{ "best_metric": null, "best_model_checkpoint": null, "epoch": 3.0, "global_step": 22086, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.07, "learning_rate": 4.8868061215249484e-05, "loss": 0.6807, "step": 500 }, { "epoch": 0.14, "learning_rate": 4.773612243049896e-05, "loss": 0.4855, "step": 1000 }, { "epoch": 0.2, "learning_rate": 4.660418364574844e-05, "loss": 0.4476, "step": 1500 }, { "epoch": 0.27, "learning_rate": 4.547224486099792e-05, "loss": 0.4386, "step": 2000 }, { "epoch": 0.34, "learning_rate": 4.43403060762474e-05, "loss": 0.4359, "step": 2500 }, { "epoch": 0.41, "learning_rate": 4.3208367291496876e-05, "loss": 0.4146, "step": 3000 }, { "epoch": 0.48, "learning_rate": 4.207642850674636e-05, "loss": 0.4143, "step": 3500 }, { "epoch": 0.54, "learning_rate": 4.094448972199584e-05, "loss": 0.3972, "step": 4000 }, { "epoch": 0.61, "learning_rate": 3.981255093724531e-05, "loss": 0.4009, "step": 4500 }, { "epoch": 0.68, "learning_rate": 3.8680612152494794e-05, "loss": 0.4085, "step": 5000 }, { "epoch": 0.75, "learning_rate": 3.7548673367744275e-05, "loss": 0.3819, "step": 5500 }, { "epoch": 0.81, "learning_rate": 3.6416734582993756e-05, "loss": 0.394, "step": 6000 }, { "epoch": 0.88, "learning_rate": 3.528479579824323e-05, "loss": 0.3901, "step": 6500 }, { "epoch": 0.95, "learning_rate": 3.415285701349271e-05, "loss": 0.3975, "step": 7000 }, { "epoch": 1.02, "learning_rate": 3.302091822874219e-05, "loss": 0.3761, "step": 7500 }, { "epoch": 1.09, "learning_rate": 3.188897944399167e-05, "loss": 0.3519, "step": 8000 }, { "epoch": 1.15, "learning_rate": 3.075704065924115e-05, "loss": 0.3385, "step": 8500 }, { "epoch": 1.22, "learning_rate": 2.962510187449063e-05, "loss": 0.3505, "step": 9000 }, { "epoch": 1.29, "learning_rate": 2.849316308974011e-05, "loss": 0.3364, "step": 9500 }, { "epoch": 1.36, "learning_rate": 2.7361224304989585e-05, "loss": 0.347, "step": 10000 }, { "epoch": 1.43, "learning_rate": 2.6229285520239066e-05, "loss": 0.3378, "step": 10500 }, { "epoch": 1.49, "learning_rate": 2.5097346735488547e-05, "loss": 0.3523, "step": 11000 }, { "epoch": 1.56, "learning_rate": 2.3965407950738025e-05, "loss": 0.3441, "step": 11500 }, { "epoch": 1.63, "learning_rate": 2.2833469165987503e-05, "loss": 0.3493, "step": 12000 }, { "epoch": 1.7, "learning_rate": 2.1701530381236984e-05, "loss": 0.3416, "step": 12500 }, { "epoch": 1.77, "learning_rate": 2.0569591596486462e-05, "loss": 0.3364, "step": 13000 }, { "epoch": 1.83, "learning_rate": 1.9437652811735943e-05, "loss": 0.3401, "step": 13500 }, { "epoch": 1.9, "learning_rate": 1.830571402698542e-05, "loss": 0.3543, "step": 14000 }, { "epoch": 1.97, "learning_rate": 1.7173775242234902e-05, "loss": 0.3269, "step": 14500 }, { "epoch": 2.04, "learning_rate": 1.604183645748438e-05, "loss": 0.3228, "step": 15000 }, { "epoch": 2.11, "learning_rate": 1.4909897672733858e-05, "loss": 0.3102, "step": 15500 }, { "epoch": 2.17, "learning_rate": 1.3777958887983339e-05, "loss": 0.3203, "step": 16000 }, { "epoch": 2.24, "learning_rate": 1.2646020103232817e-05, "loss": 0.2996, "step": 16500 }, { "epoch": 2.31, "learning_rate": 1.1514081318482298e-05, "loss": 0.3019, "step": 17000 }, { "epoch": 2.38, "learning_rate": 1.0382142533731776e-05, "loss": 0.3116, "step": 17500 }, { "epoch": 2.44, "learning_rate": 9.250203748981255e-06, "loss": 0.2996, "step": 18000 }, { "epoch": 2.51, "learning_rate": 8.118264964230735e-06, "loss": 0.3065, "step": 18500 }, { "epoch": 2.58, "learning_rate": 6.986326179480214e-06, "loss": 0.3168, "step": 19000 }, { "epoch": 2.65, "learning_rate": 5.8543873947296935e-06, "loss": 0.3124, "step": 19500 }, { "epoch": 2.72, "learning_rate": 4.722448609979173e-06, "loss": 0.2943, "step": 20000 }, { "epoch": 2.78, "learning_rate": 3.5905098252286516e-06, "loss": 0.3034, "step": 20500 }, { "epoch": 2.85, "learning_rate": 2.458571040478131e-06, "loss": 0.2954, "step": 21000 }, { "epoch": 2.92, "learning_rate": 1.3266322557276103e-06, "loss": 0.3102, "step": 21500 }, { "epoch": 2.99, "learning_rate": 1.9469347097708958e-07, "loss": 0.3105, "step": 22000 }, { "epoch": 3.0, "step": 22086, "total_flos": 1.617792737439744e+16, "train_loss": 0.36091021529403833, "train_runtime": 3139.6723, "train_samples_per_second": 56.275, "train_steps_per_second": 7.034 } ], "max_steps": 22086, "num_train_epochs": 3, "total_flos": 1.617792737439744e+16, "trial_name": null, "trial_params": null }