Technical Blog6 min readSmoothQuant: Accurate and Efficient Post-Training Quantization for Large Language Models