{"train/loss": 4.678, "train/grad_norm": 103.39625549316406, "train/learning_rate": 2.341780598043574e-05, "train/epoch": 0.0, "train/global_step": 20, "_timestamp": 1712906880.3023584, "_runtime": 23.670279502868652, "_step": 0} |
{"train/loss": 4.678, "train/grad_norm": 103.39625549316406, "train/learning_rate": 2.341780598043574e-05, "train/epoch": 0.0, "train/global_step": 20, "_timestamp": 1712906880.3023584, "_runtime": 23.670279502868652, "_step": 0} |