| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 0.9508716323296355, |
| "global_step": 600, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.08, |
| "learning_rate": 9.318541996830429e-06, |
| "loss": 6.4989, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 8.526148969889065e-06, |
| "loss": 5.7137, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 7.733755942947703e-06, |
| "loss": 5.4966, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.32, |
| "learning_rate": 6.973058637083995e-06, |
| "loss": 5.2454, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.4, |
| "learning_rate": 6.180665610142631e-06, |
| "loss": 4.7898, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 5.3882725832012685e-06, |
| "loss": 4.5389, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.55, |
| "learning_rate": 4.595879556259905e-06, |
| "loss": 4.0363, |
| "step": 350 |
| }, |
| { |
| "epoch": 0.63, |
| "learning_rate": 3.8034865293185427e-06, |
| "loss": 4.009, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.71, |
| "learning_rate": 3.0110935023771793e-06, |
| "loss": 3.855, |
| "step": 450 |
| }, |
| { |
| "epoch": 0.79, |
| "learning_rate": 2.2187004754358164e-06, |
| "loss": 3.8576, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.87, |
| "learning_rate": 1.4263074484944533e-06, |
| "loss": 4.0199, |
| "step": 550 |
| }, |
| { |
| "epoch": 0.95, |
| "learning_rate": 6.339144215530904e-07, |
| "loss": 3.577, |
| "step": 600 |
| } |
| ], |
| "max_steps": 631, |
| "num_train_epochs": 1, |
| "total_flos": 3.960666495391154e+17, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
