| { | |
| "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": 1.488160490989685, | |
| "learning_rate": 6.25e-05, | |
| "loss": 6.1623, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 0.9419446587562561, | |
| "learning_rate": 0.000125, | |
| "loss": 3.618, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 1.0650488138198853, | |
| "learning_rate": 0.0001875, | |
| "loss": 3.3416, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 1.0354753732681274, | |
| "learning_rate": 0.00025, | |
| "loss": 3.1604, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "grad_norm": 0.9344812631607056, | |
| "learning_rate": 0.0003125, | |
| "loss": 3.0469, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.43948088653247, | |
| "eval_loss": 3.010124444961548, | |
| "eval_runtime": 3.3747, | |
| "eval_samples_per_second": 1330.78, | |
| "eval_steps_per_second": 10.668, | |
| "step": 2928 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "grad_norm": 0.9709697365760803, | |
| "learning_rate": 0.000375, | |
| "loss": 2.9646, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "grad_norm": 0.8762878775596619, | |
| "learning_rate": 0.00043750000000000006, | |
| "loss": 2.8683, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "grad_norm": 0.8630770444869995, | |
| "learning_rate": 0.0005, | |
| "loss": 2.8084, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "grad_norm": 0.8325084447860718, | |
| "learning_rate": 0.0005625000000000001, | |
| "loss": 2.773, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "grad_norm": 0.7819183468818665, | |
| "learning_rate": 0.000625, | |
| "loss": 2.7315, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "grad_norm": 0.7667275667190552, | |
| "learning_rate": 0.0006875, | |
| "loss": 2.7142, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.46125496837945357, | |
| "eval_loss": 2.7790303230285645, | |
| "eval_runtime": 3.436, | |
| "eval_samples_per_second": 1307.031, | |
| "eval_steps_per_second": 10.477, | |
| "step": 5856 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "grad_norm": 0.7901221513748169, | |
| "learning_rate": 0.00075, | |
| "loss": 2.664, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "grad_norm": 0.7636890411376953, | |
| "learning_rate": 0.0008125, | |
| "loss": 2.6219, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "grad_norm": 0.722453236579895, | |
| "learning_rate": 0.0008750000000000001, | |
| "loss": 2.6068, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "grad_norm": 0.7329472303390503, | |
| "learning_rate": 0.0009375, | |
| "loss": 2.6052, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "grad_norm": 0.6657545566558838, | |
| "learning_rate": 0.001, | |
| "loss": 2.5877, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "grad_norm": 0.6002160906791687, | |
| "learning_rate": 0.0010625, | |
| "loss": 2.5841, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.4693832108661768, | |
| "eval_loss": 2.6902520656585693, | |
| "eval_runtime": 3.4276, | |
| "eval_samples_per_second": 1310.233, | |
| "eval_steps_per_second": 10.503, | |
| "step": 8784 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "grad_norm": 0.6292432546615601, | |
| "learning_rate": 0.0011250000000000001, | |
| "loss": 2.537, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "grad_norm": 0.5978102684020996, | |
| "learning_rate": 0.0011875, | |
| "loss": 2.5072, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "grad_norm": 0.5643737316131592, | |
| "learning_rate": 0.00125, | |
| "loss": 2.5165, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "grad_norm": 0.5356239080429077, | |
| "learning_rate": 0.0013125, | |
| "loss": 2.5138, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "grad_norm": 0.5515431761741638, | |
| "learning_rate": 0.001375, | |
| "loss": 2.5133, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "grad_norm": 0.47632288932800293, | |
| "learning_rate": 0.0014375000000000002, | |
| "loss": 2.5089, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.47639460899050945, | |
| "eval_loss": 2.63588547706604, | |
| "eval_runtime": 3.3985, | |
| "eval_samples_per_second": 1321.47, | |
| "eval_steps_per_second": 10.593, | |
| "step": 11712 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "grad_norm": 0.4891969561576843, | |
| "learning_rate": 0.0015, | |
| "loss": 2.4601, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "grad_norm": 0.4287746548652649, | |
| "learning_rate": 0.0015625, | |
| "loss": 2.4371, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "grad_norm": 0.43466711044311523, | |
| "learning_rate": 0.001625, | |
| "loss": 2.4523, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "grad_norm": 0.417784720659256, | |
| "learning_rate": 0.0016875, | |
| "loss": 2.4623, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "grad_norm": 0.4051859974861145, | |
| "learning_rate": 0.0017500000000000003, | |
| "loss": 2.4674, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "grad_norm": 0.40950262546539307, | |
| "learning_rate": 0.0018124999999999999, | |
| "loss": 2.4566, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.4807094503968567, | |
| "eval_loss": 2.602451801300049, | |
| "eval_runtime": 3.3747, | |
| "eval_samples_per_second": 1330.789, | |
| "eval_steps_per_second": 10.668, | |
| "step": 14640 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "grad_norm": 0.39824172854423523, | |
| "learning_rate": 0.001875, | |
| "loss": 2.4126, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 5.29, | |
| "grad_norm": 0.3892059326171875, | |
| "learning_rate": 0.0019375000000000002, | |
| "loss": 2.3928, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 5.46, | |
| "grad_norm": 0.3697751462459564, | |
