[Feature]: Support FP8 Marlin MoE for CompressedTensorsW8A8Fp8MoEMethod

[mgoin]

🚀 The feature, motivation and pitch

Like what was added in #16850 for enabling marlin in fp8.py MoE layers, we should enable FP8 Marlin MoE for compressed tensors models to support users wanting to run them on older hardware.

Basically you want to take the changes in fp8.py's moe method (https://github.com/vllm-project/vllm/pull/16850/files#diff-5511bfcc9c53f7d96517ad43e4087f6777bef21302da983f42cafae40a866644) and apply them to CompressedTensorsW8A8Fp8MoEMethod

Alternatives

No response

Additional context

No response

Before submitting a new issue...

• Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the documentation page, which can answer lots of frequently asked questions.

[gigit0000]

I’d like to give this a try if it’s available. Can I pick it up?

[gigit0000]

@mgoin Sorry for the nudge. Can I go ahead with this?

[mgoin]

My apologies for not clearing this up but we already resolved this issue. I really appreciate your interest though, and I’m happy to guide you to work on any related features if you have questions!

[mgoin]

Closed by #18026