finding
active
finding:best-run-with-grep-and-semantic-search-discovered-all-52-behaviorsBest run with grep and semantic search discovered all 52 behaviors
Best-performing run, borderline root cause.
Source paper
extracted_from(2025) · Trenton Bricken · Rowan Wang · Sam Bowman
Neighborhood — ranked by edge-count
Papers (1)
paper
Claims (1)
claim
- Interpretability tools and semantic search are key to winning the Marks et al. auditing gamesupportsAffordance insight.
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.
- Grouping similar model behaviors; the unsupervised method surfaces clusters of concerning patterns.
- Prior work studying sycophancy and desire not to be shut down in RLHF-trained models
- Study showing RLHF can exacerbate self-preservation tendencies in LLMs; key empirical support for a paper claim
- GRU behavior can be compressed to as few as 4 dimensions using DAS and MAS with comparable IIAsfinding0.719Shows that behaviorally relevant information is low-dimensional; contrasted with model stitching achieving near-perfect IIA at rank 2.
- The central research question explored, by way of examples, throughout the paper.
- Pairwise A/B choices for 4032 pairs converted to Elo.