Unverified50% confidenceFactExact time
长上下文遗忘（Lost in the Middle）是当前Transformer架构的固有局限
1
Sources
50%
Confidence
Long-term
Relevance
7/6/2026
First Seen
Sources
Meta Open-Sources Astryx: The First Design System Built for AI Agents
githubGitHub Trending7/5/2026
Related Claims
UnverifiedTransformer models exhibit a "Lost in the Middle" phenomenon, where attention to content located in the middle of the context and far from the current generation position systematically decreases.80% similarUnverifiedThe current Transformer architecture has a fundamental limitation in handling implicit constraints scattered across ultra-long contexts, leading to the introduction of hidden bugs when modifying code.79% similarUnverifiedLLM在长序列生成中因注意力稀释而遗忘之前约束的现象被称为「长上下文遗忘（Lost in the Middle）」，是当前Transformer架构的固有局限78% similarUnverified斯坦福大学2023年的研究将Transformer在处理超长上下文时对中间位置信息遗忘率显著高于首尾位置的现象命名为'Lost in the Middle'问题75% similarUnverifiedTransformer模型存在位置偏差，对中间位置信息的检索能力尤其薄弱，即lost in the middle问题72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/121500API
curl https://kongchang.com/api/v1/knowledge/claims/121500MCP
get_claim(id=121500)