| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "global_step": 6200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 5e-06, | |
| "loss": 2.5208, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 5e-06, | |
| "loss": 2.5177, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 5e-06, | |
| "loss": 2.5159, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 5e-06, | |
| "loss": 2.5165, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 5e-06, | |
| "loss": 2.5148, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5e-06, | |
| "loss": 2.5101, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 5e-06, | |
| "loss": 2.506, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5e-06, | |
| "loss": 2.504, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 5e-06, | |
| "loss": 2.4961, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 5e-06, | |
| "loss": 2.4963, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 5e-06, | |
| "loss": 2.4875, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 5e-06, | |
| "loss": 2.4879, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 6200, | |
| "total_flos": 3.231863246998733e+18, | |
| "train_runtime": 58892.1512, | |
| "train_samples_per_second": 0.105 | |
| } | |
| ], | |
| "max_steps": 6200, | |
| "num_train_epochs": 1, | |
| "total_flos": 3.231863246998733e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |