{ "best_metric": null, "best_model_checkpoint": null, "epoch": 2.0, "global_step": 4000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 4.99875e-05, "loss": 9.7097, "step": 1 }, { "epoch": 0.25, "learning_rate": 4.375e-05, "loss": 0.7791, "step": 500 }, { "epoch": 0.5, "learning_rate": 3.7500000000000003e-05, "loss": 0.5107, "step": 1000 }, { "epoch": 0.75, "learning_rate": 3.125e-05, "loss": 0.3557, "step": 1500 }, { "epoch": 1.0, "learning_rate": 2.5e-05, "loss": 0.3144, "step": 2000 }, { "epoch": 1.0, "eval_cer": 0.04733333333333333, "eval_loss": 0.24631145596504211, "eval_runtime": 55301.7054, "eval_samples_per_second": 0.072, "eval_steps_per_second": 0.009, "step": 2000 }, { "epoch": 1.25, "learning_rate": 1.8750000000000002e-05, "loss": 0.2086, "step": 2500 }, { "epoch": 1.5, "learning_rate": 1.25e-05, "loss": 0.1776, "step": 3000 }, { "epoch": 1.75, "learning_rate": 6.25e-06, "loss": 0.1802, "step": 3500 }, { "epoch": 2.0, "learning_rate": 0.0, "loss": 0.143, "step": 4000 }, { "epoch": 2.0, "eval_cer": 0.036833333333333336, "eval_loss": 0.15811479091644287, "eval_runtime": 54648.5441, "eval_samples_per_second": 0.073, "eval_steps_per_second": 0.009, "step": 4000 }, { "epoch": 2.0, "step": 4000, "total_flos": 2.3945126111870976e+19, "train_loss": 0.3358841931819916, "train_runtime": 774109.7161, "train_samples_per_second": 0.041, "train_steps_per_second": 0.005 } ], "max_steps": 4000, "num_train_epochs": 2, "total_flos": 2.3945126111870976e+19, "trial_name": null, "trial_params": null }