finding
active
finding:steering-vector-norm-jumps-after-first-shot-cosine-similarity-to-0-shot-vector-drops-suddenly-while-similarity-to-128-shot-vector-rises-slowlySteering vector norm jumps after first shot; cosine similarity to 0-shot vector drops suddenly while similarity to 128-shot vector rises slowly
Many-shot vector geometry.
Source paper
extracted_from(2025) · Eric Bigelow · Daniel Wurgaft · YingQiao Wang · Noah D. Goodman +3
Neighborhood — ranked by edge-count
Papers (1)
paper
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.
- Steering vectors computed over multiple shots and norm-normalized have weaker effect than 0-shot vectorsfinding0.865Counter-intuitive result.
- Distinct from internal emotion.
- Demonstrates averaging multiple prompt pairs reduces noise; optimal subset selection further improves performance.
- Why do steering vectors computed over multiple shots have a weaker effect than single-query vectors?question0.802Left to future work.
- Steering Vector Control maintains low unexpected rate of 0.08 in Experiment 1, comparable to baselinefinding0.801Shows that inducing deception via steering vectors preserves semantic coherence and does not cause random errors
- Evidence for two representational pathways based on cross-method activation divergence
- Experiment 3 comparison: zero-shot control shows lower semantic convergence than experimental condition
- Appendix E replication of DIM alignment finding in Qwen model