{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "global_step": 6088, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.08, "learning_rate": 2.7536136662286466e-05, "loss": 2.3225, "step": 500 }, { "epoch": 0.16, "learning_rate": 2.5072273324572932e-05, "loss": 1.4439, "step": 1000 }, { "epoch": 0.25, "learning_rate": 2.2608409986859398e-05, "loss": 1.3325, "step": 1500 }, { "epoch": 0.33, "learning_rate": 2.014454664914586e-05, "loss": 1.236, "step": 2000 }, { "epoch": 0.41, "learning_rate": 1.7680683311432326e-05, "loss": 1.1708, "step": 2500 }, { "epoch": 0.49, "learning_rate": 1.5216819973718791e-05, "loss": 1.1273, "step": 3000 }, { "epoch": 0.57, "learning_rate": 1.2752956636005257e-05, "loss": 1.097, "step": 3500 }, { "epoch": 0.66, "learning_rate": 1.0289093298291721e-05, "loss": 1.0855, "step": 4000 }, { "epoch": 0.74, "learning_rate": 7.825229960578187e-06, "loss": 1.0549, "step": 4500 }, { "epoch": 0.82, "learning_rate": 5.361366622864652e-06, "loss": 1.0084, "step": 5000 }, { "epoch": 0.9, "learning_rate": 2.8975032851511168e-06, "loss": 1.0201, "step": 5500 }, { "epoch": 0.99, "learning_rate": 4.336399474375821e-07, "loss": 1.0301, "step": 6000 }, { "epoch": 1.0, "step": 6088, "total_flos": 1.2726197240070144e+16, "train_loss": 1.240907019917191, "train_runtime": 2500.1104, "train_samples_per_second": 38.961, "train_steps_per_second": 2.435 } ], "max_steps": 6088, "num_train_epochs": 1, "total_flos": 1.2726197240070144e+16, "trial_name": null, "trial_params": null }