{ "epoch": 111.0, "learning_rate": 1.0000000000000002e-06, "total_flos": 4.1153080208666034e+20, "train_loss": 0.14764341744551096, "train_runtime": 198162.6547, "train_samples_per_second": 6.598, "train_steps_per_second": 0.207 }