skills.rest
jleechanorg
jleechanorgOFFICIAL
Explore all skills by this author →
GitHub

sparse-autoencoder-training

Train and analyze Sparse Autoencoders with SAELens to decompose neural activations.

Updated Jun 28, 2026
Search Tags:#superposition#feature discovery#sae#sparse autoencoders#mechanistic interpretability#saelens#transformer activations
One-click install
npx skills add https://github.com/jleechanorg/hermes-agent --skill sparse-autoencoder-training-jleechanorg
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: sparse-autoencoder-training
Source: https://github.com/jleechanorg/hermes-agent/tree/main/optional-skills/mlops/saelens
Command: npx skills add https://github.com/jleechanorg/hermes-agent --skill sparse-autoencoder-training-jleechanorg

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

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

Core Features & Use Cases

  • Training SAEs with SAELens to expand and compress model activations for interpretability
  • Analyzing sparse features to identify monosemantic concepts and study superposition
  • Research workflows for transformer activations, feature attribution, and steering ablations

Quick Start

Configure a training run with LanguageModelSAERunnerConfig and execute SAETrainingRunner to train, evaluate, and save a trained SAE.

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 sparse autoencoders to find interpretable features in transformer activations?▼

To train sparse autoencoders for feature discovery, you configure a training run using LanguageModelSAERunnerConfig and execute SAETrainingRunner to train, evaluate, and save a model that decomposes neural activations into sparse, interpretable features.

Can I use SAELens with TransformerLens to analyze mechanistic interpretability in language models?▼

Yes, you can use SAELens with TransformerLens to study mechanistic interpretability. The Skill provides integration patterns for analyzing transformer activations, performing feature attribution, and conducting steering ablations on language models.

Does SAELens provide pre-trained sparse autoencoders for analyzing transformer models?▼

Yes, SAELens provides access to available pre-trained SAEs. These pre-trained models allow you to immediately analyze sparse features and study monosemantic representations without needing to train a sparse autoencoder from scratch.

What are the limitations of using sparse autoencoders for feature discovery in neural networks?▼

Sparse autoencoders for feature discovery require careful configuration of training parameters to effectively expand and compress model activations. Researchers must analyze the resulting sparse features to validate monosemantic representations and ensure accurate superposition analysis.

Related Skills

Explore More Education & Research Skills→
dawsonblockdawsonblock

sparse-autoencoder-training

Train sparse autoencoders on transformer activations with SAELens.

Community
Advanced
SignmanalSignmanal
1

sparse-autoencoder-training

Decompose polysemantic transformer activations into sparse interpretable features with SAELens.

Community
Advanced
GregsGreyCodeGregsGreyCode
6

sparse-autoencoder-training

Train and analyze sparse autoencoders on transformer activations using SAELens.

Community
Advanced
AissenLiuAissenLiu

sparse-autoencoder-training

Train and analyze Sparse Autoencoders with SAELens on transformer activations.

Community
Advanced