Feature request

EETQ quantized model perform with very good quality in my case, but the loading is pretty slow. So that if the base model is quantized with EETQ already, LoRAX should load it directly without the JIT quantization, but currently will failed to find related layers.

Motivation

Speed up the EETQ model loading speed.

Your contribution

I will prepare a PR for a review, also I need some help with the implementation in someplace.