Mario Lezcano Casado

@pytorchbot merge

@pytorchbot merge

> (but rationals are bad for ValueRange strict typing). >_assert_bound_is_rational is no more, we no longer generate rational bounds This is bad. Using rationals to make inference, even if not...

Should I continue reviewing tomorrow or should I wait for CI to be green?

I think the top two prs in this stack were not intended. Make sure to keep a clean commit history before running ghstack.

@pytorchbot merge

@pytorchbot merge

it seems you'll also need to change a few other things here and there.

I left this aside for now, prioritising the mxfp stuff, as this is lower prio. It shouldn't take long too long to land tho.

It pretty much just needs to make the tests more resilient. Many tests rely on UB and things that should not be tested. I think it shouldn't be terribly difficult...