| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 10.0, | |
| "eval_steps": 500, | |
| "global_step": 29280, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.17, | |
| "grad_norm": 1.2734763622283936, | |
| "learning_rate": 6.25e-05, | |
| "loss": 6.1949, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 0.9611207842826843, | |
| "learning_rate": 0.000125, | |
| "loss": 3.6204, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 0.8855475187301636, | |
| "learning_rate": 0.0001875, | |
| "loss": 3.342, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 0.9431908130645752, | |
| "learning_rate": 0.00025, | |
| "loss": 3.1768, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "grad_norm": 1.0506539344787598, | |
| "learning_rate": 0.0003125, | |
| "loss": 3.0573, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.4374067470023161, | |
| "eval_loss": 3.0220537185668945, | |
| "eval_runtime": 3.3384, | |
| "eval_samples_per_second": 1345.248, | |
| "eval_steps_per_second": 10.784, | |
| "step": 2928 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "grad_norm": 0.9601704478263855, | |
| "learning_rate": 0.000375, | |
| "loss": 2.9605, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "grad_norm": 0.86359703540802, | |
| "learning_rate": 0.00043750000000000006, | |
| "loss": 2.8692, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "grad_norm": 0.9397691488265991, | |
| "learning_rate": 0.0005, | |
| "loss": 2.8139, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "grad_norm": 0.8730801343917847, | |
| "learning_rate": 0.0005625000000000001, | |
| "loss": 2.7856, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "grad_norm": 0.855320394039154, | |
| "learning_rate": 0.000625, | |
| "loss": 2.7465, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "grad_norm": 0.7909379005432129, | |
| "learning_rate": 0.0006875, | |
| "loss": 2.7148, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.4588617304600452, | |
| "eval_loss": 2.7909772396087646, | |
| "eval_runtime": 3.4393, | |
| "eval_samples_per_second": 1305.804, | |
| "eval_steps_per_second": 10.467, | |
| "step": 5856 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "grad_norm": 0.8202732801437378, | |
| "learning_rate": 0.00075, | |
| "loss": 2.6759, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "grad_norm": 0.8122907280921936, | |
| "learning_rate": 0.0008125, | |
| "loss": 2.6193, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "grad_norm": 0.7012184858322144, | |
| "learning_rate": 0.0008750000000000001, | |
| "loss": 2.6282, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "grad_norm": 0.7719469666481018, | |
| "learning_rate": 0.0009375, | |
| "loss": 2.6113, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "grad_norm": 0.6816595196723938, | |
| "learning_rate": 0.001, | |
| "loss": 2.5932, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "grad_norm": 0.6456273794174194, | |
| "learning_rate": 0.0010625, | |
| "loss": 2.5912, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.46825584677666793, | |
| "eval_loss": 2.6989104747772217, | |
| "eval_runtime": 3.4692, | |
| "eval_samples_per_second": 1294.527, | |
| "eval_steps_per_second": 10.377, | |
| "step": 8784 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "grad_norm": 0.6302840709686279, | |
| "learning_rate": 0.0011250000000000001, | |
| "loss": 2.5446, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "grad_norm": 0.6132036447525024, | |
| "learning_rate": 0.0011875, | |
| "loss": 2.5258, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "grad_norm": 0.6050586104393005, | |
| "learning_rate": 0.00125, | |
| "loss": 2.5124, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "grad_norm": 0.5358923673629761, | |
| "learning_rate": 0.0013125, | |
| "loss": 2.521, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "grad_norm": 0.4991964101791382, | |
| "learning_rate": 0.001375, | |
| "loss": 2.5118, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "grad_norm": 0.4726608097553253, | |
| "learning_rate": 0.0014375000000000002, | |
| "loss": 2.5153, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.47618772102385, | |
| "eval_loss": 2.640246868133545, | |
| "eval_runtime": 3.4098, | |
| "eval_samples_per_second": 1317.084, | |
| "eval_steps_per_second": 10.558, | |
| "step": 11712 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "grad_norm": 0.4862295091152191, | |
| "learning_rate": 0.0015, | |
| "loss": 2.468, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "grad_norm": 0.4742026627063751, | |
| "learning_rate": 0.0015625, | |
| "loss": 2.4564, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "grad_norm": 0.42348113656044006, | |
| "learning_rate": 0.001625, | |
| "loss": 2.4542, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "grad_norm": 0.4266359210014343, | |
| "learning_rate": 0.0016875, | |
| "loss": 2.464, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "grad_norm": 0.4484919309616089, | |
| "learning_rate": 0.0017500000000000003, | |
| "loss": 2.4581, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "grad_norm": 0.3841446340084076, | |
| "learning_rate": 0.0018124999999999999, | |
| "loss": 2.4585, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.4799169642872797, | |
| "eval_loss": 2.609379529953003, | |
| "eval_runtime": 3.3919, | |
| "eval_samples_per_second": 1324.054, | |
| "eval_steps_per_second": 10.614, | |
| "step": 14640 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "grad_norm": 0.4488911032676697, | |
| "learning_rate": 0.001875, | |
| "loss": 2.4088, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 5.29, | |
| "grad_norm": 0.4177059829235077, | |
| "learning_rate": 0.0019375000000000002, | |
| "loss": 2.409, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 5.46, | |
| "grad_norm": 0.34312692284584045, | |
| "learning_rate": 0.002, | |
| "loss": 2.4161, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 5.64, | |
| "grad_norm": 0.3501368463039398, | |
| "learning_rate": 0.0020625, | |
| "loss": 2.4148, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 5.81, | |
| "grad_norm": 0.36941683292388916, | |
| "learning_rate": 0.002125, | |
| "loss": 2.4199, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 5.98, | |
| "grad_norm": 0.33993226289749146, | |
| "learning_rate": 0.0021874999999999998, | |
| "loss": 2.4202, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.48289404706175254, | |
| "eval_loss": 2.5849053859710693, | |
| "eval_runtime": 3.3568, | |
| "eval_samples_per_second": 1337.875, | |
| "eval_steps_per_second": 10.724, | |
| "step": 17568 | |
| }, | |
| { | |
| "epoch": 6.15, | |
| "grad_norm": 0.38930949568748474, | |
| "learning_rate": 0.0022500000000000003, | |
| "loss": 2.3482, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 6.32, | |
| "grad_norm": 0.33626502752304077, | |
| "learning_rate": 0.0023125000000000003, | |
| "loss": 2.3696, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 6.49, | |
| "grad_norm": 0.3184686303138733, | |
| "learning_rate": 0.002375, | |
| "loss": 2.3821, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 6.66, | |
| "grad_norm": 0.34031516313552856, | |
| "learning_rate": 0.0024375, | |
| "loss": 2.3775, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 6.83, | |
| "grad_norm": 0.34857431054115295, | |
| "learning_rate": 0.0025, | |
| "loss": 2.395, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.48452635805293875, | |
| "eval_loss": 2.5702977180480957, | |
| "eval_runtime": 3.3619, | |
| "eval_samples_per_second": 1335.857, | |
| "eval_steps_per_second": 10.708, | |
| "step": 20496 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "grad_norm": 0.3316744863986969, | |
| "learning_rate": 0.0025625, | |
| "loss": 2.3917, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 7.17, | |
| "grad_norm": 0.45487892627716064, | |
| "learning_rate": 0.002625, | |
| "loss": 2.3159, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 7.34, | |
| "grad_norm": 0.35938137769699097, | |
| "learning_rate": 0.0026875000000000002, | |
| "loss": 2.3368, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 7.51, | |
| "grad_norm": 0.29650333523750305, | |
| "learning_rate": 0.00275, | |
| "loss": 2.3457, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 7.68, | |
| "grad_norm": 0.35303518176078796, | |
| "learning_rate": 0.0028125, | |
| "loss": 2.3632, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 7.86, | |
| "grad_norm": 0.3135327100753784, | |
| "learning_rate": 0.0028750000000000004, | |
| "loss": 2.363, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.48588866271475584, | |
| "eval_loss": 2.5576674938201904, | |
| "eval_runtime": 3.3548, | |
| "eval_samples_per_second": 1338.695, | |
| "eval_steps_per_second": 10.731, | |
| "step": 23424 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "grad_norm": 0.3409130275249481, | |
| "learning_rate": 0.0029375, | |
| "loss": 2.3464, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 8.2, | |
| "grad_norm": 0.30268123745918274, | |
| "learning_rate": 0.003, | |
| "loss": 2.2887, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 8.37, | |
| "grad_norm": 0.35700365900993347, | |
| "learning_rate": 0.002715909090909091, | |
| "loss": 2.3084, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 8.54, | |
| "grad_norm": 0.3313594162464142, | |
| "learning_rate": 0.0024318181818181817, | |
| "loss": 2.2992, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 8.71, | |
| "grad_norm": 0.3504340648651123, | |
| "learning_rate": 0.002147727272727273, | |
| "loss": 2.2959, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 8.88, | |
| "grad_norm": 0.3072393536567688, | |
| "learning_rate": 0.0018636363636363638, | |
| "loss": 2.2878, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.4939870993079773, | |
| "eval_loss": 2.509469509124756, | |
| "eval_runtime": 3.3506, | |
| "eval_samples_per_second": 1340.373, | |
| "eval_steps_per_second": 10.744, | |
| "step": 26352 | |
| }, | |
| { | |
| "epoch": 9.05, | |
| "grad_norm": 0.323819637298584, | |
| "learning_rate": 0.0015795454545454546, | |
| "loss": 2.2377, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 9.22, | |
| "grad_norm": 0.30831000208854675, | |
| "learning_rate": 0.0012954545454545456, | |
| "loss": 2.1654, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 9.39, | |
| "grad_norm": 0.33950263261795044, | |
| "learning_rate": 0.0010113636363636364, | |
| "loss": 2.1686, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 9.56, | |
| "grad_norm": 0.33820563554763794, | |
| "learning_rate": 0.0007272727272727273, | |
| "loss": 2.1555, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 9.73, | |
| "grad_norm": 0.32013264298439026, | |
| "learning_rate": 0.0004431818181818182, | |
| "loss": 2.143, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 9.9, | |
| "grad_norm": 0.3225158452987671, | |
| "learning_rate": 0.0001590909090909091, | |
| "loss": 2.1407, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.49941703178886204, | |
| "eval_loss": 2.4756977558135986, | |
| "eval_runtime": 3.3496, | |
| "eval_samples_per_second": 1340.777, | |
| "eval_steps_per_second": 10.748, | |
| "step": 29280 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "step": 29280, | |
| "total_flos": 2273237316403200.0, | |
| "train_loss": 2.5683612552496906, | |
| "train_runtime": 751.927, | |
| "train_samples_per_second": 622.999, | |
| "train_steps_per_second": 38.94 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 29280, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 10, | |
| "save_steps": 2000, | |
| "total_flos": 2273237316403200.0, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |