Unverified50% confidenceFactExact time
DSA (DeepSeek sparse attention) reduces the computational complexity of attention from quadratic to approximately linear through sparsification.
1
Sources
50%
Confidence
Long-term
Relevance
10/31/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedIn linear attention, kernel functions are used to decompose softmax attention into an associable form, reducing inference computational complexity from quadratic in sequence length to linear.79% similarVerifiedOptimization techniques such as Flash Attention and sparse attention can compress memory complexity to near-linear.79% similarVerifiedThe 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.75% similarUnverified标准Attention的显存消耗随序列长度平方增长，Flash Attention将其压缩到线性增长72% similarUnverifiedCompared with hand-written attention, SDPA avoids explicitly materializing the intermediate attention matrix, thereby reducing memory usage and improving speed.72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1120234API
curl https://kongchang.com/api/v1/knowledge/claims/1120234MCP
get_claim(id=1120234)