Unverified50% confidenceFactExact time
尽管上下文窗口很大，上下文腐化（Context Rot）依然是巨大问题，模型对上下文中不同位置信息的注意力分布不均匀（Lost in the Middle效应）
1
Sources
50%
Confidence
Long-term
Relevance
7/18/2026
First Seen
Sources
Related Claims
UnverifiedContext Rot refers to the phenomenon where, as the number of conversation turns increases, irrelevant information accumulates and the attention budget becomes filled with low-value content, causing reasoning quality to continuously decline.83% similarUnverified上下文腐烂（Context Rot）包含两种不同的故障模式：Distraction（注意力稀释）和Poisoning（因果链污染）80% similarUnverifiedContext Rot（上下文腐烂）指随着对话变长、输入 Token 增多，模型准确度会逐步下降，该现象在 Claude、OpenAI、Gemini 等主流模型上普遍存在78% similarUnverifiedRAG systems suffer from the 'Lost in the Middle' problem, where models tend to ignore information in the middle of a long context and favor the beginning and end.75% similarUnverifiedDespite Opus 4.7 already having a million-token context window, context rot remains a significant problem.74% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/552394API
curl https://kongchang.com/api/v1/knowledge/claims/552394MCP
get_claim(id=552394)