{"train/loss": 2.778, "train/grad_norm": 0.0, "train/learning_rate": 0.0, "train/epoch": 0.07, "train/global_step": 3000, "_timestamp": 1712782518.9167209, "_runtime": 32701.13459777832, "_step": 100, "train_runtime": 32702.7018, "train_samples_per_second": 2.936, "train_steps_per_second": 0.092, "total_flos": 8.629801849965773e+16, "train_loss": 2.7816907374064126, "_wandb": {"runtime": 32705}}