Add FP8 dynamically quantized MNLP_M2_mcqa_model model using llm-compressor

initial commit