{"train/loss": 4.9076, "train/grad_norm": 92.3516845703125, "train/learning_rate": 1.3000000000000001e-05, "train/epoch": 0.01, "train/global_step": 260, "_timestamp": 1712907460.001729, "_runtime": 104.2528429031372, "_step": 12} |
{"train/loss": 4.9076, "train/grad_norm": 92.3516845703125, "train/learning_rate": 1.3000000000000001e-05, "train/epoch": 0.01, "train/global_step": 260, "_timestamp": 1712907460.001729, "_runtime": 104.2528429031372, "_step": 12} |