finding
active
finding:probe-predictions-are-systematically-miscalibrated-until-isotonic-regression-corrects-biasProbe predictions are systematically miscalibrated until isotonic regression corrects bias
Calibration necessity.
Source paper
extracted_from(2026) · Eric Bigelow · Raphaël Sarfati · Daniel Wurgaft · Owen Lewis +4
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.
- Hyperparameter tuning result for probes; consistent with Hewitt and Liang 2019 finding
- Key methodological insight: introspection enables a new probe validation criterion beyond conventional separation metrics
- Key interpretive question the framework helps address through convergent validation logic
- Key interpretive claim from Case Study II distinguishing probe accuracy from causal relevance
- Framework premise.
- Motivation for causal evaluation over purely behavioural probing accuracy
- Convergent validity logic applied to LLM interpretability; probes validate self-reports and vice versa
- Shows that truth representations are not reducible to text probability representations