| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9940357852882704, | |
| "global_step": 5500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.819266220856678e-05, | |
| "loss": 3.0844, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.6385324417133565e-05, | |
| "loss": 1.8306, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.4577986625700345e-05, | |
| "loss": 1.6551, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.2770648834267125e-05, | |
| "loss": 1.5101, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.0963311042833906e-05, | |
| "loss": 1.4381, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.155973251400688e-06, | |
| "loss": 1.3683, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 7.348635459967468e-06, | |
| "loss": 1.2983, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 5.541297668534249e-06, | |
| "loss": 1.3465, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 3.7339598771010305e-06, | |
| "loss": 1.3036, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.9266220856678112e-06, | |
| "loss": 1.2562, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.1928429423459245e-07, | |
| "loss": 1.2602, | |
| "step": 5500 | |
| } | |
| ], | |
| "max_steps": 5533, | |
| "num_train_epochs": 1, | |
| "total_flos": 8623104602112000.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |