Unverified60% confidenceFactTime unknown
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
2
Sources
60%
Confidence
Long-term
Relevance
5/31/2026
First Seen
Sources
Related Entities
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.81% similarVerified斯坦福大学2023年发表了研究论文《Lost in the Middle》，实验证明当关键信息被放置在长上下文的中间位置时，模型的检索准确率会大幅下降80% 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78% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/3348API
curl https://kongchang.com/api/v1/knowledge/claims/3348MCP
get_claim(id=3348)