the writter's Transformer torch.py 's position embed has some mistake.

[zhangbo2008]

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