claim
active
claim:the-heavy-cost-of-resampling-steps-in-reasoning-is-a-modeling-problem-noise-is-exactly-multinomial-and-curves-are-smooth-with-sparse-forks-so-large-budgets-mostly-re-measure-existing-structureThe heavy cost of resampling steps in reasoning is a modeling problem: noise is exactly multinomial and curves are smooth with sparse forks, so large budgets mostly re-measure existing structure.
Central interpretation.
Source paper
extracted_from(2026) · Eric Bigelow · Amir Zur · Satchel Grant · Tal Haklay +6
Neighborhood — ranked by edge-count
Papers (1)
paper
Findings (2)
finding
- Noise matches null.
- Intervals miscalibrated.
Questions (1)
question
- Central scientific question.
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.
- Main result 1.
- Combining spaced resampling with smoothing cuts total budget by 1/8 with small error increase.finding0.815Main result 3.
- Load-bearing abstract sentence stating the paper's central discovery.
- Optimal number of features scales faster than optimal number of training steps with compute budget.finding0.772Allocation result from scaling laws.
- A better basis (e.g., factoring tokens into languages or parts of speech) may yield a sparse interpretable modelhypothesis0.770Same model can be dense in one basis and sparse in another.
- Closing sentence of the abstract stating the paper's headline conclusion.
- Interpretation of Grok 4 vs Grok 4 Fast per-koan comparison