| { | |
| "best_metric": 0.014639241620898247, | |
| "best_model_checkpoint": "./beans_outputs/checkpoint-520", | |
| "epoch": 5.0, | |
| "eval_steps": 500, | |
| "global_step": 650, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 18.207672119140625, | |
| "learning_rate": 1.9692307692307696e-05, | |
| "loss": 1.0224, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "grad_norm": 21.679859161376953, | |
| "learning_rate": 1.9384615384615386e-05, | |
| "loss": 0.7236, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "grad_norm": 16.34660530090332, | |
| "learning_rate": 1.907692307692308e-05, | |
| "loss": 0.3531, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "grad_norm": 11.42393970489502, | |
| "learning_rate": 1.876923076923077e-05, | |
| "loss": 0.2446, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "grad_norm": 11.492581367492676, | |
| "learning_rate": 1.8461538461538465e-05, | |
| "loss": 0.2049, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "grad_norm": 36.640625, | |
| "learning_rate": 1.8153846153846155e-05, | |
| "loss": 0.2194, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "grad_norm": 18.952184677124023, | |
| "learning_rate": 1.784615384615385e-05, | |
| "loss": 0.219, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "grad_norm": 0.392322301864624, | |
| "learning_rate": 1.753846153846154e-05, | |
| "loss": 0.0203, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "grad_norm": 0.2271302044391632, | |
| "learning_rate": 1.7230769230769234e-05, | |
| "loss": 0.1386, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "grad_norm": 10.620685577392578, | |
| "learning_rate": 1.6923076923076924e-05, | |
| "loss": 0.2381, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "grad_norm": 6.545742034912109, | |
| "learning_rate": 1.6615384615384618e-05, | |
| "loss": 0.2715, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "grad_norm": 13.884422302246094, | |
| "learning_rate": 1.630769230769231e-05, | |
| "loss": 0.2243, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 11.045103073120117, | |
| "learning_rate": 1.6000000000000003e-05, | |
| "loss": 0.1905, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.8872180451127819, | |
| "eval_loss": 0.36382579803466797, | |
| "eval_runtime": 10.3408, | |
| "eval_samples_per_second": 12.862, | |
| "eval_steps_per_second": 1.644, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "grad_norm": 5.467391014099121, | |
| "learning_rate": 1.5692307692307693e-05, | |
| "loss": 0.1068, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "grad_norm": 32.9152717590332, | |
| "learning_rate": 1.5384615384615387e-05, | |
| "loss": 0.2281, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "grad_norm": 0.03083522617816925, | |
| "learning_rate": 1.5076923076923078e-05, | |
| "loss": 0.1692, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "grad_norm": 3.7771084308624268, | |
| "learning_rate": 1.4769230769230772e-05, | |
| "loss": 0.1042, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "grad_norm": 0.02365809865295887, | |
| "learning_rate": 1.4461538461538462e-05, | |
| "loss": 0.0153, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "grad_norm": 33.88566589355469, | |
| "learning_rate": 1.4153846153846156e-05, | |
| "loss": 0.1443, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "grad_norm": 43.38811111450195, | |
| "learning_rate": 1.3846153846153847e-05, | |
| "loss": 0.1925, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "grad_norm": 9.77875804901123, | |
| "learning_rate": 1.353846153846154e-05, | |
| "loss": 0.1049, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "grad_norm": 0.05195164680480957, | |
| "learning_rate": 1.3230769230769231e-05, | |
| "loss": 0.3587, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "grad_norm": 22.248680114746094, | |
| "learning_rate": 1.2923076923076925e-05, | |
| "loss": 0.197, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "grad_norm": 34.09804153442383, | |
| "learning_rate": 1.2615384615384616e-05, | |
| "loss": 0.3231, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "grad_norm": 0.28886252641677856, | |
| "learning_rate": 1.230769230769231e-05, | |
| "loss": 0.1945, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 0.04396539926528931, | |
| "learning_rate": 1.2e-05, | |
| "loss": 0.1421, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.9473684210526315, | |
| "eval_loss": 0.14454808831214905, | |
| "eval_runtime": 10.0526, | |
| "eval_samples_per_second": 13.23, | |
| "eval_steps_per_second": 1.691, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "grad_norm": 3.8738396167755127, | |
| "learning_rate": 1.1692307692307694e-05, | |
| "loss": 0.0693, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "grad_norm": 0.053060416132211685, | |
| "learning_rate": 1.1384615384615385e-05, | |
| "loss": 0.1508, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "grad_norm": 1.897510290145874, | |
| "learning_rate": 1.1076923076923079e-05, | |
| "loss": 0.1713, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "grad_norm": 0.04172158241271973, | |
| "learning_rate": 1.076923076923077e-05, | |
| "loss": 0.0711, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "grad_norm": 32.68455123901367, | |
| "learning_rate": 1.0461538461538463e-05, | |
| "loss": 0.1683, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "grad_norm": 0.02692575938999653, | |
| "learning_rate": 1.0153846153846154e-05, | |
| "loss": 0.2328, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "grad_norm": 0.512816309928894, | |
| "learning_rate": 9.846153846153848e-06, | |
| "loss": 0.1052, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "grad_norm": 36.12916564941406, | |
| "learning_rate": 9.53846153846154e-06, | |
| "loss": 0.0788, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "grad_norm": 0.04400486871600151, | |
