question
active
question:do-uninformative-chain-of-thought-tokens-also-improve-performanceDo uninformative chain of thought tokens also improve performance?
Filler tokens question.
Source paper
extracted_from(2023) · Tamera Lanham · Anna Chen · Ansh Radhakrishnan · Benoit Steiner +26
Neighborhood — ranked by edge-count
Papers (1)
paper
Findings (1)
finding
- Evidence against test-time compute hypothesis.
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.
- A small number of high-quality human demonstrations of chain-of-thought reasoning could be used to improve and focus performance.hypothesis0.798Section 6 mentions high-quality human demos could improve natural language feedback.
- Wait, we don't need to Wait! Removing thinking tokens improves reasoning efficiency (Wang et al., 2025a)concept0.778Baseline comparison method paper for reflection token suppression
- CoT improves accuracy on HHH evals and makes the decision process legible.
- under what conditions does chain-of-thought reflect genuine uncertainty resolution versus a learned performance?question0.757Key question addressed by the task difficulty analysis comparing MMLU and GPQA-Diamond
- Gains on unsupervised proxy metrics do not reliably translate to better practical performanceclaim0.751Core thesis.
- Mechanism for how the model modulates representation strength.
- Poisoning results.
- Central conclusion.