| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.506265664160401, | |
| "global_step": 2000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.5822890559732666e-05, | |
| "loss": 3.1176, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 2.115739107131958, | |
| "eval_runtime": 3.995, | |
| "eval_samples_per_second": 177.722, | |
| "eval_steps_per_second": 11.264, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 1.1645781119465331e-05, | |
| "loss": 2.5326, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 7.468671679197995e-06, | |
| "loss": 2.3611, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 1.9598926305770874, | |
| "eval_runtime": 3.8777, | |
| "eval_samples_per_second": 183.099, | |
| "eval_steps_per_second": 11.605, | |
| "step": 1596 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 3.29156223893066e-06, | |
| "loss": 2.3096, | |
| "step": 2000 | |
| } | |
| ], | |
| "max_steps": 2394, | |
| "num_train_epochs": 3, | |
| "total_flos": 2114734388674560.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |