the writter's Transformer torch.py 's position embed has some mistake.
zhangbo2008 opened this issue · 0 comments
zhangbo2008 commented
i fixed it in the following
https://github.com/zhangbo2008/nlp-tutorial/blob/master/2cpu_Input_myData.py
https://github.com/zhangbo2008/nlp-tutorial/blob/master/2gpu_Input_myData.py
if you find some mistake, please comment me