Unverified50% confidenceCautionExact time
在长对话中早期的错误尝试和调试过程会持续占据上下文窗口，造成上下文污染并干扰后续推理准确性
1
Sources
50%
Confidence
Long-term
Relevance
7/15/2026
First Seen
Sources
Related Claims
UnverifiedThe drawback of the context condensation strategy is that it may lose details deemed unimportant, leading to misunderstandings or repeated mistakes in subsequent conversations.78% similarUnverifiedOverly verbose, contradictory, or ambiguous system prompts cause the model's attention to become scattered, resulting in over-thinking or instruction-following failures; streamlining agents.md can improve the determinism of instruction following.76% similarUnverifiedThe KV Cache, which grows dynamically during inference, continuously occupies VRAM and may suddenly trigger an OOM crash during long conversations.76% similarUnverifiedWhen prompts are too long, large models experience an 'attention explosion' phenomenon, where excessively long inputs cause effective information to be lost.76% similarUnverifiedError cascading refers to small judgment deviations in early steps being progressively amplified as the reasoning chain extends, ultimately producing results far from what was intended.75% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/523959API
curl https://kongchang.com/api/v1/knowledge/claims/523959MCP
get_claim(id=523959)