Created by: yanmeizhao
报错信息是:NaN or Inf found in input tensor 此时的学习率已经调到0.0001,训练很慢了,但是还会出现NaN,请问该怎么处理这个问题?