{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "global_step": 7417, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.07, "learning_rate": 4.6629364972360795e-05, "loss": 3.4842, "step": 500 }, { "epoch": 0.13, "learning_rate": 4.325872994472159e-05, "loss": 3.4358, "step": 1000 }, { "epoch": 0.2, "learning_rate": 3.988809491708238e-05, "loss": 3.4252, "step": 1500 }, { "epoch": 0.27, "learning_rate": 3.6517459889443174e-05, "loss": 3.3965, "step": 2000 }, { "epoch": 0.34, "learning_rate": 3.314682486180397e-05, "loss": 3.3921, "step": 2500 }, { "epoch": 0.4, "learning_rate": 2.977618983416476e-05, "loss": 3.3917, "step": 3000 }, { "epoch": 0.47, "learning_rate": 2.6405554806525552e-05, "loss": 3.3715, "step": 3500 }, { "epoch": 0.54, "learning_rate": 2.3034919778886342e-05, "loss": 3.3522, "step": 4000 }, { "epoch": 0.61, "learning_rate": 1.9664284751247135e-05, "loss": 3.354, "step": 4500 }, { "epoch": 0.67, "learning_rate": 1.6293649723607928e-05, "loss": 3.3391, "step": 5000 }, { "epoch": 0.74, "learning_rate": 1.2923014695968722e-05, "loss": 3.3427, "step": 5500 }, { "epoch": 0.81, "learning_rate": 9.552379668329513e-06, "loss": 3.3244, "step": 6000 }, { "epoch": 0.88, "learning_rate": 6.181744640690306e-06, "loss": 3.3286, "step": 6500 }, { "epoch": 0.94, "learning_rate": 2.811109613051099e-06, "loss": 3.3197, "step": 7000 }, { "epoch": 1.0, "step": 7417, "total_flos": 0, "train_runtime": 4429.0691, "train_samples_per_second": 1.675 } ], "max_steps": 7417, "num_train_epochs": 1, "total_flos": 0, "trial_name": null, "trial_params": null }