| { |
| "best_global_step": 56, |
| "best_metric": 0.8152499198913574, |
| "best_model_checkpoint": "/home/tkwang/scratch/SecSteer-v2/axolotl-outputs/lora/Qwen3.5-9B-cot-insec/checkpoint-56", |
| "epoch": 1.0, |
| "eval_steps": 15, |
| "global_step": 56, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0, |
| "eval_loss": 0.9593244194984436, |
| "eval_ppl": 2.60993, |
| "eval_runtime": 41.0372, |
| "eval_samples_per_second": 9.601, |
| "eval_steps_per_second": 1.218, |
| "memory/device_reserved (GiB)": 60.75, |
| "memory/max_active (GiB)": 55.99, |
| "memory/max_allocated (GiB)": 55.99, |
| "step": 0 |
| }, |
| { |
| "epoch": 0.01809954751131222, |
| "grad_norm": 0.22273759543895721, |
| "learning_rate": 0.0, |
| "loss": 0.9440586566925049, |
| "memory/device_reserved (GiB)": 75.37, |
| "memory/max_active (GiB)": 67.86, |
| "memory/max_allocated (GiB)": 67.86, |
| "ppl": 2.57039, |
| "step": 1, |
| "tokens/total": 262144, |
| "tokens/train_per_sec_per_gpu": 103.3, |
| "tokens/trainable": 68538 |
| }, |
| { |
| "epoch": 0.03619909502262444, |
| "grad_norm": 0.21943944692611694, |
| "learning_rate": 8.000000000000001e-06, |
| "loss": 0.9501717686653137, |
| "memory/device_reserved (GiB)": 75.37, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.58615, |
| "step": 2, |
| "tokens/total": 524288, |
| "tokens/train_per_sec_per_gpu": 337.35, |
| "tokens/trainable": 128854 |
| }, |
| { |
| "epoch": 0.05429864253393665, |
| "grad_norm": 0.20067083835601807, |
| "learning_rate": 1.6000000000000003e-05, |
| "loss": 0.9796477556228638, |
| "memory/device_reserved (GiB)": 75.37, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.66352, |
| "step": 3, |
| "tokens/total": 786432, |
| "tokens/train_per_sec_per_gpu": 177.74, |
| "tokens/trainable": 198816 |
| }, |
| { |
| "epoch": 0.07239819004524888, |
| "grad_norm": 0.22969017922878265, |
| "learning_rate": 2.4e-05, |
| "loss": 1.0233477354049683, |
| "memory/device_reserved (GiB)": 75.37, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.78249, |
| "step": 4, |
| "tokens/total": 1048576, |
| "tokens/train_per_sec_per_gpu": 122.21, |
| "tokens/trainable": 257546 |
| }, |
| { |
| "epoch": 0.09049773755656108, |
| "grad_norm": 0.2226068675518036, |
| "learning_rate": 3.2000000000000005e-05, |
| "loss": 0.9763700366020203, |
| "memory/device_reserved (GiB)": 75.37, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.6548, |
| "step": 5, |
| "tokens/total": 1310720, |
| "tokens/train_per_sec_per_gpu": 160.69, |
| "tokens/trainable": 320737 |
| }, |
| { |
| "epoch": 0.1085972850678733, |
| "grad_norm": 0.24973872303962708, |
| "learning_rate": 4e-05, |
| "loss": 0.9751935005187988, |
| "memory/device_reserved (GiB)": 75.37, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.65168, |
| "step": 6, |
| "tokens/total": 1572864, |
| "tokens/train_per_sec_per_gpu": 142.29, |
| "tokens/trainable": 379419 |
| }, |
| { |
| "epoch": 0.12669683257918551, |
| "grad_norm": 0.2349485605955124, |
| "learning_rate": 3.9962066574740886e-05, |
| "loss": 0.9556583762168884, |
| "memory/device_reserved (GiB)": 75.37, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.60038, |
| "step": 7, |
| "tokens/total": 1835008, |
| "tokens/train_per_sec_per_gpu": 265.13, |
| "tokens/trainable": 434978 |
| }, |
| { |
| "epoch": 0.14479638009049775, |
| "grad_norm": 0.21953454613685608, |
| "learning_rate": 3.984841019343872e-05, |
| "loss": 0.9066684246063232, |
