[Feature]: Packing format llm_compressor support INT4 (W4A16)

[noobHappylife]

Feature Description

Currently export format from doesn't support llm_compressor (compressed_tensor) if we are using INT4 (W4A16) quantization scheme.

Motivation and Use Case

Would be good to be able to export into llm_compressor format using INT4 (W4A16) scheme. We can already do that through llm_compressor + AutoRoundModifier. But now if I use autoround library directly with w4a16 scheme + packing-format llm_compressor, it's not working.

Alternatives Considered

No response

Definition of Done

No response

Additional Context

No response

[thuang6]

Thanks for raising the feature request, it was already in the plan.