Two Calls, Two Moments, and the Vote-Accuracy Curve of Repeated LLM Inference

• URL: http://arxiv.org/abs/2605.03379v2
• Date: Thu, 07 May 2026 02:23:28 GMT
• Title: Two Calls, Two Moments, and the Vote-Accuracy Curve of Repeated LLM Inference
• Abstract summary: We study the binary correctness layer of repeated LLM inference under conditional-i.i.d. calls.<n>Every fixed majority-vote budget has a sharp distribution-free two-call interval.<n>Experiments on LLM calls over QNLI and QQP show that empirical three- and five-vote accuracies are contained in the projected two-call regions.
• Score: 4.28787537081191
• License: http://creativecommons.org/licenses/by-sa/4.0/
• Abstract: Repeated sampling is a standard way to spend test-time compute, but its benefit is controlled by the latent distribution of correctness across examples, not by one-call accuracy alone. We study the binary correctness layer of repeated LLM inference under conditional-i.i.d. calls. One labeled call identifies the mean latent success probability; two labeled calls identify its second moment and hence the same-example correctness correlation that separates stable errors from recoverable call-level randomness. From these two moments, every fixed majority-vote budget has a sharp distribution-free two-call interval. The key technical reduction is that the infinite-dimensional moment problem has three-atom extremizers and quadratic dual certificates for every finite budget, so the bounds are exact rather than discretized or parametric. The first useful budget, three votes, has a closed form, width at most $1/8$, and a certified-improvement criterion. The infinite-vote endpoint is the limit of majority voting as the number of calls tends to infinity; it is also sharply bounded, but remains threshold-sensitive because it depends on latent mass around $q=1/2$. We add maximum-entropy and Latent-difficulty Gaussian-probit point completions, and experiments on LLM calls over QNLI and QQP show that empirical three- and five-vote accuracies are contained in the projected two-call regions while temperature changes and randomized model mixtures can create voting gains not ordered by one-call accuracy.

Related papers

• Constrained Online Learning with Noisy Constraint Values [55.29259818039367]
  Under common feasibility, our LEDGER algorithm attains $O(sqrt T)$ expected regret and $O(sqrtTlog(eT))$ expected budget violation.<n>Neither a Slater condition, independence between feedback channels, nor an absolute constraint-value bound is needed.
  arXiv  Detail & Related papers  (2026-09-07T01:38:41Z)
• Replicable Conformal Prediction [0.3058685580689604]
  Two analysts calibrate the same predictive model on independent samples every time.<n>No one can verify that two calibrations produced the same object.<n>We show that sharing a single random seed and rounding the calibrated threshold up to a coarse shared grid resolves the tension.
  arXiv  Detail & Related papers  (2026-08-23T17:47:40Z)
• Robust Average-Reward Markov Decision Processes: Minimax-Optimal Learning via Plug-in Reductions [51.50375419691955]
  Distributionally robust Markov decision processes provide a principled framework for sequential decision making under model uncertainty.<n>We study how many samples are necessary and sufficient to learn an $varepsilon$-optimal robust policy under the average-reward criterion.
  arXiv  Detail & Related papers  (2026-08-06T19:49:48Z)
• Squeezed-state semi-device-independent quantum randomness generation [0.0]
  This paper investigates semi-device-independent quantum randomness generation with a trusted binary pure-state source and an untrusted binary detector.<n>We derive a closed-form Shannon-rate expression for this setting, depending only on the trusted Gram overlap of the two source states.<n>We then apply the result to squeezed-coherent BPSK sources, showing how squeezing changes the trade-off between state distinguishability and certified randomness.
  arXiv  Detail & Related papers  (2026-06-02T16:54:29Z)
• When Are Trade-Off Functions Testable from Finite Samples? [4.706977254727891]
  We study finite-sample inference for the trade-off function of two unknown probability distributions.<n>By inverting the test, we also obtain simultaneous confidence bands for the whole trade-off curve.
  arXiv  Detail & Related papers  (2026-05-11T16:07:48Z)
• The Extrapolation Cliff in On-Policy Distillation of Near-Deterministic Structured Outputs [52.709361620508595]
  ListOPD brings a student to in-domain with an 8B-SFT baseline at one-fifth the parameters.<n>On Amazon Fashion, three pre-registered tests--a fine-grid cliff interval test, and a small-clip cross-prediction--fall within their locked prediction windows--with the small-clip value matching the closed-form prediction below grid resolution.
  arXiv  Detail & Related papers  (2026-05-09T06:48:00Z)
• Tail allocation for conformal prediction intervals [1.0163716169047106]
  We study split-conformal prediction for regression when the reported prediction set must be a single interval, at target marginal coverage $1-$, where $$ is the nominal miscoverage level.<n>Under this reporting constraint, the natural conditional target is the shortest interval with conditional mass at least $1-$, rather than an equal-tailed interval or a possibly disconnected high-probability set.<n>We parameterize this single-interval by a lower-tail allocation, which determines how the nominal miscoverage $$ is split between the two endpoints.
  arXiv  Detail & Related papers  (2026-04-28T04:14:27Z)
• Accelerating Quantum Materials Characterization: Hybrid Active Learning for Autonomous Spin Wave Spectroscopy [0.0]
  We present TAS-AI, a hybrid-to-physics-informed framework for autonomous triple-axis spin-wave spectroscopy.<n>In blind reconstruction benchmarks, model-agnostic methods such as random sampling, coarse grids, and Gaussian-process mappers reach a global error threshold more reliably and with fewer measurements than physics-informed planning.<n> TAS-AI reaches a decisive AIC-derived evidence ratio (>100) in fewer than 10 measurements, while motion-aware scheduling cuts wall-clock time by 32% at a fixed measurement budget.
  arXiv  Detail & Related papers  (2026-04-26T17:54:40Z)
• Correction and Corruption: A Two-Rate View of Error Flow in LLM Protocols [51.56484100374058]
  We propose a paired-outcome measurement interface for auditing a single protocol step on exact-match tasks.<n>For each instance, the interface records a baseline correctness bit and a post-step correctness bit.<n>These rates predict accuracy changes and define a reusable empirical interface testable across seeds, mixtures, and pipelines.
  arXiv  Detail & Related papers  (2026-04-20T13:25:40Z)
• Test-Time Scaling with Diffusion Language Models via Reward-Guided Stitching [66.39914384073145]
  We propose a self-consistency framework that turns cheap diffusion-sampled reasoning into a reusable pool of step-level candidates.<n>We find that step-level recombination is most beneficial on harder problems.<n>Our training-free framework improves average accuracy by up to 2 across six math and coding tasks.
  arXiv  Detail & Related papers  (2026-02-26T11:08:39Z)
• Sample Smart, Not Hard: Correctness-First Decoding for Better Reasoning in LLMs [72.82403830490084]
  We argue that the decoding rule should be calibrated by correctness, not confidence alone.<n>We propose simple strategies that achieve this goal: Greedy-Threshold makes sampling greedy at very low confidence steps.<n>Together, our findings challenge prevailings about decoding under uncertainty and show gains across math and general reasoning benchmarks.
  arXiv  Detail & Related papers  (2025-10-07T14:46:12Z)
• Ensemble Threshold Calibration for Stable Sensitivity Control [0.0]
  We present an end-to-end framework that achieves exact recall with sub-percent variance over tens of millions of geometry pairs.<n>Our approach consistently hits a recall target within a small error, decreases redundant verifications relative to other calibrations, and runs end-to-end on a single TPU v3 core.
  arXiv  Detail & Related papers  (2025-10-02T15:22:28Z)
• Local Discovery by Partitioning: Polynomial-Time Causal Discovery Around Exposure-Outcome Pairs [18.31538168213386]
  We propose local discovery by partitioning (LDP) for causal inference tasks. LDP is a constraint-based procedure that returns a VAS for an exposure-outcome pair under latent confounding. Adjustment sets from LDP yield less biased and more precise average treatment effect estimates than baseline discovery algorithms.
  arXiv  Detail & Related papers  (2023-10-25T14:53:10Z)
• Krotov Type Optimization of Coherent and Incoherent Controls for Open Two-Qubit Systems [77.34726150561087]
  This work considers two-qubit open quantum systems driven by coherent and incoherent controls. Incoherent control induces time-dependent decoherence rates via time-dependent spectral density of the environment. The system evolves according to the Gorini-Kossakowski-Sudarshan-Lindblad master equation with time-dependent coefficients.
  arXiv  Detail & Related papers  (2023-08-11T13:17:19Z)
• Optimization of Time-Dependent Decoherence Rates and Coherent Control for a Qutrit System [77.34726150561087]
  Incoherent control makes the decoherence rates depending on time in a specific controlled manner. We consider the problem of maximizing the Hilbert-Schmidt overlap between the system's final state $rho(T)$ and a given target state $rho_rm target.
  arXiv  Detail & Related papers  (2023-08-08T01:28:50Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.