Unverified50% confidenceFactExact time
A linear probe is a simple linear classifier trained on model activations, used to detect whether specific behaviors or concepts exist inside the model.
1
Sources
50%
Confidence
Long-term
Relevance
9/11/2026
First Seen
Sources
Related Entities
Related Claims
VerifiedLIME通过在目标样本周围构造扰动数据集，训练局部可解释的线性模型来近似复杂模型的局部行为69% similarUnverified执行轨迹记录模型或人类解决问题的完整中间步骤，使训练系统能识别推理链条中具体哪一步出错，从而提供更精细的学习信号68% similarUnverifiedCurrent large model evaluation procedures often provide models with complete questions and conditions, and conduct targeted training on specific question banks, making high scores easy to achieve.67% similarUnverified探针（Probing Classifier）通常是一个轻量级线性分类器，输入为模型中间层激活向量，输出为研究者关心的语义属性，探针回答模型是否知道而因果干预回答模型是否使用了该信息67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/899107API
curl https://kongchang.com/api/v1/knowledge/claims/899107MCP
get_claim(id=899107)