| "learning_rate": 9.230769230769232e-06, | |
| "loss": 0.0827, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "grad_norm": 37.217041015625, | |
| "learning_rate": 8.923076923076925e-06, | |
| "loss": 0.1229, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "grad_norm": 1.8499213457107544, | |
| "learning_rate": 8.615384615384617e-06, | |
| "loss": 0.2565, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "grad_norm": 0.16358694434165955, | |
| "learning_rate": 8.307692307692309e-06, | |
| "loss": 0.0087, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "grad_norm": 0.19832997024059296, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.0367, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.9699248120300752, | |
| "eval_loss": 0.09795637428760529, | |
| "eval_runtime": 9.8804, | |
| "eval_samples_per_second": 13.461, | |
| "eval_steps_per_second": 1.721, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "grad_norm": 71.58658599853516, | |
| "learning_rate": 7.692307692307694e-06, | |
| "loss": 0.2215, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "grad_norm": 0.8096259832382202, | |
| "learning_rate": 7.384615384615386e-06, | |
| "loss": 0.1097, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "grad_norm": 0.13009585440158844, | |
| "learning_rate": 7.076923076923078e-06, | |
| "loss": 0.0446, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "grad_norm": 19.282468795776367, | |
| "learning_rate": 6.76923076923077e-06, | |
| "loss": 0.1867, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "grad_norm": 0.07225558161735535, | |
| "learning_rate": 6.461538461538463e-06, | |
| "loss": 0.1384, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "grad_norm": 0.0648762583732605, | |
| "learning_rate": 6.153846153846155e-06, | |
| "loss": 0.0418, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "grad_norm": 0.6348621249198914, | |
| "learning_rate": 5.846153846153847e-06, | |
| "loss": 0.0934, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "grad_norm": 0.39203521609306335, | |
| "learning_rate": 5.538461538461539e-06, | |
| "loss": 0.0687, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "grad_norm": 54.287208557128906, | |
| "learning_rate": 5.230769230769232e-06, | |
| "loss": 0.3163, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "grad_norm": 9.522802352905273, | |
| "learning_rate": 4.923076923076924e-06, | |
| "loss": 0.1979, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "grad_norm": 0.052913323044776917, | |
| "learning_rate": 4.615384615384616e-06, | |
| "loss": 0.1689, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "grad_norm": 1.6295053958892822, | |
| "learning_rate": 4.307692307692308e-06, | |
| "loss": 0.0038, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "grad_norm": 0.0041633169166743755, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.0493, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.9924812030075187, | |
| "eval_loss": 0.014639241620898247, | |
| "eval_runtime": 9.7079, | |
| "eval_samples_per_second": 13.7, | |
| "eval_steps_per_second": 1.751, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "grad_norm": 0.008889425545930862, | |
| "learning_rate": 3.692307692307693e-06, | |
| "loss": 0.0523, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "grad_norm": 48.735984802246094, | |
| "learning_rate": 3.384615384615385e-06, | |
| "loss": 0.109, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "grad_norm": 0.9101144671440125, | |
| "learning_rate": 3.0769230769230774e-06, | |
| "loss": 0.0241, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 4.31, | |
| "grad_norm": 16.32082748413086, | |
| "learning_rate": 2.7692307692307697e-06, | |
| "loss": 0.0565, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "grad_norm": 0.12989751994609833, | |
| "learning_rate": 2.461538461538462e-06, | |
| "loss": 0.0043, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 4.46, | |
| "grad_norm": 0.11421233415603638, | |
| "learning_rate": 2.153846153846154e-06, | |
| "loss": 0.0482, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 4.54, | |
| "grad_norm": 55.456321716308594, | |
| "learning_rate": 1.8461538461538465e-06, | |
| "loss": 0.088, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 4.62, | |
| "grad_norm": 51.96505355834961, | |
| "learning_rate": 1.5384615384615387e-06, | |
| "loss": 0.122, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 4.69, | |
| "grad_norm": 29.100589752197266, | |
| "learning_rate": 1.230769230769231e-06, | |
| "loss": 0.0842, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 4.77, | |
| "grad_norm": 0.007961105555295944, | |
| "learning_rate": 9.230769230769232e-07, | |
| "loss": 0.1973, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 4.85, | |
| "grad_norm": 12.616386413574219, | |
| "learning_rate": 6.153846153846155e-07, | |
| "loss": 0.0081, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 4.92, | |
| "grad_norm": 0.21798954904079437, | |
| "learning_rate": 3.0769230769230774e-07, | |
| "loss": 0.0947, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "grad_norm": 0.038331467658281326, | |
| "learning_rate": 0.0, | |
| "loss": 0.0627, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.9774436090225563, | |
| "eval_loss": 0.03607732057571411, | |
| "eval_runtime": 8.348, | |
| "eval_samples_per_second": 15.932, | |
| "eval_steps_per_second": 2.036, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "step": 650, | |
| "total_flos": 4.0046594160900096e+17, | |
| "train_loss": 0.16150983500939148, | |
| "train_runtime": 1437.6272, | |
| "train_samples_per_second": 3.596, | |
| "train_steps_per_second": 0.452 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 650, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 5, | |
| "save_steps": 500, | |
| "total_flos": 4.0046594160900096e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |