{ "epoch": 112.0, "learning_rate": 1.0000000000000002e-07, "total_flos": 6.926864611971372e+20, "train_loss": 0.8085850866355923, "train_runtime": 128547.5283, "train_samples_per_second": 123.643, "train_steps_per_second": 3.865 }