Samantha Andow

Fails with error: `RuntimeError: output with shape [5, 10, 5] doesn't match the broadcast shape [3, 5, 10, 5]`

Original conversation: https://github.com/pytorch/functorch/pull/818#discussion_r878187686 Right now, all the functions that do forward over reverse testing using a decompositions don't change dtypes (if we remove the lines that turn all the jacobian...

The decomposition of layer_norm_backward returns None. When AOTAutograd encounters this, `torch_dispatch` will convert that to an undefined Tensor, getting correct behavior. However, TorchScript will still return a None and doesn't...

Add flake8 pre commit hook script

9

[PyTorch's pre commit hooks](https://github.com/pytorch/pytorch/blob/master/tools/git-pre-commit), scripts that are called are in tools Really I selfishly just want the flake8 ones so I don't have to remember to run it against my...

Currently, we're seeing that even after PRs are landed internally (and have a related commit added to the fbsync branch, the original PR stays open, [example](https://github.com/pytorch/functorch/pull/784)). This is confusing for...

test_make_fx_no_decompose is failing on master

1

Currently covered with a skip, [previously](https://app.circleci.com/pipelines/github/pytorch/functorch/2743/workflows/ceaa99f9-e14e-4984-b99b-05d7ecbf5f50/jobs/18159) it was infinitely recursing on the prim decomposition

[Modes] remove enable and rewrite mode stack (squashed)

6

Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * __->__ #84774 Based on @ezyang's suggestion, mode stack now has "one true mode" which is the _only_ mode that can ever be active...

This adds a rewrite of the simple functorch notebook to use torch dispatch modes. In doing this, we needed to add padding to vmap (i.e. we need to make `vmap(f)(torch.randn(2),...