| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.679938744257274, | |
| "global_step": 3500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 5e-05, | |
| "loss": 1.4269, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 4.268578115857227e-05, | |
| "loss": 0.8747, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 3.537156231714453e-05, | |
| "loss": 0.7293, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 2.8057343475716798e-05, | |
| "loss": 0.6426, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 2.074312463428906e-05, | |
| "loss": 0.5825, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 1.3428905792861324e-05, | |
| "loss": 0.4312, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 6.1146869514335876e-06, | |
| "loss": 0.3789, | |
| "step": 3500 | |
| } | |
| ], | |
| "max_steps": 3918, | |
| "num_train_epochs": 3, | |
| "total_flos": 1309073265973200.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |