{"train/loss": 4.5023, "train/grad_norm": 239.7991943359375, "train/learning_rate": 2e-06, "train/epoch": 0.0, "train/global_step": 40, "_timestamp": 1712907386.0967398, "_runtime": 30.347853660583496, "_step": 1} |
{"train/loss": 4.5023, "train/grad_norm": 239.7991943359375, "train/learning_rate": 2e-06, "train/epoch": 0.0, "train/global_step": 40, "_timestamp": 1712907386.0967398, "_runtime": 30.347853660583496, "_step": 1} |