Interpretability and mechanistic analysis · MLD-1139
Sparse autoencoder
An autoencoder trained with sparse internal activations, often used to discover features in model representations.
Original reference page · detailed expansion pending. Original definitions, examples, notes and citations are preserved below. New examples are illustrative, not benchmark results.
Original example
Decompose hidden states into sparse feature activations.
Original caveat
Recovered features can be incomplete, split, or hard to interpret.
Context and aliases
Level: Advanced. Stages: Research · Evaluation.
Tags: ML · DL · SLM · LLM · Interpretability · Safety.
Aliases: SAE.