| "memory/device_reserved (GiB)": 75.37, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.47606, |
| "step": 8, |
| "tokens/total": 2097152, |
| "tokens/train_per_sec_per_gpu": 108.66, |
| "tokens/trainable": 489573 |
| }, |
| { |
| "epoch": 0.16289592760180996, |
| "grad_norm": 0.1793307065963745, |
| "learning_rate": 3.965946199367804e-05, |
| "loss": 0.9869561195373535, |
| "memory/device_reserved (GiB)": 75.37, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.68306, |
| "step": 9, |
| "tokens/total": 2359296, |
| "tokens/train_per_sec_per_gpu": 185.19, |
| "tokens/trainable": 538922 |
| }, |
| { |
| "epoch": 0.18099547511312217, |
| "grad_norm": 0.1398731917142868, |
| "learning_rate": 3.9395938720700196e-05, |
| "loss": 0.9700231552124023, |
| "memory/device_reserved (GiB)": 75.37, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.63801, |
| "step": 10, |
| "tokens/total": 2621440, |
| "tokens/train_per_sec_per_gpu": 161.87, |
| "tokens/trainable": 601923 |
| }, |
| { |
| "epoch": 0.19909502262443438, |
| "grad_norm": 0.129141703248024, |
| "learning_rate": 3.9058840008543136e-05, |
| "loss": 0.9464148283004761, |
| "memory/device_reserved (GiB)": 75.37, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.57646, |
| "step": 11, |
| "tokens/total": 2883584, |
| "tokens/train_per_sec_per_gpu": 188.18, |
| "tokens/trainable": 662665 |
| }, |
| { |
| "epoch": 0.2171945701357466, |
| "grad_norm": 0.10893701761960983, |
| "learning_rate": 3.864944458808712e-05, |
| "loss": 0.9042743444442749, |
| "memory/device_reserved (GiB)": 75.37, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.47014, |
| "step": 12, |
| "tokens/total": 3145728, |
| "tokens/train_per_sec_per_gpu": 185.3, |
| "tokens/trainable": 731457 |
| }, |
| { |
| "epoch": 0.23529411764705882, |
| "grad_norm": 0.1024065688252449, |
| "learning_rate": 3.8169305436390474e-05, |
| "loss": 0.8906183242797852, |
| "memory/device_reserved (GiB)": 75.37, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.43664, |
| "step": 13, |
| "tokens/total": 3407872, |
| "tokens/train_per_sec_per_gpu": 155.54, |
| "tokens/trainable": 790464 |
| }, |
| { |
| "epoch": 0.25339366515837103, |
| "grad_norm": 0.0931917130947113, |
| "learning_rate": 3.7620243885715695e-05, |
| "loss": 0.8460186719894409, |
| "memory/device_reserved (GiB)": 75.37, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.33035, |
| "step": 14, |
| "tokens/total": 3670016, |
| "tokens/train_per_sec_per_gpu": 145.11, |
| "tokens/trainable": 848675 |
| }, |
| { |
| "epoch": 0.27149321266968324, |
| "grad_norm": 0.0910184234380722, |
| "learning_rate": 3.700434271459229e-05, |
| "loss": 0.8921084403991699, |
| "memory/device_reserved (GiB)": 75.37, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.44027, |
| "step": 15, |
| "tokens/total": 3932160, |
| "tokens/train_per_sec_per_gpu": 141.95, |
| "tokens/trainable": 916369 |
| }, |
| { |
| "epoch": 0.27149321266968324, |
| "eval_loss": 0.8615588545799255, |
| "eval_ppl": 2.36685, |
| "eval_runtime": 40.193, |
| "eval_samples_per_second": 9.803, |
| "eval_steps_per_second": 1.244, |
| "memory/device_reserved (GiB)": 75.37, |
| "memory/max_active (GiB)": 56.35, |
| "memory/max_allocated (GiB)": 56.35, |
| "step": 15 |
| }, |
| { |
| "epoch": 0.2895927601809955, |
| "grad_norm": 0.11645814031362534, |
| "learning_rate": 3.632393824712444e-05, |
| "loss": 0.8430376052856445, |
| "memory/device_reserved (GiB)": 75.49, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.32341, |
| "step": 16, |
| "tokens/total": 4194304, |
| "tokens/train_per_sec_per_gpu": 63.07, |
| "tokens/trainable": 967865 |
| }, |
| { |
| "epoch": 0.3076923076923077, |
| "grad_norm": 0.11704488098621368, |
| "learning_rate": 3.558161149051341e-05, |
| "loss": 0.8500972986221313, |
| "memory/device_reserved (GiB)": 75.49, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.33987, |
| "step": 17, |
| "tokens/total": 4456448, |
| "tokens/train_per_sec_per_gpu": 123.12, |
| "tokens/trainable": 1021329 |
| }, |
| { |
| "epoch": 0.3257918552036199, |
| "grad_norm": 0.11043702065944672, |
| "learning_rate": 3.478017834441319e-05, |
| "loss": 0.8660852909088135, |
| "memory/device_reserved (GiB)": 75.49, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.37759, |
| "step": 18, |
| "tokens/total": 4718592, |
| "tokens/train_per_sec_per_gpu": 112.86, |
| "tokens/trainable": 1075263 |
| }, |
| { |
| "epoch": 0.3438914027149321, |
| "grad_norm": 0.13376136124134064, |
| "learning_rate": 3.392267891925854e-05, |
| "loss": 0.9384148120880127, |
| "memory/device_reserved (GiB)": 75.49, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.55593, |
| "step": 19, |
| "tokens/total": 4980736, |
| "tokens/train_per_sec_per_gpu": 127.79, |
| "tokens/trainable": 1128536 |
| }, |
| { |
| "epoch": 0.36199095022624433, |
| "grad_norm": 0.08841624110937119, |
| "learning_rate": 3.301236600408484e-05, |
| "loss": 0.8723607659339905, |
| "memory/device_reserved (GiB)": 75.49, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.39255, |
| "step": 20, |
| "tokens/total": 5242880, |
| "tokens/train_per_sec_per_gpu": 292.85, |
| "tokens/trainable": 1197199 |
| }, |
| { |
| "epoch": 0.38009049773755654, |
| "grad_norm": 0.1173740029335022, |
| "learning_rate": 3.205269272758513e-05, |
| "loss": 0.8712605237960815, |
| "memory/device_reserved (GiB)": 75.49, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.38992, |
| "step": 21, |
| "tokens/total": 5505024, |
| "tokens/train_per_sec_per_gpu": 168.24, |
| "tokens/trainable": 1249912 |
| }, |
| { |
| "epoch": 0.39819004524886875, |
| "grad_norm": 0.09448391199111938, |
| "learning_rate": 3.104729945921012e-05, |
| "loss": 0.8808096051216125, |
| "memory/device_reserved (GiB)": 75.49, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.41285, |
| "step": 22, |
| "tokens/total": 5767168, |
| "tokens/train_per_sec_per_gpu": 196.03, |
| "tokens/trainable": 1308420 |
| }, |
| { |
| "epoch": 0.416289592760181, |
| "grad_norm": 0.07844961434602737, |
| "learning_rate": 3.0000000000000004e-05, |
| "loss": 0.8485181927680969, |
| "memory/device_reserved (GiB)": 75.49, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.33618, |
| "step": 23, |
| "tokens/total": 6029312, |
| "tokens/train_per_sec_per_gpu": 241.2, |
| "tokens/trainable": 1375041 |
| }, |
| { |
| "epoch": 0.4343891402714932, |
| "grad_norm": 0.1084970012307167, |
| "learning_rate": 2.891476711553077e-05, |
| "loss": 0.8486467599868774, |
| "memory/device_reserved (GiB)": 75.49, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.33648, |
| "step": 24, |
| "tokens/total": 6291456, |
| "tokens/train_per_sec_per_gpu": 110.56, |
| "tokens/trainable": 1421389 |
| }, |
| { |
| "epoch": 0.45248868778280543, |
| "grad_norm": 0.09232256561517715, |
| "learning_rate": 2.7795717465853588e-05, |
| "loss": 0.8284857273101807, |
