question
active
question:why-do-steering-vectors-computed-over-multiple-shots-have-a-weaker-effect-than-single-query-vectorsWhy do steering vectors computed over multiple shots have a weaker effect than single-query vectors?
Left to future work.
Source paper
extracted_from(2025) · Eric Bigelow · Daniel Wurgaft · YingQiao Wang · Noah D. Goodman +3
Neighborhood — ranked by edge-count
Papers (1)
paper
Findings (1)
finding
- Counter-intuitive result.
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.
- Observation from 100% accuracy on specific concept-layer-strength combinations suggesting concept-specific detectability
- Validates that steering vectors capture reflection semantics by finding tokens reported in related work.
- Demonstrates averaging multiple prompt pairs reduces noise; optimal subset selection further improves performance.
- Empirical comparison showing advantage of SAE features in low-data regime.
- Supported by the instruction discovery experiments comparing steering vs. embedding baselines.
- Many-shot vector geometry.
- Priority claim establishing novelty of the Head Cor intervention approach
- Open question arising from the 100% accuracy on specific concept-layer-strength combinations