| { | |
| "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": 1.1469449996948242, | |
| "learning_rate": 6.25e-05, | |
| "loss": 6.159, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 1.0157864093780518, | |
| "learning_rate": 0.000125, | |
| "loss": 3.6138, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 1.1782385110855103, | |
| "learning_rate": 0.0001875, | |
| "loss": 3.3436, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 1.0569684505462646, | |
| "learning_rate": 0.00025, | |
| "loss": 3.1723, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "grad_norm": 0.9641144275665283, | |
| "learning_rate": 0.0003125, | |
| "loss": 3.0397, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.4385586571217676, | |
| "eval_loss": 3.0116729736328125, | |
| "eval_runtime": 3.3975, | |
| "eval_samples_per_second": 1321.86, | |
| "eval_steps_per_second": 10.596, | |
| "step": 2928 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "grad_norm": 0.9262954592704773, | |
| "learning_rate": 0.000375, | |
| "loss": 2.9523, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "grad_norm": 0.8675119876861572, | |
| "learning_rate": 0.00043750000000000006, | |
| "loss": 2.86, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "grad_norm": 0.9038915038108826, | |
| "learning_rate": 0.0005, | |
| "loss": 2.8063, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "grad_norm": 0.8369466066360474, | |
| "learning_rate": 0.0005625000000000001, | |
| "loss": 2.7738, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "grad_norm": 0.7954362630844116, | |
| "learning_rate": 0.000625, | |
| "loss": 2.7437, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "grad_norm": 0.7949953675270081, | |
| "learning_rate": 0.0006875, | |
| "loss": 2.7146, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.45942979572443227, | |
| "eval_loss": 2.777545690536499, | |
| "eval_runtime": 3.4218, | |
| "eval_samples_per_second": 1312.463, | |
| "eval_steps_per_second": 10.521, | |
| "step": 5856 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "grad_norm": 0.8254726529121399, | |
| "learning_rate": 0.00075, | |
| "loss": 2.6703, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "grad_norm": 0.7527750730514526, | |
| "learning_rate": 0.0008125, | |
| "loss": 2.6175, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "grad_norm": 0.7464008331298828, | |
| "learning_rate": 0.0008750000000000001, | |
| "loss": 2.6076, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "grad_norm": 0.749565064907074, | |
| "learning_rate": 0.0009375, | |
| "loss": 2.6061, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "grad_norm": 0.6759158968925476, | |
| "learning_rate": 0.001, | |
| "loss": 2.5896, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "grad_norm": 0.6443790793418884, | |
| "learning_rate": 0.0010625, | |
| "loss": 2.5805, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.47120838352119815, | |
| "eval_loss": 2.6823782920837402, | |
| "eval_runtime": 3.3769, | |
| "eval_samples_per_second": 1329.919, | |
| "eval_steps_per_second": 10.661, | |
| "step": 8784 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "grad_norm": 0.599865734577179, | |
| "learning_rate": 0.0011250000000000001, | |
| "loss": 2.5413, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "grad_norm": 0.6362124085426331, | |
| "learning_rate": 0.0011875, | |
| "loss": 2.5078, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "grad_norm": 0.5414631962776184, | |
| "learning_rate": 0.00125, | |
| "loss": 2.5172, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "grad_norm": 0.5505661368370056, | |
| "learning_rate": 0.0013125, | |
| "loss": 2.5145, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "grad_norm": 0.5100904107093811, | |
| "learning_rate": 0.001375, | |
| "loss": 2.511, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "grad_norm": 0.5121994614601135, | |
| "learning_rate": 0.0014375000000000002, | |
| "loss": 2.5051, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.47740274950601114, | |
| "eval_loss": 2.6317946910858154, | |
| "eval_runtime": 3.4238, | |
| "eval_samples_per_second": 1311.689, | |
| "eval_steps_per_second": 10.515, | |
| "step": 11712 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "grad_norm": 0.44310715794563293, | |
| "learning_rate": 0.0015, | |
| "loss": 2.4617, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "grad_norm": 0.4380172789096832, | |
| "learning_rate": 0.0015625, | |
| "loss": 2.4573, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "grad_norm": 0.46541547775268555, | |
| "learning_rate": 0.001625, | |
| "loss": 2.4615, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "grad_norm": 0.4405967891216278, | |
| "learning_rate": 0.0016875, | |
| "loss": 2.4486, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "grad_norm": 0.3662794828414917, | |
| "learning_rate": 0.0017500000000000003, | |
| "loss": 2.4608, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "grad_norm": 0.4047889709472656, | |
| "learning_rate": 0.0018124999999999999, | |
| "loss": 2.4548, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.4802483356915055, | |
| "eval_loss": 2.6045305728912354, | |
| "eval_runtime": 3.4368, | |
| "eval_samples_per_second": 1306.728, | |
| "eval_steps_per_second": 10.475, | |
| "step": 14640 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "grad_norm": 0.39033982157707214, | |
| "learning_rate": 0.001875, | |
| "loss": 2.3964, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 5.29, | |
| "grad_norm": 0.4250138998031616, | |
| "learning_rate": 0.0019375000000000002, | |
| "loss": 2.4074, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 5.46, | |
| "grad_norm": 0.3768446445465088, | |
| "learning_rate": 0.002, | |
| "loss": 2.4077, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 5.64, | |
| "grad_norm": 0.42265185713768005, | |
| "learning_rate": 0.0020625, | |
| "loss": 2.4189, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 5.81, | |
| "grad_norm": 0.350845068693161, | |
| "learning_rate": 0.002125, | |
| "loss": 2.43, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 5.98, | |
| "grad_norm": 0.3043559789657593, | |
| "learning_rate": 0.0021874999999999998, | |
| "loss": 2.42, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.4842475852843044, | |
| "eval_loss": 2.577181577682495, | |
| "eval_runtime": 3.398, | |
| "eval_samples_per_second": 1321.679, | |
| "eval_steps_per_second": 10.595, | |
| "step": 17568 | |
| }, | |
| { | |
| "epoch": 6.15, | |
| "grad_norm": 0.33819225430488586, | |
| "learning_rate": 0.0022500000000000003, | |
| "loss": 2.3613, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 6.32, | |
| "grad_norm": 0.3696514368057251, | |
| "learning_rate": 0.0023125000000000003, | |
| "loss": 2.363, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 6.49, | |
| "grad_norm": 0.331849068403244, | |
| "learning_rate": 0.002375, | |
| "loss": 2.3793, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 6.66, | |
| "grad_norm": 0.31926459074020386, | |
| "learning_rate": 0.0024375, | |
| "loss": 2.3856, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 6.83, | |
| "grad_norm": 0.36476558446884155, | |
| "learning_rate": 0.0025, | |
| "loss": 2.3902, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.485290791556867, | |
| "eval_loss": 2.566494941711426, | |
| "eval_runtime": 3.4233, | |
| "eval_samples_per_second": 1311.904, | |
| "eval_steps_per_second": 10.516, | |
| "step": 20496 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "grad_norm": 0.28605133295059204, | |
| "learning_rate": 0.0025625, | |
| "loss": 2.3866, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 7.17, | |
| "grad_norm": 0.3234470784664154, | |
| "learning_rate": 0.002625, | |
| "loss": 2.3149, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 7.34, | |
| "grad_norm": 0.37032100558280945, | |
| "learning_rate": 0.0026875000000000002, | |
| "loss": 2.3354, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 7.51, | |
| "grad_norm": 0.30851054191589355, | |
| "learning_rate": 0.00275, | |
| "loss": 2.3576, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 7.68, | |
| "grad_norm": 0.34853699803352356, | |
| "learning_rate": 0.0028125, | |
| "loss": 2.362, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 7.86, | |
| "grad_norm": 0.3221457898616791, | |
| "learning_rate": 0.0028750000000000004, | |
| "loss": 2.357, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.48660751073450487, | |
| "eval_loss": 2.559926748275757, | |
| "eval_runtime": 3.4215, | |
| "eval_samples_per_second": 1312.576, | |
| "eval_steps_per_second": 10.522, | |
| "step": 23424 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "grad_norm": 0.3174827992916107, | |
| "learning_rate": 0.0029375, | |
| "loss": 2.3456, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 8.2, | |
| "grad_norm": 0.305833637714386, | |
| "learning_rate": 0.003, | |
| "loss": 2.2982, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 8.37, | |
| "grad_norm": 0.3276754915714264, | |
| "learning_rate": 0.002715909090909091, | |
| "loss": 2.3006, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 8.54, | |
| "grad_norm": 0.26479214429855347, | |
| "learning_rate": 0.0024318181818181817, | |
| "loss": 2.3056, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 8.71, | |
| "grad_norm": 0.36134952306747437, | |
| "learning_rate": 0.002147727272727273, | |
| "loss": 2.2974, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 8.88, | |
| "grad_norm": 0.3356594145298004, | |
| "learning_rate": 0.0018636363636363638, | |
| "loss": 2.2797, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.4931981197741064, | |
| "eval_loss": 2.5077412128448486, | |
| "eval_runtime": 3.3528, | |
| "eval_samples_per_second": 1339.467, | |
| "eval_steps_per_second": 10.737, | |
| "step": 26352 | |
| }, | |
| { | |
| "epoch": 9.05, | |
| "grad_norm": 0.2849366068840027, | |
| "learning_rate": 0.0015795454545454546, | |
| "loss": 2.2413, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 9.22, | |
| "grad_norm": 0.28835079073905945, | |
| "learning_rate": 0.0012954545454545456, | |
| "loss": 2.1688, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 9.39, | |
| "grad_norm": 0.3054426610469818, | |
| "learning_rate": 0.0010113636363636364, | |
| "loss": 2.1665, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 9.56, | |
| "grad_norm": 0.295037180185318, | |
| "learning_rate": 0.0007272727272727273, | |
| "loss": 2.1614, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 9.73, | |
| "grad_norm": 0.30976712703704834, | |
| "learning_rate": 0.0004431818181818182, | |
| "loss": 2.1518, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 9.9, | |
| "grad_norm": 0.2766718864440918, | |
| "learning_rate": 0.0001590909090909091, | |
| "loss": 2.1365, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.4998798997820663, | |
| "eval_loss": 2.475207805633545, | |
| "eval_runtime": 3.4298, | |
| "eval_samples_per_second": 1309.392, | |
| "eval_steps_per_second": 10.496, | |
| "step": 29280 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "step": 29280, | |
| "total_flos": 2273237316403200.0, | |
| "train_loss": 2.5651781655400177, | |
| "train_runtime": 742.1667, | |
| "train_samples_per_second": 631.192, | |
| "train_steps_per_second": 39.452 | |
| } | |
| ], | |
| "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 | |
| } | |