{ "best_metric": null, "best_model_checkpoint": null, "epoch": 14.285714285714286, "global_step": 500, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.71, "learning_rate": 4.4e-06, "loss": 0.5543, "step": 25 }, { "epoch": 1.43, "learning_rate": 9.4e-06, "loss": 0.2192, "step": 50 }, { "epoch": 2.14, "learning_rate": 9.511111111111112e-06, "loss": 0.1465, "step": 75 }, { "epoch": 2.86, "learning_rate": 8.977777777777778e-06, "loss": 0.0888, "step": 100 }, { "epoch": 3.57, "learning_rate": 8.422222222222223e-06, "loss": 0.0544, "step": 125 }, { "epoch": 4.29, "learning_rate": 7.866666666666667e-06, "loss": 0.0361, "step": 150 }, { "epoch": 5.0, "learning_rate": 7.3111111111111125e-06, "loss": 0.0309, "step": 175 }, { "epoch": 5.71, "learning_rate": 6.755555555555556e-06, "loss": 0.0201, "step": 200 }, { "epoch": 6.43, "learning_rate": 6.200000000000001e-06, "loss": 0.0139, "step": 225 }, { "epoch": 7.14, "learning_rate": 5.6444444444444445e-06, "loss": 0.0122, "step": 250 }, { "epoch": 7.86, "learning_rate": 5.088888888888889e-06, "loss": 0.0076, "step": 275 }, { "epoch": 8.57, "learning_rate": 4.533333333333334e-06, "loss": 0.0046, "step": 300 }, { "epoch": 9.29, "learning_rate": 3.977777777777778e-06, "loss": 0.0057, "step": 325 }, { "epoch": 10.0, "learning_rate": 3.4222222222222224e-06, "loss": 0.0026, "step": 350 }, { "epoch": 10.71, "learning_rate": 2.866666666666667e-06, "loss": 0.0008, "step": 375 }, { "epoch": 11.43, "learning_rate": 2.311111111111111e-06, "loss": 0.0004, "step": 400 }, { "epoch": 12.14, "learning_rate": 1.7555555555555556e-06, "loss": 0.0002, "step": 425 }, { "epoch": 12.86, "learning_rate": 1.2000000000000002e-06, "loss": 0.0002, "step": 450 }, { "epoch": 13.57, "learning_rate": 6.444444444444445e-07, "loss": 0.0002, "step": 475 }, { "epoch": 14.29, "learning_rate": 8.88888888888889e-08, "loss": 0.0002, "step": 500 }, { "epoch": 14.29, "eval_loss": 0.33821406960487366, "eval_runtime": 1147.3146, "eval_samples_per_second": 0.348, "eval_steps_per_second": 0.044, "eval_wer": 21.27310061601643, "step": 500 }, { "epoch": 14.29, "step": 500, "total_flos": 1.69853681664e+19, "train_loss": 0.059931193444877864, "train_runtime": 3771.6867, "train_samples_per_second": 2.121, "train_steps_per_second": 0.133 } ], "max_steps": 500, "num_train_epochs": 15, "total_flos": 1.69853681664e+19, "trial_name": null, "trial_params": null }