The purpose of this talk is to put a map in your hands: to tell you about the basic conceptual structure of a “tensor library that supports automatic differentiation”, and give you some tools and tricks for finding your way around the codebase.
tensor
strides,跟物理内存有关,跟实际元素的使用有关
Every tensor records an offset, but most of the time it’s zero, and I’ll omit it from my diagrams when that’s the case.
stride !=1 means we skip elements
tensor是一个logic view
extensions
分层,针对自己想要加的
autograd
source
按照这个层次和流程,通过pytorch operator来改写