Verified90% confidenceFactTime unknown
The computational complexity of the standard Self-Attention mechanism has a quadratic relationship with sequence length (O(n²)).
28
Sources
90%
Confidence
Long-term
Relevance
6/1/2026
First Seen
Sources
Related Entities
Related Claims
Verified自注意力机制的计算复杂度随序列长度呈近似二次方增长，序列长度为n时注意力矩阵计算量正比于n²84% similarUnverifiedSelf-Attention机制计算复杂度与序列长度平方成正比（O(n²)），一个128K token请求的原生注意力计算量约为4K token请求的1024倍81% similarUnverified注意力机制的计算复杂度为O(n²)，其中n为上下文长度，Claude具有200K token上下文窗口81% similarVerifiedThe self-attention mechanism allows each position to interact directly with any other position in the sequence, with a computational complexity of O(n²).79% 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.78% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/17838API
curl https://kongchang.com/api/v1/knowledge/claims/17838MCP
get_claim(id=17838)