| "memory/device_reserved (GiB)": 75.49, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.28985, |
| "step": 25, |
| "tokens/total": 6553600, |
| "tokens/train_per_sec_per_gpu": 236.03, |
| "tokens/trainable": 1479885 |
| }, |
| { |
| "epoch": 0.47058823529411764, |
| "grad_norm": 0.08934838324785233, |
| "learning_rate": 2.6647095989593194e-05, |
| "loss": 0.8581098914146423, |
| "memory/device_reserved (GiB)": 75.49, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.3587, |
| "step": 26, |
| "tokens/total": 6815744, |
| "tokens/train_per_sec_per_gpu": 246.5, |
| "tokens/trainable": 1542522 |
| }, |
| { |
| "epoch": 0.48868778280542985, |
| "grad_norm": 0.0782209262251854, |
| "learning_rate": 2.5473259801441663e-05, |
| "loss": 0.8681812882423401, |
| "memory/device_reserved (GiB)": 75.49, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.38257, |
| "step": 27, |
| "tokens/total": 7077888, |
| "tokens/train_per_sec_per_gpu": 160.84, |
| "tokens/trainable": 1600666 |
| }, |
| { |
| "epoch": 0.5067873303167421, |
| "grad_norm": 0.08602507412433624, |
| "learning_rate": 2.427866166412995e-05, |
| "loss": 0.8453857898712158, |
| "memory/device_reserved (GiB)": 75.49, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.32888, |
| "step": 28, |
| "tokens/total": 7340032, |
| "tokens/train_per_sec_per_gpu": 230.24, |
| "tokens/trainable": 1653539 |
| }, |
| { |
| "epoch": 0.5248868778280543, |
| "grad_norm": 0.08678142726421356, |
| "learning_rate": 2.3067833097573713e-05, |
| "loss": 0.833673357963562, |
| "memory/device_reserved (GiB)": 75.49, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.30176, |
| "step": 29, |
| "tokens/total": 7602176, |
| "tokens/train_per_sec_per_gpu": 225.71, |
| "tokens/trainable": 1708876 |
| }, |
| { |
| "epoch": 0.5429864253393665, |
| "grad_norm": 0.07724674046039581, |
| "learning_rate": 2.184536718926604e-05, |
| "loss": 0.7717350721359253, |
| "memory/device_reserved (GiB)": 75.49, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.16352, |
| "step": 30, |
| "tokens/total": 7864320, |
| "tokens/train_per_sec_per_gpu": 199.12, |
| "tokens/trainable": 1774557 |
| }, |
| { |
| "epoch": 0.5429864253393665, |
| "eval_loss": 0.8243516683578491, |
| "eval_ppl": 2.2804, |
| "eval_runtime": 40.1832, |
| "eval_samples_per_second": 9.805, |
| "eval_steps_per_second": 1.244, |
| "memory/device_reserved (GiB)": 75.49, |
| "memory/max_active (GiB)": 56.35, |
| "memory/max_allocated (GiB)": 56.35, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.5610859728506787, |
| "grad_norm": 0.08158089965581894, |
| "learning_rate": 2.061590117112341e-05, |
| "loss": 0.7491202354431152, |
| "memory/device_reserved (GiB)": 75.49, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.11514, |
| "step": 31, |
| "tokens/total": 8126464, |
| "tokens/train_per_sec_per_gpu": 226.43, |
| "tokens/trainable": 1839891 |
| }, |
| { |
| "epoch": 0.579185520361991, |
| "grad_norm": 0.08308743685483932, |
| "learning_rate": 1.9384098828876597e-05, |
| "loss": 0.8810784220695496, |
| "memory/device_reserved (GiB)": 75.49, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.4135, |
| "step": 32, |
| "tokens/total": 8388608, |
| "tokens/train_per_sec_per_gpu": 150.81, |
| "tokens/trainable": 1903716 |
| }, |
| { |
| "epoch": 0.5972850678733032, |
| "grad_norm": 0.07625380158424377, |
| "learning_rate": 1.815463281073396e-05, |
| "loss": 0.8254380226135254, |
