mlsanityData to 7 Oct 2026

Optimal Bayesian Stopping for Efficient Inference of Consistent LLM Answers

arXiv 2602.05395 ↗First posted 5 Feb 2026stat.ML cs.AI cs.LGcited by 7

Not on the map. Why?

Abstract (arXiv)

A simple strategy for improving LLM accuracy, especially in math and reasoning problems, is to sample multiple responses and submit the answer most consistently reached. In this paper we leverage Bayesian prior information to save on sampling costs, stopping once sufficient consistency is reached. Although the exact posterior is computationally intractable, we further introduce an efficient "L-aggregated" stopping policy that tracks only the L-1 most frequent answer counts. Theoretically, we prove that L=3 is all you need: this coarse approximation is sufficient to achieve asymptotic optimality, and strictly dominates prior-free baselines, while having a fast posterior computation. Empirically, this identifies the most consistent (i.e., mode) LLM answer using fewer samples, and can achieve similar answer accuracy while cutting the number of LLM calls (i.e., saving on LLM inference costs) by up to 50%.

Key facts. Cited by 7 papers in the map (data through 7 Oct 2026). Builds on 19 of its 25 references that are in the map. First posted on arXiv 5 Feb 2026.

Builds on

19 of 25 references.

References · 19

  1. 2020Language Models are Few-Shot Learnerscited by 32,199
  2. 2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Modelscited by 12,302
  3. 2022Self-Consistency Improves Chain of Thought Reasoning in Language Modelscited by 5,601
  4. 2022Large Language Models are Zero-Shot Reasonerscited by 4,323
  5. 2023Let's Sample Step by Step: Adaptive-Consistency for Efficient Reasoning and Coding with LLMscited by 111
  6. 2024Escape Sky-high Cost: Early-stopping Self-Consistency for Multi-step Reasoningcited by 105
  7. 2024Calibrating Large Language Models with Sample Consistencycited by 72
  8. 2024Reasoning Aware Self-Consistency: Leveraging Reasoning Paths for Efficient LLM Samplingcited by 63
  9. 2024Make Every Penny Count: Difficulty-Adaptive Self-Consistency for Cost-Efficient Reasoningcited by 61
  10. 2025Online Scheduling for LLM Inference with KV Cache Constraintscited by 17
  11. 2025Optimizing LLM Inference: Fluid-Guided Online Scheduling with Memory Constraintscited by 16
  12. 2025Sample Complexity and Representation Ability of Test-time Scaling Paradigmscited by 12
7 more
  1. 2025Optimal Self-Consistency for Efficient Reasoning with Large Language Modelscited by 8
  2. 2025BEACON: Bayesian Optimal Stopping for Efficient LLM Samplingcited by 6
  3. 2025CGES: Confidence-Guided Early Stopping for Efficient and Accurate Self-Consistencycited by 6
  4. 2019Sequential Mode Estimation with Oracle Queriescited by 5
  5. 2021PAC Mode Estimation using PPR Martingale Confidence Sequencescited by 5
  6. 2024Distributed Load Balancing with Workload-Dependent Service Ratescited by 2
  7. 2025FEval-TTC: Fair Evaluation Protocol for Test-Time Computecited by 1

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 7.

7 papers.

  1. 2026ATLAS: Agentic Test-time Learning-to-Allocate Scalingcited by 3
  2. 2026MARS: Margin-Adversarial Risk-controlled Stopping for Parallel LLM Test-time Scalingcited by 2
  3. 2026Adaptive Generate-Rank-Verify: Inference-Time Search with Costly Verificationcited by 1
  4. 2026How Many Posterior Samples? Calibrated Stopping for Adaptive Sensingcited by 0
  5. 2026Strategic Self-Consistencycited by 0
  6. 2026Adaptive Self-Consistency: From Black-Box Sampling to Distribution-Valued Feedbackcited by 0
  7. 2026From Discovery to Decision: Finite-Budget Recoverability in LLM Votingcited by 0