{"train/loss": 0.1257, "train/grad_norm": 0.30121245980262756, "train/learning_rate": 0.0, "train/epoch": 50.0, "train/global_step": 100, "_timestamp": 1711034092.04044, "_runtime": 3579.477168083191, "_step": 100, "eval/loss": 0.497196763753891, "eval/runtime": 6.0321, "eval/samples_per_second": 1.658, "eval/steps_per_second": 0.332, "train/train_runtime": 3589.5773, "train/train_samples_per_second": 0.543, "train/train_steps_per_second": 0.028, "train/total_flos": 795304292106240.0, "train/train_loss": 0.5981726232171058}