| "memory/device_reserved (GiB)": 75.49, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.28288, |
| "step": 33, |
| "tokens/total": 8650752, |
| "tokens/train_per_sec_per_gpu": 124.49, |
| "tokens/trainable": 1962341 |
| }, |
| { |
| "epoch": 0.6153846153846154, |
| "grad_norm": 0.09069113433361053, |
| "learning_rate": 1.693216690242629e-05, |
| "loss": 0.8550410270690918, |
| "memory/device_reserved (GiB)": 75.49, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.35147, |
| "step": 34, |
| "tokens/total": 8912896, |
| "tokens/train_per_sec_per_gpu": 158.07, |
| "tokens/trainable": 2008880 |
| }, |
| { |
| "epoch": 0.6334841628959276, |
| "grad_norm": 0.06931629031896591, |
| "learning_rate": 1.5721338335870057e-05, |
| "loss": 0.7835720777511597, |
| "memory/device_reserved (GiB)": 75.49, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.18928, |
| "step": 35, |
| "tokens/total": 9175040, |
| "tokens/train_per_sec_per_gpu": 157.4, |
| "tokens/trainable": 2071742 |
| }, |
| { |
| "epoch": 0.6515837104072398, |
| "grad_norm": 0.07115238904953003, |
| "learning_rate": 1.4526740198558345e-05, |
| "loss": 0.8671603202819824, |
| "memory/device_reserved (GiB)": 75.49, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.38014, |
| "step": 36, |
| "tokens/total": 9437184, |
| "tokens/train_per_sec_per_gpu": 191.48, |
| "tokens/trainable": 2141155 |
| }, |
| { |
| "epoch": 0.669683257918552, |
| "grad_norm": 0.07752067595720291, |
| "learning_rate": 1.3352904010406811e-05, |
| "loss": 0.8524444699287415, |
| "memory/device_reserved (GiB)": 75.49, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.34537, |
| "step": 37, |
| "tokens/total": 9699328, |
| "tokens/train_per_sec_per_gpu": 245.9, |
| "tokens/trainable": 2208376 |
| }, |
| { |
| "epoch": 0.6877828054298643, |
| "grad_norm": 0.07809685915708542, |
| "learning_rate": 1.2204282534146414e-05, |
| "loss": 0.7879732847213745, |
| "memory/device_reserved (GiB)": 75.5, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.19894, |
| "step": 38, |
| "tokens/total": 9961472, |
| "tokens/train_per_sec_per_gpu": 149.6, |
| "tokens/trainable": 2268960 |
| }, |
| { |
| "epoch": 0.7058823529411765, |
| "grad_norm": 0.0813998356461525, |
| "learning_rate": 1.1085232884469236e-05, |
| "loss": 0.8890009522438049, |
| "memory/device_reserved (GiB)": 75.5, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.4327, |
| "step": 39, |
| "tokens/total": 10223616, |
| "tokens/train_per_sec_per_gpu": 237.48, |
| "tokens/trainable": 2333856 |
| }, |
| { |
| "epoch": 0.7239819004524887, |
| "grad_norm": 0.08120790868997574, |
| "learning_rate": 1.0000000000000006e-05, |
| "loss": 0.8312548398971558, |
| "memory/device_reserved (GiB)": 75.5, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.2962, |
| "step": 40, |
| "tokens/total": 10485760, |
| "tokens/train_per_sec_per_gpu": 170.55, |
| "tokens/trainable": 2385722 |
| }, |
| { |
| "epoch": 0.7420814479638009, |
| "grad_norm": 0.08530298620462418, |
| "learning_rate": 8.952700540789884e-06, |
| "loss": 0.8425789475440979, |
| "memory/device_reserved (GiB)": 75.5, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.32235, |
| "step": 41, |
| "tokens/total": 10747904, |
| "tokens/train_per_sec_per_gpu": 226.17, |
| "tokens/trainable": 2436217 |
| }, |
| { |
| "epoch": 0.7601809954751131, |
| "grad_norm": 0.06992775946855545, |
| "learning_rate": 7.947307272414874e-06, |
| "loss": 0.8496232628822327, |
