{"train/loss": 3.6265, "train/grad_norm": 29.243453979492188, "train/learning_rate": 2.812583271253125e-05, "train/epoch": 0.3, "train/global_step": 3720, "_timestamp": 1712912198.9971228, "_runtime": 4554.489404678345, "_step": 185} |
{"train/loss": 3.6265, "train/grad_norm": 29.243453979492188, "train/learning_rate": 2.812583271253125e-05, "train/epoch": 0.3, "train/global_step": 3720, "_timestamp": 1712912198.9971228, "_runtime": 4554.489404678345, "_step": 185} |