{"train/loss": 3.4266, "train/grad_norm": 32.5091667175293, "train/learning_rate": 2.775776814817928e-05, "train/epoch": 0.32, "train/global_step": 4020, "_timestamp": 1712912563.941801, "_runtime": 4919.434082984924, "_step": 200} |
{"train/loss": 3.4266, "train/grad_norm": 32.5091667175293, "train/learning_rate": 2.775776814817928e-05, "train/epoch": 0.32, "train/global_step": 4020, "_timestamp": 1712912563.941801, "_runtime": 4919.434082984924, "_step": 200} |