec": 8.237969601992518, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.99660062789917, "lr": 0.00019364603083722845, "elapsed_sec": 79362.69052815437, "step_time_sec": 8.239265197014902, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9249799251556396, "lr": 0.0001936410857386984, "elapsed_sec": 79370.92794132233, "step_time_sec": 8.237208902020939, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.094992160797119, "lr": 0.00019363613878147366, "elapsed_sec": 79379.16707062721, "step_time_sec": 8.238994769053534, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9466636180877686, "lr": 0.00019363118996565504, "elapsed_sec": 79387.40615534782, "step_time_sec": 8.238845597021282, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.140856742858887, "lr": 0.00019362623929134357, "elapsed_sec": 79395.6432454586, "step_time_sec": 8.23691170202801, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.029986381530762, "lr": 0.00019362128675864016, "elapsed_sec": 79403.88346195221, "step_time_sec": 8.240083681012038, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1303391456604, "lr": 0.00019361633236764577, "elapsed_sec": 79412.1209886074, "step_time_sec": 8.23736764898058, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.051626682281494, "lr": 0.00019361137611846152, "elapsed_sec": 79420.35800123215, "step_time_sec": 8.23681932099862, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.017646312713623, "lr": 0.00019360641801118844, "elapsed_sec": 79428.59671378136, "step_time_sec": 8.238561610050965, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9032676219940186, "lr": 0.00019360145804592764, "elapsed_sec": 79436.83385682106, "step_time_sec": 8.23702185699949, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9831669330596924, "lr": 0.00019359649622278027, "elapsed_sec": 79445.07265210152, "step_time_sec": 8.238590848981403, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.964970111846924, "lr": 0.00019359153254184757, "elapsed_sec": 79453.30931210518, "step_time_sec": 8.236528453009669, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.920358657836914, "lr": 0.0001935865670032307, "elapsed_sec": 79461.54808783531, "step_time_sec": 8.23857968201628, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.018582820892334, "lr": 0.00019358159960703095, "elapsed_sec": 79469.78643417358, "step_time_sec": 8.238211519026663, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.095297813415527, "lr": 0.00019357663035334961, "elapsed_sec": 79478.02443766594, "step_time_sec": 8.237838850996923, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0322184562683105, "lr": 0.00019357165924228806, "elapsed_sec": 79486.26222872734, "step_time_sec": 8.237584076996427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.983210325241089, "lr": 0.00019356668627394766, "elapsed_sec": 79494.49999856949, "step_time_sec": 8.23767714999849, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.039614677429199, "lr": 0.0001935617114484298, "elapsed_sec": 79502.74009656906, "step_time_sec": 8.2398931219941, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.076391696929932, "lr": 0.00019355673476583596, "elapsed_sec": 79510.97888803482, "step_time_sec": 8.238623979035765, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.998866319656372, "lr": 0.00019355175622626762, "elapsed_sec": 79519.21652722359, "step_time_sec": 8.237452438974287, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0982561111450195, "lr": 0.00019354677582982636, "elapsed_sec": 79527.45497536659, "step_time_sec": 8.238307122024707, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1054487228393555, "lr": 0.00019354179357661367, "elapsed_sec": 79535.6922044754, "step_time_sec": 8.237055346020497, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9175426959991455, "lr": 0.00019353680946673123, "elapsed_sec": 79543.93214917183, "step_time_sec": 8.239781540003605, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9301669597625732, "lr": 0.0001935318235002806, "elapsed_sec": 79552.17083334923, "step_time_sec": 8.238545920990873, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9435441493988037, "lr": 0.00019352683567736356, "elapsed_sec": 79560.40928220749, "step_time_sec": 8.238228660018649, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.866020679473877, "lr": 0.00019352184599808175, "elapsed_sec": 79568.64860868454, "step_time_sec": 8.239212089974899, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.020045280456543, "lr": 0.00019351685446253693, "elapsed_sec": 79576.88672304153, "step_time_sec": 8.23790943500353, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.096805572509766, "lr": 0.00019351186107083095, "elapsed_sec": 79585.12552332878, "step_time_sec": 8.238656535977498, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9238452911376953, "lr": 0.00019350686582306561, "elapsed_sec": 79593.36329078674, "step_time_sec": 8.237592167046387, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9932029247283936, "lr": 0.00019350186871934283, "elapsed_sec": 79601.600202322, "step_time_sec": 8.236754807003308, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.943302869796753, "lr": 0.0001934968697597644, "elapsed_sec": 79609.83779001236, "step_time_sec": 8.237433645990677, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.786747455596924, "lr": 0.0001934918689444324, "elapsed_sec": 79618.07630515099, "step_time_sec": 8.238370581995696, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7776100635528564, "lr": 0.00019348686627344874, "elapsed_sec": 79626.31582212448, "step_time_sec": 8.239364081993699, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.926527500152588, "lr": 0.00019348186174691546, "elapsed_sec": 79634.55500674248, "step_time_sec": 8.23902983602602, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0948381423950195, "lr": 0.0001934768553649346, "elapsed_sec": 79642.79277300835, "step_time_sec": 8.2375252659549, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.856095552444458, "lr": 0.0001934718471276083, "elapsed_sec": 79651.03272986412, "step_time_sec": 8.239820990012959, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.022377014160156, "lr": 0.00019346683703503863, "elapsed_sec": 79659.27041649818, "step_time_sec": 8.23749054595828, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9853107929229736, "lr": 0.00019346182508732786, "elapsed_sec": 79667.50799965858, "step_time_sec": 8.237474658002611, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9904897212982178, "lr": 0.00019345681128457814, "elapsed_sec": 79675.74591493607, "step_time_sec": 8.237735219998285, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1638288497924805, "lr": 0.00019345179562689168, "elapsed_sec": 79683.983979702, "step_time_sec": 8.237890529970173, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.152799129486084, "lr": 0.00019344677811437085, "elapsed_sec": 79692.22223114967, "step_time_sec": 8.238156067964155, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.898634910583496, "lr": 0.00019344175874711796, "elapsed_sec": 79700.45841860771, "step_time_sec": 8.235947633045726, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9096596240997314, "lr": 0.00019343673752523534, "elapsed_sec": 79708.69707894325, "step_time_sec": 8.238552086986601, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.10472297668457, "lr": 0.00019343171444882542, "elapsed_sec": 79716.93397378922, "step_time_sec": 8.236669346049894, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.058689594268799, "lr": 0.00019342668951799057, "elapsed_sec": 79725.17404651642, "step_time_sec": 8.239919904968701, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.997375011444092, "lr": 0.00019342166273283336, "elapsed_sec": 79733.41267156601, "step_time_sec": 8.238483472028747, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.149354934692383, "lr": 0.00019341663409345626, "elapsed_sec": 79741.65221571922, "step_time_sec": 8.239345240988769, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.125895977020264, "lr": 0.00019341160359996184, "elapsed_sec": 79749.89244365692, "step_time_sec": 8.24006612697849, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0969462394714355, "lr": 0.00019340657125245265, "elapsed_sec": 79758.1302947998, "step_time_sec": 8.237698481010739, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.086864948272705, "lr": 0.00019340153705103133, "elapsed_sec": 79766.36747145653, "step_time_sec": 8.237036497041117, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9943273067474365, "lr": 0.0001933965009958006, "elapsed_sec": 79774.60595273972, "step_time_sec": 8.238321660959627, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.068560600280762, "lr": 0.00019339146308686313, "elapsed_sec": 79782.84409475327, "step_time_sec": 8.237947296991479, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.930948257446289, "lr": 0.0001933864233243216, "elapsed_sec": 79791.08321928978, "step_time_sec": 8.23894853098318, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.917050838470459, "lr": 0.00019338138170827887, "elapsed_sec": 79799.32036685944, "step_time_sec": 8.237044120032806, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.961559772491455, "lr": 0.00019337633823883775, "elapsed_sec": 79807.55915427208, "step_time_sec": 8.23859991505742, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8875179290771484, "lr": 0.00019337129291610103, "elapsed_sec": 79815.79790639877, "step_time_sec": 8.238543749030214, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.085836410522461, "lr": 0.0001933662457401717, "elapsed_sec": 79824.03605747223, "step_time_sec": 8.238014216010924, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.07122802734375, "lr": 0.0001933611967111526, "elapsed_sec": 79832.27581357956, "step_time_sec": 8.239627485978417, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.015724182128906, "lr": 0.00019335614582914671, "elapsed_sec": 79840.51267576218, "step_time_sec": 8.236655788030475, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8465120792388916, "lr": 0.00019335109309425707, "elapsed_sec": 79848.75230073929, "step_time_sec": 8.23950176796643, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2462005615234375, "lr": 0.00019334603850658673, "elapsed_sec": 79856.99171185493, "step_time_sec": 8.239303472975735, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.037384033203125, "lr": 0.00019334098206623876, "elapsed_sec": 79865.22995138168, "step_time_sec": 8.23801580897998, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9764022827148438, "lr": 0.00019333592377331627, "elapsed_sec": 79873.4686703682, "step_time_sec": 8.238608881947584, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.188136100769043, "lr": 0.00019333086362792243, "elapsed_sec": 79881.70585179329, "step_time_sec": 8.236995375016704, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.062334060668945, "lr": 0.0001933258016301604, "elapsed_sec": 79889.94458413124, "step_time_sec": 8.238566983025521, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.19846248626709, "lr": 0.00019332073778013343, "elapsed_sec": 79898.18290543556, "step_time_sec": 8.238195611978881, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1893157958984375, "lr": 0.0001933156720779448, "elapsed_sec": 79906.4211666584, "step_time_sec": 8.238036205002572, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.034911155700684, "lr": 0.00019331060452369784, "elapsed_sec": 79914.6605284214, "step_time_sec": 8.239280409994535, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.293686389923096, "lr": 0.00019330553511749584, "elapsed_sec": 79922.89804911613, "step_time_sec": 8.23730631096987, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.125295639038086, "lr": 0.00019330046385944224, "elapsed_sec": 79931.13707828522, "step_time_sec": 8.238871517009102, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.085137367248535, "lr": 0.00019329539074964042, "elapsed_sec": 79939.37528514862, "step_time_sec": 8.238057787995785, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.083503723144531, "lr": 0.00019329031578819387, "elapsed_sec": 79947.61292290688, "step_time_sec": 8.237488024984486, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.131006240844727, "lr": 0.00019328523897520604, "elapsed_sec": 79955.8510735035, "step_time_sec": 8.23798691201955, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.065422058105469, "lr": 0.0001932801603107805, "elapsed_sec": 79964.0881152153, "step_time_sec": 8.236916391993873, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.047354221343994, "lr": 0.00019327507979502087, "elapsed_sec": 79972.32666754723, "step_time_sec": 8.238375946995802, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.055856227874756, "lr": 0.00019326999742803064, "elapsed_sec": 79980.56502127647, "step_time_sec": 8.238150348013733, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9240639209747314, "lr": 0.00019326491320991357, "elapsed_sec": 79988.8033657074, "step_time_sec": 8.238177705963608, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7378575801849365, "lr": 0.0001932598271407733, "elapsed_sec": 79997.04245233536, "step_time_sec": 8.238974936015438, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.922900676727295, "lr": 0.00019325473922071352, "elapsed_sec": 80005.27985239029, "step_time_sec": 8.237214114982635, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.082484245300293, "lr": 0.00019324964944983806, "elapsed_sec": 80013.51800227165, "step_time_sec": 8.237988990964368, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1057515144348145, "lr": 0.00019324455782825065, "elapsed_sec": 80021.75569343567, "step_time_sec": 8.237503668002319, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8580851554870605, "lr": 0.00019323946435605518, "elapsed_sec": 80029.99550747871, "step_time_sec": 8.239655309007503, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.040069103240967, "lr": 0.0001932343690333555, "elapsed_sec": 80038.2338244915, "step_time_sec": 8.238185021968093, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.962113380432129, "lr": 0.00019322927186025553, "elapsed_sec": 80046.47178959846, "step_time_sec": 8.237807393015828, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9792797565460205, "lr": 0.0001932241728368592, "elapsed_sec": 80054.70882892609, "step_time_sec": 8.236869095009752, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.214203357696533, "lr": 0.00019321907196327053, "elapsed_sec": 80062.94611096382, "step_time_sec": 8.23705668997718, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.108558177947998, "lr": 0.00019321396923959353, "elapsed_sec": 80071.18562841415, "step_time_sec": 8.239380387996789, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1612229347229, "lr": 0.00019320886466593228, "elapsed_sec": 80079.42346262932, "step_time_sec": 8.237660875951406, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.10445499420166, "lr": 0.0001932037582423908, "elapsed_sec": 80087.66174983978, "step_time_sec": 8.238115159037989, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.01649284362793, "lr": 0.00019319864996907333, "elapsed_sec": 80095.89988088608, "step_time_sec": 8.237938745995052, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0388712882995605, "lr": 0.00019319353984608402, "elapsed_sec": 80104.13718175888, "step_time_sec": 8.237129157001618, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.064483642578125, "lr": 0.00019318842787352706, "elapsed_sec": 80112.37461256981, "step_time_sec": 8.237281116016675, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.987250328063965, "lr": 0.00019318331405150668, "elapsed_sec": 80120.61181950569, "step_time_sec": 8.237034294987097, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0985107421875, "lr": 0.00019317819838012725, "elapsed_sec": 80128.8505654335, "step_time_sec": 8.238613096997142, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.055643558502197, "lr": 0.00019317308085949303, "elapsed_sec": 80137.08783364296, "step_time_sec": 8.237063818029128, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9699480533599854, "lr": 0.00019316796148970838, "elapsed_sec": 80145.32432889938, "step_time_sec": 8.236364305019379, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.971113443374634, "lr": 0.00019316284027087776, "elapsed_sec": 80153.56194591522, "step_time_sec": 8.237442951998673, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.107799053192139, "lr": 0.00019315771720310556, "elapsed_sec": 80161.79926013947, "step_time_sec": 8.237133337010164, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0778279304504395, "lr": 0.00019315259228649628, "elapsed_sec": 80170.03611564636, "step_time_sec": 8.236717132036574, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8884060382843018, "lr": 0.0001931474655211544, "elapsed_sec": 80178.27353191376, "step_time_sec": 8.237242110015359, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.085160732269287, "lr": 0.00019314233690718454, "elapsed_sec": 80186.51259946823, "step_time_sec": 8.238927907019388, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.11271858215332, "lr": 0.00019313720644469122, "elapsed_sec": 80194.75093841553, "step_time_sec": 8.238202159991488, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.104538917541504, "lr": 0.0001931320741337791, "elapsed_sec": 80202.98816037178, "step_time_sec": 8.237043872009963, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.106601715087891, "lr": 0.00019312693997455287, "elapsed_sec": 80211.22666072845, "step_time_sec": 8.238308960979339, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.045216083526611, "lr": 0.0001931218039671172, "elapsed_sec": 80219.46304011345, "step_time_sec": 8.236200056038797, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.103730201721191, "lr": 0.00019311666611157685, "elapsed_sec": 80227.7021741867, "step_time_sec": 8.239004664996173, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.960418462753296, "lr": 0.00019311152640803661, "elapsed_sec": 80235.94085741043, "step_time_sec": 8.238462372974027, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.062371253967285, "lr": 0.00019310638485660125, "elapsed_sec": 80244.17935991287, "step_time_sec": 8.23833816201659, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.145965099334717, "lr": 0.00019310124145737568, "elapsed_sec": 80252.41794347763, "step_time_sec": 8.238424088049214, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.961998224258423, "lr": 0.00019309609621046475, "elapsed_sec": 80260.656406641, "step_time_sec": 8.238300890021492, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9905529022216797, "lr": 0.00019309094911597338, "elapsed_sec": 80268.89453911781, "step_time_sec": 8.238027120009065, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.940453052520752, "lr": 0.0001930858001740066, "elapsed_sec": 80277.1322259903, "step_time_sec": 8.237508188991342, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.214277744293213, "lr": 0.00019308064938466937, "elapsed_sec": 80285.37204098701, "step_time_sec": 8.23961440299172, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1109466552734375, "lr": 0.00019307549674806674, "elapsed_sec": 80293.61121535301, "step_time_sec": 8.239039328007493, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2224931716918945, "lr": 0.0001930703422643038, "elapsed_sec": 80301.84870052338, "step_time_sec": 8.237265675968956, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.093591690063477, "lr": 0.00019306518593348564, "elapsed_sec": 80310.08784985542, "step_time_sec": 8.238990060985088, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9745333194732666, "lr": 0.00019306002775571748, "elapsed_sec": 80318.32591366768, "step_time_sec": 8.237909283023328, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.234649181365967, "lr": 0.00019305486773110446, "elapsed_sec": 80326.56415104866, "step_time_sec": 8.238069340994116, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.017579555511475, "lr": 0.0001930497058597518, "elapsed_sec": 80334.802942276, "step_time_sec": 8.238678155990783, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.055793285369873, "lr": 0.00019304454214176482, "elapsed_sec": 80343.04104328156, "step_time_sec": 8.237911036005244, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.923875331878662, "lr": 0.00019303937657724876, "elapsed_sec": 80351.27938723564, "step_time_sec": 8.23817325301934, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.208470821380615, "lr": 0.00019303420916630902, "elapsed_sec": 80359.5168428421, "step_time_sec": 8.237303380970843, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.893948554992676, "lr": 0.000193029039909051, "elapsed_sec": 80367.75585079193, "step_time_sec": 8.23888724599965, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.049568176269531, "lr": 0.00019302386880558006, "elapsed_sec": 80375.99187636375, "step_time_sec": 8.235825724026654, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.213693141937256, "lr": 0.0001930186958560017, "elapsed_sec": 80384.22926783562, "step_time_sec": 8.237204704026226, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.041090488433838, "lr": 0.00019301352106042136, "elapsed_sec": 80392.46808719635, "step_time_sec": 8.23866860702401, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.007084369659424, "lr": 0.00019300834441894464, "elapsed_sec": 80400.70455741882, "step_time_sec": 8.236305153986905, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.983680009841919, "lr": 0.0001930031659316771, "elapsed_sec": 80408.94363093376, "step_time_sec": 8.238961384049617, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9064719676971436, "lr": 0.0001929979855987243, "elapsed_sec": 80417.18182706833, "step_time_sec": 8.238037933013402, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.882408618927002, "lr": 0.0001929928034201919, "elapsed_sec": 80425.41967558861, "step_time_sec": 8.237634733028244, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.912353992462158, "lr": 0.00019298761939618564, "elapsed_sec": 80433.65702724457, "step_time_sec": 8.23720927699469, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.011369228363037, "lr": 0.0001929824335268112, "elapsed_sec": 80441.89455342293, "step_time_sec": 8.237341100990307, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.115172863006592, "lr": 0.0001929772458121743, "elapsed_sec": 80450.13236403465, "step_time_sec": 8.237642149964813, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0731024742126465, "lr": 0.0001929720562523808, "elapsed_sec": 80458.37016296387, "step_time_sec": 8.237615556980018, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.939178705215454, "lr": 0.00019296686484753648, "elapsed_sec": 80466.6110868454, "step_time_sec": 8.240808659000322, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9814445972442627, "lr": 0.0001929616715977473, "elapsed_sec": 80474.85073065758, "step_time_sec": 