hypothesis
active
hypothesis:sampling-noise-decreases-in-proportion-to-1-s-giving-a-linear-relation-between-log-tvd-and-log-sSampling noise decreases in proportion to 1/√S, giving a linear relation between log TVD and log S.
Derived from multinomial sampling.
Source paper
extracted_from(2026) · Eric Bigelow · Amir Zur · Satchel Grant · Tal Haklay +6
Neighborhood — ranked by edge-count
Papers (1)
paper
Findings (5)
finding
- Log-log slope of −0.4903 of pooled replicate TVD vs S through S=200 on two development questions, near −1/2.associated_withsupportsNoise scaling.
- Noise matches null.
- Noise law.
- Noise scaling.
- Noise scaling.
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.
- At S=200 variation across runs (TVD 0.0451) nearly equals sampling variation within a run (TVD 0.0442).finding0.815Supports multinomial model.
- Fig. 7d,e.
- log P(M(α) complies) is approximately linear in α when probability is not close to 1 on Qwen 3 4B pathfinding0.774Empirical basis for the extrapolation.
- Main result 1.
- Robustness check ruling out that any perturbation would decrease type hint rate due to brittleness.
- Abstract claim.
- Quantifies the information gain from using logit-based expected value over greedy or sampled decoding
- Appendix B.