finding
active
finding:at-threshold-0-7-monitoring-achieves-85-precision-56-recall-on-inline-and-88-precision-61-recall-on-not-inline-interventionsAt threshold 0.7, monitoring achieves .85 precision/.56 recall on inline and .88 precision/.61 recall on not-inline interventions
Test-time probe metrics.
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
- Design justification.
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.
- Benchmark identification rate rises from 0.08 (SFT) to 0.25 (DPO); interventions barely reduce itfinding0.764Table 11.
- Human data fine-tuning effect is distinct from synthetic emergent misalignment and likely caused by off-policy training
- Binary detection adjusted accuracy reaches 97.3% at layer 0 with α=5 before baseline control is appliedfinding0.762The misleadingly high result that prior paradigm would report as evidence of introspection
- Comparative prediction motivating future work contrasting different approaches to LLM self-knowledge
- More cautious claims.
- Base corrections: 72.9% Failed Fix, 14.7% Fixed; retractions: 69.4% Correct Retract, 30.0% Incorrect Retractfinding0.760Label distributions.
- Task-dependent optimal sparsity.
- Design principle with implications for AI and consciousness-UX; architectural requirement for self-directed cognition.