Texonom
Texonom
/
Engineering
Engineering
/Data Engineering/Artificial Intelligence/AI Risk/AI Alignment/Explainable AI/Interpretable AI/Mechanistic interpretability/Activation Engineering/Activation Decomposition/SAE/
SAE as an Embedding
Search

SAE as an Embedding

Creator
Creator
Seonglae ChoSeonglae Cho
Created
Created
2025 Dec 18 16:22
Editor
Editor
Seonglae ChoSeonglae Cho
Edited
Edited
2025 Dec 18 16:30
Refs
Refs
SAE Probing

Interpretable Embedding

Usually max pooling

 
 
 
 
 
 

Max pooling

arxiv.org
https://arxiv.org/pdf/2508.12535

data classification

aclanthology.org
https://aclanthology.org/2025.emnlp-main.1521.pdf
SAE as a
Text embedding
→
Data Classification
Dataset Diffing
arxiv.org
https://arxiv.org/pdf/2512.10092
 
 

Table of Contents
Interpretable EmbeddingUsually max poolingMax poolingdata classification

Recommendations

Texonom
Texonom
/
Engineering
Engineering
/Data Engineering/Artificial Intelligence/AI Risk/AI Alignment/Explainable AI/Interpretable AI/Mechanistic interpretability/Activation Engineering/Activation Decomposition/SAE/
SAE as an Embedding
Copyright Seonglae Cho