{"train/loss": 4.4875, "train/grad_norm": 2.1512844562530518, "train/learning_rate": 1.5833333333333336e-06, "train/epoch": 0.0, "train/global_step": 20, "_timestamp": 1713074516.2810621, "_runtime": 65.4094672203064, "_step": 0}