| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 9.999908653156034, | |
| "global_step": 95780, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 9.048181989924434e-05, | |
| "loss": 3.734, | |
| "step": 9578 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 8.043939546599497e-05, | |
| "loss": 3.5549, | |
| "step": 19156 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 7.03969710327456e-05, | |
| "loss": 3.5027, | |
| "step": 28734 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 6.0354546599496234e-05, | |
| "loss": 3.4695, | |
| "step": 38312 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 5.0312122166246854e-05, | |
| "loss": 3.4447, | |
| "step": 47890 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 4.026969773299749e-05, | |
| "loss": 3.4219, | |
| "step": 57468 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 3.0227273299748105e-05, | |
| "loss": 3.3985, | |
| "step": 67046 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "learning_rate": 2.0184848866498742e-05, | |
| "loss": 3.3824, | |
| "step": 76624 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "learning_rate": 1.0142424433249365e-05, | |
| "loss": 3.3701, | |
| "step": 86202 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "learning_rate": 9.999999999999998e-08, | |
| "loss": 3.3609, | |
| "step": 95780 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "step": 95780, | |
| "total_flos": 5.86608789194907e+18, | |
| "train_loss": 3.463967107237941, | |
| "train_runtime": 128255.8433, | |
| "train_samples_per_second": 23.899, | |
| "train_steps_per_second": 0.747 | |
| } | |
| ], | |
| "max_steps": 95780, | |
| "num_train_epochs": 10, | |
| "total_flos": 5.86608789194907e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |