{ | |
"best_metric": null, | |
"best_model_checkpoint": null, | |
"epoch": 0.9997238331952499, | |
"global_step": 905, | |
"is_hyper_param_search": false, | |
"is_local_process_zero": true, | |
"is_world_process_zero": true, | |
"log_history": [ | |
{ | |
"epoch": 0.01, | |
"learning_rate": 8.859191006777897e-06, | |
"loss": 14.7346, | |
"step": 10 | |
}, | |
{ | |
"epoch": 0.02, | |
"learning_rate": 1e-05, | |
"loss": 13.1229, | |
"step": 20 | |
}, | |
{ | |
"epoch": 0.03, | |
"learning_rate": 1e-05, | |
"loss": 12.3266, | |
"step": 30 | |
}, | |
{ | |
"epoch": 0.04, | |
"learning_rate": 1e-05, | |
"loss": 11.6279, | |
"step": 40 | |
}, | |
{ | |
"epoch": 0.06, | |
"learning_rate": 1e-05, | |
"loss": 11.9859, | |
"step": 50 | |
}, | |
{ | |
"epoch": 0.07, | |
"learning_rate": 1e-05, | |
"loss": 11.5059, | |
"step": 60 | |
}, | |
{ | |
"epoch": 0.08, | |
"learning_rate": 1e-05, | |
"loss": 11.6195, | |
"step": 70 | |
}, | |
{ | |
"epoch": 0.09, | |
"learning_rate": 1e-05, | |
"loss": 11.2809, | |
"step": 80 | |
}, | |
{ | |
"epoch": 0.1, | |
"learning_rate": 1e-05, | |
"loss": 11.1093, | |
"step": 90 | |
}, | |
{ | |
"epoch": 0.11, | |
"learning_rate": 1e-05, | |
"loss": 11.2859, | |
"step": 100 | |
}, | |
{ | |
"epoch": 0.12, | |
"learning_rate": 1e-05, | |
"loss": 11.1785, | |
"step": 110 | |
}, | |
{ | |
"epoch": 0.13, | |
"learning_rate": 1e-05, | |
"loss": 11.4462, | |
"step": 120 | |
}, | |
{ | |
"epoch": 0.14, | |
"learning_rate": 1e-05, | |
"loss": 10.7203, | |
"step": 130 | |
}, | |
{ | |
"epoch": 0.15, | |
"learning_rate": 1e-05, | |
"loss": 11.1789, | |
"step": 140 | |
}, | |
{ | |
"epoch": 0.17, | |
"learning_rate": 1e-05, | |
"loss": 11.0246, | |
"step": 150 | |
}, | |
{ | |
"epoch": 0.18, | |
"learning_rate": 1e-05, | |
"loss": 10.7604, | |
"step": 160 | |
}, | |
{ | |
"epoch": 0.19, | |
"learning_rate": 1e-05, | |
"loss": 11.0006, | |
"step": 170 | |
}, | |
{ | |
"epoch": 0.2, | |
"learning_rate": 1e-05, | |
"loss": 11.1139, | |
"step": 180 | |
}, | |
{ | |
"epoch": 0.21, | |
"learning_rate": 1e-05, | |
"loss": 10.8658, | |
"step": 190 | |
}, | |
{ | |
"epoch": 0.22, | |
"learning_rate": 1e-05, | |
"loss": 11.1844, | |
"step": 200 | |
}, | |
{ | |
"epoch": 0.23, | |
"learning_rate": 1e-05, | |
"loss": 10.9141, | |
"step": 210 | |
}, | |
{ | |
"epoch": 0.24, | |
"learning_rate": 1e-05, | |
"loss": 10.9406, | |
"step": 220 | |
}, | |
{ | |
"epoch": 0.25, | |
"learning_rate": 1e-05, | |
"loss": 10.9014, | |
"step": 230 | |
}, | |
{ | |
"epoch": 0.27, | |
"learning_rate": 1e-05, | |
"loss": 11.0768, | |
"step": 240 | |
}, | |
{ | |
"epoch": 0.28, | |
"learning_rate": 1e-05, | |
"loss": 10.8308, | |
"step": 250 | |
}, | |
{ | |
"epoch": 0.29, | |
"learning_rate": 1e-05, | |
"loss": 11.0662, | |
"step": 260 | |
}, | |
{ | |
"epoch": 0.3, | |
"learning_rate": 1e-05, | |
"loss": 10.7037, | |
"step": 270 | |
}, | |
{ | |
"epoch": 0.31, | |
"learning_rate": 1e-05, | |
"loss": 10.9682, | |
"step": 280 | |
}, | |
{ | |
"epoch": 0.32, | |
"learning_rate": 1e-05, | |
"loss": 10.9742, | |
"step": 290 | |
}, | |
{ | |
"epoch": 0.33, | |
"learning_rate": 1e-05, | |
"loss": 10.9057, | |
"step": 300 | |
}, | |
{ | |
"epoch": 0.34, | |
"learning_rate": 1e-05, | |
"loss": 10.7487, | |
"step": 310 | |
}, | |
{ | |
"epoch": 0.35, | |
"learning_rate": 1e-05, | |
"loss": 10.7682, | |
"step": 320 | |
}, | |
{ | |
"epoch": 0.36, | |
"learning_rate": 1e-05, | |
"loss": 11.0381, | |
"step": 330 | |
}, | |
{ | |
"epoch": 0.38, | |
"learning_rate": 1e-05, | |
"loss": 10.8182, | |
"step": 340 | |
}, | |
{ | |
"epoch": 0.39, | |
"learning_rate": 1e-05, | |
"loss": 10.6824, | |
"step": 350 | |
}, | |
{ | |
"epoch": 0.4, | |
"learning_rate": 1e-05, | |
"loss": 10.9189, | |
"step": 360 | |
}, | |
{ | |
"epoch": 0.41, | |
"learning_rate": 1e-05, | |
"loss": 10.9768, | |
"step": 370 | |
}, | |
{ | |
"epoch": 0.42, | |
"learning_rate": 1e-05, | |
"loss": 10.6872, | |
"step": 380 | |
}, | |
{ | |
"epoch": 0.43, | |
"learning_rate": 1e-05, | |
"loss": 10.7874, | |
"step": 390 | |
}, | |
{ | |
"epoch": 0.44, | |
"learning_rate": 1e-05, | |
"loss": 10.9916, | |
"step": 400 | |
}, | |
{ | |
"epoch": 0.45, | |
"learning_rate": 1e-05, | |
"loss": 10.8531, | |
"step": 410 | |
}, | |
{ | |
"epoch": 0.46, | |
"learning_rate": 1e-05, | |
"loss": 11.0098, | |
"step": 420 | |
}, | |
{ | |
"epoch": 0.48, | |
"learning_rate": 1e-05, | |
"loss": 10.7342, | |
"step": 430 | |
}, | |
{ | |
"epoch": 0.49, | |
"learning_rate": 1e-05, | |
"loss": 10.7775, | |
"step": 440 | |
}, | |
{ | |
"epoch": 0.5, | |
"learning_rate": 1e-05, | |
"loss": 10.7689, | |
"step": 450 | |
}, | |
{ | |
"epoch": 0.51, | |
"learning_rate": 1e-05, | |
"loss": 11.0508, | |
"step": 460 | |
}, | |
{ | |
"epoch": 0.52, | |
"learning_rate": 1e-05, | |
"loss": 10.739, | |
"step": 470 | |
}, | |
{ | |
"epoch": 0.53, | |
"learning_rate": 1e-05, | |
"loss": 10.715, | |
"step": 480 | |
}, | |
{ | |
"epoch": 0.54, | |
"learning_rate": 1e-05, | |
"loss": 10.5805, | |
"step": 490 | |
}, | |
{ | |
"epoch": 0.55, | |
"learning_rate": 1e-05, | |
"loss": 10.7307, | |
"step": 500 | |
}, | |
{ | |
"epoch": 0.56, | |
"learning_rate": 1e-05, | |
"loss": 10.8336, | |
"step": 510 | |
}, | |
{ | |
"epoch": 0.57, | |
"learning_rate": 1e-05, | |
"loss": 11.1557, | |
"step": 520 | |
}, | |
{ | |
"epoch": 0.59, | |
"learning_rate": 1e-05, | |
"loss": 10.7799, | |
"step": 530 | |
}, | |
{ | |
"epoch": 0.6, | |
"learning_rate": 1e-05, | |
"loss": 10.7711, | |
"step": 540 | |
}, | |
{ | |
"epoch": 0.61, | |
"learning_rate": 1e-05, | |
"loss": 11.2973, | |
"step": 550 | |
}, | |
{ | |
"epoch": 0.62, | |
"learning_rate": 1e-05, | |
"loss": 11.0799, | |
"step": 560 | |
}, | |
{ | |
"epoch": 0.63, | |
"learning_rate": 1e-05, | |
"loss": 10.472, | |
"step": 570 | |
}, | |
{ | |
"epoch": 0.64, | |
"learning_rate": 1e-05, | |
"loss": 10.7779, | |
"step": 580 | |
}, | |
{ | |
"epoch": 0.65, | |
"learning_rate": 1e-05, | |
"loss": 10.8314, | |
"step": 590 | |
}, | |
{ | |
"epoch": 0.66, | |
"learning_rate": 1e-05, | |
"loss": 10.8986, | |
"step": 600 | |
}, | |
{ | |
"epoch": 0.67, | |
"learning_rate": 1e-05, | |
"loss": 10.8828, | |
"step": 610 | |
}, | |
{ | |
"epoch": 0.68, | |
"learning_rate": 1e-05, | |
"loss": 11.0666, | |
"step": 620 | |
}, | |
{ | |
"epoch": 0.7, | |
"learning_rate": 1e-05, | |
"loss": 11.0117, | |
"step": 630 | |
}, | |
{ | |
"epoch": 0.71, | |
"learning_rate": 1e-05, | |
"loss": 10.828, | |
"step": 640 | |
}, | |
{ | |
"epoch": 0.72, | |
"learning_rate": 1e-05, | |
"loss": 10.8252, | |
"step": 650 | |
}, | |
{ | |
"epoch": 0.73, | |
"learning_rate": 1e-05, | |
"loss": 10.4189, | |
"step": 660 | |
}, | |
{ | |
"epoch": 0.74, | |
"learning_rate": 1e-05, | |
"loss": 10.6005, | |
"step": 670 | |
}, | |
{ | |
"epoch": 0.75, | |
"learning_rate": 1e-05, | |
"loss": 10.7747, | |
"step": 680 | |
}, | |
{ | |
"epoch": 0.76, | |
"learning_rate": 1e-05, | |
"loss": 10.6799, | |
"step": 690 | |
}, | |
{ | |
"epoch": 0.77, | |
"learning_rate": 1e-05, | |
"loss": 10.7008, | |
"step": 700 | |
}, | |
{ | |
"epoch": 0.78, | |
"learning_rate": 1e-05, | |
"loss": 10.6438, | |
"step": 710 | |
}, | |
{ | |
"epoch": 0.8, | |
"learning_rate": 1e-05, | |
"loss": 10.6026, | |
"step": 720 | |
}, | |
{ | |
"epoch": 0.81, | |
"learning_rate": 1e-05, | |
"loss": 10.8184, | |
"step": 730 | |
}, | |
{ | |
"epoch": 0.82, | |
"learning_rate": 1e-05, | |
"loss": 10.3796, | |
"step": 740 | |
}, | |
{ | |
"epoch": 0.83, | |
"learning_rate": 1e-05, | |
"loss": 10.3816, | |
"step": 750 | |
}, | |
{ | |
"epoch": 0.84, | |
"learning_rate": 1e-05, | |
"loss": 10.4938, | |
"step": 760 | |
}, | |
{ | |
"epoch": 0.85, | |
"learning_rate": 1e-05, | |
"loss": 10.4465, | |
"step": 770 | |
}, | |
{ | |
"epoch": 0.86, | |
"learning_rate": 1e-05, | |
"loss": 10.5486, | |
"step": 780 | |
}, | |
{ | |
"epoch": 0.87, | |
"learning_rate": 1e-05, | |
"loss": 10.2929, | |
"step": 790 | |
}, | |
{ | |
"epoch": 0.88, | |
"learning_rate": 1e-05, | |
"loss": 10.7909, | |
"step": 800 | |
}, | |
{ | |
"epoch": 0.89, | |
"learning_rate": 1e-05, | |
"loss": 10.7009, | |
"step": 810 | |
}, | |
{ | |
"epoch": 0.91, | |
"learning_rate": 1e-05, | |
"loss": 10.5376, | |
"step": 820 | |
}, | |
{ | |
"epoch": 0.92, | |
"learning_rate": 1e-05, | |
"loss": 10.6126, | |
"step": 830 | |
}, | |
{ | |
"epoch": 0.93, | |
"learning_rate": 1e-05, | |
"loss": 10.6541, | |
"step": 840 | |
}, | |
{ | |
"epoch": 0.94, | |
"learning_rate": 1e-05, | |
"loss": 10.8826, | |
"step": 850 | |
}, | |
{ | |
"epoch": 0.95, | |
"learning_rate": 1e-05, | |
"loss": 10.3846, | |
"step": 860 | |
}, | |
{ | |
"epoch": 0.96, | |
"learning_rate": 1e-05, | |
"loss": 10.5594, | |
"step": 870 | |
}, | |
{ | |
"epoch": 0.97, | |
"learning_rate": 1e-05, | |
"loss": 10.4418, | |
"step": 880 | |
}, | |
{ | |
"epoch": 0.98, | |
"learning_rate": 1e-05, | |
"loss": 10.6198, | |
"step": 890 | |
}, | |
{ | |
"epoch": 0.99, | |
"learning_rate": 1e-05, | |
"loss": 10.7552, | |
"step": 900 | |
} | |
], | |
"max_steps": 905, | |
"num_train_epochs": 1, | |
"total_flos": 0.0, | |
"trial_name": null, | |
"trial_params": null | |
} | |