| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.29239053629297534, | |
| "global_step": 1500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.980506822612086e-05, | |
| "loss": 8.7556, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.961013645224172e-05, | |
| "loss": 5.7897, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.941520467836257e-05, | |
| "loss": 5.3918, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.922027290448344e-05, | |
| "loss": 5.1445, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.902534113060428e-05, | |
| "loss": 4.9598, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.883040935672515e-05, | |
| "loss": 4.7922, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.8635477582846e-05, | |
| "loss": 4.7801, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.844054580896687e-05, | |
| "loss": 4.6361, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.824561403508771e-05, | |
| "loss": 4.5077, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.805068226120859e-05, | |
| "loss": 4.3726, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.785575048732943e-05, | |
| "loss": 4.3713, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.76608187134503e-05, | |
| "loss": 4.3171, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.746588693957115e-05, | |
| "loss": 4.1769, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.727095516569202e-05, | |
| "loss": 4.1894, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.707602339181286e-05, | |
| "loss": 4.1183, | |
| "step": 1500 | |
| } | |
| ], | |
| "max_steps": 51300, | |
| "num_train_epochs": 10, | |
| "total_flos": 6264825017401344.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |