Unverified50% confidenceSolutionExact time
对大语言模型而言，不要做X的约束有时比要做Y的指令更有效，因为模型在开放式创作任务中更容易被负样本锚定
1
Sources
50%
Confidence
Long-term
Relevance
7/20/2026
First Seen
Sources
Related Claims
UnverifiedLarge language models cannot reliably distinguish the boundary between system instructions and user/external content during inference—this is a fundamental limitation of prompt injection, similar in principle to SQL injection.72% similarUnverifiedLarge language models have inherent non-determinism — even with identical prompt inputs, the model may return results that differ in format, content, or even logic at different times.70% similarUnverified大型语言模型在无约束空间中会倾向于创造性发散，而非工程所需的收敛执行70% similarUnverifiedPerplexity measures a language model's ability to predict test text; the lower the value, the less surprised the model is by the text.70% similarVerifiedCurrent large language models exhibit a pattern in the systems programming domain of being strong on high-frequency knowledge and weak on long-tail knowledge, tending to make reasonable inferences rather than strictly following specifications for edge-case semantics and version-specific compatibility behavior.69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/570020API
curl https://kongchang.com/api/v1/knowledge/claims/570020MCP
get_claim(id=570020)