| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.0, | |
| "global_step": 5952, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.831989247311828e-05, | |
| "loss": 0.6085, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.663978494623656e-05, | |
| "loss": 0.6656, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.4959677419354839e-05, | |
| "loss": 0.6649, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.3279569892473118e-05, | |
| "loss": 0.6354, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.1599462365591398e-05, | |
| "loss": 0.6208, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.919354838709679e-06, | |
| "loss": 0.6138, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 8.239247311827959e-06, | |
| "loss": 0.5066, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 6.5591397849462365e-06, | |
| "loss": 0.4947, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 4.879032258064517e-06, | |
| "loss": 0.4856, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 3.198924731182796e-06, | |
| "loss": 0.4776, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 1.5188172043010754e-06, | |
| "loss": 0.4842, | |
| "step": 5500 | |
| } | |
| ], | |
| "max_steps": 5952, | |
| "num_train_epochs": 2, | |
| "total_flos": 1129973529784770.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |