{ "best_metric": null, "best_model_checkpoint": null, "epoch": 6.25, "eval_steps": 500, "global_step": 500, "is_hyper_param_search": true, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 1.0, "eval_accuracy": 0.0064516129032258064, "eval_loss": NaN, "eval_runtime": 1.0576, "eval_samples_per_second": 2931.121, "eval_steps_per_second": 16.074, "step": 80 }, { "epoch": 2.0, "eval_accuracy": 0.0064516129032258064, "eval_loss": NaN, "eval_runtime": 1.0529, "eval_samples_per_second": 2944.294, "eval_steps_per_second": 16.146, "step": 160 }, { "epoch": 3.0, "eval_accuracy": 0.0064516129032258064, "eval_loss": NaN, "eval_runtime": 1.0558, "eval_samples_per_second": 2936.237, "eval_steps_per_second": 16.102, "step": 240 }, { "epoch": 4.0, "eval_accuracy": 0.0064516129032258064, "eval_loss": NaN, "eval_runtime": 1.0539, "eval_samples_per_second": 2941.384, "eval_steps_per_second": 16.13, "step": 320 }, { "epoch": 5.0, "eval_accuracy": 0.0064516129032258064, "eval_loss": NaN, "eval_runtime": 1.0568, "eval_samples_per_second": 2933.442, "eval_steps_per_second": 16.087, "step": 400 }, { "epoch": 6.0, "eval_accuracy": 0.0064516129032258064, "eval_loss": NaN, "eval_runtime": 1.0949, "eval_samples_per_second": 2831.419, "eval_steps_per_second": 15.527, "step": 480 }, { "epoch": 6.25, "learning_rate": 7.500000000000001e-06, "loss": 0.0, "step": 500 } ], "logging_steps": 500, "max_steps": 800, "num_input_tokens_seen": 0, "num_train_epochs": 10, "save_steps": 500, "total_flos": 591309336775836.0, "train_batch_size": 192, "trial_name": null, "trial_params": { "alpha": 0.7802678398155426, "num_train_epochs": 10, "temperature": 8 } }