Unverified50% confidenceFactExact time
SR-SWA constrains the active capacity of reference tokens, whereas ordinary sliding window attention typically operates on a positional window of adjacent tokens.
1
Sources
50%
Confidence
Long-term
Relevance
10/30/2026
First Seen
Sources
Related Entities
Related Claims
Unverified在自回归解码中，注意力层通常缓存所有历史 token 的 Key 和 Value（KV Cache），显存随序列长度线性增长，但每步只需追加、无需重算；SSM 层则只保留固定大小的状态。67% similarUnverifiedSWA bounded replay means replaying or retaining a bounded length of historical context outside the window under the sliding window attention mechanism, in order to maintain long-context consistency within a limited memory and compute budget.66% similarUnverifiedRAG (Retrieval-Augmented Generation) can effectively mitigate the inherent limitations of LLMs regarding limited context windows and knowledge cutoff dates.59% similarUnverifiedEffective guidance of DLM depends on the compatibility between the content of the control token and its placement position; mismatched anchors interfere with generation capability.59% similarVerifiedCurrent LLMs' memory is primarily limited by the context window, making it impossible to truly retain information across multiple independent conversations.59% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1113981API
curl https://kongchang.com/api/v1/knowledge/claims/1113981MCP
get_claim(id=1113981)