8.239477751019876, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.128602027893066, "lr": 0.00019295647650311904, "elapsed_sec": 80483.08849835396, "step_time_sec": 8.237588717020117, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.036975383758545, "lr": 0.00019295127956375774, "elapsed_sec": 80491.32762145996, "step_time_sec": 8.238931890053209, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0694899559021, "lr": 0.00019294608077976934, "elapsed_sec": 80499.56470513344, "step_time_sec": 8.23693044797983, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1899614334106445, "lr": 0.00019294088015125993, "elapsed_sec": 80507.80354523659, "step_time_sec": 8.23870290897321, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.007253646850586, "lr": 0.0001929356776783355, "elapsed_sec": 80516.0421655178, "step_time_sec": 8.238418556051329, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.928091049194336, "lr": 0.00019293047336110214, "elapsed_sec": 80524.28110861778, "step_time_sec": 8.238750451011583, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8946077823638916, "lr": 0.00019292526719966603, "elapsed_sec": 80532.5214176178, "step_time_sec": 8.240164698974695, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9927289485931396, "lr": 0.0001929200591941333, "elapsed_sec": 80540.75906944275, "step_time_sec": 8.237495523993857, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.072525978088379, "lr": 0.00019291484934461024, "elapsed_sec": 80548.99634885788, "step_time_sec": 8.237158829986583, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.958284378051758, "lr": 0.000192909637651203, "elapsed_sec": 80557.23423051834, "step_time_sec": 8.237678847974166, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.069849967956543, "lr": 0.00019290442411401792, "elapsed_sec": 80565.472158432, "step_time_sec": 8.237786735000554, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.936246395111084, "lr": 0.00019289920873316133, "elapsed_sec": 80573.71060037613, "step_time_sec": 8.23827963398071, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9920170307159424, "lr": 0.00019289399150873956, "elapsed_sec": 80581.94710874557, "step_time_sec": 8.236376489978284, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.073294639587402, "lr": 0.000192888772440859, "elapsed_sec": 80590.1858458519, "step_time_sec": 8.238497980986722, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0257344245910645, "lr": 0.00019288355152962612, "elapsed_sec": 80598.42331647873, "step_time_sec": 8.237340096035041, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.940077543258667, "lr": 0.00019287832877514738, "elapsed_sec": 80606.66113042831, "step_time_sec": 8.237667451961897, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9476370811462402, "lr": 0.00019287310417752927, "elapsed_sec": 80614.89865541458, "step_time_sec": 8.237389383953996, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.065324783325195, "lr": 0.00019286787773687837, "elapsed_sec": 80623.13549566269, "step_time_sec": 8.23666262300685, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.026164531707764, "lr": 0.00019286264945330128, "elapsed_sec": 80631.37413215637, "step_time_sec": 8.238452341989614, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.991807699203491, "lr": 0.00019285741932690455, "elapsed_sec": 80639.61302757263, "step_time_sec": 8.238736510043964, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9082841873168945, "lr": 0.00019285218735779492, "elapsed_sec": 80647.85253095627, "step_time_sec": 8.239372304000426, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.067265033721924, "lr": 0.00019284695354607903, "elapsed_sec": 80656.09237289429, "step_time_sec": 8.239658007980324, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.099637985229492, "lr": 0.00019284171789186365, "elapsed_sec": 80664.32888484001, "step_time_sec": 8.236320046009496, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.986858606338501, "lr": 0.00019283648039525555, "elapsed_sec": 80672.5666718483, "step_time_sec": 8.237618101993576, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.136595726013184, "lr": 0.00019283124105636156, "elapsed_sec": 80680.80404233932, "step_time_sec": 8.237193143984769, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.026248455047607, "lr": 0.00019282599987528848, "elapsed_sec": 80689.04420924187, "step_time_sec": 8.240066953003407, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.075804710388184, "lr": 0.00019282075685214323, "elapsed_sec": 80697.28326392174, "step_time_sec": 8.238843987986911, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.167138576507568, "lr": 0.0001928155119870327, "elapsed_sec": 80705.52077960968, "step_time_sec": 8.237366716028191, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.007096290588379, "lr": 0.0001928102652800639, "elapsed_sec": 80713.75887703896, "step_time_sec": 8.237966256972868, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.052764892578125, "lr": 0.0001928050167313438, "elapsed_sec": 80721.9962387085, "step_time_sec": 8.237163791025523, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9226691722869873, "lr": 0.00019279976634097948, "elapsed_sec": 80730.23498368263, "step_time_sec": 8.238609716994688, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9361302852630615, "lr": 0.00019279451410907797, "elapsed_sec": 80738.4724855423, "step_time_sec": 8.237315497011878, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9853250980377197, "lr": 0.00019278926003574634, "elapsed_sec": 80746.70993304253, "step_time_sec": 8.237312286975794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9046618938446045, "lr": 0.00019278400412109183, "elapsed_sec": 80754.94885206223, "step_time_sec": 8.238783183973283, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.941755771636963, "lr": 0.00019277874636522158, "elapsed_sec": 80763.1862347126, "step_time_sec": 8.237194771005306, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.226777076721191, "lr": 0.00019277348676824283, "elapsed_sec": 80771.42467617989, "step_time_sec": 8.238237375044264, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9695029258728027, "lr": 0.00019276822533026284, "elapsed_sec": 80779.66302871704, "step_time_sec": 8.238178613013588, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.058963298797607, "lr": 0.00019276296205138888, "elapsed_sec": 80787.90122842789, "step_time_sec": 8.238043869961984, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.128184795379639, "lr": 0.00019275769693172834, "elapsed_sec": 80796.13895606995, "step_time_sec": 8.23756624100497, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9674365520477295, "lr": 0.00019275242997138852, "elapsed_sec": 80804.37729859352, "step_time_sec": 8.23817839397816, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.143337726593018, "lr": 0.00019274716117047692, "elapsed_sec": 80812.61661338806, "step_time_sec": 8.239142581995111, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.936249017715454, "lr": 0.00019274189052910094, "elapsed_sec": 80820.85483193398, "step_time_sec": 8.238103524025064, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.035462856292725, "lr": 0.00019273661804736808, "elapsed_sec": 80829.09374308586, "step_time_sec": 8.238749023003038, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.027517795562744, "lr": 0.00019273134372538585, "elapsed_sec": 80837.33279395103, "step_time_sec": 8.238862781960052, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.982637882232666, "lr": 0.0001927260675632618, "elapsed_sec": 80845.57025241852, "step_time_sec": 8.237307936011348, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.023317337036133, "lr": 0.00019272078956110358, "elapsed_sec": 80853.80975008011, "step_time_sec": 8.239332195953466, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.054037570953369, "lr": 0.0001927155097190188, "elapsed_sec": 80862.04666137695, "step_time_sec": 8.236751792021096, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1725568771362305, "lr": 0.00019271022803711514, "elapsed_sec": 80870.28631472588, "step_time_sec": 8.239479093987029, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9143919944763184, "lr": 0.00019270494451550027, "elapsed_sec": 80878.52394199371, "step_time_sec": 8.23746348102577, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1295037269592285, "lr": 0.00019269965915428203, "elapsed_sec": 80886.7612361908, "step_time_sec": 8.23715609399369, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8993258476257324, "lr": 0.0001926943719535681, "elapsed_sec": 80894.99900913239, "step_time_sec": 8.237650605966337, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.04590368270874, "lr": 0.00019268908291346641, "elapsed_sec": 80903.23624277115, "step_time_sec": 8.237048377050087, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.073896408081055, "lr": 0.00019268379203408475, "elapsed_sec": 80911.47549939156, "step_time_sec": 8.239072149037383, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.113365650177002, "lr": 0.00019267849931553103, "elapsed_sec": 80919.714209795, "step_time_sec": 8.238531556969974, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9646122455596924, "lr": 0.0001926732047579132, "elapsed_sec": 80939.55738711357, "step_time_sec": 19.842989577970002, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.098631381988525, "lr": 0.00019266790836133926, "elapsed_sec": 80947.78841233253, "step_time_sec": 8.23085735697532, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.958432912826538, "lr": 0.0001926626101259172, "elapsed_sec": 80956.02483010292, "step_time_sec": 8.236290015978739, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0714030265808105, "lr": 0.00019265731005175504, "elapsed_sec": 80964.26281523705, "step_time_sec": 8.237836337008048, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.011840343475342, "lr": 0.0001926520081389609, "elapsed_sec": 80972.50122022629, "step_time_sec": 8.238224781991448, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.146703243255615, "lr": 0.00019264670438764295, "elapsed_sec": 80980.73909640312, "step_time_sec": 8.237738605996128, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.94699764251709, "lr": 0.00019264139879790926, "elapsed_sec": 80988.97682285309, "step_time_sec": 8.237549943034537, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9698572158813477, "lr": 0.00019263609136986807, "elapsed_sec": 80997.21622753143, "step_time_sec": 8.239224524004385, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.014700889587402, "lr": 0.00019263078210362763, "elapsed_sec": 81005.45346164703, "step_time_sec": 8.237057099991944, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.008903980255127, "lr": 0.0001926254709992962, "elapsed_sec": 81013.69243597984, "step_time_sec": 8.238798981998116, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.13778829574585, "lr": 0.0001926201580569821, "elapsed_sec": 81021.93009400368, "step_time_sec": 8.237483686010819, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9758527278900146, "lr": 0.00019261484327679369, "elapsed_sec": 81030.16856884956, "step_time_sec": 8.23832503304584, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9498543739318848, "lr": 0.0001926095266588393, "elapsed_sec": 81038.40698814392, "step_time_sec": 8.238294131006114, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.23842716217041, "lr": 0.00019260420820322747, "elapsed_sec": 81046.64449477196, "step_time_sec": 8.237304494017735, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.281835556030273, "lr": 0.00019259888791006656, "elapsed_sec": 81054.88509726524, "step_time_sec": 8.240504176996183, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.184907913208008, "lr": 0.0001925935657794651, "elapsed_sec": 81063.12403535843, "step_time_sec": 8.238725936040282, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9900341033935547, "lr": 0.00019258824181153162, "elapsed_sec": 81071.3627603054, "step_time_sec": 8.23858001199551, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.087440013885498, "lr": 0.0001925829160063747, "elapsed_sec": 81079.60035824776, "step_time_sec": 8.237402716011275, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9286646842956543, "lr": 0.000192577588364103, "elapsed_sec": 81087.8367729187, "step_time_sec": 8.236279657983687, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9278430938720703, "lr": 0.00019257225888482513, "elapsed_sec": 81096.07653570175, "step_time_sec": 8.239581155008636, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.148642063140869, "lr": 0.00019256692756864974, "elapsed_sec": 81104.31490802765, "step_time_sec": 8.238223625987303, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.952131748199463, "lr": 0.0001925615944156856, "elapsed_sec": 81112.55459403992, "step_time_sec": 8.23948378500063, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9838781356811523, "lr": 0.00019255625942604144, "elapsed_sec": 81120.7925722599, "step_time_sec": 8.237833935010713, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.975062370300293, "lr": 0.0001925509225998261, "elapsed_sec": 81129.03005242348, "step_time_sec": 8.237375002005138, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9911088943481445, "lr": 0.0001925455839371484, "elapsed_sec": 81137.26818656921, "step_time_sec": 8.237923344015144, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.078800678253174, "lr": 0.0001925402434381172, "elapsed_sec": 81145.5042104721, "step_time_sec": 8.235868707997724, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.892991065979004, "lr": 0.00019253490110284144, "elapsed_sec": 81153.74318623543, "step_time_sec": 8.238799615006428, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9838175773620605, "lr": 0.00019252955693143005, "elapsed_sec": 81161.98101758957, "step_time_sec": 8.23764912202023, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.213398456573486, "lr": 0.00019252421092399205, "elapsed_sec": 81170.2185895443, "step_time_sec": 8.237412605027203, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.006890773773193, "lr": 0.0001925188630806364, "elapsed_sec": 81178.45638942719, "step_time_sec": 8.237714238988701, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.067975997924805, "lr": 0.0001925135134014722, "elapsed_sec": 81186.69386076927, "step_time_sec": 8.23727613396477, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2316436767578125, "lr": 0.00019250816188660854, "elapsed_sec": 81194.93309497833, "step_time_sec": 8.239066005975474, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8745455741882324, "lr": 0.0001925028085361546, "elapsed_sec": 81203.16991090775, "step_time_sec": 8.236601371958386, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9838078022003174, "lr": 0.00019249745335021948, "elapsed_sec": 81211.40887498856, "step_time_sec": 8.238833329000045, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9532623291015625, "lr": 0.00019249209632891246, "elapsed_sec": 81219.64776611328, "step_time_sec": 8.238692292012274, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.958359479904175, "lr": 0.00019248673747234273, "elapsed_sec": 81227.88579201698, "step_time_sec": 8.237839649023954, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9882073402404785, "lr": 0.0001924813767806196, "elapsed_sec": 81236.12340140343, "step_time_sec": 8.237481070042122, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.911086320877075, "lr": 0.0001924760142538524, "elapsed_sec": 81244.3608443737, "step_time_sec": 8.237373031966854, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8580520153045654, "lr": 0.00019247064989215052, "elapsed_sec": 81252.59826207161, "step_time_sec": 8.237147924024612, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.945997476577759, "lr": 0.00019246528369562332, "elapsed_sec": 81260.8375377655, "step_time_sec": 8.239122411003336, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.911012887954712, "lr": 0.00019245991566438022, "elapsed_sec": 81269.07371783257, "step_time_sec": 8.236044761026278, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.967747449874878, "lr": 0.00019245454579853072, "elapsed_sec": 81277.3131737709, "step_time_sec": 8.2393012259854, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.965517997741699, "lr": 0.00019244917409818433, "elapsed_sec": 81285.54997134209, "step_time_sec": 8.236611125001218, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.938063859939575, "lr": 0.00019244380056345061, "elapsed_sec": 81293.78805732727, "step_time_sec": 8.237906366994139, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.899322986602783, "lr": 0.00019243842519443914, "elapsed_sec": 81302.02751159668, "step_time_sec": 8.239308531978168, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.972172975540161, "lr": 0.0001924330479912595, "elapsed_sec": 81310.26601576805, "step_time_sec": 8.238356993009802, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.995936632156372, "lr": 0.00019242766895402144, "elapsed_sec": 81318.50429296494, "step_time_sec": 8.23816650401568, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.838344097137451, "lr": 0.00019242228808283456, "elapsed_sec": 81326.74108219147, "step_time_sec": 8.236589725012891, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.075559139251709, "lr": 0.00019241690537780864, "elapsed_sec": 81334.98052668571, "step_time_sec": 8.239325773029123, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0271382331848145, "lr": 0.00019241152083905347, "elapsed_sec": 81343.21826505661, "step_time_sec": 8.237540511006955, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9998228549957275, "lr": 0.0001924061344666788, "elapsed_sec": 81351.45605254173, "step_time_sec": 8.237621302949265, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.928591728210449, "lr": 0.00019240074626079458, "elapsed_sec": 81359.69545125961, "step_time_sec": 8.239236157038249, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8974976539611816, "lr": 0.00019239535622151064, "elapsed_sec": 81367.9319550991, "step_time_sec": 8.236315396032296, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8718597888946533, "lr": 0.00019238996434893684, "elapsed_sec": 81376.16996455193, "step_time_sec": 8.23785457201302, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.043395042419434, "lr": 0.0001923845706431832, "elapsed_sec": 81384.40769720078, "step_time_sec": 8.23761861299863, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.06742000579834, "lr": 0.00019237917510435976, "elapsed_sec": 81392.64647078514, "step_time_sec": 8.238547693996225, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.957425117492676, "lr": 0.0001923737777325765, "elapsed_sec": 81400.8846321106, "step_time_sec": 8.238011776993517, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.041167736053467, "lr": 0.00019236837852794347, "elapsed_sec": 81409.12332034111, "step_time_sec": 8.238523620006163, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1227288246154785, "lr": 0.0001923629774905708, "elapsed_sec": 81417.36201477051, "step_time_sec": 8.238540844002273, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.979853391647339, "lr": 0.00019235757462056867, "elapsed_sec": 81425.59842252731, "step_time_sec": 8.236285783001222, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9046993255615234, "lr": 0.00019235216991804723, "elapsed_sec": 81433.83571982384, "step_time_sec": 8.237077545025386, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8927955627441406, "lr": 0.00019234676338311673, "elapsed_sec": 81442.0726518631, "step_time_sec": 8.236801406019367, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9251348972320557, "lr": 0.0001923413550158874, "elapsed_sec": 81450.31119728088, "step_time_sec": 8.238355033972766, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.04105806350708, "lr": 0.00019233594481646955, "elapsed_sec": 81458.54969763756, "step_time_sec": 8.238373444008175, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.032926559448242, "lr": 0.00019233053278497347, "elapsed_sec": 81466.7876996994, "step_time_sec": 8.237850645033177, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.066949844360352, "lr": 0.00019232511892150963, "elapsed_sec": 81475.0267944336, "step_time_sec": 8.238943564007059, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9776501655578613, "lr": 0.00019231970322618837, "elapsed_sec": 81483.26378178596, "step_time_sec": 8.236801237042528, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.045254707336426, "lr": 0.0001923142856991201, "elapsed_sec": 81491.5013051033, "step_time_sec": 8.237388436042238, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9861037731170654, "lr": 0.00019230886634041534, "elapsed_sec": 81499.73972916603, "step_time_sec": 8.238282036036253, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9524810314178467, "lr": 0.00019230344515018468, "elapsed_sec": 81507.97566771507, "step_time_sec": 8.23573001101613, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9509642124176025, "lr": 0.00019229802212853857, "elapsed_sec": 81516.21321678162, "step_time_sec": 8.237437564996071, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.353965759277344, "lr": 0.00019229259727558765, "elapsed_sec": 81524.44961643219, "step_time_sec": 8.236220918945037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.574540615081787, "lr": 0.00019228717059144255, "elapsed_sec": 81532.68792843819, "step_time_sec": 8.238086677971296, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.306381702423096, "lr": 0.00019228174207621397, "elapsed_sec": 81540.92723727226, "step_time_sec": 8.239156915980857, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.039281845092773, "lr": 0.00019227631173001256, "elapsed_sec": 81549.16546082497, "step_time_sec": 8.238057723036036, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.275813102722168, "lr": 0.0001922708795529491, "elapsed_sec": 81557.40423631668, "step_time_sec": 8.238613552006427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0455756187438965, "lr": 0.00019226544554513437, "elapsed_sec": 81565.64056038857, "step_time_sec": 8.23618114402052, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.263769149780273, "lr": 0.00019226000970667918, "elapsed_sec": 81573.87968564034, "step_time_sec": 8.238979015033692, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.147359848022461, "lr": 0.0001922545720376944, "elapsed_sec": 81582.11627221107, "step_time_sec": 8.236380673013628, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.178154945373535, "lr": 0.0001922491325382909, "elapsed_sec": 81590.35540676117, "step_time_sec": 8.23896737198811, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.031535625457764, "lr": 0.0001922436912085796, "elapsed_sec": 81598.59383749962, "step_time_sec": 8.238328358042054, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.218774318695068, "lr": 0.00019223824804867154, "elapsed_sec": 81606.83242726326, "step_time_sec": 8.238368411024567, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.002299785614014, "lr": 