Results 1 issues of Guohe

hi, @taokong ,great work! I want to re-implement your experiment, and I have three small issues? 1. What is the normal train loss when the model converge well? mark: When...