| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 24.217264791464597, | |
| "global_step": 3100, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 10.1419, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 9.2788, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 3e-06, | |
| "loss": 8.8011, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 8.3846, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 5e-06, | |
| "loss": 7.8214, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 6e-06, | |
| "loss": 7.3094, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 5.47, | |
| "learning_rate": 7.000000000000001e-06, | |
| "loss": 6.7911, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 6.25, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 6.4144, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "learning_rate": 9e-06, | |
| "loss": 6.2168, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 7.81, | |
| "learning_rate": 1e-05, | |
| "loss": 6.0642, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 8.59, | |
| "learning_rate": 1.1000000000000001e-05, | |
| "loss": 6.0511, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 9.37, | |
| "learning_rate": 1.2e-05, | |
| "loss": 5.995, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 10.16, | |
| "learning_rate": 1.3000000000000001e-05, | |
| "loss": 5.9539, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 10.93, | |
| "learning_rate": 1.4000000000000001e-05, | |
| "loss": 5.8675, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 11.71, | |
| "learning_rate": 1.5e-05, | |
| "loss": 5.8874, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 12.5, | |
| "learning_rate": 1.6000000000000003e-05, | |
| "loss": 5.8578, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 13.28, | |
| "learning_rate": 1.7000000000000003e-05, | |
| "loss": 5.8386, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 14.06, | |
| "learning_rate": 1.8e-05, | |
| "loss": 5.8138, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 14.84, | |
| "learning_rate": 1.9e-05, | |
| "loss": 5.7399, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 15.62, | |
| "learning_rate": 2e-05, | |
| "loss": 5.7753, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 16.4, | |
| "learning_rate": 2.1e-05, | |
| "loss": 5.7564, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 17.19, | |
| "learning_rate": 2.2000000000000003e-05, | |
| "loss": 5.738, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 17.96, | |
| "learning_rate": 2.3000000000000003e-05, | |
| "loss": 5.6753, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 18.74, | |
| "learning_rate": 2.4e-05, | |
| "loss": 5.7082, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 19.53, | |
| "learning_rate": 2.5e-05, | |
| "loss": 5.6991, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 20.31, | |
| "learning_rate": 2.6000000000000002e-05, | |
| "loss": 5.6801, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 21.09, | |
| "learning_rate": 2.7000000000000002e-05, | |
| "loss": 5.6692, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 21.87, | |
| "learning_rate": 2.8000000000000003e-05, | |
| "loss": 5.6063, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 22.65, | |
| "learning_rate": 2.9e-05, | |
| "loss": 5.6445, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 23.43, | |
| "learning_rate": 3e-05, | |
| "loss": 5.6328, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 24.22, | |
| "learning_rate": 3.1e-05, | |
| "loss": 5.6217, | |
| "step": 3100 | |
| } | |
| ], | |
| "max_steps": 20480, | |
| "num_train_epochs": 160, | |
| "total_flos": 2.101463476076544e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |