| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.998773507767784, | |
| "global_step": 7335, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1737, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 2.8384317104696252e-05, | |
| "loss": 0.067, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 2.6768634209392504e-05, | |
| "loss": 0.063, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 2.5152951314088755e-05, | |
| "loss": 0.0614, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 2.3537268418785007e-05, | |
| "loss": 0.0592, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 2.1921585523481258e-05, | |
| "loss": 0.0563, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 2.030590262817751e-05, | |
| "loss": 0.0543, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 1.8690219732873765e-05, | |
| "loss": 0.0545, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 1.7074536837570013e-05, | |
| "loss": 0.0546, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 1.5458853942266264e-05, | |
| "loss": 0.0536, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 1.3843171046962518e-05, | |
| "loss": 0.0512, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 1.2227488151658767e-05, | |
| "loss": 0.0506, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 1.0611805256355019e-05, | |
| "loss": 0.0498, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 8.996122361051272e-06, | |
| "loss": 0.0496, | |
| "step": 7000 | |
| } | |
| ], | |
| "max_steps": 9784, | |
| "num_train_epochs": 4, | |
| "total_flos": 6.357201350295552e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |