Fetching the paper…

SKVQ: Sliding-window Key and Value Cache Quantization for Large Language Models · Around