优化 MetaX 平台 HD64 GQA Paged Attention Decode 的 K/V 数据复用 - #1557

优化 MetaX 平台 HD64 GQA Paged Attention Decode 的 K/V 数据复用#1557

Qzong2020 wants to merge 2 commits into

InfiniTensor:InfiniLM-v0.2.9cInfiniTensor/InfiniCore:InfiniLM-v0.2.9cfrom

Qzong2020:metax-paged-attention-hd64-gqaQzong2020/InfiniCore:metax-paged-attention-hd64-gqaCopy head branch name to clipboard