| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.0, | |
| "global_step": 4042, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 2e-05, | |
| "loss": 0.4775, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.717673630717109e-05, | |
| "loss": 0.2671, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.435347261434218e-05, | |
| "loss": 0.2199, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.1530208921513271e-05, | |
| "loss": 0.217, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 8.70694522868436e-06, | |
| "loss": 0.1433, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 5.883681535855449e-06, | |
| "loss": 0.1388, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 3.0604178430265392e-06, | |
| "loss": 0.1332, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 2.3715415019762845e-07, | |
| "loss": 0.1405, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "step": 4042, | |
| "total_flos": 8507585434963968.0, | |
| "train_loss": 0.21652541509776466, | |
| "train_runtime": 3978.4582, | |
| "train_samples_per_second": 16.255, | |
| "train_steps_per_second": 1.016 | |
| } | |
| ], | |
| "max_steps": 4042, | |
| "num_train_epochs": 2, | |
| "total_flos": 8507585434963968.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |