Verified90% confidenceFactExact time
Sliding window attention reduces the complexity of attention computation from O(n²) to O(n·w), linearizing GPU memory overhead for long texts.
6
Sources
90%
Confidence
Long-term
Relevance
7/21/2026
First Seen
Sources
Related Claims
VerifiedInformer reduces the self-attention computation complexity from O(n²) to O(n log n) through ProbSparse sparse attention.84% similarUnverifiedLongformer、BigBird通过局部滑动窗口注意力加少量全局token的混合注意力策略，将计算复杂度从O(n²)降至O(n)84% similarUnverified滑动窗口注意力（如Mistral采用的方案）通过限制每个token只关注固定窗口内的历史，将复杂度从O(n²)降至O(n·w)82% similarUnverifiedThe computational complexity of standard self-attention is quadratic with respect to sequence length, and achieving a context of one million tokens requires introducing techniques such as sparse attention, sliding window attention, and KV Cache compression.79% similarUnverifiedLongformer由Allen AI提出，采用滑动窗口局部注意力加全局注意力token的组合策略，将复杂度从O(n²)降至O(n)79% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/580785API
curl https://kongchang.com/api/v1/knowledge/claims/580785MCP
get_claim(id=580785)