question
active
question:do-latents-for-precise-distinctions-in-types-of-behaviour-have-distinct-steering-effectsDo latents for precise distinctions in types of behaviour have distinct steering effects?
Authors have not verified this.
Neighborhood — ranked by edge-count
Papers (1)
paper
Findings (1)
finding
- Feature cluster examples from the UMAP.
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.
- Nuanced interpretive claim about the limits of steering as a mechanism for reflection enhancement.
- Mechanism claim supported by transcript analysis and the fact that the steering vector was extracted from a model that never writes type hints.
- Core question motivating the shift from linear to geometry-aware steering; answered via manifold alignment analysis.
- Supported by the instruction discovery experiments comparing steering vs. embedding baselines.
- Central claim motivating the product.
- Off-target steering effect vanishes for concepts outside the top-level cluster in Emotions and Genresfinding0.773Hierarchy explains entanglement.
- Central interpretive claim organizing the entire paper's results
- Steering changes behavior regardless of input x, as a constant shift in log posterior oddsclaim0.772Derived from LRH.