0.00019223280305867763, "elapsed_sec": 81615.07108664513, "step_time_sec": 8.238498680992052, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0582780838012695, "lr": 0.000192227356238709, "elapsed_sec": 81623.30806279182, "step_time_sec": 8.236817847006023, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.140584945678711, "lr": 0.00019222190758887665, "elapsed_sec": 81631.5462782383, "step_time_sec": 8.238060050993226, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.118666648864746, "lr": 0.0001922164571092918, "elapsed_sec": 81639.78544926643, "step_time_sec": 8.23902952298522, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.213080883026123, "lr": 0.00019221100480006548, "elapsed_sec": 81648.02401995659, "step_time_sec": 8.238371725950856, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.034117221832275, "lr": 0.00019220555066130898, "elapsed_sec": 81656.26323866844, "step_time_sec": 8.239120093989186, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.164903163909912, "lr": 0.00019220009469313347, "elapsed_sec": 81664.50090098381, "step_time_sec": 8.23750014801044, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.233118057250977, "lr": 0.00019219463689565028, "elapsed_sec": 81672.7403151989, "step_time_sec": 8.239230567007326, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.051111221313477, "lr": 0.00019218917726897066, "elapsed_sec": 81680.97827744484, "step_time_sec": 8.237767569953576, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.077457427978516, "lr": 0.00019218371581320595, "elapsed_sec": 81689.2155957222, "step_time_sec": 8.237168141989969, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2131757736206055, "lr": 0.00019217825252846758, "elapsed_sec": 81697.45367193222, "step_time_sec": 8.237899258034304, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.190993785858154, "lr": 0.0001921727874148669, "elapsed_sec": 81705.69199299812, "step_time_sec": 8.238173048011959, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.107558727264404, "lr": 0.0001921673204725154, "elapsed_sec": 81713.93120265007, "step_time_sec": 8.239061383996159, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.080270767211914, "lr": 0.0001921618517015246, "elapsed_sec": 81722.17061853409, "step_time_sec": 8.2392249800032, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.161979675292969, "lr": 0.00019215638110200593, "elapsed_sec": 81730.40934228897, "step_time_sec": 8.238562861981336, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.006246089935303, "lr": 0.00019215090867407105, "elapsed_sec": 81738.64787650108, "step_time_sec": 8.23839928797679, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.059370517730713, "lr": 0.00019214543441783155, "elapsed_sec": 81746.88346791267, "step_time_sec": 8.235407189989928, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.207503795623779, "lr": 0.00019213995833339904, "elapsed_sec": 81755.12279486656, "step_time_sec": 8.239180842996575, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.108272552490234, "lr": 0.0001921344804208852, "elapsed_sec": 81763.36032509804, "step_time_sec": 8.237369338981807, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.32493257522583, "lr": 0.00019212900068040176, "elapsed_sec": 81771.6003472805, "step_time_sec": 8.23988124600146, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.144838333129883, "lr": 0.00019212351911206048, "elapsed_sec": 81779.83846855164, "step_time_sec": 8.237941456027329, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.175337791442871, "lr": 0.00019211803571597314, "elapsed_sec": 81788.07488250732, "step_time_sec": 8.236229183967225, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.101302146911621, "lr": 0.00019211255049225153, "elapsed_sec": 81796.31357598305, "step_time_sec": 8.238520084007178, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1551384925842285, "lr": 0.00019210706344100753, "elapsed_sec": 81804.5518283844, "step_time_sec": 8.238163154979702, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.033614158630371, "lr": 0.0001921015745623531, "elapsed_sec": 81812.7922604084, "step_time_sec": 8.240264086052775, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1837592124938965, "lr": 0.0001920960838564001, "elapsed_sec": 81821.02953982353, "step_time_sec": 8.237049703020602, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0131964683532715, "lr": 0.00019209059132326052, "elapsed_sec": 81829.2679336071, "step_time_sec": 8.238202160049696, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.018459320068359, "lr": 0.00019208509696304644, "elapsed_sec": 81837.50542593002, "step_time_sec": 8.237354010983836, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9821231365203857, "lr": 0.00019207960077586984, "elapsed_sec": 81845.74176120758, "step_time_sec": 8.236175414989702, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.185795783996582, "lr": 0.0001920741027618428, "elapsed_sec": 81853.98141312599, "step_time_sec": 8.239534454012755, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8737823963165283, "lr": 0.0001920686029210775, "elapsed_sec": 81862.21894288063, "step_time_sec": 8.237325942027383, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.989945411682129, "lr": 0.00019206310125368607, "elapsed_sec": 81870.45660710335, "step_time_sec": 8.237488681974355, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9916489124298096, "lr": 0.0001920575977597807, "elapsed_sec": 81878.69429445267, "step_time_sec": 8.237544174015056, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0573554039001465, "lr": 0.00019205209243947365, "elapsed_sec": 81886.93052148819, "step_time_sec": 8.23610037000617, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9459667205810547, "lr": 0.00019204658529287715, "elapsed_sec": 81895.16905498505, "step_time_sec": 8.238378615991678, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.102558612823486, "lr": 0.00019204107632010356, "elapsed_sec": 81903.40661764145, "step_time_sec": 8.237348413967993, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9231417179107666, "lr": 0.00019203556552126522, "elapsed_sec": 81911.64555549622, "step_time_sec": 8.23878698196495, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.045125484466553, "lr": 0.0001920300528964745, "elapsed_sec": 81919.88464450836, "step_time_sec": 8.238939417002257, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.143263816833496, "lr": 0.00019202453844584382, "elapsed_sec": 81928.12219429016, "step_time_sec": 8.237400320009328, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.038560390472412, "lr": 0.00019201902216948563, "elapsed_sec": 81936.36063408852, "step_time_sec": 8.238244890992064, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.052868366241455, "lr": 0.0001920135040675125, "elapsed_sec": 81944.59791660309, "step_time_sec": 8.237192332977429, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.024606704711914, "lr": 0.00019200798414003683, "elapsed_sec": 81952.83774137497, "step_time_sec": 8.239587827993091, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.140092372894287, "lr": 0.0001920024623871713, "elapsed_sec": 81961.0756623745, "step_time_sec": 8.237769244995434, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.022269248962402, "lr": 0.0001919969388090285, "elapsed_sec": 81969.31323552132, "step_time_sec": 8.23740889300825, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.073953628540039, "lr": 0.00019199141340572106, "elapsed_sec": 81977.55161070824, "step_time_sec": 8.238197472994216, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.167761325836182, "lr": 0.00019198588617736165, "elapsed_sec": 81985.78856897354, "step_time_sec": 8.23679334501503, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.10692024230957, "lr": 0.000191980357124063, "elapsed_sec": 81994.02841806412, "step_time_sec": 8.239740124961827, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.03058385848999, "lr": 0.0001919748262459379, "elapsed_sec": 82002.26749777794, "step_time_sec": 8.238889363012277, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.126662731170654, "lr": 0.0001919692935430991, "elapsed_sec": 82010.50670146942, "step_time_sec": 8.239035609993152, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.009223937988281, "lr": 0.00019196375901565942, "elapsed_sec": 82018.7456035614, "step_time_sec": 8.238778596976772, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.095700263977051, "lr": 0.0001919582226637318, "elapsed_sec": 82026.98161673546, "step_time_sec": 8.235847421979997, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.032721996307373, "lr": 0.0001919526844874291, "elapsed_sec": 82035.22103404999, "step_time_sec": 8.239295492996462, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.141481876373291, "lr": 0.00019194714448686424, "elapsed_sec": 82043.45875644684, "step_time_sec": 8.237549415032845, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.937169075012207, "lr": 0.00019194160266215024, "elapsed_sec": 82051.69789457321, "step_time_sec": 8.238967571989633, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9902591705322266, "lr": 0.00019193605901340007, "elapsed_sec": 82059.93681550026, "step_time_sec": 8.238724271010142, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.074170112609863, "lr": 0.00019193051354072686, "elapsed_sec": 82068.17336320877, "step_time_sec": 8.236391888989601, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.126210689544678, "lr": 0.00019192496624424366, "elapsed_sec": 82076.41122961044, "step_time_sec": 8.23775635601487, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0742011070251465, "lr": 0.0001919194171240636, "elapsed_sec": 82084.64880180359, "step_time_sec": 8.237398714001756, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9762701988220215, "lr": 0.00019191386618029984, "elapsed_sec": 82092.88762378693, "step_time_sec": 8.238608943996951, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1315598487854, "lr": 0.0001919083134130656, "elapsed_sec": 82101.12580227852, "step_time_sec": 8.238015586975962, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.975426197052002, "lr": 0.0001919027588224741, "elapsed_sec": 82109.36476612091, "step_time_sec": 8.23884986602934, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.134293556213379, "lr": 0.00019189720240863863, "elapsed_sec": 82117.60446166992, "step_time_sec": 8.239480096963234, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.094882011413574, "lr": 0.00019189164417167252, "elapsed_sec": 82125.84183812141, "step_time_sec": 8.237205806013662, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.180209636688232, "lr": 0.0001918860841116891, "elapsed_sec": 82134.08012914658, "step_time_sec": 8.238174623984378, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9773168563842773, "lr": 0.00019188052222880175, "elapsed_sec": 82142.31800866127, "step_time_sec": 8.23765199497575, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9685025215148926, "lr": 0.00019187495852312394, "elapsed_sec": 82150.55632162094, "step_time_sec": 8.238145571027417, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.854296922683716, "lr": 0.0001918693929947691, "elapsed_sec": 82158.79371476173, "step_time_sec": 8.237249783996958, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.238953590393066, "lr": 0.00019186382564385075, "elapsed_sec": 82167.03148937225, "step_time_sec": 8.237609133997466, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.976095199584961, "lr": 0.0001918582564704824, "elapsed_sec": 82175.27091884613, "step_time_sec": 8.239275066007394, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.02016019821167, "lr": 0.00019185268547477764, "elapsed_sec": 82183.50936031342, "step_time_sec": 8.238268257002346, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.965592384338379, "lr": 0.0001918471126568501, "elapsed_sec": 82191.74811244011, "step_time_sec": 8.238581015961245, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.853408098220825, "lr": 0.00019184153801681341, "elapsed_sec": 82199.98776721954, "step_time_sec": 8.239490916021168, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9043257236480713, "lr": 0.00019183596155478126, "elapsed_sec": 82208.2243475914, "step_time_sec": 8.236424781032838, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.003634452819824, "lr": 0.0001918303832708674, "elapsed_sec": 82216.46406698227, "step_time_sec": 8.23952627700055, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.021040916442871, "lr": 0.00019182480316518553, "elapsed_sec": 82224.70233297348, "step_time_sec": 8.238173531019129, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9373984336853027, "lr": 0.00019181922123784952, "elapsed_sec": 82232.94059062004, "step_time_sec": 8.238028465013485, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9563472270965576, "lr": 0.00019181363748897314, "elapsed_sec": 82241.17864608765, "step_time_sec": 8.237912077980582, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9174821376800537, "lr": 0.00019180805191867034, "elapsed_sec": 82249.41513156891, "step_time_sec": 8.236329186009243, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.020397186279297, "lr": 0.00019180246452705496, "elapsed_sec": 82257.65397453308, "step_time_sec": 8.238630537001882, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9514589309692383, "lr": 0.00019179687531424095, "elapsed_sec": 82265.89019083977, "step_time_sec": 8.23607806902146, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1646528244018555, "lr": 0.00019179128428034234, "elapsed_sec": 82274.12989425659, "step_time_sec": 8.239575869054534, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9395294189453125, "lr": 0.00019178569142547314, "elapsed_sec": 82282.3685309887, "step_time_sec": 8.238436628016643, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9563851356506348, "lr": 0.00019178009674974736, "elapsed_sec": 82290.60588741302, "step_time_sec": 8.23719221202191, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9114537239074707, "lr": 0.00019177450025327915, "elapsed_sec": 82298.84502649307, "step_time_sec": 8.238974205043633, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.981903076171875, "lr": 0.00019176890193618263, "elapsed_sec": 82307.08239030838, "step_time_sec": 8.237262166047003, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9480228424072266, "lr": 0.00019176330179857196, "elapsed_sec": 82315.32202887535, "step_time_sec": 8.239474079979118, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.130337238311768, "lr": 0.00019175769984056134, "elapsed_sec": 82323.55945563316, "step_time_sec": 8.23723114299355, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.207751750946045, "lr": 0.00019175209606226505, "elapsed_sec": 82331.7971713543, "step_time_sec": 8.237548981967848, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.963447332382202, "lr": 0.0001917464904637973, "elapsed_sec": 82340.03641104698, "step_time_sec": 8.239130403031595, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.025612831115723, "lr": 0.00019174088304527251, "elapsed_sec": 82348.27362036705, "step_time_sec": 8.236990839010105, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.969538450241089, "lr": 0.00019173527380680498, "elapsed_sec": 82356.51307630539, "step_time_sec": 8.239303251961246, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.052482604980469, "lr": 0.00019172966274850906, "elapsed_sec": 82364.75208830833, "step_time_sec": 8.238894514041021, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.094959735870361, "lr": 0.00019172404987049925, "elapsed_sec": 82372.99020218849, "step_time_sec": 8.237859190034214, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.054653644561768, "lr": 0.00019171843517288995, "elapsed_sec": 82381.2294356823, "step_time_sec": 8.239053032011725, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.068478107452393, "lr": 0.00019171281865579574, "elapsed_sec": 82389.46607899666, "step_time_sec": 8.236483563028742, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.015346527099609, "lr": 0.0001917072003193311, "elapsed_sec": 82397.70575976372, "step_time_sec": 8.239452534005977, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.108223915100098, "lr": 0.00019170158016361066, "elapsed_sec": 82405.9436211586, "step_time_sec": 8.237718902993947, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9339869022369385, "lr": 0.000191695958188749, "elapsed_sec": 82414.1816534996, "step_time_sec": 8.23791958700167, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9173836708068848, "lr": 0.00019169033439486076, "elapsed_sec": 82422.41848301888, "step_time_sec": 8.23662590701133, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.039262771606445, "lr": 0.00019168470878206067, "elapsed_sec": 82430.65486574173, "step_time_sec": 8.236210021015722, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9023094177246094, "lr": 0.0001916790813504634, "elapsed_sec": 82438.89291191101, "step_time_sec": 8.237945483997464, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.247567176818848, "lr": 0.0001916734521001838, "elapsed_sec": 82447.13023424149, "step_time_sec": 8.23711791698588, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.092025279998779, "lr": 0.0001916678210313366, "elapsed_sec": 82455.369515419, "step_time_sec": 8.239119667967316, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0430169105529785, "lr": 0.00019166218814403664, "elapsed_sec": 82463.60801053047, "step_time_sec": 8.238342437020037, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.003664970397949, "lr": 0.00019165655343839885, "elapsed_sec": 82471.84605789185, "step_time_sec": 8.237841046997346, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.012452602386475, "lr": 0.00019165091691453805, "elapsed_sec": 82480.08492040634, "step_time_sec": 8.23873124999227, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.057352066040039, "lr": 0.00019164527857256928, "elapsed_sec": 82488.32180762291, "step_time_sec": 8.236741325003095, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.324770450592041, "lr": 0.0001916396384126075, "elapsed_sec": 82496.56071400642, "step_time_sec": 8.238788160961121, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.133102893829346, "lr": 0.0001916339964347677, "elapsed_sec": 82504.79808068275, "step_time_sec": 8.237199132971, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.944047451019287, "lr": 0.00019162835263916497, "elapsed_sec": 82513.0364370346, "step_time_sec": 8.238139313005377, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9275941848754883, "lr": 0.00019162270702591437, "elapsed_sec": 82521.2757384777, "step_time_sec": 8.239135743002407, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.057222843170166, "lr": 0.0001916170595951311, "elapsed_sec": 82529.51274418831, "step_time_sec": 8.236877617950086, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.888355255126953, "lr": 0.00019161141034693027, "elapsed_sec": 82537.75143480301, "step_time_sec": 8.238478352956008, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9419167041778564, "lr": 0.00019160575928142714, "elapsed_sec": 82545.99000620842, "step_time_sec": 8.238401019014418, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.062867164611816, "lr": 0.00019160010639873693, "elapsed_sec": 82554.22980546951, "step_time_sec": 8.239663073967677, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9909393787384033, "lr": 0.0001915944516989749, "elapsed_sec": 82562.46903467178, "step_time_sec": 8.239069770032074, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.035398483276367, "lr": 0.0001915887951822564, "elapsed_sec": 82570.70638132095, "step_time_sec": 8.237190730986185, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.039313316345215, "lr": 0.0001915831368486968, "elapsed_sec": 82578.94516444206, "step_time_sec": 8.238607598003, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.084568023681641, "lr": 0.00019157747669841144, "elapsed_sec": 82587.18299293518, "step_time_sec": 8.237677101045847, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.991988182067871, "lr": 0.0001915718147315158, "elapsed_sec": 82595.42148852348, "step_time_sec": 8.238347570993938, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.097661972045898, "lr": 0.00019156615094812534, "elapsed_sec": 82603.65978717804, "step_time_sec": 8.23808675800683, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.139642715454102, "lr": 0.00019156048534835554, "elapsed_sec": 82611.89708638191, "step_time_sec": 8.237154603004456, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.170443534851074, "lr": 0.00019155481793232195, "elapsed_sec": 82620.13620352745, "step_time_sec": 8.238956640998367, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.073251247406006, "lr": 0.00019154914870014018, "elapsed_sec": 82628.37217903137, "step_time_sec": 8.235801336006261, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.001989841461182, "lr": 0.00019154347765192582, "elapsed_sec": 82636.61072397232, "step_time_sec": 8.238370493985713, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.919586420059204, "lr": 0.0001915378047877945, "elapsed_sec": 82644.84817028046, "step_time_sec": 8.23732494696742, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.915848970413208, "lr": 0.00019153213010786197, "elapsed_sec": 82653.08615922928, "step_time_sec": 8.23777672596043, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.001662731170654, "lr": 0.0001915264536122439, "elapsed_sec": 82661.32292675972, "step_time_sec": 8.236655342043377, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.091907978057861, "lr": 0.00019152077530105608, "elapsed_sec": 82669.55974411964, "step_time_sec": 8.236635270994157, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.112037658691406, "lr": 0.00019151509517441431, "elapsed_sec": 82677.79745006561, "step_time_sec": 8.23750112199923, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.989927291870117, "lr": 0.00019150941323243446, "elapsed_sec": 82686.03418922424, "step_time_sec": 8.23658784996951, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.896411657333374, "lr": 0.00019150372947523235, "elapsed_sec": 82694.27256393433, "step_time_sec": 8.238190489995759, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.179344654083252, "lr": 0.00019149804390292387, "elapsed_sec": 82702.51092410088, "step_time_sec": 8.238206437032204, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.137372016906738, "lr": 0.0001914923565156251, "elapsed_sec": 82710.74896597862, "step_time_sec": 8.237886302988045, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.037925720214844, "lr": 0.00019148666731345185, "elapsed_sec": 82718.98725533485, "step_time_sec": 8.238118053996004, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.001117706298828, "lr": 0.0001914809762965203, "elapsed_sec": 82727.22508502007, "step_time_sec": 8.237685977015644, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9189071655273438, "lr": 0.0001914752834649464, "elapsed_sec": 82735.46343946457, "step_time_sec": 8.23817305197008, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.055800914764404, "lr": 0.00019146958881884632, "elapsed_sec": 82743.70186400414, "step_time_sec": 8.238279498997144, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.004274845123291, "lr": 0.00019146389235833616, "elapsed_sec": 82751.93971943855, "step_time_sec": 8.237698730023112, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8350412845611572, "lr": 0.0001914581940835321, "elapsed_sec": 82760.1769759655, "step_time_sec": 8.237070182978641, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.144650459289551, "lr": 0.00019145249399455034, "elapsed_sec": 82768.41373634338, "step_time_sec": 8.236610616964754, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.16151762008667, "lr": 0.0001914467920915071, "elapsed_sec": 82776.72081065178, "step_time_sec": 8.243866225995589, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.995901584625244, "lr": 0.00019144108837451873, "elapsed_sec": 82784.96044611931, "step_time_sec": 8.239514652988873, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.081582546234131, "lr": 0.00019143538284370154, "elapsed_sec": 82793.19800329208, "step_time_sec": 8.237325810012408, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.009800434112549, "lr": 0.0001914296754991718, "elapsed_sec": 82801.4375243187, "step_time_sec": 8.239387488982175, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.851801872253418, "lr": 0.000191423966341046, "elapsed_sec": 82809.6752166748, "step_time_sec": 8.237535719003063, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.938469171524048, "lr": 0.0001914182553694405, "elapsed_sec": 82817.9112713337, "step_time_sec": 8.23588071600534, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8571178913116455, "lr": 0.00019141254258447183, "elapsed_sec": 82826.14999556541, "step_time_sec": 8.238522209983785, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9798388481140137, "lr": 0.00019140682798625644, "elapsed_sec": 82834.38736987114, "step_time_sec": 8.237208503996953, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.078954696655273, "lr": 0.0001914011115749109, "elapsed_sec": 82842.62590765953, "step_time_sec": 8.238374091975857, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.041671276092529, "lr": 0.0001913953933505518, "elapsed_sec": 82850.86532187462, "step_time_sec": 8.23926721396856, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.025303840637207, "lr": 0.00019138967331329572, "elapsed_sec": 82859.10282802582, "step_time_sec": 8.23738141200738, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.136791229248047, "lr": 0.00019138395146325933, "elapsed_sec": 82867.34159445763, "step_time_sec": 8.23855203797575, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0876054763793945, "lr": 0.0001913782278005593, "elapsed_sec": 82875.57899451256, "step_time_sec": 8.237263133982196, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.105787754058838, "lr": 0.0001913725023253124, "elapsed_sec": 82883.81813430786, "step_time_sec": 8.238926485995762, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9918935298919678, "lr": 0.00019136677503763539, "elapsed_sec": 82892.05529546738, "step_time_sec": 8.237008651020005, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.950561285018921, "lr": 0.00019136104593764502, "elapsed_sec": 82900.29402256012, "step_time_sec": 8.238558045006357, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.074565410614014, "lr": 0.00019135531502545818, "elapsed_sec": 82908.53173542023, "step_time_sec": 8.237561794987414, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.122226715087891, "lr": 0.00019134958230119167, "elapsed_sec": 82916.76864004135, "step_time_sec": 8.236800011945888, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.00410270690918, "lr": 0.00019134384776496247, "elapsed_sec": 82925.0063765049, "step_time_sec": 8.237584079965018, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.890531063079834, "lr": 0.00019133811141688753, "elapsed_sec": 82933.24422645569, "step_time_sec": 8.23762288899161, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.142597198486328, "lr": 0.00019133237325708383, "elapsed_sec": 82941.48385739326, "step_time_sec": 8.239480792020913, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.090185165405273, "lr": 0.00019132663328566834, "elapsed_sec": 82949.72170329094, "step_time_sec": 8.237718197982758, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.7845752239227295, "lr": 0.00019132089150275815, "elapsed_sec": 82957.95971798897, "step_time_sec": 8.237838896980975, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0357465744018555, "lr": 0.00019131514790847036, "elapsed_sec": 82966.19751167297, "step_time_sec": 8.237609559029806, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.024628639221191, "lr": 0.0001913094025029221, "elapsed_sec": 82974.43468141556, "step_time_sec": 8.237002206034958, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0194220542907715, "lr": 0.00019130365528623053, "elapsed_sec": 82982.67275452614, "step_time_sec": 8.237901955959387, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0523271560668945, "lr": 0.00019129790625851292, "elapsed_sec": 82990.9096968174, "step_time_sec": 8.236822951002978, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.051191806793213, "lr": 0.0001912921554198864, "elapsed_sec": 82999.1477150917, "step_time_sec": 8.23785457300255, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9360463619232178, "lr": 0.00019128640277046836, "elapsed_sec": 83007.38799071312, "step_time_sec": 8.240136575012002, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.2638750076293945, "lr": 0.00019128064831037606, "elapsed_sec": 83015.62534761429, "step_time_sec": 8.237241447030101, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.153387546539307, "lr": 0.00019127489203972685, "elapsed_sec": 83023.86474776268, "step_time_sec": 8.239178313000593, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.997875213623047, "lr": 0.00019126913395863812, "elapsed_sec": 83032.10322332382, "step_time_sec": 8.238319612981286, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.191596031188965, "lr": 0.0001912633740672273, "elapsed_sec": 83040.34276366234, "step_time_sec": 8.239372321986593, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.05554723739624, "lr": 0.0001912576123656119, "elapsed_sec": 83048.58068609238, "step_time_sec": 8.237743705976754, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.035055160522461, "lr": 0.0001912518488539094, "elapsed_sec": 83056.81875181198, "step_time_sec": 8.237926998990588, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1643524169921875, "lr": 0.00019124608353223728, "elapsed_sec": 83065.05798649788, "step_time_sec": 8.23908988799667, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.107550621032715, "lr": 0.00019124031640071318, "elapsed_sec": 83073.2959830761, "step_time_sec": 8.237789703998715, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9616122245788574, "lr": 0.00019123454745945473, "elapsed_sec": 83081.5342028141, "step_time_sec": 8.238060395000502, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.977832794189453, "lr": 0.0001912287767085795, "elapsed_sec": 83089.77291345596, "step_time_sec": 8.238557666016277, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0699005126953125, "lr": 0.00019122300414820524, "elapsed_sec": 83098.00947356224, "step_time_sec": 8.236375293985475, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.057428359985352, "lr": 0.00019121722977844967, "elapsed_sec": 83106.24899935722, "step_time_sec": 8.239366905007046, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1038737297058105, "lr": 0.0001912114535994305, "elapsed_sec": 83114.4861433506, "step_time_sec": 8.23701165302191, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.109582424163818, "lr": 0.00019120567561126556, "elapsed_sec": 83122.72587823868, "step_time_sec": 8.239478956966195, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.968003273010254, "lr": 0.0001911998958140727, "elapsed_sec": 83130.96497225761, "step_time_sec": 8.238972057995852, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.135181427001953, "lr": 0.0001911941142079698, "elapsed_sec": 83139.20301961899, "step_time_sec": 53.50190171203576, "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": 0.9993711480055936, "warmup_excluded": false, "peak_vram_bytes": 13269960192}
{"step": 10001, "loss": 3.9482884407043457, "lr": 0.00019118833079307472, "elapsed_sec": 83192.71182155609, "step_time_sec": 8.244674568006303, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.07896614074707, "lr": 0.00019118254556950543, "elapsed_sec": 83200.94533061981, "step_time_sec": 8.233356285025366, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9529337882995605, "lr": 0.00019117675853737992, "elapsed_sec": 83209.18270015717, "step_time_sec": 8.237221672025044, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.982306957244873, "lr": 0.00019117096969681617, "elapsed_sec": 83217.42126488686, "step_time_sec": 8.238396380038466, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.033736228942871, "lr": 0.0001911651790479323, "elapsed_sec": 83225.65966176987, "step_time_sec": 8.23821423802292, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9986815452575684, "lr": 0.00019115938659084635, "elapsed_sec": 83233.89877676964, "step_time_sec": 8.238956127024721, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.8582539558410645, "lr": 0.00019115359232567647, "elapsed_sec": 83242.13772535324, "step_time_sec": 8.23882238403894, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9369211196899414, "lr": 0.0001911477962525408, "elapsed_sec": 83250.37509655952, "step_time_sec": 8.237185824022163, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.052097797393799, "lr": 0.00019114199837155758, "elapsed_sec": 83258.61519360542, "step_time_sec": 8.239930022042245, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.117123603820801, "lr": 0.00019113619868284503, "elapsed_sec": 83266.85181093216, "step_time_sec": 8.236474048986565, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.869288444519043, "lr": 0.00019113039718652143, "elapsed_sec": 83275.09153962135, "step_time_sec": 8.239581741974689, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.518129348754883, "lr": 0.00019112459388270512, "elapsed_sec": 83283.33053040504, "step_time_sec": 8.238790049974341, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.07343053817749, "lr": 0.00019111878877151438, "elapsed_sec": 83291.56840825081, "step_time_sec": 8.237737096962519, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.094842433929443, "lr": 0.00019111298185306766, "elapsed_sec": 83299.80736112595, "step_time_sec": 8.238806817040313, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.037281513214111, "lr": 0.0001911071731274834, "elapsed_sec": 83308.04567480087, "step_time_sec": 8.238130179990549, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0469970703125, "lr": 0.00019110136259487998, "elapsed_sec": 83316.28534340858, "step_time_sec": 8.239553134015296, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9455175399780273, "lr": 0.00019109555025537594, "elapsed_sec": 83324.5234439373, "step_time_sec": 8.237942576990463, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9469358921051025, "lr": 0.00019108973610908983, "elapsed_sec": 83332.76111221313, "step_time_sec": 8.237434710026719, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.037265300750732, "lr": 0.00019108392015614023, "elapsed_sec": 83341.00011777878, "step_time_sec": 8.238850969006307, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.049010276794434, "lr": 0.00019107810239664568, "elapsed_sec": 83349.23824858665, "step_time_sec": 8.237965081003495, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.099522113800049, "lr": 0.0001910722828307249, "elapsed_sec": 83357.477247715, "step_time_sec": 8.238822096027434, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9730167388916016, "lr": 0.00019106646145849657, "elapsed_sec": 83365.71568870544, "step_time_sec": 8.238261356018484, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.115188121795654, "lr": 0.00019106063828007935, "elapsed_sec": 83373.9535150528, "step_time_sec": 8.237669677997474, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.001940727233887, "lr": 0.00019105481329559202, "elapsed_sec": 83382.19152283669, "step_time_sec": 8.237848303979263, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.6939852237701416, "lr": 0.0001910489865051534, "elapsed_sec": 83390.42946362495, "step_time_sec": 8.2377841679845, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.964116334915161, "lr": 0.00019104315790888232, "elapsed_sec": 83398.66941380501, "step_time_sec": 8.239769774023443, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.088866233825684, "lr": 0.0001910373275068976, "elapsed_sec": 83406.90773940086, "step_time_sec": 8.238153208978474, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.042151927947998, "lr": 0.00019103149529931817, "elapsed_sec": 83415.14556670189, "step_time_sec": 8.237741889955942, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9364964962005615, "lr": 0.00019102566128626297, "elapsed_sec": 83423.38384747505, "step_time_sec": 8.23804480698891, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.025852203369141, "lr": 0.00019101982546785098, "elapsed_sec": 83431.62170100212, "step_time_sec": 8.237711964000482, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.030050754547119, "lr": 0.00019101398784420118, "elapsed_sec": 83439.86087298393, "step_time_sec": 8.23898567800643, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.0222249031066895, "lr": 0.0001910081484154327, "elapsed_sec": 83448.09928011894, "step_time_sec": 8.238268481043633, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.973449230194092, "lr": 0.00019100230718166456, "elapsed_sec": 83456.33767437935, "step_time_sec": 8.238273622002453, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.056419849395752, "lr": 0.00019099646414301588, "elapsed_sec": 83464.57588505745, "step_time_sec": 8.238059565017466, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1966447830200195, "lr": 0.00019099061929960583, "elapsed_sec": 83472.81304740906, "step_time_sec": 8.236931368010119, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.98378849029541, "lr": 0.00019098477265155366, "elapsed_sec": 83481.052744627, "step_time_sec": 8.239559353038203, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.985765218734741, "lr": 0.00019097892419897854, "elapsed_sec": 83489.29139780998, "step_time_sec": 8.238542714971118, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.043617248535156, "lr": 0.00019097307394199976, "elapsed_sec": 83497.5300731659, "step_time_sec": 8.238471249002032, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 3.9847371578216553, "lr": 0.00019096722188073663, "elapsed_sec": 83505.76870560646, "step_time_sec": 8.238462059991434, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.151092052459717, "lr": 0.0001909613680153085, "elapsed_sec": 83514.00559735298, "step_time_sec": 8.236725730996113, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.223673343658447, "lr": 0.00019095551234583473, "elapsed_sec": 83522.24626803398, "step_time_sec": 8.240547374996822, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.98874044418335, "lr": 0.00019094965487243478, "elapsed_sec": 83530.48446512222, "step_time_sec": 8.238028126012068, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.271914482116699, "lr": 0.00019094379559522805, "elapsed_sec": 83538.72324872017, "step_time_sec": 8.238590674998704, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4355597496032715, "lr": 0.0001909379345143341, "elapsed_sec": 83546.96169781685, "step_time_sec": 8.23828174796654, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.485095024108887, "lr": 0.00019093207162987238, "elapsed_sec": 83555.19908452034, "step_time_sec": 8.237221996008884, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.553922653198242, "lr": 0.00019092620694196254, "elapsed_sec": 83563.4366800785, "step_time_sec": 8.237399094970897, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.407025337219238, "lr": 0.0001909203404507241, "elapsed_sec": 83571.67507433891, "step_time_sec": 8.238265790976584, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.440546035766602, "lr": 0.00019091447215627676, "elapsed_sec": 83579.9146990776, "step_time_sec": 8.239466298022307, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.376701354980469, "lr": 0.00019090860205874015, "elapsed_sec": 83588.15405488014, "step_time_sec": 8.23916325403843, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.4860429763793945, "lr": 0.00019090273015823401, "elapsed_sec": 83596.39123511314, "step_time_sec": 8.237042471009772, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.317348480224609, "lr": 0.00019089685645487807, "elapsed_sec": 83604.6298854351, "step_time_sec": 8.238498700025957, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.138047695159912, "lr": 0.00019089098094879215, "elapsed_sec": 83612.8687415123, "step_time_sec": 8.238611819047946, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.1017632484436035, "lr": 0.00019088510364009606, "elapsed_sec": 83621.10800743103, "step_time_sec": 8.23910393496044, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.228966236114502, "lr": 0.00019087922452890962, "elapsed_sec": 83629.34539556503, "step_time_sec": 8.237238794972654, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.495240688323975, "lr": 0.0001908733436153528, "elapsed_sec": 83637.58338737488, "step_time_sec": 8.237840013985988, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 4.867888927459717, "lr": 0.00019086746089954547, "elapsed_sec": 83645.82296466827, "step_time_sec": 8.239413581963163, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.023404121398926, "lr": 0.0001908615763816076, "elapsed_sec": 83654.05987620354, "step_time_sec": 8.236742335022427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 5.962111949920654, "lr": 0.00019085569006165924, "elapsed_sec": 83662.29916071892, "step_time_sec": 8.239162211015355, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.879826068878174, "lr": 0.0001908498019398204, "elapsed_sec": 83670.53548955917, "step_time_sec": 8.236190929019358, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.44487190246582, "lr": 0.00019084391201621117, "elapsed_sec": 83678.77301955223, "step_time_sec": 8.237341127009131, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 10.283035278320312, "lr": 0.00019083802029095168, "elapsed_sec": 83687.00909662247, "step_time_sec": 8.235930971975904, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.872790336608887, "lr": 0.00019083212676416207, "elapsed_sec": 83695.2452082634, "step_time_sec": 8.235941519029438, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 9.296001434326172, "lr": 0.0001908262314359625, "elapsed_sec": 83703.48226213455, "step_time_sec": 8.236870930995792, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.640596389770508, "lr": 0.00019082033430647326, "elapsed_sec": 83711.71841669083, "step_time_sec": 8.235997335985303, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.5929536819458, "lr": 0.0001908144353758146, "elapsed_sec": 83719.9530775547, "step_time_sec": 8.23445091600297, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.417941093444824, "lr": 0.00019080853464410676, "elapsed_sec": 83728.18808817863, "step_time_sec": 8.234846832987387, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.334635734558105, "lr": 0.00019080263211147013, "elapsed_sec": 83736.42290639877, "step_time_sec": 8.234698447980918, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.320586204528809, "lr": 0.00019079672777802507, "elapsed_sec": 83744.65790081024, "step_time_sec": 8.23486684000818, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.276265144348145, "lr": 0.00019079082164389198, "elapsed_sec": 83752.8935649395, "step_time_sec": 8.235425547987688, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.339071273803711, "lr": 0.00019078491370919136, "elapsed_sec": 83761.12920188904, "step_time_sec": 8.23550954501843, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.228540420532227, "lr": 0.00019077900397404364, "elapsed_sec": 83769.36437129974, "step_time_sec": 8.234987860021647, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.194204330444336, "lr": 0.00019077309243856932, "elapsed_sec": 83777.59937238693, "step_time_sec": 8.234871829976328, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.183781623840332, "lr": 0.00019076717910288904, "elapsed_sec": 83785.83457636833, "step_time_sec": 8.234999332984444, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.34713077545166, "lr": 0.00019076126396712335, "elapsed_sec": 83794.0692999363, "step_time_sec": 8.234564482991118, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.205544471740723, "lr": 0.00019075534703139285, "elapsed_sec": 83802.30431365967, "step_time_sec": 8.234873184992466, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.242338180541992, "lr": 0.00019074942829581823, "elapsed_sec": 83810.539021492, "step_time_sec": 8.234563688980415, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.174230575561523, "lr": 0.00019074350776052023, "elapsed_sec": 83818.77413535118, "step_time_sec": 8.234982259979006, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.238214492797852, "lr": 0.00019073758542561953, "elapsed_sec": 83827.00748944283, "step_time_sec": 8.23313164699357, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.155972480773926, "lr": 0.00019073166129123697, "elapsed_sec": 83835.24178957939, "step_time_sec": 8.23413386096945, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.199116706848145, "lr": 0.0001907257353574933, "elapsed_sec": 83843.47705364227, "step_time_sec": 8.23510841402458, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.087191581726074, "lr": 0.0001907198076245094, "elapsed_sec": 83851.71222281456, "step_time_sec": 8.23503687698394, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.172494888305664, "lr": 0.0001907138780924062, "elapsed_sec": 83859.94730186462, "step_time_sec": 8.234946504002437, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.183817863464355, "lr": 0.00019070794676130458, "elapsed_sec": 83868.181579113, "step_time_sec": 8.234048145008273, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.222917556762695, "lr": 0.0001907020136313255, "elapsed_sec": 83876.41733145714, "step_time_sec": 8.235623034008313, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.145392417907715, "lr": 0.00019069607870258996, "elapsed_sec": 83884.65292239189, "step_time_sec": 8.235418910975568, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.159838676452637, "lr": 0.00019069014197521897, "elapsed_sec": 83892.8880803585, "step_time_sec": 8.234965471958276, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.224316596984863, "lr": 0.00019068420344933368, "elapsed_sec": 83901.12310862541, "step_time_sec": 8.234854502021335, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.169584274291992, "lr": 0.00019067826312505514, "elapsed_sec": 83909.35822486877, "step_time_sec": 8.234965407056734, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.180732727050781, "lr": 0.0001906723210025045, "elapsed_sec": 83917.59362435341, "step_time_sec": 8.235245432995725, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.207062721252441, "lr": 0.00019066637708180294, "elapsed_sec": 83925.8289244175, "step_time_sec": 8.235131895984523, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.174744606018066, "lr": 0.0001906604313630717, "elapsed_sec": 83934.06440901756, "step_time_sec": 8.235303925001062, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.166647911071777, "lr": 0.000190654483846432, "elapsed_sec": 83942.29996299744, "step_time_sec": 8.235376217053272, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.232281684875488, "lr": 0.00019064853453200513, "elapsed_sec": 83950.53551316261, "step_time_sec": 8.235413693997543, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.267609596252441, "lr": 0.0001906425834199125, "elapsed_sec": 83958.77110314369, "step_time_sec": 8.235390138986986, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.107993125915527, "lr": 0.00019063663051027533, "elapsed_sec": 83967.00601196289, "step_time_sec": 8.234772848023567, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.141724586486816, "lr": 0.00019063067580321516, "elapsed_sec": 83975.24105978012, "step_time_sec": 8.234854728973005, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.12427806854248, "lr": 0.00019062471929885336, "elapsed_sec": 83983.47635865211, "step_time_sec": 8.235147639003117, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.140813827514648, "lr": 0.0001906187609973114, "elapsed_sec": 83991.71132731438, "step_time_sec": 8.234862776007503, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.19096851348877, "lr": 0.00019061280089871082, "elapsed_sec": 83999.94682502747, "step_time_sec": 8.23528805701062, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.141700744628906, "lr": 0.00019060683900317313, "elapsed_sec": 84008.18208861351, "step_time_sec": 8.23514302604599, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.170361518859863, "lr": 0.00019060087531081997, "elapsed_sec": 84016.41753435135, "step_time_sec": 8.235245488001965, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.130664825439453, "lr": 0.00019059490982177292, "elapsed_sec": 84024.65167689323, "step_time_sec": 8.234022255986929, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.187122344970703, "lr": 0.00019058894253615364, "elapsed_sec": 84032.88718342781, "step_time_sec": 8.235272053978406, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.340298652648926, "lr": 0.00019058297345408385, "elapsed_sec": 84041.12221980095, "step_time_sec": 8.23486981797032, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.09383487701416, "lr": 0.00019057700257568523, "elapsed_sec": 84049.35695552826, "step_time_sec": 8.234609376988374, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.105202674865723, "lr": 0.0001905710299010796, "elapsed_sec": 84057.5930929184, "step_time_sec": 8.235945316031575, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.149148941040039, "lr": 0.00019056505543038876, "elapsed_sec": 84065.82787942886, "step_time_sec": 8.234661488968413, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.26845932006836, "lr": 0.00019055907916373452, "elapsed_sec": 84074.06341004372, "step_time_sec": 8.235337661986705, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.162212371826172, "lr": 0.00019055310110123878, "elapsed_sec": 84082.29834079742, "step_time_sec": 8.234755092009436, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.142759323120117, "lr": 0.00019054712124302344, "elapsed_sec": 84090.53338265419, "step_time_sec": 8.234927812998649, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.195907592773438, "lr": 0.00019054113958921046, "elapsed_sec": 84098.76824593544, "step_time_sec": 8.234655353997368, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.127182960510254, "lr": 0.00019053515613992182, "elapsed_sec": 84107.00248217583, "step_time_sec": 8.23409659898607, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.22438907623291, "lr": 0.00019052917089527958, "elapsed_sec": 84115.23719978333, "step_time_sec": 8.234504629974253, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.074326515197754, "lr": 0.00019052318385540576, "elapsed_sec": 84123.4728012085, "step_time_sec": 8.235428261978086, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.20630931854248, "lr": 0.00019051719502042244, "elapsed_sec": 84131.70164346695, "step_time_sec": 8.228679838008247, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.161916732788086, "lr": 0.0001905112043904518, "elapsed_sec": 84139.92988824844, "step_time_sec": 8.228091361990664, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.17294692993164, "lr": 0.000190505211965616, "elapsed_sec": 84148.15943908691, "step_time_sec": 8.229432080988772, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.25866413116455, "lr": 0.00019049921774603725, "elapsed_sec": 84156.39430189133, "step_time_sec": 8.234662134025712, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.159319877624512, "lr": 0.00019049322173183775, "elapsed_sec": 84164.62949728966, "step_time_sec": 8.23510032298509, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.1702299118042, "lr": 0.00019048722392313983, "elapsed_sec": 84172.86481118202, "step_time_sec": 8.235096684016753, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.079987525939941, "lr": 0.0001904812243200658, "elapsed_sec": 84181.1004126072, "step_time_sec": 8.235439664975274, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.282600402832031, "lr": 0.000190475222922738, "elapsed_sec": 84189.33560419083, "step_time_sec": 8.235039920953568, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.199130058288574, "lr": 0.00019046921973127882, "elapsed_sec": 84197.57097125053, "step_time_sec": 8.2352101940196, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.184002876281738, "lr": 0.0001904632147458107, "elapsed_sec": 84205.80568885803, "step_time_sec": 8.234538184013218, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.089818954467773, "lr": 0.0001904572079664561, "elapsed_sec": 84214.04074525833, "step_time_sec": 8.234886536025442, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.168233871459961, "lr": 0.0001904511993933375, "elapsed_sec": 84222.27564120293, "step_time_sec": 8.234713312005624, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.143152236938477, "lr": 0.00019044518902657743, "elapsed_sec": 84230.504581213, "step_time_sec": 8.228749561007135, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.193625450134277, "lr": 0.0001904391768662985, "elapsed_sec": 84238.7332804203, "step_time_sec": 8.228540429030545, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.293821334838867, "lr": 0.0001904331629126233, "elapsed_sec": 84246.96189045906, "step_time_sec": 8.228472115006298, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.138982772827148, "lr": 0.0001904271471656745, "elapsed_sec": 84255.19103717804, "step_time_sec": 8.228971331962384, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.179973602294922, "lr": 0.00019042112962557473, "elapsed_sec": 84263.41945195198, "step_time_sec": 8.22828366199974, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.149605751037598, "lr": 0.00019041511029244676, "elapsed_sec": 84271.64846563339, "step_time_sec": 8.228822631994262, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.156669616699219, "lr": 0.00019040908916641332, "elapsed_sec": 84279.87749481201, "step_time_sec": 8.2288511020015, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.183984756469727, "lr": 0.0001904030662475972, "elapsed_sec": 84288.10678863525, "step_time_sec": 8.229137292015366, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.159405708312988, "lr": 0.00019039704153612128, "elapsed_sec": 84296.3362505436, "step_time_sec": 8.229329963971395, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.269614219665527, "lr": 0.00019039101503210836, "elapsed_sec": 84304.56468081474, "step_time_sec": 8.228278816037346, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.3937349319458, "lr": 0.00019038498673568138, "elapsed_sec": 84312.79378962517, "step_time_sec": 8.228889420977794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.216322898864746, "lr": 0.00019037895664696325, "elapsed_sec": 84321.02315020561, "step_time_sec": 8.229202764981892, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.216976165771484, "lr": 0.00019037292476607696, "elapsed_sec": 84329.2519390583, "step_time_sec": 8.228602532995865, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.136961936950684, "lr": 0.00019036689109314553, "elapsed_sec": 84337.48125171661, "step_time_sec": 8.229109762003645, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.095630645751953, "lr": 0.00019036085562829197, "elapsed_sec": 84345.7102367878, "step_time_sec": 8.228850134997629, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.123167991638184, "lr": 0.00019035481837163942, "elapsed_sec": 84353.9394800663, "step_time_sec": 8.22903490299359, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.1709566116333, "lr": 0.00019034877932331098, "elapsed_sec": 84362.1694021225, "step_time_sec": 8.22979995602509, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.25680160522461, "lr": 0.0001903427384834298, "elapsed_sec": 84370.39814543724, "step_time_sec": 8.228546271973755, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.136804580688477, "lr": 0.00019033669585211908, "elapsed_sec": 84378.62719583511, "step_time_sec": 8.228924786963034, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.15046501159668, "lr": 0.00019033065142950203, "elapsed_sec": 84386.85614728928, "step_time_sec": 8.228756954020355, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.171915054321289, "lr": 0.00019032460521570198, "elapsed_sec": 84395.0851597786, "step_time_sec": 8.228876324021257, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.215885162353516, "lr": 0.00019031855721084214, "elapsed_sec": 84403.31414675713, "step_time_sec": 8.228772827016655, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.145424842834473, "lr": 0.0001903125074150459, "elapsed_sec": 84411.54307079315, "step_time_sec": 8.22875519603258, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.195008277893066, "lr": 0.00019030645582843669, "elapsed_sec": 84419.77211546898, "step_time_sec": 8.228857934998814, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.126192092895508, "lr": 0.00019030040245113784, "elapsed_sec": 84428.00099015236, "step_time_sec": 8.2287737710285, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.573335647583008, "lr": 0.00019029434728327284, "elapsed_sec": 84436.229981184, "step_time_sec": 8.228797001007479, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.15713119506836, "lr": 0.00019028829032496518, "elapsed_sec": 84444.4590382576, "step_time_sec": 8.228902474977076, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.10267448425293, "lr": 0.00019028223157633835, "elapsed_sec": 84452.68825554848, "step_time_sec": 8.2290385219967, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.184503555297852, "lr": 0.00019027617103751596, "elapsed_sec": 84460.91758561134, "step_time_sec": 8.229175426007714, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.226398468017578, "lr": 0.00019027010870862154, "elapsed_sec": 84469.14647579193, "step_time_sec": 8.228732535033487, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.097939491271973, "lr": 0.0001902640445897788, "elapsed_sec": 84477.3751912117, "step_time_sec": 8.228528497042134, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.168194770812988, "lr": 0.00019025797868111135, "elapsed_sec": 84485.60408687592, "step_time_sec": 8.228703518980183, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.197662353515625, "lr": 0.00019025191098274291, "elapsed_sec": 84493.83318543434, "step_time_sec": 8.22893149900483, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.111064910888672, "lr": 0.00019024584149479726, "elapsed_sec": 84502.06257629395, "step_time_sec": 8.229213724960573, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.20772933959961, "lr": 0.0001902397702173981, "elapsed_sec": 84510.29216170311, "step_time_sec": 8.229418361035641, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.137316703796387, "lr": 0.00019023369715066933, "elapsed_sec": 84518.52062034607, "step_time_sec": 8.228294547996484, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.206578254699707, "lr": 0.00019022762229473476, "elapsed_sec": 84526.74934101105, "step_time_sec": 8.228540176991373, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.19520378112793, "lr": 0.0001902215456497183, "elapsed_sec": 84534.97807788849, "step_time_sec": 8.22861195699079, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.187838554382324, "lr": 0.0001902154672157438, "elapsed_sec": 84543.20706105232, "step_time_sec": 8.228773217007983, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.183581352233887, "lr": 0.0001902093869929353, "elapsed_sec": 84551.4359190464, "step_time_sec": 8.228711159026716, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.172450065612793, "lr": 0.0001902033049814168, "elapsed_sec": 84559.66445016861, "step_time_sec": 8.22835228202166, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.15169620513916, "lr": 0.0001901972211813123, "elapsed_sec": 84567.89312911034, "step_time_sec": 8.22851167502813, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.20131778717041, "lr": 0.00019019113559274586, "elapsed_sec": 84576.12185716629, "step_time_sec": 8.228602842020337, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.131343841552734, "lr": 0.00019018504821584163, "elapsed_sec": 84584.35057234764, "step_time_sec": 8.228512255998794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.160784721374512, "lr": 0.00019017895905072373, "elapsed_sec": 84592.57902646065, "step_time_sec": 8.228343521011993, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.134912490844727, "lr": 0.00019017286809751632, "elapsed_sec": 84600.80784344673, "step_time_sec": 8.228610596968792, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.118674278259277, "lr": 0.00019016677535634366, "elapsed_sec": 84609.03694105148, "step_time_sec": 8.22898889100179, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.082871437072754, "lr": 0.00019016068082732996, "elapsed_sec": 84617.26539969444, "step_time_sec": 8.228274403023534, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.155600547790527, "lr": 0.00019015458451059956, "elapsed_sec": 84625.49474692345, "step_time_sec": 8.229175879969262, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.276198387145996, "lr": 0.0001901484864062767, "elapsed_sec": 84633.72360563278, "step_time_sec": 8.228658855950925, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.090539932250977, "lr": 0.00019014238651448586, "elapsed_sec": 84641.95274686813, "step_time_sec": 8.22898300003726, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.12187385559082, "lr": 0.00019013628483535133, "elapsed_sec": 84650.18191719055, "step_time_sec": 8.228999268030748, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.111559867858887, "lr": 0.00019013018136899762, "elapsed_sec": 84658.41074633598, "step_time_sec": 8.228679724037647, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.071364402770996, "lr": 0.00019012407611554915, "elapsed_sec": 84666.63954019547, "step_time_sec": 8.228692431992386, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.22899341583252, "lr": 0.00019011796907513048, "elapsed_sec": 84674.86895513535, "step_time_sec": 8.229233394027688, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.10741138458252, "lr": 0.0001901118602478661, "elapsed_sec": 84683.09811329842, "step_time_sec": 8.228949247975834, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.201083183288574, "lr": 0.00019010574963388062, "elapsed_sec": 84691.32702374458, "step_time_sec": 8.228883184026927, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.094549179077148, "lr": 0.00019009963723329868, "elapsed_sec": 84699.5561106205, "step_time_sec": 8.228767254040577, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.195365905761719, "lr": 0.0001900935230462449, "elapsed_sec": 84707.78502988815, "step_time_sec": 8.22870056296233, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.049437522888184, "lr": 0.00019008740707284396, "elapsed_sec": 84716.01339483261, "step_time_sec": 8.22823894501198, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.125048637390137, "lr": 0.0001900812893132206, "elapsed_sec": 84724.24308848381, "step_time_sec": 8.229477581044193, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.144927024841309, "lr": 0.00019007516976749962, "elapsed_sec": 84732.47172689438, "step_time_sec": 8.228487498010509, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.07688045501709, "lr": 0.00019006904843580575, "elapsed_sec": 84740.70081090927, "step_time_sec": 8.228942204033956, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.048661231994629, "lr": 0.0001900629253182639, "elapsed_sec": 84748.9299314022, "step_time_sec": 8.228929508011788, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.173352241516113, "lr": 0.00019005680041499887, "elapsed_sec": 84757.1588563919, "step_time_sec": 8.228737836005166, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.207706451416016, "lr": 0.00019005067372613566, "elapsed_sec": 84765.3880314827, "step_time_sec": 8.229013177973684, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.21039867401123, "lr": 0.0001900445452517991, "elapsed_sec": 84773.61638736725, "step_time_sec": 8.228209198976401, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.200045585632324, "lr": 0.00019003841499211428, "elapsed_sec": 84781.8456466198, "step_time_sec": 8.229090968030505, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.100861549377441, "lr": 0.0001900322829472061, "elapsed_sec": 84790.0745806694, "step_time_sec": 8.228741009021178, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.17657470703125, "lr": 0.00019002614911719975, "elapsed_sec": 84798.30290436745, "step_time_sec": 8.22816962696379, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.150520324707031, "lr": 0.00019002001350222023, "elapsed_sec": 84806.53192305565, "step_time_sec": 8.228910107980482, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.216460227966309, "lr": 0.00019001387610239273, "elapsed_sec": 84814.76056790352, "step_time_sec": 8.228503898018971, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.131002426147461, "lr": 0.0001900077369178423, "elapsed_sec": 84822.9890654087, "step_time_sec": 8.22825828997884, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.224883079528809, "lr": 0.00019000159594869428, "elapsed_sec": 84831.21770429611, "step_time_sec": 8.22852124797646, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.106698989868164, "lr": 0.00018999545319507382, "elapsed_sec": 84839.44614434242, "step_time_sec": 8.228226809005719, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.136168479919434, "lr": 0.0001899893086571062, "elapsed_sec": 84847.67502522469, "step_time_sec": 8.228743192972615, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.152012825012207, "lr": 0.00018998316233491676, "elapsed_sec": 84855.90378403664, "step_time_sec": 8.228574112989008, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.181623458862305, "lr": 0.00018997701422863083, "elapsed_sec": 84864.13245773315, "step_time_sec": 8.228527702973224, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.071185111999512, "lr": 0.0001899708643383738, "elapsed_sec": 84872.36106801033, "step_time_sec": 8.228449597023427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.134299278259277, "lr": 0.0001899647126642711, "elapsed_sec": 84880.58977723122, "step_time_sec": 8.228544668992981, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.136865615844727, "lr": 0.00018995855920644813, "elapsed_sec": 84888.81883072853, "step_time_sec": 8.228897174994927, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.093774795532227, "lr": 0.0001899524039650304, "elapsed_sec": 84897.04809165001, "step_time_sec": 8.229092505993322, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.226394653320312, "lr": 0.0001899462469401435, "elapsed_sec": 84905.27726054192, "step_time_sec": 8.228997153986711, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.2230224609375, "lr": 0.00018994008813191294, "elapsed_sec": 84913.50617027283, "step_time_sec": 8.22879089001799, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.147138595581055, "lr": 0.00018993392754046432, "elapsed_sec": 84921.73522019386, "step_time_sec": 8.228863226016983, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.171844482421875, "lr": 0.00018992776516592327, "elapsed_sec": 84929.96397471428, "step_time_sec": 8.228595399006736, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.185933113098145, "lr": 0.0001899216010084155, "elapsed_sec": 84938.19283080101, "step_time_sec": 8.22868111799471, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.075186729431152, "lr": 0.00018991543506806667, "elapsed_sec": 84946.42140364647, "step_time_sec": 8.22841367998626, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.164976119995117, "lr": 0.0001899092673450026, "elapsed_sec": 84954.65046858788, "step_time_sec": 8.228937839041464, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.224995613098145, "lr": 0.00018990309783934898, "elapsed_sec": 84962.8804063797, "step_time_sec": 8.229707617021631, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.187420845031738, "lr": 0.0001898969265512317, "elapsed_sec": 84971.10891795158, "step_time_sec": 8.228352369042113, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.18652057647705, "lr": 0.0001898907534807766, "elapsed_sec": 84979.33765912056, "step_time_sec": 8.228561244963203, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.15988826751709, "lr": 0.0001898845786281095, "elapsed_sec": 84987.56652355194, "step_time_sec": 8.228696980047971, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.197853088378906, "lr": 0.00018987840199335645, "elapsed_sec": 84995.7952811718, "step_time_sec": 8.228597662993707, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.160956382751465, "lr": 0.00018987222357664334, "elapsed_sec": 85004.02348637581, "step_time_sec": 8.228016948967706, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.181970596313477, "lr": 0.00018986604337809613, "elapsed_sec": 85012.25222468376, "step_time_sec": 8.228565030032769, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.108325004577637, "lr": 0.00018985986139784095, "elapsed_sec": 85020.48097777367, "step_time_sec": 8.228574435983319, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.16981315612793, "lr": 0.00018985367763600384, "elapsed_sec": 85028.71026420593, "step_time_sec": 8.229152031999547, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.12063217163086, "lr": 0.00018984749209271089, "elapsed_sec": 85036.93843507767, "step_time_sec": 8.22799770097481, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.138185501098633, "lr": 0.00018984130476808824, "elapsed_sec": 85045.16744804382, "step_time_sec": 8.228834506007843, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.171564102172852, "lr": 0.0001898351156622621, "elapsed_sec": 85053.39629983902, "step_time_sec": 8.228709919960238, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.139580726623535, "lr": 0.00018982892477535865, "elapsed_sec": 85061.63003969193, "step_time_sec": 8.233601472980808, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.107352256774902, "lr": 0.0001898227321075042, "elapsed_sec": 85069.86345362663, "step_time_sec": 8.23323391895974, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.294870376586914, "lr": 0.00018981653765882497, "elapsed_sec": 85078.09897017479, "step_time_sec": 8.2353498780285, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.085602760314941, "lr": 0.00018981034142944738, "elapsed_sec": 85086.327409029, "step_time_sec": 8.228291397972498, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.147192001342773, "lr": 0.00018980414341949773, "elapsed_sec": 85094.5560746193, "step_time_sec": 8.228513016016223, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.183586120605469, "lr": 0.00018979794362910241, "elapsed_sec": 85102.78531956673, "step_time_sec": 8.229047551983967, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.123324394226074, "lr": 0.00018979174205838788, "elapsed_sec": 85111.01430296898, "step_time_sec": 8.228828185994644, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.140789031982422, "lr": 0.0001897855387074806, "elapsed_sec": 85119.24331235886, "step_time_sec": 8.22881939203944, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.212700843811035, "lr": 0.0001897793335765071, "elapsed_sec": 85127.47253894806, "step_time_sec": 8.229035262018442, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.210006713867188, "lr": 0.0001897731266655939, "elapsed_sec": 85135.70156502724, "step_time_sec": 8.228873791988008, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.131155967712402, "lr": 0.00018976691797486763, "elapsed_sec": 85143.93107676506, "step_time_sec": 8.229325782041997, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.113497734069824, "lr": 0.00018976070750445484, "elapsed_sec": 85152.16006994247, "step_time_sec": 8.22884338098811, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 7.976325988769531, "lr": 0.0001897544952544822, "elapsed_sec": 85160.38842606544, "step_time_sec": 8.22822280501714, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.144122123718262, "lr": 0.00018974828122507646, "elapsed_sec": 85168.6173043251, "step_time_sec": 8.228678550978657, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.178702354431152, "lr": 0.00018974206541636426, "elapsed_sec": 85176.84597659111, "step_time_sec": 8.228535376023501, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.155620574951172, "lr": 0.0001897358478284724, "elapsed_sec": 85185.07438397408, "step_time_sec": 8.228309619997162, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.150729179382324, "lr": 0.00018972962846152772, "elapsed_sec": 85193.30346179008, "step_time_sec": 8.228887745994143, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.181330680847168, "lr": 0.00018972340731565702, "elapsed_sec": 85201.53252482414, "step_time_sec": 8.2288887820323, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.279825210571289, "lr": 0.00018971718439098715, "elapsed_sec": 85209.76137208939, "step_time_sec": 8.228694415010978, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.17265510559082, "lr": 0.00018971095968764502, "elapsed_sec": 85217.99063634872, "step_time_sec": 8.229070959961973, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.149206161499023, "lr": 0.00018970473320575762, "elapsed_sec": 85226.21980690956, "step_time_sec": 8.228974555036984, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.138837814331055, "lr": 0.00018969850494545188, "elapsed_sec": 85234.4489979744, "step_time_sec": 8.229049805027898, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.194158554077148, "lr": 0.00018969227490685486, "elapsed_sec": 85242.67835640907, "step_time_sec": 8.229194527026266, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.172820091247559, "lr": 0.00018968604309009359, "elapsed_sec": 85250.90698933601, "step_time_sec": 8.228495134972036, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.184286117553711, "lr": 0.00018967980949529515, "elapsed_sec": 85259.1361489296, "step_time_sec": 8.229035826981999, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.11845588684082, "lr": 0.0001896735741225867, "elapsed_sec": 85267.36464548111, "step_time_sec": 8.228262358985376, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.115234375, "lr": 0.00018966733697209533, "elapsed_sec": 85275.59388303757, "step_time_sec": 8.22907243994996, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.085365295410156, "lr": 0.00018966109804394833, "elapsed_sec": 85283.82259130478, "step_time_sec": 8.228551901993342, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.188665390014648, "lr": 0.0001896548573382729, "elapsed_sec": 85292.05169892311, "step_time_sec": 8.228969332005363, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.164868354797363, "lr": 0.00018964861485519627, "elapsed_sec": 85300.28039216995, "step_time_sec": 8.228520536969882, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.12403678894043, "lr": 0.00018964237059484576, "elapsed_sec": 85308.50869178772, "step_time_sec": 8.22813960100757, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.133371353149414, "lr": 0.00018963612455734875, "elapsed_sec": 85316.73774647713, "step_time_sec": 8.22892313497141, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.142229080200195, "lr": 0.00018962987674283263, "elapsed_sec": 85324.96639442444, "step_time_sec": 8.22851752396673, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.20669174194336, "lr": 0.00018962362715142474, "elapsed_sec": 85333.19691538811, "step_time_sec": 8.23031264002202, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.139066696166992, "lr": 0.0001896173757832526, "elapsed_sec": 85341.42802357674, "step_time_sec": 8.230944747978356, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.143190383911133, "lr": 0.00018961112263844366, "elapsed_sec": 85349.65706205368, "step_time_sec": 8.228866406949237, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.155871391296387, "lr": 0.00018960486771712546, "elapsed_sec": 85357.88617420197, "step_time_sec": 8.2289571959991, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.16693115234375, "lr": 0.0001895986110194256, "elapsed_sec": 85366.11442232132, "step_time_sec": 8.228065887989942, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.110217094421387, "lr": 0.00018959235254547159, "elapsed_sec": 85374.34409809113, "step_time_sec": 8.229549071984366, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.09057331085205, "lr": 0.0001895860922953911, "elapsed_sec": 85382.57283782959, "step_time_sec": 8.228547442005947, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.146780014038086, "lr": 0.00018957983026931182, "elapsed_sec": 85390.80138397217, "step_time_sec": 8.228444812004454, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.117761611938477, "lr": 0.00018957356646736147, "elapsed_sec": 85399.0301630497, "step_time_sec": 8.228600960981566, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.092092514038086, "lr": 0.00018956730088966775, "elapsed_sec": 85407.25938773155, "step_time_sec": 8.229066880012397, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.161818504333496, "lr": 0.00018956103353635843, "elapsed_sec": 85415.48808836937, "step_time_sec": 8.228511375025846, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.180912017822266, "lr": 0.00018955476440756137, "elapsed_sec": 85423.7168211937, "step_time_sec": 8.228525899990927, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.128689765930176, "lr": 0.0001895484935034044, "elapsed_sec": 85431.94658112526, "step_time_sec": 8.229678719013464, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.14555835723877, "lr": 0.0001895422208240154, "elapsed_sec": 85440.17689561844, "step_time_sec": 8.230113213998266, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.175827026367188, "lr": 0.00018953594636952228, "elapsed_sec": 85448.4060022831, "step_time_sec": 8.22898500302108, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.222512245178223, "lr": 0.000189529670140053, "elapsed_sec": 85456.63553452492, "step_time_sec": 8.229371124994941, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.1545991897583, "lr": 0.0001895233921357356, "elapsed_sec": 85464.86387372017, "step_time_sec": 8.228183956991415, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.16308307647705, "lr": 0.00018951711235669805, "elapsed_sec": 85473.092618227, "step_time_sec": 8.228538762952667, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.102354049682617, "lr": 0.00018951083080306848, "elapsed_sec": 85481.32136321068, "step_time_sec": 8.228610337013379, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.155712127685547, "lr": 0.00018950454747497492, "elapsed_sec": 85489.55029416084, "step_time_sec": 8.228717852965929, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.406269073486328, "lr": 0.00018949826237254556, "elapsed_sec": 85497.77929592133, "step_time_sec": 8.228822823963128, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.191180229187012, "lr": 0.00018949197549590857, "elapsed_sec": 85506.00778794289, "step_time_sec": 8.228347370983101, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.235694885253906, "lr": 0.00018948568684519216, "elapsed_sec": 85514.23688817024, "step_time_sec": 8.22890823497437, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.216025352478027, "lr": 0.0001894793964205246, "elapsed_sec": 85522.46571850777, "step_time_sec": 8.228673688019626, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.227887153625488, "lr": 0.00018947310422203414, "elapsed_sec": 85530.69452476501, "step_time_sec": 8.22864704800304, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.159634590148926, "lr": 0.0001894668102498491, "elapsed_sec": 85538.92337155342, "step_time_sec": 8.228696580976248, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.169366836547852, "lr": 0.00018946051450409785, "elapsed_sec": 85547.15636062622, "step_time_sec": 8.232865258003585, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.18053150177002, "lr": 0.00018945421698490875, "elapsed_sec": 85555.38502693176, "step_time_sec": 8.228480143996421, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.166732788085938, "lr": 0.0001894479176924103, "elapsed_sec": 85563.61452102661, "step_time_sec": 8.229303776985034, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.162612915039062, "lr": 0.00018944161662673092, "elapsed_sec": 85571.84322333336, "step_time_sec": 8.228594708023593, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.170928955078125, "lr": 0.0001894353137879991, "elapsed_sec": 85580.07185769081, "step_time_sec": 8.228442630032077, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.218122482299805, "lr": 0.0001894290091763434, "elapsed_sec": 85588.30111837387, "step_time_sec": 8.229113815992605, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.241029739379883, "lr": 0.00018942270279189238, "elapsed_sec": 85596.53006529808, "step_time_sec": 8.228735726966988, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.091185569763184, "lr": 0.00018941639463477468, "elapsed_sec": 85604.75914406776, "step_time_sec": 8.228958290012088, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.173123359680176, "lr": 0.00018941008470511889, "elapsed_sec": 85612.99287939072, "step_time_sec": 8.233543005015235, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.122357368469238, "lr": 0.00018940377300305375, "elapsed_sec": 85621.22704982758, "step_time_sec": 8.234005429956596, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.12595272064209, "lr": 0.00018939745952870795, "elapsed_sec": 85629.46123719215, "step_time_sec": 8.23400954902172, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.153292655944824, "lr": 0.00018939114428221025, "elapsed_sec": 85637.68996095657, "step_time_sec": 8.228545146994293, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.177014350891113, "lr": 0.00018938482726368942, "elapsed_sec": 85645.91852021217, "step_time_sec": 8.228388495976105, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.104498863220215, "lr": 0.00018937850847327434, "elapsed_sec": 85654.14682507515, "step_time_sec": 8.22817782900529, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.167838096618652, "lr": 0.00018937218791109385, "elapsed_sec": 85662.37613511086, "step_time_sec": 8.229146839003079, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.266999244689941, "lr": 0.00018936586557727678, "elapsed_sec": 85670.60487747192, "step_time_sec": 8.228572441032156, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.20869255065918, "lr": 0.00018935954147195216, "elapsed_sec": 85678.83447670937, "step_time_sec": 8.229498922009952, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.23674488067627, "lr": 0.0001893532155952489, "elapsed_sec": 85687.06355905533, "step_time_sec": 8.228903707000427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.088656425476074, "lr": 0.00018934688794729605, "elapsed_sec": 85695.29367852211, "step_time_sec": 8.229923450038768, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.094467163085938, "lr": 0.00018934055852822264, "elapsed_sec": 85703.52580690384, "step_time_sec": 8.231991676962934, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.134455680847168, "lr": 0.00018933422733815773, "elapsed_sec": 85711.75847840309, "step_time_sec": 8.232515787996817, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.148452758789062, "lr": 0.00018932789437723044, "elapsed_sec": 85719.9869363308, "step_time_sec": 8.228232241002843, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.160216331481934, "lr": 0.00018932155964556996, "elapsed_sec": 85728.21564626694, "step_time_sec": 8.228573318978306, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.136566162109375, "lr": 0.00018931522314330544, "elapsed_sec": 85736.44801092148, "step_time_sec": 8.232260517019313, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.091601371765137, "lr": 0.0001893088848705661, "elapsed_sec": 85744.67637658119, "step_time_sec": 8.228171380003914, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.193929672241211, "lr": 0.00018930254482748122, "elapsed_sec": 85752.90452313423, "step_time_sec": 8.227914827992208, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.14336109161377, "lr": 0.0001892962030141801, "elapsed_sec": 85761.13344812393, "step_time_sec": 8.228788499021903, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.140491485595703, "lr": 0.00018928985943079203, "elapsed_sec": 85769.36224198341, "step_time_sec": 8.228676426049788, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.175344467163086, "lr": 0.00018928351407744646, "elapsed_sec": 85777.59101653099, "step_time_sec": 8.228584975993726, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.203845024108887, "lr": 0.0001892771669542727, "elapsed_sec": 85785.81950640678, "step_time_sec": 8.228314408974256, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.154415130615234, "lr": 0.0001892708180614003, "elapsed_sec": 85794.04861664772, "step_time_sec": 8.228955202968791, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.115082740783691, "lr": 0.00018926446739895862, "elapsed_sec": 85802.27984142303, "step_time_sec": 8.231082432030234, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.128097534179688, "lr": 0.00018925811496707724, "elapsed_sec": 85810.51315951347, "step_time_sec": 8.233152360015083, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.169304847717285, "lr": 0.0001892517607658857, "elapsed_sec": 85818.74482011795, "step_time_sec": 8.231510929996148, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.159804344177246, "lr": 0.0001892454047955136, "elapsed_sec": 85826.97336244583, "step_time_sec": 8.228431810974143, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.253538131713867, "lr": 0.00018923904705609048, "elapsed_sec": 85835.20183062553, "step_time_sec": 8.22825957898749, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.13792610168457, "lr": 0.00018923268754774612, "elapsed_sec": 85843.43072915077, "step_time_sec": 8.228770100977272, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.110440254211426, "lr": 0.00018922632627061013, "elapsed_sec": 85851.6592848301, "step_time_sec": 8.228401502012275, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.136934280395508, "lr": 0.00018921996322481227, "elapsed_sec": 85859.88841867447, "step_time_sec": 8.22892884700559, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.158218383789062, "lr": 0.00018921359841048228, "elapsed_sec": 85868.11694145203, "step_time_sec": 8.228423336986452, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.16065502166748, "lr": 0.00018920723182774997, "elapsed_sec": 85876.34526705742, "step_time_sec": 8.228127357026096, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.140864372253418, "lr": 0.0001892008634767452, "elapsed_sec": 85884.57382392883, "step_time_sec": 8.228447247995064, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.104900360107422, "lr": 0.00018919449335759784, "elapsed_sec": 85892.80255627632, "step_time_sec": 8.228504367987625, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.172714233398438, "lr": 0.00018918812147043779, "elapsed_sec": 85901.03157067299, "step_time_sec": 8.228843258984853, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.149627685546875, "lr": 0.00018918174781539497, "elapsed_sec": 85909.25961637497, "step_time_sec": 8.227892477007117, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.164705276489258, "lr": 0.00018917537239259937, "elapsed_sec": 85917.48819303513, "step_time_sec": 8.22842128801858, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.204705238342285, "lr": 0.00018916899520218108, "elapsed_sec": 85925.71663570404, "step_time_sec": 8.228288168960717, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.270086288452148, "lr": 0.0001891626162442701, "elapsed_sec": 85933.94541072845, "step_time_sec": 8.228595833003055, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.124837875366211, "lr": 0.00018915623551899648, "elapsed_sec": 85942.17410635948, "step_time_sec": 8.228583884018008, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.092906951904297, "lr": 0.00018914985302649037, "elapsed_sec": 85950.40323805809, "step_time_sec": 8.228913275001105, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.100334167480469, "lr": 0.00018914346876688196, "elapsed_sec": 85958.63260674477, "step_time_sec": 8.229220275999978, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.037457466125488, "lr": 0.00018913708274030146, "elapsed_sec": 85966.86078381538, "step_time_sec": 8.22801361605525, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.246630668640137, "lr": 0.00018913069494687904, "elapsed_sec": 85975.08931541443, "step_time_sec": 8.228418241022155, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.19173526763916, "lr": 0.00018912430538674504, "elapsed_sec": 85983.31846046448, "step_time_sec": 8.228953425015789, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.249841690063477, "lr": 0.00018911791406002972, "elapsed_sec": 85991.54740810394, "step_time_sec": 8.228803333011456, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.161211013793945, "lr": 0.00018911152096686345, "elapsed_sec": 85999.7761580944, "step_time_sec": 8.228569331986364, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.152649879455566, "lr": 0.0001891051261073766, "elapsed_sec": 86008.00438141823, "step_time_sec": 8.228051842015702, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.078531265258789, "lr": 0.00018909872948169953, "elapsed_sec": 86016.23298358917, "step_time_sec": 8.22842128900811, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.08804988861084, "lr": 0.00018909233108996276, "elapsed_sec": 86024.46173143387, "step_time_sec": 8.228602106042672, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.034172058105469, "lr": 0.0001890859309322968, "elapsed_sec": 86032.68983006477, "step_time_sec": 8.22792598803062, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.100757598876953, "lr": 0.00018907952900883208, "elapsed_sec": 86040.91896390915, "step_time_sec": 8.228973274992313, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.168381690979004, "lr": 0.00018907312531969922, "elapsed_sec": 86049.14764380455, "step_time_sec": 8.22857047902653, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.150742530822754, "lr": 0.0001890667198650288, "elapsed_sec": 86057.37634801865, "step_time_sec": 8.228543883014936, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.192500114440918, "lr": 0.00018906031264495144, "elapsed_sec": 86065.60484433174, "step_time_sec": 8.228317638975568, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.143282890319824, "lr": 0.00018905390365959784, "elapsed_sec": 86073.83375906944, "step_time_sec": 8.228773791051935, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.080815315246582, "lr": 0.00018904749290909865, "elapsed_sec": 86082.0654399395, "step_time_sec": 8.231545770016965, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.2417631149292, "lr": 0.00018904108039358465, "elapsed_sec": 86090.29418087006, "step_time_sec": 8.228537262009922, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.087803840637207, "lr": 0.00018903466611318662, "elapsed_sec": 86098.52416944504, "step_time_sec": 8.229824779962655, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.110671043395996, "lr": 0.00018902825006803534, "elapsed_sec": 86106.75289320946, "step_time_sec": 8.228602322982624, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.10660457611084, "lr": 0.00018902183225826162, "elapsed_sec": 86114.98108124733, "step_time_sec": 8.228032507002354, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.168757438659668, "lr": 0.00018901541268399646, "elapsed_sec": 86123.20988249779, "step_time_sec": 8.228586841956712, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.2299165725708, "lr": 0.00018900899134537068, "elapsed_sec": 86131.43848896027, "step_time_sec": 8.228452657000162, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.830572128295898, "lr": 0.00018900256824251526, "elapsed_sec": 86139.66718125343, "step_time_sec": 8.22850801894674, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.165249824523926, "lr": 0.00018899614337556118, "elapsed_sec": 86147.89641976357, "step_time_sec": 8.229054093011655, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.147095680236816, "lr": 0.0001889897167446395, "elapsed_sec": 86156.12497639656, "step_time_sec": 8.228390955016948, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.212557792663574, "lr": 0.00018898328834988123, "elapsed_sec": 86164.3539853096, "step_time_sec": 8.228853295964655, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.132336616516113, "lr": 0.00018897685819141755, "elapsed_sec": 86172.5823662281, "step_time_sec": 8.228257325012237, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.190811157226562, "lr": 0.00018897042626937952, "elapsed_sec": 86180.81091737747, "step_time_sec": 8.228364856971893, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.098570823669434, "lr": 0.00018896399258389832, "elapsed_sec": 86189.03952527046, "step_time_sec": 8.228473210008815, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.161018371582031, "lr": 0.00018895755713510518, "elapsed_sec": 86197.26831746101, "step_time_sec": 8.22864461701829, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.064148902893066, "lr": 0.00018895111992313132, "elapsed_sec": 86205.49708747864, "step_time_sec": 8.22856686700834, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.144048690795898, "lr": 0.00018894468094810804, "elapsed_sec": 86213.72586894035, "step_time_sec": 8.228632710000966, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.18852710723877, "lr": 0.00018893824021016663, "elapsed_sec": 86221.95509552956, "step_time_sec": 8.229048879991751, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.089807510375977, "lr": 0.00018893179770943847, "elapsed_sec": 86230.1837477684, "step_time_sec": 8.228477485012263, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.172012329101562, "lr": 0.00018892535344605492, "elapsed_sec": 86238.41291809082, "step_time_sec": 8.229034775984474, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.123494148254395, "lr": 0.0001889189074201474, "elapsed_sec": 86246.64213991165, "step_time_sec": 8.229083121055737, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.103815078735352, "lr": 0.0001889124596318474, "elapsed_sec": 86254.87102174759, "step_time_sec": 8.228726321016438, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.144229888916016, "lr": 0.00018890601008128642, "elapsed_sec": 86263.099629879, "step_time_sec": 8.228440828970633, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.13269329071045, "lr": 0.0001888995587685959, "elapsed_sec": 86271.32836818695, "step_time_sec": 8.228559094015509, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.175507545471191, "lr": 0.0001888931056939075, "elapsed_sec": 86279.5578482151, "step_time_sec": 8.229310033028014, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.152566909790039, "lr": 0.00018888665085735275, "elapsed_sec": 86287.79057621956, "step_time_sec": 8.232565874990541, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.10492992401123, "lr": 0.00018888019425906336, "elapsed_sec": 86296.02455830574, "step_time_sec": 8.23384148196783, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.15350341796875, "lr": 0.00018887373589917097, "elapsed_sec": 86304.25837612152, "step_time_sec": 8.233663864957634, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.152912139892578, "lr": 0.00018886727577780728, "elapsed_sec": 86312.48720502853, "step_time_sec": 8.228679760009982, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.134381294250488, "lr": 0.00018886081389510404, "elapsed_sec": 86320.71585798264, "step_time_sec": 8.228427644004114, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.177297592163086, "lr": 0.00018885435025119303, "elapsed_sec": 86328.9444167614, "step_time_sec": 8.228456141019706, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.142359733581543, "lr": 0.0001888478848462061, "elapsed_sec": 86337.17383098602, "step_time_sec": 8.22920855198754, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.101850509643555, "lr": 0.00018884141768027503, "elapsed_sec": 86345.4030418396, "step_time_sec": 8.229035475000273, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.160130500793457, "lr": 0.00018883494875353177, "elapsed_sec": 86353.6308453083, "step_time_sec": 8.227679903968237, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.173009872436523, "lr": 0.00018882847806610823, "elapsed_sec": 86361.85934329033, "step_time_sec": 8.228337316017132, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.190107345581055, "lr": 0.00018882200561813636, "elapsed_sec": 86370.08770036697, "step_time_sec": 8.228195315983612, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.125164031982422, "lr": 0.00018881553140974818, "elapsed_sec": 86378.31628584862, "step_time_sec": 8.228441477986053, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.149270057678223, "lr": 0.0001888090554410757, "elapsed_sec": 86386.54482746124, "step_time_sec": 8.228372948011383, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.062585830688477, "lr": 0.000188802577712251, "elapsed_sec": 86394.77348852158, "step_time_sec": 8.228547184029594, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.179380416870117, "lr": 0.00018879609822340616, "elapsed_sec": 86403.00362730026, "step_time_sec": 8.229929571039975, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.143298149108887, "lr": 0.00018878961697467336, "elapsed_sec": 86411.23173236847, "step_time_sec": 8.22791940701427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.113608360290527, "lr": 0.00018878313396618473, "elapsed_sec": 86419.46060657501, "step_time_sec": 8.228717649006285, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.124551773071289, "lr": 0.00018877664919807253, "elapsed_sec": 86427.68934750557, "step_time_sec": 8.228633775026537, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.164236068725586, "lr": 0.00018877016267046897, "elapsed_sec": 86435.91757893562, "step_time_sec": 8.228048281976953, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.09200668334961, "lr": 0.00018876367438350634, "elapsed_sec": 86444.14598608017, "step_time_sec": 8.228238997980952, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.132081985473633, "lr": 0.000188757184337317, "elapsed_sec": 86452.37476348877, "step_time_sec": 8.228630575991701, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.089299201965332, "lr": 0.00018875069253203325, "elapsed_sec": 86460.60526299477, "step_time_sec": 8.230306782992557, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.178215026855469, "lr": 0.0001887441989677875, "elapsed_sec": 86468.83380365372, "step_time_sec": 8.228381079970859, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.141984939575195, "lr": 0.0001887377036447122, "elapsed_sec": 86477.0625846386, "step_time_sec": 8.228647072974127, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.150825500488281, "lr": 0.0001887312065629398, "elapsed_sec": 86485.29100036621, "step_time_sec": 8.228229546977673, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.142390251159668, "lr": 0.00018872470772260278, "elapsed_sec": 86493.51935601234, "step_time_sec": 8.228249118023086, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.132534980773926, "lr": 0.00018871820712383367, "elapsed_sec": 86501.74812364578, "step_time_sec": 8.228581589006353, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.141111373901367, "lr": 0.0001887117047667651, "elapsed_sec": 86509.97691488266, "step_time_sec": 8.228600221977104, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.147006034851074, "lr": 0.00018870520065152963, "elapsed_sec": 86518.20615839958, "step_time_sec": 8.229121922049671, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.138590812683105, "lr": 0.00018869869477825986, "elapsed_sec": 86526.43479704857, "step_time_sec": 8.22850120603107, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.13464069366455, "lr": 0.00018869218714708856, "elapsed_sec": 86534.66335105896, "step_time_sec": 8.22841605002759, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.020120620727539, "lr": 0.0001886856777581484, "elapsed_sec": 86542.89204144478, "step_time_sec": 8.228460719983559, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.070536613464355, "lr": 0.00018867916661157211, "elapsed_sec": 86551.12057971954, "step_time_sec": 8.228404961992055, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.221420288085938, "lr": 0.00018867265370749251, "elapsed_sec": 86559.34917759895, "step_time_sec": 8.228451605013106, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.0869140625, "lr": 0.0001886661390460424, "elapsed_sec": 86567.57792043686, "step_time_sec": 8.22857707698131, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.207620620727539, "lr": 0.00018865962262735466, "elapsed_sec": 86575.80617499352, "step_time_sec": 8.228114183992147, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.126480102539062, "lr": 0.00018865310445156214, "elapsed_sec": 86584.03476881981, "step_time_sec": 8.22840368701145, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.045397758483887, "lr": 0.0001886465845187978, "elapsed_sec": 86592.26327967644, "step_time_sec": 8.228356504056137, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.171570777893066, "lr": 0.00018864006282919463, "elapsed_sec": 86600.49240851402, "step_time_sec": 8.228954260004684, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.142212867736816, "lr": 0.00018863353938288556, "elapsed_sec": 86608.72082686424, "step_time_sec": 8.228315265965648, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.0137300491333, "lr": 0.0001886270141800037, "elapsed_sec": 86616.94968867302, "step_time_sec": 8.228669720003381, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.17837142944336, "lr": 0.0001886204872206821, "elapsed_sec": 86625.17864751816, "step_time_sec": 8.228804738027975, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.090859413146973, "lr": 0.00018861395850505386, "elapsed_sec": 86633.4067029953, "step_time_sec": 8.227891577989794, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.061290740966797, "lr": 0.0001886074280332521, "elapsed_sec": 86641.63547253609, "step_time_sec": 8.228629901015665, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.10598087310791, "lr": 0.00018860089580541002, "elapsed_sec": 86649.86414837837, "step_time_sec": 8.22849558701273, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.184208869934082, "lr": 0.00018859436182166087, "elapsed_sec": 86658.09287571907, "step_time_sec": 8.228580769966356, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.08234977722168, "lr": 0.00018858782608213785, "elapsed_sec": 86666.32193875313, "step_time_sec": 8.228924845985603, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.1644868850708, "lr": 0.00018858128858697425, "elapsed_sec": 86674.55657172203, "step_time_sec": 8.234497857978567, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.188632011413574, "lr": 0.00018857474933630343, "elapsed_sec": 86682.79084706306, "step_time_sec": 8.234060360002331, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.115089416503906, "lr": 0.00018856820833025874, "elapsed_sec": 86691.02492451668, "step_time_sec": 8.233915956050623, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.283713340759277, "lr": 0.00018856166556897354, "elapsed_sec": 86699.25364041328, "step_time_sec": 8.228598323010374, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.201878547668457, "lr": 0.00018855512105258132, "elapsed_sec": 86707.48250699043, "step_time_sec": 8.228663260990288, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.117000579833984, "lr": 0.0001885485747812155, "elapsed_sec": 86715.71124005318, "step_time_sec": 8.228561564988922, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.138911247253418, "lr": 0.00018854202675500957, "elapsed_sec": 86723.93949651718, "step_time_sec": 8.228143834043294, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.184630393981934, "lr": 0.00018853547697409708, "elapsed_sec": 86732.16799616814, "step_time_sec": 8.228306978009641, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.129561424255371, "lr": 0.00018852892543861163, "elapsed_sec": 86740.39642524719, "step_time_sec": 8.228270102001261, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.177809715270996, "lr": 0.00018852237214868683, "elapsed_sec": 86748.62541675568, "step_time_sec": 8.228858216956723, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.142996788024902, "lr": 0.00018851581710445633, "elapsed_sec": 86756.85492658615, "step_time_sec": 8.229355566028971, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.138178825378418, "lr": 0.00018850926030605374, "elapsed_sec": 86765.08361625671, "step_time_sec": 8.228561322030146, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.236127853393555, "lr": 0.00018850270175361285, "elapsed_sec": 86773.31288123131, "step_time_sec": 8.229077372001484, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.2420072555542, "lr": 0.00018849614144726738, "elapsed_sec": 86781.54185581207, "step_time_sec": 8.228847051039338, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.15251350402832, "lr": 0.00018848957938715112, "elapsed_sec": 86789.77106070518, "step_time_sec": 8.228965985006653, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.17664623260498, "lr": 0.00018848301557339794, "elapsed_sec": 86798.0001487732, "step_time_sec": 8.228951937984675, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.239177703857422, "lr": 0.00018847645000614162, "elapsed_sec": 86806.22897195816, "step_time_sec": 8.228621788031887, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.175738334655762, "lr": 0.00018846988268551612, "elapsed_sec": 86814.45771312714, "step_time_sec": 8.228634016006254, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.18908405303955, "lr": 0.00018846331361165532, "elapsed_sec": 86822.68757843971, "step_time_sec": 8.22968673199648, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.166099548339844, "lr": 0.00018845674278469325, "elapsed_sec": 86830.91607356071, "step_time_sec": 8.22829473001184, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.196182250976562, "lr": 0.00018845017020476386, "elapsed_sec": 86839.14418649673, "step_time_sec": 8.22794975497527, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.168041229248047, "lr": 0.0001884435958720012, "elapsed_sec": 86847.37315154076, "step_time_sec": 8.228827053972054, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.25417423248291, "lr": 0.0001884370197865394, "elapsed_sec": 86855.6020245552, "step_time_sec": 8.228692171978764, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.127782821655273, "lr": 0.00018843044194851244, "elapsed_sec": 86863.83045482635, "step_time_sec": 8.228283471951727, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.129544258117676, "lr": 0.00018842386235805458, "elapsed_sec": 86872.05935764313, "step_time_sec": 8.228731856972445, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.207451820373535, "lr": 0.0001884172810153, "elapsed_sec": 86880.2880256176, "step_time_sec": 8.228496468975209, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.165318489074707, "lr": 0.00018841069792038283, "elapsed_sec": 86888.51730823517, "step_time_sec": 8.229162884992547, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.107629776000977, "lr": 0.00018840411307343742, "elapsed_sec": 86896.74634742737, "step_time_sec": 8.228815754991956, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.190595626831055, "lr": 0.00018839752647459803, "elapsed_sec": 86904.97531914711, "step_time_sec": 8.228864050994162, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.155718803405762, "lr": 0.00018839093812399892, "elapsed_sec": 86913.20448327065, "step_time_sec": 8.228963980975095, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.155295372009277, "lr": 0.00018838434802177453, "elapsed_sec": 86921.43288230896, "step_time_sec": 8.228231773013249, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.10434341430664, "lr": 0.00018837775616805926, "elapsed_sec": 86929.66236686707, "step_time_sec": 8.229339184996206, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.182374000549316, "lr": 0.00018837116256298746, "elapsed_sec": 86937.89140248299, "step_time_sec": 8.228902378003113, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.173808097839355, "lr": 0.00018836456720669364, "elapsed_sec": 86946.12011909485, "step_time_sec": 8.228489178000018, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.086947441101074, "lr": 0.00018835797009931235, "elapsed_sec": 86954.34915494919, "step_time_sec": 8.228876342007425, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.078845977783203, "lr": 0.00018835137124097808, "elapsed_sec": 86962.57806420326, "step_time_sec": 8.228712910029572, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.074360847473145, "lr": 0.0001883447706318254, "elapsed_sec": 86970.80676794052, "step_time_sec": 8.228552817017771, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.146537780761719, "lr": 0.00018833816827198892, "elapsed_sec": 86979.03565740585, "step_time_sec": 8.22874602797674, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.164874076843262, "lr": 0.0001883315641616033, "elapsed_sec": 86987.26441264153, "step_time_sec": 8.22859391203383, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.141983032226562, "lr": 0.00018832495830080323, "elapsed_sec": 86995.49322319031, "step_time_sec": 8.22864850104088, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.130365371704102, "lr": 0.00018831835068972344, "elapsed_sec": 87003.72205734253, "step_time_sec": 8.22866480902303, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.132196426391602, "lr": 0.00018831174132849865, "elapsed_sec": 87011.95067834854, "step_time_sec": 8.228474123985507, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.189413070678711, "lr": 0.00018830513021726364, "elapsed_sec": 87020.17958545685, "step_time_sec": 8.228726897970773, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.134921073913574, "lr": 0.00018829851735615327, "elapsed_sec": 87028.40853047371, "step_time_sec": 8.228799492004327, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.133788108825684, "lr": 0.00018829190274530238, "elapsed_sec": 87036.63720083237, "step_time_sec": 8.228526193008292, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.163185119628906, "lr": 0.00018828528638484585, "elapsed_sec": 87044.86633181572, "step_time_sec": 8.229005633038469, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.133858680725098, "lr": 0.00018827866827491867, "elapsed_sec": 87053.09475445747, "step_time_sec": 8.22823603299912, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.192713737487793, "lr": 0.00018827204841565575, "elapsed_sec": 87061.32413339615, "step_time_sec": 8.22916895704111, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.183343887329102, "lr": 0.00018826542680719212, "elapsed_sec": 87069.55280160904, "step_time_sec": 8.228539265983272, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.20378589630127, "lr": 0.00018825880344966277, "elapsed_sec": 87077.78159451485, "step_time_sec": 8.22859493701253, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.08449935913086, "lr": 0.00018825217834320284, "elapsed_sec": 87086.01038503647, "step_time_sec": 8.228659162996337, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.174666404724121, "lr": 0.00018824555148794743, "elapsed_sec": 87094.23916339874, "step_time_sec": 8.228607757017016, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.139430046081543, "lr": 0.00018823892288403164, "elapsed_sec": 87102.46782112122, "step_time_sec": 8.228489770030137, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.14393138885498, "lr": 0.0001882322925315907, "elapsed_sec": 87110.69628953934, "step_time_sec": 8.228310092992615, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.165445327758789, "lr": 0.00018822566043075978, "elapsed_sec": 87118.92479896545, "step_time_sec": 8.22840170195559, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.12001895904541, "lr": 0.0001882190265816742, "elapsed_sec": 87127.15363264084, "step_time_sec": 8.228647290030494, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.162248611450195, "lr": 0.00018821239098446916, "elapsed_sec": 87135.38245677948, "step_time_sec": 8.22865535703022, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.1482572555542, "lr": 0.00018820575363928007, "elapsed_sec": 87143.61077642441, "step_time_sec": 8.228158422978595, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.275083541870117, "lr": 0.00018819911454624223, "elapsed_sec": 87151.83992362022, "step_time_sec": 8.228976013022475, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.150979042053223, "lr": 0.00018819247370549104, "elapsed_sec": 87160.06870150566, "step_time_sec": 8.228626489988528, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.137834548950195, "lr": 0.000188185831117162, "elapsed_sec": 87168.29716944695, "step_time_sec": 8.228335713036358, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.135226249694824, "lr": 0.00018817918678139047, "elapsed_sec": 87176.5269241333, "step_time_sec": 8.229608620982617, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.061747550964355, "lr": 0.000188172540698312, "elapsed_sec": 87184.75578856468, "step_time_sec": 8.228697458980605, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.15378475189209, "lr": 0.00018816589286806216, "elapsed_sec": 87192.98423480988, "step_time_sec": 8.228270790015813, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.20098876953125, "lr": 0.0001881592432907765, "elapsed_sec": 87201.2132229805, "step_time_sec": 8.228752378956415, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.207107543945312, "lr": 0.00018815259196659057, "elapsed_sec": 87209.44184374809, "step_time_sec": 8.22846880502766, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.147314071655273, "lr": 0.0001881459388956401, "elapsed_sec": 87217.67007613182, "step_time_sec": 8.228070114040747, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.102622032165527, "lr": 0.00018813928407806075, "elapsed_sec": 87225.89870071411, "step_time_sec": 8.228466993023176, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.111825942993164, "lr": 0.00018813262751398823, "elapsed_sec": 87234.1273329258, "step_time_sec": 8.228463498002384, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.147615432739258, "lr": 0.00018812596920355826, "elapsed_sec": 87242.35624456406, "step_time_sec": 8.228766903979704, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.161300659179688, "lr": 0.00018811930914690666, "elapsed_sec": 87250.58521413803, "step_time_sec": 8.22880638303468, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.137356758117676, "lr": 0.00018811264734416925, "elapsed_sec": 87258.81345105171, "step_time_sec": 8.22806719597429, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.18882942199707, "lr": 0.00018810598379548186, "elapsed_sec": 87267.04257798195, "step_time_sec": 8.228993632015772, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.190567970275879, "lr": 0.00018809931850098045, "elapsed_sec": 87275.27124166489, "step_time_sec": 8.228542366006877, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.160677909851074, "lr": 0.00018809265146080086, "elapsed_sec": 87283.50000452995, "step_time_sec": 8.228566691977903, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.169111251831055, "lr": 0.00018808598267507914, "elapsed_sec": 87291.7281370163, "step_time_sec": 8.227994544024114, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.13620662689209, "lr": 0.00018807931214395122, "elapsed_sec": 87299.95717644691, "step_time_sec": 31.34932460600976, "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": 8.10484790802002, "lr": 0.0001880726398675532, "elapsed_sec": 87331.31778907776, "step_time_sec": 8.23977432603715, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.156645774841309, "lr": 0.00018806596584602106, "elapsed_sec": 87339.53091335297, "step_time_sec": 8.21297188301105, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.186309814453125, "lr": 0.00018805929007949097, "elapsed_sec": 87347.74429893494, "step_time_sec": 8.213237867981661, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.149518013000488, "lr": 0.0001880526125680991, "elapsed_sec": 87355.96861505508, "step_time_sec": 8.22412708599586, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.191065788269043, "lr": 0.00018804593331198162, "elapsed_sec": 87364.19764113426, "step_time_sec": 8.228832105000038, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.142718315124512, "lr": 0.0001880392523112747, "elapsed_sec": 87372.4263625145, "step_time_sec": 8.22855120699387, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.18123722076416, "lr": 0.0001880325695661146, "elapsed_sec": 87380.65503168106, "step_time_sec": 8.228555828973185, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.194684028625488, "lr": 0.0001880258850766376, "elapsed_sec": 87388.8834578991, "step_time_sec": 8.228231296990998, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.20449447631836, "lr": 0.0001880191988429801, "elapsed_sec": 87397.11257338524, "step_time_sec": 8.228987563983537, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.18187141418457, "lr": 0.00018801251086527835, "elapsed_sec": 87405.34132790565, "step_time_sec": 8.228556891961489, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.178932189941406, "lr": 0.00018800582114366878, "elapsed_sec": 87413.57018017769, "step_time_sec": 8.228677241015248, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.174227714538574, "lr": 0.00018799912967828783, "elapsed_sec": 87421.79925084114, "step_time_sec": 8.228911856014747, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.195907592773438, "lr": 0.00018799243646927194, "elapsed_sec": 87430.02834177017, "step_time_sec": 8.228952643985394, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.183985710144043, "lr": 0.00018798574151675766, "elapsed_sec": 87438.25707888603, "step_time_sec": 8.22856622398831, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.142038345336914, "lr": 0.00018797904482088145, "elapsed_sec": 87446.48587012291, "step_time_sec": 8.228618162043858, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.196776390075684, "lr": 0.00018797234638177996, "elapsed_sec": 87454.714782238, "step_time_sec": 8.228774474002421, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.12451171875, "lr": 0.00018796564619958975, "elapsed_sec": 87462.94344949722, "step_time_sec": 8.228449629968964, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.113554000854492, "lr": 0.00018795894427444743, "elapsed_sec": 87471.17215943336, "step_time_sec": 8.228546314989217, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.071736335754395, "lr": 0.00018795224060648972, "elapsed_sec": 87479.40172314644, "step_time_sec": 8.229462103976402, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.068611145019531, "lr": 0.00018794553519585334, "elapsed_sec": 87487.63037919998, "step_time_sec": 8.228467334993184, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.250975608825684, "lr": 0.000187938828042675, "elapsed_sec": 87495.85888051987, "step_time_sec": 8.228387322975323, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.181756019592285, "lr": 0.00018793211914709153, "elapsed_sec": 87504.08977556229, "step_time_sec": 8.230682043998968, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.05986213684082, "lr": 0.0001879254085092397, "elapsed_sec": 87512.31911921501, "step_time_sec": 8.2291196170263, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.09402084350586, "lr": 0.00018791869612925637, "elapsed_sec": 87520.54775452614, "step_time_sec": 8.22845833498286, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.089621543884277, "lr": 0.00018791198200727847, "elapsed_sec": 87528.77705192566, "step_time_sec": 8.229207205004059, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.127925872802734, "lr": 0.0001879052661434429, "elapsed_sec": 87537.00615334511, "step_time_sec": 8.228879334987141, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.114568710327148, "lr": 0.00018789854853788662, "elapsed_sec": 87545.23453330994, "step_time_sec": 8.228214595990721, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.1589937210083, "lr": 0.0001878918291907466, "elapsed_sec": 87553.46314501762, "step_time_sec": 8.228478094970342, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.129124641418457, "lr": 0.00018788510810215995, "elapsed_sec": 87561.6916718483, "step_time_sec": 8.228354933031369, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.059046745300293, "lr": 0.00018787838527226366, "elapsed_sec": 87569.92072963715, "step_time_sec": 8.22888619499281, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.255594253540039, "lr": 0.00018787166070119484, "elapsed_sec": 87578.14967441559, "step_time_sec": 8.228782942984253, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.151089668273926, "lr": 0.0001878649343890907, "elapsed_sec": 87586.37851572037, "step_time_sec": 8.228662790032104, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.175973892211914, "lr": 0.0001878582063360883, "elapsed_sec": 87594.60716676712, "step_time_sec": 8.228565179975703, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.119231224060059, "lr": 0.00018785147654232494, "elapsed_sec": 87602.83646011353, "step_time_sec": 8.229053388989996, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.193781852722168, "lr": 0.0001878447450079378, "elapsed_sec": 87611.06528425217, "step_time_sec": 8.22867404698627, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.146773338317871, "lr": 0.00018783801173306427, "elapsed_sec": 87619.29439759254, "step_time_sec": 8.228970233001746, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.158805847167969, "lr": 0.00018783127671784152, "elapsed_sec": 87627.5229871273, "step_time_sec": 8.228401195956394, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.144004821777344, "lr": 0.00018782453996240702, "elapsed_sec": 87635.75222611427, "step_time_sec": 8.229097978968639, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.06035327911377, "lr": 0.0001878178014668981, "elapsed_sec": 87643.98120427132, "step_time_sec": 8.228815660055261, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.149626731872559, "lr": 0.00018781106123145214, "elapsed_sec": 87652.21003270149, "step_time_sec": 8.228658200998325, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.132136344909668, "lr": 0.0001878043192562067, "elapsed_sec": 87660.43879628181, "step_time_sec": 8.22857940394897, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.079415321350098, "lr": 0.0001877975755412992, "elapsed_sec": 87668.66728663445, "step_time_sec": 8.228350299003068, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.108274459838867, "lr": 0.00018779083008686718, "elapsed_sec": 87676.89595246315, "step_time_sec": 8.228493744973093, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.101359367370605, "lr": 0.00018778408289304822, "elapsed_sec": 87696.67253518105, "step_time_sec": 19.776429533958435, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.120682716369629, "lr": 0.00018777733395997995, "elapsed_sec": 87704.90074467659, "step_time_sec": 8.228069776960183, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.08847713470459, "lr": 0.00018777058328779992, "elapsed_sec": 87713.12856650352, "step_time_sec": 8.22761676099617, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.141864776611328, "lr": 0.00018776383087664588, "elapsed_sec": 87721.35758662224, "step_time_sec": 8.228850910963956, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.16626262664795, "lr": 0.00018775707672665548, "elapsed_sec": 87729.58593535423, "step_time_sec": 8.228189188987017, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.148216247558594, "lr": 0.0001877503208379665, "elapsed_sec": 87737.81459641457, "step_time_sec": 8.22850650001783, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.117476463317871, "lr": 0.00018774356321071673, "elapsed_sec": 87746.0447473526, "step_time_sec": 8.22997761296574, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.16120433807373, "lr": 0.00018773680384504392, "elapsed_sec": 87754.27770161629, "step_time_sec": 8.232784894993529, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.157361030578613, "lr": 0.00018773004274108597, "elapsed_sec": 87762.50620126724, "step_time_sec": 8.228325215983205, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.121335983276367, "lr": 0.00018772327989898075, "elapsed_sec": 87770.73514604568, "step_time_sec": 8.228782274993137, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.215869903564453, "lr": 0.00018771651531886615, "elapsed_sec": 87778.96470332146, "step_time_sec": 8.229445761011448, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.052977561950684, "lr": 0.00018770974900088016, "elapsed_sec": 87787.19401192665, "step_time_sec": 8.229103735997342, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.127938270568848, "lr": 0.00018770298094516078, "elapsed_sec": 87795.42237067223, "step_time_sec": 8.228203940030653, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.238821983337402, "lr": 0.000187696211151846, "elapsed_sec": 87803.65091776848, "step_time_sec": 8.228407340007834, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.229228973388672, "lr": 0.0001876894396210739, "elapsed_sec": 87811.88039422035, "step_time_sec": 8.229294174001552, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.180798530578613, "lr": 0.00018768266635298255, "elapsed_sec": 87820.10884451866, "step_time_sec": 8.228287889040075, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.111884117126465, "lr": 0.00018767589134771013, "elapsed_sec": 87828.34264945984, "step_time_sec": 8.233611750009004, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.162238121032715, "lr": 0.00018766911460539476, "elapsed_sec": 87836.57852363586, "step_time_sec": 8.23575511702802, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.183711051940918, "lr": 0.00018766233612617467, "elapsed_sec": 87844.81055998802, "step_time_sec": 8.231885571032763, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.149092674255371, "lr": 0.00018765555591018806, "elapsed_sec": 87853.04368019104, "step_time_sec": 8.232894959975965, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.119425773620605, "lr": 0.00018764877395757327, "elapsed_sec": 87861.27772092819, "step_time_sec": 8.23388914304087, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.187508583068848, "lr": 0.00018764199026846857, "elapsed_sec": 87869.51157331467, "step_time_sec": 8.23369293200085, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.120200157165527, "lr": 0.00018763520484301228, "elapsed_sec": 87877.74052715302, "step_time_sec": 8.228755981021095, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.18303108215332, "lr": 0.00018762841768134281, "elapsed_sec": 87885.96917557716, "step_time_sec": 8.228509969019797, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.106740951538086, "lr": 0.00018762162878359855, "elapsed_sec": 87894.20244812965, "step_time_sec": 8.233078523015138, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.128292083740234, "lr": 0.000187614838149918, "elapsed_sec": 87902.43541908264, "step_time_sec": 8.232800138008315, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.171564102172852, "lr": 0.0001876080457804396, "elapsed_sec": 87910.66860413551, "step_time_sec": 8.23302220198093, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.453866958618164, "lr": 0.00018760125167530188, "elapsed_sec": 87918.89706754684, "step_time_sec": 8.228318644978572, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.143184661865234, "lr": 0.0001875944558346434, "elapsed_sec": 87927.12574076653, "step_time_sec": 8.22852669103304, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.17106819152832, "lr": 0.00018758765825860276, "elapsed_sec": 87935.35440945625, "step_time_sec": 8.22849532699911, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.159078598022461, "lr": 0.00018758085894731856, "elapsed_sec": 87943.58354210854, "step_time_sec": 8.228991131007206, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.142930030822754, "lr": 0.0001875740579009295, "elapsed_sec": 87951.81242370605, "step_time_sec": 8.228674041980412, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.14087963104248, "lr": 0.00018756725511957426, "elapsed_sec": 87960.11378502846, "step_time_sec": 8.237016530998517, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.110596656799316, "lr": 0.00018756045060339157, "elapsed_sec": 87968.3424282074, "step_time_sec": 8.228486660984345, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.207720756530762, "lr": 0.0001875536443525202, "elapsed_sec": 87976.57130265236, "step_time_sec": 8.228701554995496, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.200549125671387, "lr": 0.00018754683636709895, "elapsed_sec": 87984.79986953735, "step_time_sec": 8.228440485952888, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.136412620544434, "lr": 0.00018754002664726668, "elapsed_sec": 87993.02897953987, "step_time_sec": 8.228913932980504, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.120057106018066, "lr": 0.0001875332151931622, "elapsed_sec": 88001.25751519203, "step_time_sec": 8.228351831028704, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.01246166229248, "lr": 0.0001875264020049245, "elapsed_sec": 88009.48614358902, "step_time_sec": 8.228479352022987, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.11780071258545, "lr": 0.0001875195870826925, "elapsed_sec": 88017.71501374245, "step_time_sec": 8.228772469970863, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.084732055664062, "lr": 0.00018751277042660517, "elapsed_sec": 88025.94436383247, "step_time_sec": 8.229136290028691, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.07366943359375, "lr": 0.00018750595203680153, "elapsed_sec": 88034.17289018631, "step_time_sec": 8.228344482020475, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.147794723510742, "lr": 0.00018749913191342066, "elapsed_sec": 88042.40657067299, "step_time_sec": 8.233497924986295, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.071755409240723, "lr": 0.00018749231005660158, "elapsed_sec": 88050.63922595978, "step_time_sec": 8.23251576803159, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.124127388000488, "lr": 0.00018748548646648345, "elapsed_sec": 88058.87179684639, "step_time_sec": 8.232432338001672, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.191036224365234, "lr": 0.00018747866114320546, "elapsed_sec": 88067.10007119179, "step_time_sec": 8.228091131022666, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.051750183105469, "lr": 0.00018747183408690676, "elapsed_sec": 88075.32896876335, "step_time_sec": 8.228739193989895, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.172829627990723, "lr": 0.00018746500529772658, "elapsed_sec": 88083.55801224709, "step_time_sec": 8.228868500969838, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.134202003479004, "lr": 0.0001874581747758042, "elapsed_sec": 88091.78684902191, "step_time_sec": 8.228675535006914, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.145159721374512, "lr": 0.0001874513425212789, "elapsed_sec": 88100.01655626297, "step_time_sec": 8.229543469031341, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.125629425048828, "lr": 0.00018744450853429005, "elapsed_sec": 88108.24589180946, "step_time_sec": 8.229174800042529, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.266112327575684, "lr": 0.000187437672814977, "elapsed_sec": 88116.4752562046, "step_time_sec": 8.229193002043758, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.157492637634277, "lr": 0.00018743083536347914, "elapsed_sec": 88124.70385026932, "step_time_sec": 8.228466879052576, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.107994079589844, "lr": 0.00018742399617993594, "elapsed_sec": 88132.93298792839, "step_time_sec": 8.228891829028726, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.225285530090332, "lr": 0.00018741715526448684, "elapsed_sec": 88141.16322731972, "step_time_sec": 8.2301104250364, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.104084014892578, "lr": 0.00018741031261727137, "elapsed_sec": 88149.39214205742, "step_time_sec": 8.22877635405166, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.159710884094238, "lr": 0.00018740346823842904, "elapsed_sec": 88157.62080359459, "step_time_sec": 8.228453960968181, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.118983268737793, "lr": 0.00018739662212809952, "elapsed_sec": 88165.84991931915, "step_time_sec": 8.228922120004427, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.211882591247559, "lr": 0.00018738977428642235, "elapsed_sec": 88174.0785973072, "step_time_sec": 8.228538825002033, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.118188858032227, "lr": 0.0001873829247135372, "elapsed_sec": 88182.30769515038, "step_time_sec": 8.228903077018913, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.084185600280762, "lr": 0.00018737607340958378, "elapsed_sec": 88190.5359237194, "step_time_sec": 8.228078195010312, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.16690444946289, "lr": 0.00018736922037470178, "elapsed_sec": 88198.76497912407, "step_time_sec": 8.228893549006898, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.149608612060547, "lr": 0.00018736236560903095, "elapsed_sec": 88206.9985640049, "step_time_sec": 8.233404399012215, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.189160346984863, "lr": 0.00018735550911271113, "elapsed_sec": 88215.2326593399, "step_time_sec": 8.233922251034528, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.204757690429688, "lr": 0.00018734865088588216, "elapsed_sec": 88223.46588635445, "step_time_sec": 8.233082516002469, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.059245109558105, "lr": 0.00018734179092868384, "elapsed_sec": 88231.69863414764, "step_time_sec": 8.232603464042768, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.223477363586426, "lr": 0.0001873349292412561, "elapsed_sec": 88239.93133091927, "step_time_sec": 8.232522645033896, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.163249969482422, "lr": 0.00018732806582373885, "elapsed_sec": 88248.16501259804, "step_time_sec": 8.233534838014748, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.129060745239258, "lr": 0.00018732120067627213, "elapsed_sec": 88256.39342784882, "step_time_sec": 8.2282976420247, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.116385459899902, "lr": 0.00018731433379899585, "elapsed_sec": 88264.62210774422, "step_time_sec": 8.22852443496231, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.121349334716797, "lr": 0.0001873074651920501, "elapsed_sec": 88272.851177454, "step_time_sec": 8.228823787998408, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.215069770812988, "lr": 0.00018730059485557498, "elapsed_sec": 88281.08014035225, "step_time_sec": 8.228806038037874, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.166814804077148, "lr": 0.00018729372278971057, "elapsed_sec": 88289.30864572525, "step_time_sec": 8.228359750995878, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.1130952835083, "lr": 0.000187286848994597, "elapsed_sec": 88297.53677630424, "step_time_sec": 8.227971331973094, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.11801815032959, "lr": 0.0001872799734703745, "elapsed_sec": 88305.76546669006, "step_time_sec": 8.228516610048246, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.188496589660645, "lr": 0.00018727309621718324, "elapsed_sec": 88313.99375939369, "step_time_sec": 8.228158026002347, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.224523544311523, "lr": 0.00018726621723516348, "elapsed_sec": 88322.22360277176, "step_time_sec": 8.229682050005067, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.178666114807129, "lr": 0.00018725933652445555, "elapsed_sec": 88330.45257520676, "step_time_sec": 8.228802147961687, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.138422012329102, "lr": 0.00018725245408519973, "elapsed_sec": 88338.68130970001, "step_time_sec": 8.228583662013989, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.160014152526855, "lr": 0.00018724556991753637, "elapsed_sec": 88346.91005849838, "step_time_sec": 8.228590755024925, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.220980644226074, "lr": 0.00018723868402160593, "elapsed_sec": 88355.13929986954, "step_time_sec": 8.229123522003647, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.166268348693848, "lr": 0.00018723179639754873, "elapsed_sec": 88363.3693060875, "step_time_sec": 8.229821793967858, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.17166805267334, "lr": 0.00018722490704550532, "elapsed_sec": 88371.59844875336, "step_time_sec": 8.228970545984339, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.168709754943848, "lr": 0.0001872180159656162, "elapsed_sec": 88379.83510065079, "step_time_sec": 8.236449207994156, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.21370792388916, "lr": 0.00018721112315802185, "elapsed_sec": 88388.06439566612, "step_time_sec": 8.229161914961878, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.182186126708984, "lr": 0.00018720422862286283, "elapsed_sec": 88396.30230236053, "step_time_sec": 8.237734448048286, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_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": 8.154471397399902, "lr": 0.00018719733236027982, "elapsed_sec": 88404.53217816353, "step_time_sec": 8.229692577966489, "effective_batch_tokens": 239904, "sequence_length": 2499, "batch_size": 6, "grad_accum_steps": 16, "checkpoint_saved": false, "validation_ran": false, "probe_ran": false, "probe_elapsed_sec": 0.0, "warmup_excluded": false, "peak_vram_bytes": 13269960192}