{"train/loss": 4.5777, "train/grad_norm": 62.10089111328125, "train/learning_rate": 1e-06, "train/epoch": 0.0, "train/global_step": 20, "_timestamp": 1712907682.0631418, "_runtime": 37.555423736572266, "_step": 0} |
{"train/loss": 4.5777, "train/grad_norm": 62.10089111328125, "train/learning_rate": 1e-06, "train/epoch": 0.0, "train/global_step": 20, "_timestamp": 1712907682.0631418, "_runtime": 37.555423736572266, "_step": 0} |