convert : avoid dequantizing mxfp4 for GPT-OSS#16756

compilade merged 1 commit intomasterggml-org/llama.cpp:masterfrom

compilade/fix-prequant-mxfp4-gpt-ossggml-org/llama.cpp:compilade/fix-prequant-mxfp4-gpt-ossCopy head branch name to clipboard