| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 11.0225, |
| "global_step": 500, |
| "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": 10.9692, |
| "step": 1 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 0.00025, |
| "loss": 8.063, |
| "step": 50 |
| }, |
| { |
| "epoch": 2.01, |
| "learning_rate": 0.0005, |
| "loss": 4.9432, |
| "step": 100 |
| }, |
| { |
| "epoch": 3.02, |
| "learning_rate": 0.0005833333333333333, |
| "loss": 3.9936, |
| "step": 150 |
| }, |
| { |
| "epoch": 4.02, |
| "learning_rate": 0.0005555555555555556, |
| "loss": 3.7529, |
| "step": 200 |
| }, |
| { |
| "epoch": 5.03, |
| "learning_rate": 0.0005277777777777777, |
| "loss": 3.6008, |
| "step": 250 |
| }, |
| { |
| "epoch": 6.04, |
| "learning_rate": 0.0005, |
| "loss": 3.3332, |
| "step": 300 |
| }, |
| { |
| "epoch": 8.01, |
| "learning_rate": 0.00047222222222222224, |
| "loss": 3.089, |
| "step": 350 |
| }, |
| { |
| "epoch": 9.01, |
| "learning_rate": 0.00044444444444444436, |
| "loss": 2.8687, |
| "step": 400 |
| }, |
| { |
| "epoch": 10.02, |
| "learning_rate": 0.00041666666666666664, |
| "loss": 2.7458, |
| "step": 450 |
| }, |
| { |
| "epoch": 11.02, |
| "learning_rate": 0.00038888888888888887, |
| "loss": 2.6512, |
| "step": 500 |
| } |
| ], |
| "max_steps": 1200, |
| "num_train_epochs": 9223372036854775807, |
| "total_flos": 1.35776223756288e+17, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
