concept
active
concept:interpretability-driven-steeringInterpretability-driven steering
General approach of using interpretability feedback to steer model generation.
Neighborhood — ranked by edge-count
Methods (1)
method
- Self-Correcting SearchimplementsTechnique using internal model representations as feedback loops to steer diffusion-based materials generation toward target properties.
Concepts (2)
concept
- Framework of using internal-state representations to control or steer generative models; conceptually parallel to manifold steering in language models.
- Manifold Steeringassociated_withCentral framework: steering neural networks by intervening along the curved manifold where a concept lives, rather than in straight lines through activation space.
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.
- Paradigm of finding the right direction in activation space (e.g., linear steering).
- The capability to explain model predictions; a central theme of the paper, with disruption profiles as vehicle.
- Parent concept; the practice of controlling neural network outputs by manipulating internal representations.
- Central interpretive claim organizing the entire paper's results
- Method using large language models (Claude) to generate and test explanations of features at scale
- Novel method that applies intervention only when the model begins a new thinking step (at the \n\n delimiter) rather than at every token
- Using interventions to guide model generation behavior, e.g., adding sentiment vectors at inference time
- The central phenomenon introduced by this paper: inference-time recovery from irrelevant activation steering in LLMs