{ "best_metric": null, "best_model_checkpoint": null, "epoch": 15.0, "global_step": 5250, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 1.43, "learning_rate": 9.047619047619048e-05, "loss": 5.2409, "step": 500 }, { "epoch": 2.86, "learning_rate": 8.095238095238096e-05, "loss": 2.9606, "step": 1000 }, { "epoch": 4.29, "learning_rate": 7.142857142857143e-05, "loss": 2.033, "step": 1500 }, { "epoch": 5.71, "learning_rate": 6.19047619047619e-05, "loss": 1.7955, "step": 2000 }, { "epoch": 7.14, "learning_rate": 5.2380952380952384e-05, "loss": 1.6764, "step": 2500 }, { "epoch": 8.57, "learning_rate": 4.2857142857142856e-05, "loss": 1.6116, "step": 3000 }, { "epoch": 10.0, "learning_rate": 3.3333333333333335e-05, "loss": 1.566, "step": 3500 }, { "epoch": 11.43, "learning_rate": 2.380952380952381e-05, "loss": 1.5403, "step": 4000 }, { "epoch": 12.86, "learning_rate": 1.4285714285714285e-05, "loss": 1.524, "step": 4500 }, { "epoch": 14.29, "learning_rate": 4.7619047619047615e-06, "loss": 1.5067, "step": 5000 }, { "epoch": 15.0, "step": 5250, "total_flos": 2.6285793201251942e+17, "train_loss": 2.113718255905878, "train_runtime": 1777.2067, "train_samples_per_second": 754.192, "train_steps_per_second": 2.954 } ], "max_steps": 5250, "num_train_epochs": 15, "total_flos": 2.6285793201251942e+17, "trial_name": null, "trial_params": null }