Richard Zou

[functorch] vjpvjp inplace testing

1

Stack from [ghstack](https://github.com/ezyang/ghstack): * __->__ #83119 * #83114 * #83077 * #83106 * #83105 * #83080 Test Plan: - run tests

[functorch] in-place testing for test_vjp

1

Stack from [ghstack](https://github.com/ezyang/ghstack): * #83119 * __->__ #83114 * #83077 * #83106 * #83105 * #83080 This is relatively simple; we just test that `input.clone().inplace_(...)` gives us the correct gradients...

[functorch] in-place jvp testing

1

Stack from [ghstack](https://github.com/ezyang/ghstack): * #83119 * #83114 * __->__ #83077 * #83106 * #83105 * #83080 Testing code is starting to trend toward entropy. Not sure if the in-place tests...

[functorch] remove some testing hacks

1

Stack from [ghstack](https://github.com/ezyang/ghstack): * __->__ #83079 * #83073 I'm not sure why I called this a hack in the first place (perhaps I wanted to use tree_map and pytrees didn't...

Move vmap's OrderedDict pytree support to torch.utils._pytree

1

Stack from [ghstack](https://github.com/ezyang/ghstack): * #83079 * __->__ #83073 There's no reason why it should just apply when you import vmap Test Plan: - added a new test

Reland "[functorch] add error inputs check to vmap test"

1

Stack from [ghstack](https://github.com/ezyang/ghstack): * #83119 * #83114 * #83077 * __->__ #83106 * #83105 * #83080 Reland of #83017. It was reverted because it was stacked on top of a...

Fix error_inputs for linalg.lstsq; assert SampleInput args are tuple, take2

1

Stack from [ghstack](https://github.com/ezyang/ghstack): * #83119 * #83114 * #83077 * #83106 * __->__ #83105 * #83080 The previous merge attempt failed some trunk tests. Test Plan: - wait for tests

autograd.Function is silently incorrect on functorch transforms

11

We should fix it. Not quite sure how though. https://gist.github.com/zou3519/73a6189e21561f6ef5b42874e8a4826f

vmap and forward-mode AD fail sometimes on in-place views

9

## The Problem ```py import torch from functorch import jvp, vmap from functools import partial B = 2 def f(x, y): x = x.clone() view = x[0] x.copy_(y) return view,...

This happens a lot to us, we should add a test to prevent it from happening. Test probably needs to shell out