concept
active
concept:model-self-repairModel Self-Repair
Later layers correct single-layer steering interventions.
Neighborhood — ranked by edge-count
Papers (2)
paper
Concepts (1)
concept
- Steering Entanglementassociated_withSteering one concept shifts beliefs in related unintended concepts.
Related by similarity (8)
cosine ≥ 0.65 · no typed edgeEntities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.
- Ability of a model to predict its own outputs or behavior, sometimes distinguished from introspection.
- The transformer's model of itself as a predictive text engine, developed through in-context learning.
- Technique to measure representational compatibility by integrating intermediate representations of one model into another
- Robots capable of building internal models of their own body and unexpected changes, blurring the embodied/non-embodied AI distinction
- Metzinger's concept of the self as a sustained representation; distinguished from consciousness itself in CIMC's framework
- A representation that captures relevant aspects of a system; according to the theorem, the regulator must embody this.
- Edits MLP weights for all layers to modify model behavior; used by Abdelnabi & Salem to decrease verbalized evaluation awareness.