Texonom
Texonom
/
Engineering
Engineering
/Data Engineering/Artificial Intelligence/AI Problem/AI Alignment/Explainable AI/Interpretable AI/Mechanistic interpretability/Activation Engineering/Neuron SAE/SAE Feature/
SAE Feature Specificity
Search

SAE Feature Specificity

Creator
Creator
Seonglae Cho
Created
Created
2025 Feb 4 15:27
Editor
Editor
Seonglae Cho
Edited
Edited
2025 Feb 19 13:25
Refs
Refs
When the feature is activated, the related concept reliably exists in the context
  • Binary Detection
 
 
Scaling Monosemanticity: Extracting Interpretable Features from Claude 3 Sonnet
We find a diversity of highly abstract features. They both respond to and behaviorally cause abstract behaviors. Examples of features we find include features for famous people, features for countries and cities, and features tracking type signatures in code. Many features are multilingual (responding to the same concept across languages) and multimodal (responding to the same concept in both text and images), as well as encompassing both abstract and concrete instantiations of the same idea (such as code with security vulnerabilities, and abstract discussion of security vulnerabilities).
Scaling Monosemanticity: Extracting Interpretable Features from Claude 3 Sonnet
https://transformer-circuits.pub/2024/scaling-monosemanticity/index.html#assessing-tour-specificity
 
 

Recommendations

Texonom
Texonom
/
Engineering
Engineering
/Data Engineering/Artificial Intelligence/AI Problem/AI Alignment/Explainable AI/Interpretable AI/Mechanistic interpretability/Activation Engineering/Neuron SAE/SAE Feature/
SAE Feature Specificity
Copyright Seonglae Cho