finding
active
finding:french-word-end-feature-3013-harmful-weights-remain-intermingled-with-helpful-across-high-effectiveness-rangeFrench word-end feature 3013: harmful weights remain intermingled with helpful across high-effectiveness range
Effectiveness doesn't fully clean.
Source paper
extracted_fromNicholas L. Turner · Jeffrey Wu · Joshua Batson
Neighborhood — ranked by edge-count
Papers (1)
paper
Claims (1)
claim
- Further sharpening of the effectiveness metric is not high-leverage compared to finding better basessupportsHelpfulness is direct measure and Fisher already prunes similarly.
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.
- Three regimes by effectiveness.
- Helpful and harmful weights are scattered across the entire range of virtual weight magnitudesfinding0.755IN Tokens→Logits shows weak VW-helpfulness relationship.
- Operation component is language-independent.
- Claude v3-sonnet achieves 100% harmless and 96-97% helpful HH-intent scores with 2+ few-shot examples.finding0.737Numerical result from Table 3 for Claude sonnet.
- QK attribution of antonym heads.
- Discussion.
- Interpretation of motivating example.
- Motivating example.