tracked author

Xunhao Lai (赖勋豪)

First author of MiniMax Sparse Attention. Homepage identifies him as a Peking University master's student working on long-context models and efficient attention; GitHub links the same homepage and FlexPrefill / sparse-attention projects. No high-confidence personal X account was found.

1 archived notes X: not-found HomepageGitHub

Related Notes

按论文归档时间排序,展示该作者在本站已经出现的材料。

Jun 24, 2026

MiniMax Sparse Attention

MiniMax Sparse Attention (MSA) 是 MiniMax M3 的长上下文技术核心:它在 GQA (Grouped Query Attention,分组查询注意力) 之上增加轻量 Index Branch,为每个 query token 和每个 GQA group 选择少量 KV blocks,再由 Main Branch 对被选 block 做精确 softmax attention;论文的强证据在于 109...

2606.13392-minimax-sparse-attention-m3 Sparse AttentionLong ContextMoE Architecture