Hi,Excellent work! However, i cannot find traing details in your paper. Could you give more training details such as hyper parameters or GPU memory?