arxiv OliVe: Accelerating Large Language Models via Hardware-friendly Outlier-Victim Pair Quantization

阅 读

Star 0