Verified70% confidenceFactTime unknown
Context pollution refers to irrelevant or harmful information occupying the limited context window space of a model, causing the model's attention to be diverted and reasoning quality to degrade.
4
Sources
70%
Confidence
Long-term
Relevance
6/3/2026
First Seen
Sources
Related Entities
Related Claims
VerifiedContext pollution causes the model's attention to become scattered, response quality to decline, and costs to spiral out of control.85% similarVerifiedIrrelevant information occupying context space (context pollution) can significantly degrade the model's performance on core tasks.80% similarUnverifiedWhen multiple unrelated tasks share the same context, it leads to context contamination, degrading the quality of model outputs.76% similarUnverified过度膨胀的上下文会稀释模型注意力、抬高推理成本，甚至引入噪声导致幻觉72% similarVerified研究表明当上下文过长时，模型对中间部分信息的注意力会显著下降，即"Lost in the Middle"现象71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/39147API
curl https://kongchang.com/api/v1/knowledge/claims/39147MCP
get_claim(id=39147)