Unverified50% confidenceFactExact time

反事实解释回答的问题是：最少需要改变输入的哪些特征，才能让模型的预测结果翻转

1

Sources

50%

Confidence

Long-term

Relevance

8/23/2026

First Seen

Sources

CounterDistill: An XAI Engineering Approach to Distilling Counterfactual Explanations into Global Rules

Related Entities

反事实解释concept

Related Claims

Unverified循环论证偏差指当特征本身隐含标签信息时模型看似准确率极高实则用答案预测答案75% similar Unverified模型基于前文预测后文，Prompt中任何一个字的变化都会改变后续生成概率，而标点符号的影响权重较小74% similar Unverified多项实验表明，模型的表面推理与其实际决策机制之间存在显著偏差，通过激活修补可发现真正影响输出的内部路径与书面推理不吻合73% similar Unverified要求模型列出反直觉假设、从批评者角度反驳主流观点或连接两个看似无关领域，可以激活模型更边缘的知识表示73% similar

Cite This Claim

Stable URI

https://kongchang.com/claim/790031

API

curl https://kongchang.com/api/v1/knowledge/claims/790031

MCP

get_claim(id=790031)