Unverified50% confidenceTradeoffExact time
Many linear attention or sparse attention methods lose the ability to precisely retrieve historical information while compressing computation.
1
Sources
50%
Confidence
Long-term
Relevance
10/10/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedStuffing large amounts of historical records into a prompt not only incurs token costs, but may also cause important information to be drowned out due to attention dilution, producing worse results than a curated short summary.74% similarUnverified斯帕罗的实验发现，当人们预期能在网上找到信息时，对信息本身的记忆编码会显著减弱，但对去哪里找到信息的记忆会增强72% similarVerifiedWhen conversation history exceeds the context window limit, platforms use engineering methods such as truncation and summary compression to trim the context, a process that inevitably loses detail information.72% similarUnverifiedSummary-based memory compresses original interactions into summaries, causing irreversible information loss.72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/995072API
curl https://kongchang.com/api/v1/knowledge/claims/995072MCP
get_claim(id=995072)