claim
active
claim:interpretability-could-audit-models-for-concerning-hidden-thought-processes-by-distinguishing-faithful-from-unfaithful-cotInterpretability could audit models for concerning hidden thought processes by distinguishing faithful from unfaithful CoT.
Proof of concept.
Neighborhood — ranked by edge-count
Papers (1)
paper
Findings (1)
finding
- Mechanistic discrimination of CoT faithfulness.
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.
- Interpretation of §3.
- Forward-looking suggestion for how inner interpretability could extend the indicator method
- Authors' outlook.
- Practical implication.
- Generality of methods.
- Limitations.
- Theoretical framing establishing why CoT models are uniquely suited to exhibit strategic deception