Fetching the paper…

Norm Tweaking: High-performance Low-bit Quantization of Large Language Models · Around