{ "best_metric": 0.4746718108654022, "best_model_checkpoint": "sohail3/checkpoint-200", "epoch": 4.733727810650888, "eval_steps": 50, "global_step": 200, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.59, "learning_rate": 0.0001, "loss": 0.6978, "step": 25 }, { "epoch": 1.18, "learning_rate": 0.0002, "loss": 0.5717, "step": 50 }, { "epoch": 1.18, "eval_loss": 0.5111950039863586, "eval_runtime": 11.5977, "eval_samples_per_second": 29.833, "eval_steps_per_second": 14.917, "step": 50 }, { "epoch": 1.78, "learning_rate": 0.000175, "loss": 0.5435, "step": 75 }, { "epoch": 2.37, "learning_rate": 0.00015000000000000001, "loss": 0.5235, "step": 100 }, { "epoch": 2.37, "eval_loss": 0.4941622316837311, "eval_runtime": 11.6258, "eval_samples_per_second": 29.761, "eval_steps_per_second": 14.881, "step": 100 }, { "epoch": 2.96, "learning_rate": 0.000125, "loss": 0.5137, "step": 125 }, { "epoch": 3.55, "learning_rate": 0.0001, "loss": 0.5013, "step": 150 }, { "epoch": 3.55, "eval_loss": 0.47987258434295654, "eval_runtime": 11.8958, "eval_samples_per_second": 29.086, "eval_steps_per_second": 14.543, "step": 150 }, { "epoch": 4.14, "learning_rate": 7.500000000000001e-05, "loss": 0.4931, "step": 175 }, { "epoch": 4.73, "learning_rate": 5e-05, "loss": 0.4858, "step": 200 }, { "epoch": 4.73, "eval_loss": 0.4746718108654022, "eval_runtime": 11.8752, "eval_samples_per_second": 29.136, "eval_steps_per_second": 14.568, "step": 200 } ], "logging_steps": 25, "max_steps": 250, "num_train_epochs": 6, "save_steps": 100, "total_flos": 1775052675060192.0, "trial_name": null, "trial_params": null }