{ "best_metric": 1.303820013999939, "epoch": 1.0, "global_step": 6230, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.08, "learning_rate": 9.598715890850723e-06, "loss": 1.528, "step": 500 }, { "epoch": 0.16, "learning_rate": 9.197431781701446e-06, "loss": 1.3696, "step": 1000 }, { "epoch": 0.24, "learning_rate": 8.796147672552167e-06, "loss": 1.2899, "step": 1500 }, { "epoch": 0.32, "learning_rate": 8.39486356340289e-06, "loss": 1.2996, "step": 2000 }, { "epoch": 0.4, "learning_rate": 7.993579454253612e-06, "loss": 1.2637, "step": 2500 }, { "epoch": 0.48, "learning_rate": 7.592295345104335e-06, "loss": 1.2596, "step": 3000 }, { "epoch": 0.56, "learning_rate": 7.191011235955056e-06, "loss": 1.2513, "step": 3500 }, { "epoch": 0.64, "learning_rate": 6.789727126805778e-06, "loss": 1.2211, "step": 4000 }, { "epoch": 0.72, "learning_rate": 6.3884430176565015e-06, "loss": 1.2108, "step": 4500 }, { "epoch": 0.8, "learning_rate": 5.987158908507223e-06, "loss": 1.1853, "step": 5000 }, { "epoch": 0.88, "learning_rate": 5.585874799357945e-06, "loss": 1.2138, "step": 5500 }, { "epoch": 0.96, "learning_rate": 5.1845906902086685e-06, "loss": 1.1879, "step": 6000 }, { "epoch": 1.0, "eval_gen_len": 18.758, "eval_loss": 1.303820013999939, "eval_rouge1": 38.7078, "eval_rouge2": 14.3583, "eval_rougeL": 32.2229, "eval_rougeLsum": 34.3647, "eval_runtime": 739.1192, "eval_samples_per_second": 2.029, "eval_steps_per_second": 1.015, "step": 6230 } ], "max_steps": 12460, "num_train_epochs": 2, "total_flos": 1.48195611648e+16, "trial_name": null, "trial_params": null }