| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 10.0, | |
| "global_step": 5460, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 2.7252747252747255e-05, | |
| "loss": 1.7633, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 2.4505494505494506e-05, | |
| "loss": 1.1786, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 2.175824175824176e-05, | |
| "loss": 0.9744, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 1.901098901098901e-05, | |
| "loss": 0.8272, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 4.58, | |
| "learning_rate": 1.6263736263736265e-05, | |
| "loss": 0.6982, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 5.49, | |
| "learning_rate": 1.3516483516483517e-05, | |
| "loss": 0.603, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 6.41, | |
| "learning_rate": 1.076923076923077e-05, | |
| "loss": 0.5212, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 7.33, | |
| "learning_rate": 8.021978021978022e-06, | |
| "loss": 0.451, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 8.24, | |
| "learning_rate": 5.274725274725275e-06, | |
| "loss": 0.4001, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 9.16, | |
| "learning_rate": 2.527472527472528e-06, | |
| "loss": 0.3608, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "step": 5460, | |
| "total_flos": 2.2620408196051763e+17, | |
| "train_loss": 0.7400637085184509, | |
| "train_runtime": 1828.1814, | |
| "train_samples_per_second": 715.755, | |
| "train_steps_per_second": 2.987 | |
| } | |
| ], | |
| "max_steps": 5460, | |
| "num_train_epochs": 10, | |
| "total_flos": 2.2620408196051763e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |