hypothesis
active
hypothesis:stylistic-elements-of-the-cot-sample-elicit-higher-quality-answers-by-prompting-the-model-to-imitate-a-more-thoughtful-agentStylistic elements of the CoT sample elicit higher-quality answers by prompting the model to imitate a more thoughtful agent.
Fourth hypothesis not investigated here.
Source paper
extracted_from(2023) · Tamera Lanham · Anna Chen · Ansh Radhakrishnan · Benoit Steiner +26
Neighborhood — ranked by edge-count
Papers (1)
paper
- Measuring Faithfulness in Chain-of-Thought Reasoningassociated_withmentions
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.
- Question resolved by the correlation between inflection points and probe-detected belief shifts
- Theoretical framing establishing why CoT models are uniquely suited to exhibit strategic deception
- Interpretation of early answering/mistakes results.
- The central empirical claim of the paper, supported by activation probing evidence
- Practical implication.
- In motivated reasoning the CoT derives from the human hint, but not in faithful reasoning.hypothesis0.769Tested by varying the suggested answer.
- All models exhibit above-baseline representation of the think word when instructed to think about itfinding0.769In the intentional control experiment, all tested models show above-zero cosine similarity to the think word's concept vector.
- Empirical finding from creative writing experiments showing VS variants achieve higher diversity without sacrificing quality