SAEVerbalizer: Generating Explanations for Sparse Autoencoder Features via Representation Verbalization
Sparse autoencoders (SAEs) are proposed to extract numerous features from large language model (LLM) representations, yet explaining these features still relies primarily on external observation. This reliance leads to superficial explanations inferred from observed model behavior and computational inefficiency from collecting such behavioral evidence at scale. We introduce SAEVerbalizer, a framework that injects SAE decoder directions into an LLM's representations and fine-tunes the LLM's downstream layers to generate natural-language explanations of the injected features. Once trained, the r
Lineage graph
Paper → model → repo connections mined from source citations (Tier-1 exact match).
Why these links exist
Every edge carries a method, confidence, and the source snippet that justified it — so bad links are debuggable.
- FuzzyOverlapping authors or contributors · 62%mudler/LocalAI →
“Shared author/contributor keys: guo”
- FuzzyOverlapping authors or contributors · 62%bytedance/deer-flow →
“Shared author/contributor keys: wang”
- FuzzyOverlapping authors or contributors · 62%modular/modular →
“Shared author/contributor keys: liu”
- FuzzyOverlapping authors or contributors · 62%ray-project/ray →
“Shared author/contributor keys: wang”
- LinkedLinked via arxiv author · 85%Weihan Meng →
“SAEVerbalizer: Generating Explanations for Sparse Autoencoder Features via Representation Verbalization”
- LinkedLinked via arxiv author · 85%Hongzhu Guo →
“SAEVerbalizer: Generating Explanations for Sparse Autoencoder Features via Representation Verbalization”
- LinkedLinked via arxiv author · 85%Yi Jing →
“SAEVerbalizer: Generating Explanations for Sparse Autoencoder Features via Representation Verbalization”
- LinkedLinked via arxiv author · 85%Dewen Liu →
“SAEVerbalizer: Generating Explanations for Sparse Autoencoder Features via Representation Verbalization”