| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9778612219353829, | |
| "global_step": 12500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.804427755612923e-05, | |
| "loss": 5.6625, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.608855511225847e-05, | |
| "loss": 5.0889, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.41328326683877e-05, | |
| "loss": 4.7108, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.217711022451694e-05, | |
| "loss": 4.5723, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.0221387780646176e-05, | |
| "loss": 4.3708, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 3.826566533677541e-05, | |
| "loss": 4.2174, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 3.6309942892904644e-05, | |
| "loss": 4.1822, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 3.4354220449033875e-05, | |
| "loss": 4.1232, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 3.2398498005163106e-05, | |
| "loss": 3.9563, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.0442775561292343e-05, | |
| "loss": 3.9201, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 2.8487053117421574e-05, | |
| "loss": 3.7911, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 2.6531330673550808e-05, | |
| "loss": 3.7768, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.4575608229680045e-05, | |
| "loss": 3.8178, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.2619885785809276e-05, | |
| "loss": 3.6894, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.0664163341938514e-05, | |
| "loss": 3.5751, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.8708440898067748e-05, | |
| "loss": 3.5514, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.6752718454196982e-05, | |
| "loss": 3.5458, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.4796996010326214e-05, | |
| "loss": 3.5009, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.284127356645545e-05, | |
| "loss": 3.4645, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.0885551122584684e-05, | |
| "loss": 3.5183, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 8.929828678713916e-06, | |
| "loss": 3.4722, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 6.974106234843151e-06, | |
| "loss": 3.4618, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 5.018383790972385e-06, | |
| "loss": 3.4238, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 3.0626613471016194e-06, | |
| "loss": 3.4192, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.1069389032308535e-06, | |
| "loss": 3.4197, | |
| "step": 12500 | |
| } | |
| ], | |
| "max_steps": 12783, | |
| "num_train_epochs": 1, | |
| "total_flos": 974676043775232, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |