I want to see the source code of “torch.cumsum”.
I want to understand how it is implemented and optimized.
I search the “pytorch/aten” fold, and print all files which contain the string “cumsum”.
This is the results:
/src/ATen/core/TensorMethods.h
/src/ATen/core/aten_interned_strings.h
/src/ATen/core/Type.h
/src/ATen/core/Tensor.h
/src/ATen/native/EmbeddingBag.cpp
/src/ATen/native/ReduceOps.cpp
I read all of those files. However, I still can’t find the implementation details of cumsum.