FlashAttention (2022)
研究论文
系统
The kernel trick that made long context affordable.
您正在查看一条研究论文记录：一篇里程碑式的论文，以及它为该领域带来了什么。
Tri Dao's FlashAttention reordered attention computation around GPU memory hierarchy — exact results, several times faster, memory linear instead of quadratic. Versions 2 and 3 chase each hardware generation; virtually every serious training and inference stack ships it. Systems research as leverage: one algorithm changed the economics of every cluster.
它如何与整个技术栈相连
本记录的关联链接一览——点击分组可直接跳转。
来源
记录审阅截至 2026-01-15 — 查看 方法论.