| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 35.0, | |
| "global_step": 4480, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 1.7767857142857143e-05, | |
| "loss": 3.4911, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 7.81, | |
| "learning_rate": 1.553571428571429e-05, | |
| "loss": 1.1908, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 11.72, | |
| "learning_rate": 1.3303571428571429e-05, | |
| "loss": 0.8102, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 15.62, | |
| "learning_rate": 1.1071428571428572e-05, | |
| "loss": 0.632, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 19.53, | |
| "learning_rate": 8.839285714285714e-06, | |
| "loss": 0.508, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 23.44, | |
| "learning_rate": 6.607142857142858e-06, | |
| "loss": 0.4371, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 27.34, | |
| "learning_rate": 4.3750000000000005e-06, | |
| "loss": 0.3877, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 31.25, | |
| "learning_rate": 2.1428571428571427e-06, | |
| "loss": 0.3677, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "step": 4480, | |
| "total_flos": 6656119413719040.0, | |
| "train_loss": 0.9091585227421352, | |
| "train_runtime": 611.6069, | |
| "train_samples_per_second": 29.3, | |
| "train_steps_per_second": 7.325 | |
| } | |
| ], | |
| "max_steps": 4480, | |
| "num_train_epochs": 35, | |
| "total_flos": 6656119413719040.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |