| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 14.319809069212411, | |
| "global_step": 6000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 4.701670644391408e-05, | |
| "loss": 0.0411, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 4.4033412887828165e-05, | |
| "loss": 0.003, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 4.105011933174224e-05, | |
| "loss": 0.0014, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 4.77, | |
| "learning_rate": 3.806682577565633e-05, | |
| "loss": 0.0001, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 5.97, | |
| "learning_rate": 3.5083532219570406e-05, | |
| "loss": 0.002, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 7.16, | |
| "learning_rate": 3.210023866348449e-05, | |
| "loss": 0.0036, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 8.35, | |
| "learning_rate": 2.911694510739857e-05, | |
| "loss": 0.0021, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 9.55, | |
| "learning_rate": 2.613365155131265e-05, | |
| "loss": 0.0015, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 10.74, | |
| "learning_rate": 2.3150357995226732e-05, | |
| "loss": 0.0011, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 11.93, | |
| "learning_rate": 2.0167064439140813e-05, | |
| "loss": 0.0015, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 13.13, | |
| "learning_rate": 1.718377088305489e-05, | |
| "loss": 0.0002, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 14.32, | |
| "learning_rate": 1.4200477326968974e-05, | |
| "loss": 0.0003, | |
| "step": 6000 | |
| } | |
| ], | |
| "max_steps": 8380, | |
| "num_train_epochs": 20, | |
| "total_flos": 555202280044620.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |