finding
active
finding:classification-probe-auc-0-94-on-validation-monitoring-auc-88-for-localizationClassification probe AUC 0.94 on validation; monitoring AUC .88 for localization
Probe quality on held-out set.
Source paper
extracted_from(2026) · Aaditya Vikram Prasad · Connor Watts · Jack Merullo · Dhruvil Gala +3
Neighborhood — ranked by edge-count
Papers (1)
paper
Claims (1)
claim
- Calibration enables repurposing.
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.
- Demonstrates the sharp drop in factual truth generalization at the counting boundary.
- Causal transformer probe predicting whether a token is in the same Entity as its predecessor.
- Attention probe predicting whether an Entity is hallucinated; AUC .94 on validation.
- Hypothesis generation.
- Discovery claim.
- Key improvement in cross-task generalization enabled by explicit instruction framing.
- Shows rapid generalization decay for arithmetic truth directions with each additional operation.
- Demonstrates that early-layer probes capture sentence polarity rather than truth.