| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 93.61702127659575, | |
| "global_step": 4400, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 10.64, | |
| "learning_rate": 4.9444444444444446e-05, | |
| "loss": 1.6101, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 21.28, | |
| "learning_rate": 4.666666666666667e-05, | |
| "loss": 0.2288, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 31.91, | |
| "learning_rate": 4.388888888888889e-05, | |
| "loss": 0.1346, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 42.55, | |
| "learning_rate": 4.111111111111111e-05, | |
| "loss": 0.1033, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 53.19, | |
| "learning_rate": 3.8333333333333334e-05, | |
| "loss": 0.0754, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 63.83, | |
| "learning_rate": 3.555555555555556e-05, | |
| "loss": 0.0536, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 74.47, | |
| "learning_rate": 3.277777777777778e-05, | |
| "loss": 0.0433, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 85.11, | |
| "learning_rate": 3e-05, | |
| "loss": 0.0455, | |
| "step": 4000 | |
| } | |
| ], | |
| "max_steps": 9400, | |
| "num_train_epochs": 200, | |
| "total_flos": 5549594939136000.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |