It's great to see such excellent work! This work achieves good performance compared to other deep learning models. However, we found that the MAE and MSE metrics used for comparison were log-processed and not inversely normalized. Therefore, we would like to ask whether the inversely normalized results should be compared.