PyTorch ops to native function mapper

I am looking for a mapping table between PyTorch ops like (sigmoid, narrow, lstm, etc) to its native function. I agree there are some ops which invoke multiple native functions like
empty_strided → empty and stride. Where can I have a detailed table where I can see the mapping. In other words, if I have a log of native functions how can I reverse map to it’s corresponding pytorch operator

I don’t think there exists a mapping as the internal implementations could change and could also dispatch to different operations/methods for different backends. The best way would be to check the implementation in the source code and see which operations are used.