concept
active
concept:cot-faithfulnessCoT faithfulness
Whether stated reasoning accurately represents the reasoning process behind the prediction.
Neighborhood — ranked by edge-count
Papers (1)
paper
- Measuring Faithfulness in Chain-of-Thought Reasoningcitesmentions
Thinkers (1)
thinker
- Tamera Lanhamstudies
Methods (1)
method
- Chain-of-thought promptingassociated_withTechnique by which LLMs generate intermediate reasoning steps before final output; used by ChatGPT o3.
Concepts (1)
concept
- Encoded reasoning (steganography)associated_withInformation hidden in CoT phrasing not understandable by humans.
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.
- it seems hard to get an exact estimate of CoT faithfulness without solving interpretabilityquote0.842Limit on measurement.
- Framing of defense-in-depth.
- From decoupling of AOC and accuracy gain.
- Mechanism explanation.
- The condition that commitments are fulfilled.
- Low faithfulness.
- In motivated reasoning the CoT derives from the human hint, but not in faithful reasoning.hypothesis0.768Tested by varying the suggested answer.
- Idea that decomposed components must causally reflect the model's computation; enforced by adversarial ablation in VPD.