SAE Feature Steering

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

• SAE featuresconcept0.855
  The individual, supposedly monosemantic directions learned by SAEs; argued here to fragment manifolds into disconnected pieces.
• Experiment 2: SAE Deception Feature Steeringconcept0.836
  Tests whether deception- and roleplay-related features causally gate consciousness self-reports in LLaMA 3.3 70B
• SAE feature steering in history, conceptual, and zero-shot control conditions produces zero experience reports under either suppression or amplificationfinding0.787
  Shows gating effect is specific to the self-referential computational regime, not a general feature effect
• Feature steering (clamping feature activations)method0.772
  Modifying model behavior by clamping SAE feature activations to specific values during forward pass.
• Model Steeringconcept0.769
  Using interventions to guide model generation behavior, e.g., adding sentiment vectors at inference time
• Representation Steeringconcept0.750
  Parent concept; the practice of controlling neural network outputs by manipulating internal representations.
• If agentic self-steering evaluation proves robust, it might be used to better explain and interpret SAE features in generalhypothesis0.747
  Speculative claim about scaling introspective access to general SAE feature interpretation
• Stepwise steeringmethod0.747
  Novel method that applies intervention only when the model begins a new thinking step (at the \n\n delimiter) rather than at every token