Accelerate LLM with low-bit (FP4 / INT4 / FP8 / INT8) optimizations using ipex-llm - View it on GitHub