| "learning_rate": 0.002, | |
| "loss": 2.4102, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 5.64, | |
| "grad_norm": 0.38204559683799744, | |
| "learning_rate": 0.0020625, | |
| "loss": 2.416, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 5.81, | |
| "grad_norm": 0.3960261642932892, | |
| "learning_rate": 0.002125, | |
| "loss": 2.4218, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 5.98, | |
| "grad_norm": 0.30173224210739136, | |
| "learning_rate": 0.0021874999999999998, | |
| "loss": 2.4168, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.4828274221233368, | |
| "eval_loss": 2.5854403972625732, | |
| "eval_runtime": 3.3792, | |
| "eval_samples_per_second": 1329.014, | |
| "eval_steps_per_second": 10.653, | |
| "step": 17568 | |
| }, | |
| { | |
| "epoch": 6.15, | |
| "grad_norm": 0.30894598364830017, | |
| "learning_rate": 0.0022500000000000003, | |
| "loss": 2.357, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 6.32, | |
| "grad_norm": 0.36697593331336975, | |
| "learning_rate": 0.0023125000000000003, | |
| "loss": 2.36, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 6.49, | |
| "grad_norm": 0.3668546974658966, | |
| "learning_rate": 0.002375, | |
| "loss": 2.3825, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 6.66, | |
| "grad_norm": 0.3038611114025116, | |
| "learning_rate": 0.0024375, | |
| "loss": 2.3875, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 6.83, | |
| "grad_norm": 0.290789395570755, | |
| "learning_rate": 0.0025, | |
| "loss": 2.3886, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.4851137094837093, | |
| "eval_loss": 2.5672802925109863, | |
| "eval_runtime": 3.3776, | |
| "eval_samples_per_second": 1329.659, | |
| "eval_steps_per_second": 10.659, | |
| "step": 20496 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "grad_norm": 0.2979215681552887, | |
| "learning_rate": 0.0025625, | |
| "loss": 2.3835, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 7.17, | |
| "grad_norm": 0.2849644422531128, | |
| "learning_rate": 0.002625, | |
| "loss": 2.3134, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 7.34, | |
| "grad_norm": 0.37557193636894226, | |
| "learning_rate": 0.0026875000000000002, | |
| "loss": 2.3387, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 7.51, | |
| "grad_norm": 0.3276946544647217, | |
| "learning_rate": 0.00275, | |
| "loss": 2.3492, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 7.68, | |
| "grad_norm": 0.30562373995780945, | |
| "learning_rate": 0.0028125, | |
| "loss": 2.3541, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 7.86, | |
| "grad_norm": 0.32304540276527405, | |
| "learning_rate": 0.0028750000000000004, | |
| "loss": 2.3618, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.4873631777991679, | |
| "eval_loss": 2.556323528289795, | |
| "eval_runtime": 3.3854, | |
| "eval_samples_per_second": 1326.589, | |
| "eval_steps_per_second": 10.634, | |
| "step": 23424 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "grad_norm": 0.30651578307151794, | |
| "learning_rate": 0.0029375, | |
| "loss": 2.3431, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 8.2, | |
| "grad_norm": 0.33351045846939087, | |
| "learning_rate": 0.003, | |
| "loss": 2.2909, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 8.37, | |
| "grad_norm": 0.36140263080596924, | |
| "learning_rate": 0.002715909090909091, | |
| "loss": 2.3174, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 8.54, | |
| "grad_norm": 0.32251298427581787, | |
| "learning_rate": 0.0024318181818181817, | |
| "loss": 2.2999, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 8.71, | |
| "grad_norm": 0.3153398931026459, | |
| "learning_rate": 0.002147727272727273, | |
| "loss": 2.2891, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 8.88, | |
| "grad_norm": 0.2726821005344391, | |
| "learning_rate": 0.0018636363636363638, | |
| "loss": 2.2757, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.493145521138515, | |
| "eval_loss": 2.502424955368042, | |
| "eval_runtime": 3.3675, | |
| "eval_samples_per_second": 1333.647, | |
| "eval_steps_per_second": 10.691, | |
| "step": 26352 | |
| }, | |
| { | |
| "epoch": 9.05, | |
| "grad_norm": 0.29384592175483704, | |
| "learning_rate": 0.0015795454545454546, | |
| "loss": 2.2495, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 9.22, | |
| "grad_norm": 0.343348890542984, | |
| "learning_rate": 0.0012954545454545456, | |
| "loss": 2.176, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 9.39, | |
| "grad_norm": 0.31881004571914673, | |
| "learning_rate": 0.0010113636363636364, | |
| "loss": 2.1687, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 9.56, | |
| "grad_norm": 0.3788747787475586, | |
| "learning_rate": 0.0007272727272727273, | |
| "loss": 2.1582, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 9.73, | |
| "grad_norm": 0.3564055860042572, | |
| "learning_rate": 0.0004431818181818182, | |
| "loss": 2.1505, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 9.9, | |
| "grad_norm": 0.28436481952667236, | |
| "learning_rate": 0.0001590909090909091, | |
| "loss": 2.139, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.5000622417187831, | |
| "eval_loss": 2.4731357097625732, | |
| "eval_runtime": 3.4247, | |
| "eval_samples_per_second": 1311.358, | |
| "eval_steps_per_second": 10.512, | |
| "step": 29280 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "step": 29280, | |
| "total_flos": 2273237316403200.0, | |
| "train_loss": 2.5646611927636984, | |
| "train_runtime": 771.972, | |
| "train_samples_per_second": 606.822, | |
| "train_steps_per_second": 37.929 | |
| } | |
| ], | |
| "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 | |
| } | |