Unverified50% confidenceFactExact time
负面指令起到类似分类器自由引导（Classifier-Free Guidance）的作用，显式压低某些输出模式的概率
1
Sources
50%
Confidence
Long-term
Relevance
8/26/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedNegative constraints narrow the probability space by explicitly excluding undesired outputs, are equally important as positive instructions, and can reduce deviations caused by the model's creative improvisation.77% similarUnverified在代码生成Prompt中，否定性约束（说明不需要什么）能缩小模型搜索空间，比增加期望描述更能提高准确率71% similarUnverified对大语言模型而言，「不要做X」的负向约束有时比「要做Y」的正向指令更有效，因为负样本相当于在概率空间中设置了禁区70% similarUnverified覆盖率本质上是负向指标，能告诉你哪些代码没有被执行，却无法保证被执行的代码被正确断言69% similarUnverified研究表明模型在处理约束性指令时的准确率显著低于处理生成性指令，这被称为'约束遵循缺陷'68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/802828API
curl https://kongchang.com/api/v1/knowledge/claims/802828MCP
get_claim(id=802828)