| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 4.775549188156638, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.8089780324737347e-05, | |
| "loss": 1.2572, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.617956064947469e-05, | |
| "loss": 1.0388, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 1.4269340974212036e-05, | |
| "loss": 0.9388, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 1.235912129894938e-05, | |
| "loss": 0.8806, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 1.0448901623686724e-05, | |
| "loss": 0.8384, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 8.53868194842407e-06, | |
| "loss": 0.8012, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 6.628462273161415e-06, | |
| "loss": 0.7839, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "learning_rate": 4.718242597898759e-06, | |
| "loss": 0.7627, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 4.3, | |
| "learning_rate": 2.8080229226361035e-06, | |
| "loss": 0.7479, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "learning_rate": 8.97803247373448e-07, | |
| "loss": 0.7422, | |
| "step": 5000 | |
| } | |
| ], | |
| "max_steps": 5235, | |
| "num_train_epochs": 5, | |
| "total_flos": 1.3155901731827712e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |