finding
active
finding:turn-averaged-features-have-highest-embedding-similarity-0-663-at-k-3-per-token-max-lowest-at-0-617Turn-averaged features have highest embedding similarity 0.663 at k=3; per-token max lowest at 0.617
Consistent across k=1,2,3,5.
Source paper
extracted_from(2026) · Kevin Der · Harish Kamath · Thompson, Ben
Neighborhood — ranked by edge-count
Papers (1)
paper
Claims (1)
claim
- Training SAEs on mean hidden state yields features capturing high-level turn characteristics (topic, style, behavior) rather than lexical patternsassociated_withsupportsCentral claim.
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.
- Pairwise coverage.
- 5-way ranking.
- Prediction from coverage results.
- Figure 1.
- 20,000 trials.
- Example A.
- §5.5.
- LongBench v2 example D.