{"train/loss": 3.7613, "train/grad_norm": 32.12275314331055, "train/learning_rate": 2.9994965304402304e-05, "train/epoch": 0.06, "train/global_step": 760, "_timestamp": 1712908593.453055, "_runtime": 948.9453368186951, "_step": 37} |
{"train/loss": 3.7613, "train/grad_norm": 32.12275314331055, "train/learning_rate": 2.9994965304402304e-05, "train/epoch": 0.06, "train/global_step": 760, "_timestamp": 1712908593.453055, "_runtime": 948.9453368186951, "_step": 37} |