| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 24.02, |
| "global_step": 600, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.9999999999999996e-06, |
| "loss": 10.8568, |
| "step": 1 |
| }, |
| { |
| "epoch": 2.0, |
| "learning_rate": 0.00025, |
| "loss": 8.4903, |
| "step": 50 |
| }, |
| { |
| "epoch": 4.0, |
| "learning_rate": 0.0005, |
| "loss": 7.2157, |
| "step": 100 |
| }, |
| { |
| "epoch": 6.0, |
| "learning_rate": 0.0005833333333333333, |
| "loss": 6.6121, |
| "step": 150 |
| }, |
| { |
| "epoch": 8.01, |
| "learning_rate": 0.0005555555555555556, |
| "loss": 6.0811, |
| "step": 200 |
| }, |
| { |
| "epoch": 10.01, |
| "learning_rate": 0.0005277777777777777, |
| "loss": 5.6744, |
| "step": 250 |
| }, |
| { |
| "epoch": 12.01, |
| "learning_rate": 0.0005, |
| "loss": 5.4147, |
| "step": 300 |
| }, |
| { |
| "epoch": 14.01, |
| "learning_rate": 0.00047222222222222224, |
| "loss": 5.246, |
| "step": 350 |
| }, |
| { |
| "epoch": 16.01, |
| "learning_rate": 0.00044444444444444436, |
| "loss": 5.1141, |
| "step": 400 |
| }, |
| { |
| "epoch": 18.02, |
| "learning_rate": 0.00041666666666666664, |
| "loss": 4.9877, |
| "step": 450 |
| }, |
| { |
| "epoch": 20.02, |
| "learning_rate": 0.00038888888888888887, |
| "loss": 4.8652, |
| "step": 500 |
| }, |
| { |
| "epoch": 22.02, |
| "learning_rate": 0.0003611111111111111, |
| "loss": 4.7263, |
| "step": 550 |
| }, |
| { |
| "epoch": 24.02, |
| "learning_rate": 0.0003333333333333333, |
| "loss": 4.5808, |
| "step": 600 |
| } |
| ], |
| "max_steps": 1200, |
| "num_train_epochs": 9223372036854775807, |
| "total_flos": 1.60951711039488e+17, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
