| { | |
| "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": 1.3676248788833618, | |
| "learning_rate": 6.25e-05, | |
| "loss": 6.1929, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 0.8271423578262329, | |
| "learning_rate": 0.000125, | |
| "loss": 3.6201, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 1.0618003606796265, | |
| "learning_rate": 0.0001875, | |
| "loss": 3.3367, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 1.0157318115234375, | |
| "learning_rate": 0.00025, | |
| "loss": 3.1626, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "grad_norm": 0.9570010304450989, | |
| "learning_rate": 0.0003125, | |
| "loss": 3.0383, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.4371174545065634, | |
| "eval_loss": 3.0183205604553223, | |
| "eval_runtime": 3.2884, | |
| "eval_samples_per_second": 1365.698, | |
| "eval_steps_per_second": 10.947, | |
| "step": 2928 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "grad_norm": 0.9757611751556396, | |
| "learning_rate": 0.000375, | |
| "loss": 2.9504, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "grad_norm": 0.9592107534408569, | |
| "learning_rate": 0.00043750000000000006, | |
| "loss": 2.8615, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "grad_norm": 0.9070651531219482, | |
| "learning_rate": 0.0005, | |
| "loss": 2.8223, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "grad_norm": 0.8300113677978516, | |
| "learning_rate": 0.0005625000000000001, | |
| "loss": 2.7679, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "grad_norm": 0.8314893245697021, | |
| "learning_rate": 0.000625, | |
| "loss": 2.7455, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "grad_norm": 0.8281814455986023, | |
| "learning_rate": 0.0006875, | |
| "loss": 2.7202, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.4601416305927691, | |
| "eval_loss": 2.784088611602783, | |
| "eval_runtime": 3.3962, | |
| "eval_samples_per_second": 1322.357, | |
| "eval_steps_per_second": 10.6, | |
| "step": 5856 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "grad_norm": 0.796658456325531, | |
| "learning_rate": 0.00075, | |
| "loss": 2.6877, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "grad_norm": 0.8038381934165955, | |
| "learning_rate": 0.0008125, | |
| "loss": 2.6236, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "grad_norm": 0.7780303359031677, | |
| "learning_rate": 0.0008750000000000001, | |
| "loss": 2.6091, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "grad_norm": 0.7377750277519226, | |
| "learning_rate": 0.0009375, | |
| "loss": 2.6178, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "grad_norm": 0.6352299451828003, | |
| "learning_rate": 0.001, | |
| "loss": 2.5939, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "grad_norm": 0.664020836353302, | |
| "learning_rate": 0.0010625, | |
| "loss": 2.5954, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.46891332972156036, | |
| "eval_loss": 2.701345443725586, | |
| "eval_runtime": 3.4037, | |
| "eval_samples_per_second": 1319.439, | |
| "eval_steps_per_second": 10.577, | |
| "step": 8784 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "grad_norm": 0.5852430462837219, | |
| "learning_rate": 0.0011250000000000001, | |
| "loss": 2.5482, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "grad_norm": 0.5592005252838135, | |
| "learning_rate": 0.0011875, | |
| "loss": 2.5147, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "grad_norm": 0.5826843976974487, | |
| "learning_rate": 0.00125, | |
| "loss": 2.5198, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "grad_norm": 0.560430645942688, | |
| "learning_rate": 0.0013125, | |
| "loss": 2.5193, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "grad_norm": 0.5494054555892944, | |
| "learning_rate": 0.001375, | |
| "loss": 2.5194, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "grad_norm": 0.49182966351509094, | |
| "learning_rate": 0.0014375000000000002, | |
| "loss": 2.5123, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.4754934190340436, | |
| "eval_loss": 2.6403415203094482, | |
| "eval_runtime": 3.3923, | |
| "eval_samples_per_second": 1323.894, | |
| "eval_steps_per_second": 10.612, | |
| "step": 11712 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "grad_norm": 0.5087066292762756, | |
| "learning_rate": 0.0015, | |
| "loss": 2.4619, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "grad_norm": 0.4928716719150543, | |
| "learning_rate": 0.0015625, | |
| "loss": 2.4478, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "grad_norm": 0.41928255558013916, | |
| "learning_rate": 0.001625, | |
| "loss": 2.4609, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "grad_norm": 0.4840702414512634, | |
| "learning_rate": 0.0016875, | |
| "loss": 2.4698, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "grad_norm": 0.397327721118927, | |
| "learning_rate": 0.0017500000000000003, | |
| "loss": 2.4584, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "grad_norm": 0.37015002965927124, | |
| "learning_rate": 0.0018124999999999999, | |
| "loss": 2.4613, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.48022028308585674, | |
| "eval_loss": 2.604732036590576, | |
| "eval_runtime": 3.3941, | |
| "eval_samples_per_second": 1323.178, | |
| "eval_steps_per_second": 10.607, | |
| "step": 14640 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "grad_norm": 0.3754883408546448, | |
| "learning_rate": 0.001875, | |
| "loss": 2.411, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 5.29, | |
| "grad_norm": 0.40850552916526794, | |
| "learning_rate": 0.0019375000000000002, | |
| "loss": 2.4024, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 5.46, | |
| "grad_norm": 0.3658270537853241, | |
| "learning_rate": 0.002, | |
| "loss": 2.4077, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 5.64, | |
| "grad_norm": 0.4083290696144104, | |
| "learning_rate": 0.0020625, | |
| "loss": 2.4324, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 5.81, | |
| "grad_norm": 0.388753205537796, | |
| "learning_rate": 0.002125, | |
| "loss": 2.425, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 5.98, | |
| "grad_norm": 0.30247732996940613, | |
| "learning_rate": 0.0021874999999999998, | |
| "loss": 2.4261, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.48243293235640133, | |
| "eval_loss": 2.5919439792633057, | |
| "eval_runtime": 3.3927, | |
| "eval_samples_per_second": 1323.74, | |
| "eval_steps_per_second": 10.611, | |
| "step": 17568 | |
| }, | |
| { | |
| "epoch": 6.15, | |
| "grad_norm": 0.34406954050064087, | |
| "learning_rate": 0.0022500000000000003, | |
| "loss": 2.3551, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 6.32, | |
| "grad_norm": 0.3480839729309082, | |
| "learning_rate": 0.0023125000000000003, | |
| "loss": 2.3652, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 6.49, | |
| "grad_norm": 0.3208664059638977, | |
| "learning_rate": 0.002375, | |
| "loss": 2.3746, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 6.66, | |
| "grad_norm": 0.30590417981147766, | |
| "learning_rate": 0.0024375, | |
| "loss": 2.3918, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 6.83, | |
| "grad_norm": 0.33193260431289673, | |
| "learning_rate": 0.0025, | |
| "loss": 2.3906, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.48501201878823264, | |
| "eval_loss": 2.5665977001190186, | |
| "eval_runtime": 3.4053, | |
| "eval_samples_per_second": 1318.819, | |
| "eval_steps_per_second": 10.572, | |
| "step": 20496 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "grad_norm": 0.31459102034568787, | |
| "learning_rate": 0.0025625, | |
| "loss": 2.3979, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 7.17, | |
| "grad_norm": 0.3610854148864746, | |
| "learning_rate": 0.002625, | |
| "loss": 2.3098, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 7.34, | |
| "grad_norm": 0.3405974209308624, | |
| "learning_rate": 0.0026875000000000002, | |
| "loss": 2.3376, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 7.51, | |
| "grad_norm": 0.32892337441444397, | |
| "learning_rate": 0.00275, | |
| "loss": 2.3508, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 7.68, | |
| "grad_norm": 0.33222299814224243, | |
| "learning_rate": 0.0028125, | |
| "loss": 2.3573, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 7.86, | |
| "grad_norm": 0.31080856919288635, | |
| "learning_rate": 0.0028750000000000004, | |
| "loss": 2.3601, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.48613061643847627, | |
| "eval_loss": 2.5550413131713867, | |
| "eval_runtime": 3.4064, | |
| "eval_samples_per_second": 1318.419, | |
| "eval_steps_per_second": 10.568, | |
| "step": 23424 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "grad_norm": 0.35419604182243347, | |
| "learning_rate": 0.0029375, | |
| "loss": 2.3599, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 8.2, | |
| "grad_norm": 0.3225402235984802, | |
| "learning_rate": 0.003, | |
| "loss": 2.2923, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 8.37, | |
| "grad_norm": 0.3462129235267639, | |
| "learning_rate": 0.002715909090909091, | |
| "loss": 2.3112, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 8.54, | |
| "grad_norm": 0.3057979941368103, | |
| "learning_rate": 0.0024318181818181817, | |
| "loss": 2.3082, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 8.71, | |
| "grad_norm": 0.3206704258918762, | |
| "learning_rate": 0.002147727272727273, | |
| "loss": 2.2996, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 8.88, | |
| "grad_norm": 0.2941910922527313, | |
| "learning_rate": 0.0018636363636363638, | |
| "loss": 2.2843, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.4939643065658877, | |
| "eval_loss": 2.5031280517578125, | |
| "eval_runtime": 3.3967, | |
| "eval_samples_per_second": 1322.152, | |
| "eval_steps_per_second": 10.598, | |
| "step": 26352 | |
| }, | |
| { | |
| "epoch": 9.05, | |
| "grad_norm": 0.3082716464996338, | |
| "learning_rate": 0.0015795454545454546, | |
| "loss": 2.2365, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 9.22, | |
| "grad_norm": 0.29701223969459534, | |
| "learning_rate": 0.0012954545454545456, | |
| "loss": 2.1701, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 9.39, | |
| "grad_norm": 0.3021478056907654, | |
| "learning_rate": 0.0010113636363636364, | |
| "loss": 2.1685, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 9.56, | |
| "grad_norm": 0.3144983649253845, | |
| "learning_rate": 0.0007272727272727273, | |
| "loss": 2.1654, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 9.73, | |
| "grad_norm": 0.32024458050727844, | |
| "learning_rate": 0.0004431818181818182, | |
| "loss": 2.1508, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 9.9, | |
| "grad_norm": 0.27385374903678894, | |
| "learning_rate": 0.0001590909090909091, | |
| "loss": 2.1392, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.499075140657518, | |
| "eval_loss": 2.47538685798645, | |
| "eval_runtime": 3.3927, | |
| "eval_samples_per_second": 1323.721, | |
| "eval_steps_per_second": 10.611, | |
| "step": 29280 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "step": 29280, | |
| "total_flos": 2273237316403200.0, | |
| "train_loss": 2.568481111787056, | |
| "train_runtime": 765.1195, | |
| "train_samples_per_second": 612.257, | |
| "train_steps_per_second": 38.269 | |
| } | |
| ], | |
| "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 | |
| } | |