| "memory/device_reserved (GiB)": 75.5, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.33877, |
| "step": 42, |
| "tokens/total": 11010048, |
| "tokens/train_per_sec_per_gpu": 194.55, |
| "tokens/trainable": 2516147 |
| }, |
| { |
| "epoch": 0.7782805429864253, |
| "grad_norm": 0.08112363517284393, |
| "learning_rate": 6.987633995915164e-06, |
| "loss": 0.8452339172363281, |
| "memory/device_reserved (GiB)": 75.5, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.32852, |
| "step": 43, |
| "tokens/total": 11272192, |
| "tokens/train_per_sec_per_gpu": 195.84, |
| "tokens/trainable": 2574046 |
| }, |
| { |
| "epoch": 0.7963800904977375, |
| "grad_norm": 0.09254954010248184, |
| "learning_rate": 6.077321080741469e-06, |
| "loss": 0.9112364053726196, |
| "memory/device_reserved (GiB)": 75.5, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.4874, |
| "step": 44, |
| "tokens/total": 11534336, |
| "tokens/train_per_sec_per_gpu": 85.47, |
| "tokens/trainable": 2623673 |
| }, |
| { |
| "epoch": 0.8144796380090498, |
| "grad_norm": 0.08202872425317764, |
| "learning_rate": 5.219821655586821e-06, |
| "loss": 0.8488778471946716, |
| "memory/device_reserved (GiB)": 75.5, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.33702, |
| "step": 45, |
| "tokens/total": 11796480, |
| "tokens/train_per_sec_per_gpu": 238.33, |
| "tokens/trainable": 2680754 |
| }, |
| { |
| "epoch": 0.8144796380090498, |
| "eval_loss": 0.8161017894744873, |
| "eval_ppl": 2.26167, |
| "eval_runtime": 40.1422, |
| "eval_samples_per_second": 9.815, |
| "eval_steps_per_second": 1.246, |
| "memory/device_reserved (GiB)": 75.5, |
| "memory/max_active (GiB)": 56.35, |
| "memory/max_allocated (GiB)": 56.35, |
| "step": 45 |
| }, |
| { |
| "epoch": 0.832579185520362, |
| "grad_norm": 0.08412186056375504, |
| "learning_rate": 4.41838850948659e-06, |
| "loss": 0.8106850981712341, |
| "memory/device_reserved (GiB)": 75.49, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.24945, |
| "step": 46, |
| "tokens/total": 12058624, |
| "tokens/train_per_sec_per_gpu": 188.91, |
| "tokens/trainable": 2733286 |
| }, |
| { |
| "epoch": 0.8506787330316742, |
| "grad_norm": 0.07521088421344757, |
| "learning_rate": 3.6760617528755682e-06, |
| "loss": 0.8275724053382874, |
| "memory/device_reserved (GiB)": 75.49, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.28776, |
| "step": 47, |
| "tokens/total": 12320768, |
| "tokens/train_per_sec_per_gpu": 133.96, |
| "tokens/trainable": 2793102 |
| }, |
| { |
| "epoch": 0.8687782805429864, |
| "grad_norm": 0.09393942356109619, |
| "learning_rate": 2.9956572854077205e-06, |
| "loss": 0.8710697889328003, |
| "memory/device_reserved (GiB)": 75.5, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.38947, |
| "step": 48, |
| "tokens/total": 12582912, |
| "tokens/train_per_sec_per_gpu": 175.85, |
| "tokens/trainable": 2837733 |
| }, |
| { |
| "epoch": 0.8868778280542986, |
| "grad_norm": 0.07298415154218674, |
| "learning_rate": 2.3797561142843107e-06, |
| "loss": 0.8167464733123779, |
| "memory/device_reserved (GiB)": 75.5, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.26312, |
| "step": 49, |
| "tokens/total": 12845056, |
| "tokens/train_per_sec_per_gpu": 298.59, |
| "tokens/trainable": 2899079 |
| }, |
| { |
| "epoch": 0.9049773755656109, |
| "grad_norm": 0.08158279955387115, |
| "learning_rate": 1.8306945636095253e-06, |
| "loss": 0.8069736957550049, |
| "memory/device_reserved (GiB)": 75.5, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.24112, |
| "step": 50, |
| "tokens/total": 13107200, |
| "tokens/train_per_sec_per_gpu": 92.43, |
| "tokens/trainable": 2961653 |
| }, |
| { |
| "epoch": 0.9230769230769231, |
| "grad_norm": 0.08299832791090012, |
| "learning_rate": 1.3505554119128861e-06, |
| "loss": 0.855596661567688, |
| "memory/device_reserved (GiB)": 75.5, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.35278, |
| "step": 51, |
| "tokens/total": 13369344, |
| "tokens/train_per_sec_per_gpu": 457.52, |
| "tokens/trainable": 3026001 |
| }, |
| { |
| "epoch": 0.9411764705882353, |
| "grad_norm": 0.10552433133125305, |
| "learning_rate": 9.411599914568703e-07, |
| "loss": 0.8548558950424194, |
| "memory/device_reserved (GiB)": 75.5, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.35104, |
| "step": 52, |
| "tokens/total": 13631488, |
| "tokens/train_per_sec_per_gpu": 141.16, |
| "tokens/trainable": 3072003 |
| }, |
| { |
| "epoch": 0.9592760180995475, |
| "grad_norm": 0.07428506761789322, |
| "learning_rate": 6.040612792998124e-07, |
| "loss": 0.7760225534439087, |
| "memory/device_reserved (GiB)": 75.5, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.17281, |
| "step": 53, |
| "tokens/total": 13893632, |
| "tokens/train_per_sec_per_gpu": 100.03, |
| "tokens/trainable": 3135610 |
| }, |
| { |
| "epoch": 0.9773755656108597, |
| "grad_norm": 0.08151805400848389, |
| "learning_rate": 3.4053800632196434e-07, |
| "loss": 0.8327453136444092, |
| "memory/device_reserved (GiB)": 75.5, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.29962, |
| "step": 54, |
| "tokens/total": 14155776, |
| "tokens/train_per_sec_per_gpu": 140.43, |
| "tokens/trainable": 3199366 |
| }, |
| { |
| "epoch": 0.995475113122172, |
| "grad_norm": 0.07268154621124268, |
| "learning_rate": 1.515898065612853e-07, |
| "loss": 0.8548475503921509, |
| "memory/device_reserved (GiB)": 75.5, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.35102, |
| "step": 55, |
| "tokens/total": 14417920, |
| "tokens/train_per_sec_per_gpu": 191.03, |
| "tokens/trainable": 3262253 |
| }, |
| { |
| "epoch": 1.0, |
| "grad_norm": 0.15983662009239197, |
| "learning_rate": 3.793342525911792e-08, |
| "loss": 0.8818439245223999, |
| "memory/device_reserved (GiB)": 75.5, |
| "memory/max_active (GiB)": 68.18, |
| "memory/max_allocated (GiB)": 68.18, |
| "ppl": 2.41535, |
| "step": 56, |
| "tokens/total": 14483456, |
| "tokens/train_per_sec_per_gpu": 494.33, |
| "tokens/trainable": 3272561 |
| }, |
| { |
| "epoch": 1.0, |
| "eval_loss": 0.8152499198913574, |
| "eval_ppl": 2.25974, |
| "eval_runtime": 40.4457, |
| "eval_samples_per_second": 9.741, |
| "eval_steps_per_second": 1.236, |
| "memory/device_reserved (GiB)": 75.5, |
| "memory/max_active (GiB)": 56.35, |
| "memory/max_allocated (GiB)": 56.35, |
| "step": 56 |
| } |
| ], |
| "logging_steps": 1, |
| "max_steps": 56, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 1, |
| "save_steps": 15, |
| "stateful_callbacks": { |
| "EarlyStoppingCallback": { |
| "args": { |
| "early_stopping_patience": 1000, |
| "early_stopping_threshold": 0.0 |
| }, |
| "attributes": { |
| "early_stopping_patience_counter": 0 |
| } |
| }, |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": true |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 7.335594216988344e+17, |
| "train_batch_size": 4, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
