EMWQ: An Efficient Mixed Precision Weight Quantization Method for Large Language Models | IEEE Journals & Magazine | IEEE Xplore