Samantha Andow

Thanks for the thorough writeup @Benjamin-eecs! To check a couple things, is this a current bottleneck for your examples? We had been under the assumption that the training would be...

> Also, in our specific usage where we want to repeat the same training process with different seeds or hyperparamters using functorch.vmap, we think it would be better if user...

From looking at this a bit, I think what happened is: - pytorch changes unsqueeze + mm -> mv in case where first element is vector - this example hits...

Yep! I have changes for all of these functions locally so let me pick them onto the new repo and put up some PRs

Not quite! Discussed a bit offline but @AlphaBetaGamma96 I'm hoping to get https://github.com/pytorch/pytorch/pull/82814 in soon and then I'll double check that we see speedups for your example (thanks for the...

Hi @AlphaBetaGamma96! Just wanted to let you know that linalg_solve just landed https://github.com/pytorch/pytorch/pull/82814. I tested locally that this example ran significantly faster after the fix than before the fix (exact...

Not a silly question, there's been a lot of change in the past couple of weeks. Yes the main development of functorch is being done in pytorch/pytorch. So if you're...

@pop-cultural-reference could you share your code?

@pop-cultural-reference Thanks for the context! Could you also help me understand what sort of matrix you're trying to get in the end? It's a little confusing to me since the...

Full Stack Trace ``` Traceback (most recent call last): File "/home/circleci/project/env/lib/python3.8/site-packages/torch/testing/_internal/common_device_type.py", line 389, in instantiated_test raise rte File "/home/circleci/project/env/lib/python3.8/site-packages/torch/testing/_internal/common_device_type.py", line 376, in instantiated_test result = test(self, **param_kwargs) File "test/test_pythonkey.py", line...