🔥 remove auto-marked xfail for fp8, include fp8 tests by default, add xfail manually

[prashantgupta24]

Description

Fp8 should now be fully supported (almost). Some tests are failing due to output mismatch, those are marked xfail manually.

Related PRs

fix: static batching with FP8 - #457
and
fix: tests for graph comparison with FP8 - #462

[github-actions]

👋 Hi! Thank you for contributing to vLLM support on Spyre.
Just a reminder: Make sure that your code passes all the linting checks, otherwise your PR won't be able to be merged. To do so, first install the linting requirements, then run format.sh and commit the changes. This can be done with uv directly:

uv sync --frozen --group lint --active --inexact

Or this can be done with pip:

uv pip compile --group lint > requirements-lint.txt
pip install -r requirements-lint.txt
bash format.sh

Now you are good to go 🚀

[maxdebayser]

bot:test

[joerunde]

@prashantgupta24 I see we also still have _skip_quantized_by_default- do you think it's time to remove that too?

If we're confident in the fp8 tests (except the ones manually marked xfail here) then it may be less confusing to have them all run by default. We would have to update our pytest invocations though to e.g. skip fp8 on most of the test matrix here in GHA since the cpus are too slow to run the fp8 models

[prashantgupta24]

@prashantgupta24 I see we also still have _skip_quantized_by_default- do you think it's time to remove that too?

If we're confident in the fp8 tests (except the ones manually marked xfail here) then it may be less confusing to have them all run by default. We would have to update our pytest invocations though to e.g. skip fp8 on most of the test matrix here in GHA since the cpus are too slow to run the fp8 models

Trying now...

[wallashss]

LPGTM thanks!

[prashantgupta24]

bot:test

[prashantgupta24]

bot:test

[prashantgupta24]

bot:test