pytorch

3186e366 - Support `0`s in `out_size` of `FractionalMaxPoolNd`

Support `0`s in `out_size` of `FractionalMaxPoolNd` Fixes #73624 CUDA implementation was correct :), only CPU had an out of bounds memory access Pull Request resolved: https://github.com/pytorch/pytorch/pull/73634 Approved by: jbschlosser