skills.rest
founderphantom
founderphantomCOMMUNITY
Explore all skills by this author →
GitHub

sparse-autoencoder-training

Train and analyze Sparse Autoencoders with SAELens to decompose activations into interpretable features.

Updated Apr 1, 2026
Search Tags:#training#feature-discovery#mechanistic-interpretability#sae#transformer-lens#sae-lens#sparse-autoencoder
One-click install
npx skills add https://github.com/founderphantom/zola-agent --skill sparse-autoencoder-training-founderphantom
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: sparse-autoencoder-training
Source: https://github.com/founderphantom/zola-agent/tree/main/optional-skills/mlops/saelens
Command: npx skills add https://github.com/founderphantom/zola-agent --skill sparse-autoencoder-training-founderphantom

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Provides guidance for training and analyzing Sparse Autoencoders (SAEs) using SAELens to decompose neural network activations into interpretable features. Use when discovering interpretable features, analyzing superposition, or studying monosemantic representations in language models.

Core Features & Use Cases

  • Train and evaluate SAEs with configurable architectures and sparsity settings
  • Analyze learned features to interpret model activations and superposition
  • Apply SAE-based steering and ablation experiments to model behavior

Quick Start

Train and analyze a Sparse Autoencoder using SAELens to decompose activations into interpretable features.

Frequently Asked Questions about sparse-autoencoder-training

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I train a sparse autoencoder to find interpretable features in a language model?▼

Train a sparse autoencoder using SAELens to decompose neural network activations into interpretable features. You can configure architectures and sparsity settings to analyze superposition and discover monosemantic representations.

What is the best way to analyze model activations and superposition with SAEs?▼

Analyze model activations and superposition by training sparse autoencoders with SAELens. The Skill evaluates learned features to interpret monosemantic representations within transformer language models.

Can I use TransformerLens with SAELens for mechanistic interpretability research?▼

Yes, SAELens works with TransformerLens to support mechanistic interpretability research. You can train sparse autoencoders and apply SAE-based steering and ablation experiments to study model behavior.

How do I configure sparse autoencoder training hyperparameters for feature discovery?▼

Configure sparse autoencoder training by adjusting architectures and sparsity settings in SAELens. These hyperparameters control how the model decomposes activations to reveal specific interpretable features.

When should I use sparse autoencoders for debugging model activations?▼

Use sparse autoencoders for debugging model activations when you need to study monosemantic representations or analyze superposition. SAELens helps decompose these activations to reveal distinct interpretable features.

Related Skills

Explore More Software Engineering Skills→
GregsGreyCodeGregsGreyCode
6

sparse-autoencoder-training

Train and analyze sparse autoencoders on transformer activations using SAELens.

Community
Advanced
SignmanalSignmanal
1

sparse-autoencoder-training

Decompose polysemantic transformer activations into sparse interpretable features with SAELens.

Community
Advanced
sheawinklersheawinkler
78

sparse-autoencoder-training

Train sparse autoencoders on transformer activations using SAELens configurations.

Community
Advanced
zerwizzerwiz

sparse-autoencoder-training

Train and analyze sparse autoencoders on transformer activations with SAELens.

Community
Advanced