| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9994739610731194, | |
| "global_step": 950, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02, | |
| "eval_loss": 2.40625, | |
| "eval_runtime": 1.9288, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_loss": 2.35546875, | |
| "eval_runtime": 1.9286, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.519, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "eval_loss": 2.341796875, | |
| "eval_runtime": 1.9286, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.519, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "eval_loss": 2.33203125, | |
| "eval_runtime": 1.9285, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.519, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "eval_loss": 2.32421875, | |
| "eval_runtime": 1.9288, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "eval_loss": 2.318359375, | |
| "eval_runtime": 1.9291, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "eval_loss": 2.314453125, | |
| "eval_runtime": 1.9288, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "eval_loss": 2.3125, | |
| "eval_runtime": 1.9287, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "eval_loss": 2.3125, | |
| "eval_runtime": 1.9289, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "eval_loss": 2.306640625, | |
| "eval_runtime": 1.9287, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "eval_loss": 2.30078125, | |
| "eval_runtime": 1.9284, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.519, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "eval_loss": 2.30078125, | |
| "eval_runtime": 1.9291, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "eval_loss": 2.302734375, | |
| "eval_runtime": 1.9292, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "eval_loss": 2.294921875, | |
| "eval_runtime": 1.9287, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "eval_loss": 2.287109375, | |
| "eval_runtime": 1.9291, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "eval_loss": 2.283203125, | |
| "eval_runtime": 1.9283, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.519, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "eval_loss": 2.279296875, | |
| "eval_runtime": 1.9288, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "eval_loss": 2.27734375, | |
| "eval_runtime": 1.929, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "eval_loss": 2.248046875, | |
| "eval_runtime": 1.9292, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "eval_loss": 2.2265625, | |
| "eval_runtime": 1.9288, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "eval_loss": 2.2265625, | |
| "eval_runtime": 1.9288, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "eval_loss": 2.2265625, | |
| "eval_runtime": 1.9287, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "eval_loss": 2.2265625, | |
| "eval_runtime": 1.9316, | |
| "eval_samples_per_second": 1.035, | |
| "eval_steps_per_second": 0.518, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "eval_loss": 2.228515625, | |
| "eval_runtime": 1.9288, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "eval_loss": 2.2265625, | |
| "eval_runtime": 1.9286, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.519, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "eval_loss": 2.224609375, | |
| "eval_runtime": 1.9289, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "eval_loss": 2.22265625, | |
| "eval_runtime": 1.9294, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "eval_loss": 2.220703125, | |
| "eval_runtime": 1.9284, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.519, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "eval_loss": 2.22265625, | |
| "eval_runtime": 1.9289, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "eval_loss": 2.22265625, | |
| "eval_runtime": 1.9295, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "eval_loss": 2.220703125, | |
| "eval_runtime": 1.9284, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.519, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "eval_loss": 2.220703125, | |
| "eval_runtime": 1.9291, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "eval_loss": 2.21875, | |
| "eval_runtime": 1.9289, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 5e-06, | |
| "loss": 2.374, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "eval_loss": 2.203125, | |
| "eval_runtime": 1.937, | |
| "eval_samples_per_second": 1.033, | |
| "eval_steps_per_second": 0.516, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "eval_loss": 2.1875, | |
| "eval_runtime": 1.9291, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "eval_loss": 2.185546875, | |
| "eval_runtime": 1.9291, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "eval_loss": 2.140625, | |
| "eval_runtime": 1.9294, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "eval_loss": 2.134765625, | |
| "eval_runtime": 1.9289, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "eval_loss": 2.134765625, | |
| "eval_runtime": 1.9285, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.519, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "eval_loss": 2.134765625, | |
| "eval_runtime": 1.929, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "eval_loss": 2.1328125, | |
| "eval_runtime": 1.929, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "eval_loss": 2.1328125, | |
| "eval_runtime": 1.9284, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.519, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "eval_loss": 2.130859375, | |
| "eval_runtime": 1.929, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "eval_loss": 2.12890625, | |
| "eval_runtime": 1.9293, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "eval_loss": 2.12890625, | |
| "eval_runtime": 1.9283, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.519, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "eval_loss": 2.1328125, | |
| "eval_runtime": 1.9291, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "eval_loss": 2.1328125, | |
| "eval_runtime": 1.9288, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "eval_loss": 2.130859375, | |
| "eval_runtime": 1.9294, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "eval_loss": 2.134765625, | |
| "eval_runtime": 1.929, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "eval_loss": 2.1328125, | |
| "eval_runtime": 1.9295, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "eval_loss": 2.134765625, | |
| "eval_runtime": 1.9289, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "eval_loss": 2.134765625, | |
| "eval_runtime": 1.9291, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "eval_loss": 2.134765625, | |
| "eval_runtime": 1.9289, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "eval_loss": 2.1328125, | |
| "eval_runtime": 1.9291, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "eval_loss": 2.12890625, | |
| "eval_runtime": 1.9286, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.519, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "eval_loss": 2.119140625, | |
| "eval_runtime": 1.9289, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "eval_loss": 2.12109375, | |
| "eval_runtime": 1.9287, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "eval_loss": 2.1171875, | |
| "eval_runtime": 1.9289, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "eval_loss": 2.119140625, | |
| "eval_runtime": 1.9291, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "eval_loss": 2.123046875, | |
| "eval_runtime": 1.9294, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "eval_loss": 2.12109375, | |
| "eval_runtime": 1.9292, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "eval_loss": 2.119140625, | |
| "eval_runtime": 1.9287, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "eval_loss": 2.12109375, | |
| "eval_runtime": 1.9291, | |
| "eval_samples_per_second": 1.037, | |
| "eval_steps_per_second": 0.518, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 950, | |
| "total_flos": 1.2380121253648794e+17, | |
| "train_loss": 2.3266365131578945, | |
| "train_runtime": 29836.7817, | |
| "train_samples_per_second": 0.255, | |
| "train_steps_per_second": 0.032 | |
| } | |
| ], | |
| "max_steps": 950, | |
| "num_train_epochs": 1, | |
| "total_flos": 1.2380121253648794e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |