{"train/loss": 4.8834, "train/grad_norm": 86.45769500732422, "train/learning_rate": 0.0, "train/epoch": 0.0, "train/global_step": 60, "_timestamp": 1712907251.181722, "_runtime": 394.54964303970337, "_step": 7, "train_runtime": 20.5872, "train_samples_per_second": 5.829, "train_steps_per_second": 2.914, "total_flos": 19812939669504.0, "train_loss": 4.715130106608073, "_wandb": {"runtime": 86}}