method
active
method:correction-probeCorrection probe
Reward probe grading correction interventions (5-class).
Neighborhood — ranked by edge-count
Papers (1)
paper
Frameworks (1)
framework
- Core pipeline turning probe-read features into scalable reward functions for open-ended tasks.
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.
- Attention probe predicting whether an Entity is hallucinated; AUC .94 on validation.
- Causal transformer probe predicting whether a token is in the same Entity as its predecessor.
- Interpretability tools that decode information from internal model activations; here, linear probes are used for data attribution.
- Standard interpretability technique used to read feature beliefs.
- The adaptive, incremental nature of living process, allowing small steps with continuous evaluation and adjustment.
- Joint firing raises TPR at slightly higher FPR.
- Reasoning pattern where model reverses a compliance tendency and returns toward refusal, associated with better safety in QwQ-32B
- Reward probe grading retraction interventions.