Why is PyTorch split into so many libraries? What's the point of these splits? What do Windows, mobile and CUDA have to do with the library splits?Further reading:<ul><li>c10 folder architecture description <a href="https://github.com/pytorch/pytorch/wiki/Software-Architecture-for-c10">https://github.com/.../wiki/Software-Architecture-for-c10</a></li><li>Implementation of the TORCH_API visibility macros <a href="https://github.com/pytorch/pytorch/blob/master/c10/macros/Export.h">https://github.com/.../blob/master/c10/macros/Export.h</a></li><li>An example of virtual call based hook to break library structure <a href="https://github.com/pytorch/pytorch/blob/master/c10/core/impl/DeviceGuardImplInterface.h">https://github.com/pytorch/pytorch/blob/master/c10/core/impl/DeviceGuardImplInterface.h</a></li></ul>

Vectorization

The PyTorch Developer Podcast is a place for the PyTorch dev team to do bite sized (10-20 min) topics about all sorts of internal development topics in PyTorch.