AI 2-Bit VPTQ: 6.5x Smaller LLMs While Preserving 95% Accuracy

Сверху Снизу