##article.return## Layer-Wise High-Impact Parameter Ratio Optimization in Post-Training Quantization for Large Language Models Download Download PDF