| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 17.018333333333334, |
| "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": 10.1251, |
| "step": 1 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 0.00025, |
| "loss": 7.2523, |
| "step": 50 |
| }, |
| { |
| "epoch": 2.03, |
| "learning_rate": 0.0005, |
| "loss": 5.2491, |
| "step": 100 |
| }, |
| { |
| "epoch": 4.01, |
| "learning_rate": 0.0005833333333333333, |
| "loss": 4.5664, |
| "step": 150 |
| }, |
| { |
| "epoch": 5.03, |
| "learning_rate": 0.0005555555555555556, |
| "loss": 4.2636, |
| "step": 200 |
| }, |
| { |
| "epoch": 7.01, |
| "learning_rate": 0.0005277777777777777, |
| "loss": 4.1491, |
| "step": 250 |
| }, |
| { |
| "epoch": 8.02, |
| "learning_rate": 0.0005, |
| "loss": 3.9977, |
| "step": 300 |
| }, |
| { |
| "epoch": 10.01, |
| "learning_rate": 0.00047222222222222224, |
| "loss": 3.8366, |
| "step": 350 |
| }, |
| { |
| "epoch": 11.02, |
| "learning_rate": 0.00044444444444444436, |
| "loss": 3.6473, |
| "step": 400 |
| }, |
| { |
| "epoch": 13.01, |
| "learning_rate": 0.00041666666666666664, |
| "loss": 3.5171, |
| "step": 450 |
| }, |
| { |
| "epoch": 14.02, |
| "learning_rate": 0.00038888888888888887, |
| "loss": 3.4001, |
| "step": 500 |
| }, |
| { |
| "epoch": 16.0, |
| "learning_rate": 0.0003611111111111111, |
| "loss": 3.3159, |
| "step": 550 |
| }, |
| { |
| "epoch": 17.02, |
| "learning_rate": 0.0003333333333333333, |
| "loss": 3.2335, |
| "step": 600 |
| } |
| ], |
| "max_steps": 1200, |
| "num_train_epochs": 9223372036854775807, |
| "total_flos": 1.60875413766144e+17, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
