[Feature request] Eager mode support for a8w4dq

[mikekgfb]

https://github.com/pytorch/torchchat/edit/main/docs/quantization.md

Does a8w4dq also work for eager mode generate for testing before export without Executorch?

cc: @digantdesai @kimishpatel

[kimishpatel]

I think I tested it and it did work as in it ran. I added this to CI https://github.com/pytorch/torchchat/blob/main/.github/workflows/pull.yml#L300 so hoping it is not really broken. Will check and make sure

[digantdesai]

I am assuming if new q/dq are working in eager mode then it should.

[mikekgfb]

@swolchok working with @digantdesai to implement. Thank you!

[digantdesai]

I am assuming if new q/dq are working in eager mode then it should.

I should have clarified, "in a non-performant way" :)