| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 6.0, | |
| "eval_steps": 500, | |
| "global_step": 324, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.08641975308642e-06, | |
| "loss": 4.0718, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 6.17283950617284e-06, | |
| "loss": 3.9483, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 9.25925925925926e-06, | |
| "loss": 3.8844, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 1.234567901234568e-05, | |
| "loss": 3.8757, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 1.54320987654321e-05, | |
| "loss": 3.8588, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 1.851851851851852e-05, | |
| "loss": 3.8919, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 1.971677559912854e-05, | |
| "loss": 3.8707, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 1.917211328976035e-05, | |
| "loss": 3.8775, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 1.862745098039216e-05, | |
| "loss": 3.8929, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 4.63, | |
| "learning_rate": 1.8082788671023966e-05, | |
| "loss": 3.8813, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 1.7538126361655776e-05, | |
| "loss": 3.8796, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 5.56, | |
| "learning_rate": 1.6993464052287582e-05, | |
| "loss": 3.8681, | |
| "step": 300 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 1080, | |
| "num_train_epochs": 20, | |
| "save_steps": 500, | |
| "total_flos": 1.5808841019347927e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |