Unverified50% confidenceFactExact time
'Lost in the Middle' refers to the tendency of models to focus on content at the beginning and end of the context while ignoring the middle portion.
1
Sources
50%
Confidence
Long-term
Relevance
9/8/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedIn extremely long contexts, the model's attention gradually dilutes, which is the technical root cause of AI gradually forgetting the requirements stated at the very beginning.61% similarUnverified在叙事场景中，传统基于语义相似度的检索策略容易遗漏因果隐含的关键节点，因为伏笔与爆发点在向量空间中相似度不一定高59% similarUnverified专项优化的AI模型在特定领域可能表现不及通用基础模型，这与深度学习中的灾难性遗忘现象有关55% similarVerifiedThe core challenge of internalizing knowledge into model parameters is the Catastrophic Forgetting problem, which is a difficulty not yet fully resolved in the field of continual learning.55% similarUnverifiedIf a model can learn directly from a single real-world demonstration, it largely bypasses the sim-to-real gap problem.55% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/879847API
curl https://kongchang.com/api/v1/knowledge/claims/879847MCP
get_claim(id=879847)