Interpretability and mechanistic analysis · MLD-1169
Steering vector
A direction in activation space used to alter a model's behavior when added or otherwise applied.
Original reference page · detailed expansion pending. Original definitions, examples, notes and citations are preserved below. New examples are illustrative, not benchmark results.
Original example
Shift an internal representation along a studied feature.
Original caveat
A vector's effect depends on layer, scale, and context.
Context and aliases
Level: Advanced. Stages: Research · Evaluation.
Tags: ML · DL · SLM · LLM · Interpretability · Safety.
Aliases: None recorded.
Related concepts
Concepts linking here
Original references
- Captum model interpretability documentation ↗ · Reference link
- Tracing the thoughts of a large language model ↗ · Consulted
- Towards Monosemanticity: Decomposing Language Models With Dictionary Learning ↗ · Consulted
- SHAP documentation ↗ · Reference link
- Original reference URL ↗