finding
active
finding:edge-weight-intervention-correlations-decrease-with-conversation-lengthEdge-weight/intervention correlations decrease with conversation length
Figure 6.
Source paper
extracted_from(2026) · Kevin Der · Harish Kamath · Thompson, Ben
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.
- Edge weight sign matches observed intervention effect ~100% of the time, minimum 95% on any graphfinding0.79536,000 interventions.
- Figure 7.
- Discriminant validity: composite scores are not reducible to verbosity
- Left to future work.
- Provides discriminant evidence: if battery rewarded verbosity, prompted responses should be longer
- Demonstrates reflection redundancy in stronger model on harder math benchmark
- Interest introspection improves from 1B to 3B: ρ from 0.19 to 0.80, R² from 0.05 to 0.66finding0.720Largest single-step scaling improvement; demonstrates dramatic introspection gain between 1B and 3B models for interest
- Wellbeing introspection improves from 1B to 3B: ρ from 0.48 to 0.66, R² from 0.26 to 0.45finding0.718Confirms scaling trend for wellbeing concept between smallest and middle model size