Joint Consistency: A Unified Test-Time Aggregation Framework via Energy Minimization
On the map: LLM reasoning → Test-time compute and efficient thinking
Abstract (arXiv)
This paper studies test-time aggregation, an approach that generates multiple reasoning traces and aggregates them into a final answer. Most existing methods rely on evaluation signals collected from candidate traces in isolation or answer frequencies, while ignoring comparative interactions among candidates. We propose Joint Consistency (JC), formulated as a constrained Ising-type energy minimization problem, where independent evaluation signals act as external fields and pairwise comparisons act as interactions. JC provides a unified framework for test-time aggregation that subsumes existing voting and weighted aggregation methods as special cases. Our construction of the interaction matrix leverages LLM-as-a-judge comparisons, and admits a theoretical interpretation under answer-level homogeneity assumptions. Moreover, we develop an efficient approximation strategy that makes interaction modeling practical for large-scale test-time aggregation. Experiments on math and code reasoning benchmarks show that JC consistently outperforms existing baselines across tasks, judge models, trace budgets, and trace-generation settings.
Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in Test-time compute and efficient thinking, one of 17 directions in LLM reasoning. Builds on 41 of its 50 references that are in the map. First posted on arXiv 7 May 2026.
Builds on
41 of 50 references.
Closest ancestors · 17
Same direction, newest first
- 2025Deep Think with Confidence21 Aug
- 2025Inference-Time Scaling for Complex Tasks: Where We Stand and What Lies Ahead31 Mar
- 2025Scalable Best-of-N Selection for Large Language Models via Self-Certainty25 Feb
- 2025Confidence Improves Self-Consistency in LLMs10 Feb
- 2025Sample, Scrutinize and Scale: Effective Inference-Time Search by Scaling Verification3 Feb
- 2025SETS: Leveraging Self-Verification and Self-Correction for Improved Test-Time Scaling31 Jan
- 2025Evolving Deeper LLM Thinking17 Jan
- 2024Semantic Self-Consistency: Enhancing Language Model Reasoning via Semantic Weighting10 Oct
- 2024LLM Self-Correction with DeCRIM: Decompose, Critique, and Refine for Enhanced Following of Instructions with Multiple Constraints9 Oct
- 2024Reasoning Aware Self-Consistency: Leveraging Reasoning Paths for Efficient LLM Sampling30 Aug
- 2024Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameters6 Aug
- 2023Alphazero-like Tree-Search can Guide Large Language Model Decoding and Training29 Sep
5 more
- 2023CRITIC: Large Language Models Can Self-Correct with Tool-Interactive Critiquing19 May
- 2023Tree of Thoughts: Deliberate Problem Solving with Large Language Models17 May
- 2023Large Language Model Guided Tree-of-Thought15 May
- 2023Self-Refine: Iterative Refinement with Self-Feedback30 Mar
- 2022Language Models (Mostly) Know What They Know11 Jul
Other references · 24
- 2022Training language models to follow instructions with human feedbackcited by 15,712
- 2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Modelscited by 12,302
- 2021Training Verifiers to Solve Math Word Problemscited by 9,743
- 2023Judging LLM-as-a-Judge with MT-Bench and Chatbot Arenacited by 8,393
- 2025Qwen3 Technical Reportcited by 8,195
- 2022Self-Consistency Improves Chain of Thought Reasoning in Language Modelscited by 5,601
- 2017Deep reinforcement learning from human preferencescited by 3,846
- 2024A Survey on LLM-as-a-Judgecited by 1,384
- 2023Can LLMs Express Their Uncertainty? An Empirical Evaluation of Confidence Elicitation in LLMscited by 946
- 2024Large Language Monkeys: Scaling Inference Compute with Repeated Samplingcited by 946
- 2026DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligencecited by 877
- 2023Reasoning with Language Model is Planning with World Modelcited by 846
12 more
- 2023Just Ask for Calibration: Strategies for Eliciting Calibrated Confidence Scores from Language Models Fine-Tuned with Human Feedbackcited by 788
- 2022Large Language Models are Better Reasoners with Self-Verificationcited by 304
- 2024CRUXEval: A Benchmark for Code Reasoning, Understanding and Executioncited by 294
- 2025MathArena: Evaluating LLMs on Uncontaminated Math Competitionscited by 275
- 2013Ising formulations of many NP problemscited by 204
- 2024Recursive Introspection: Teaching Language Model Agents How to Self-Improvecited by 162
- 2024From Decoding to Meta-Generation: Inference-time Algorithms for Large Language Modelscited by 136
- 2025Is Best-of-N the Best of Them? Coverage, Scaling, and Optimality in Inference-Time Alignmentcited by 91
- 2025Evaluating Judges as Evaluators: The JETTS Benchmark of LLM-as-Judges as Test-Time Scaling Evaluatorscited by 42
- 2025How Do Large Language Monkeys Get Their Power (Laws)?cited by 40
- 2024Regularized Best-of-N Sampling with Minimum Bayes Risk Objective for Language Model Alignmentcited by 24
- 2025Logical Reasoning with Outcome Reward Models for Test-Time Scalingcited by 3
Built on by
Not cited yet.
Similar, not cited
Close in wording, no citation either way
- 2026Efficient Test-Time Scaling via Temporal Reasoning Aggregation
- 2026When to Vote, When to Rewrite: Disagreement-Guided Strategy Routing for Test-Time Scaling
- 2025Representation Consistency for Accurate and Coherent LLM Answer Aggregation
- 2025Revisiting Self-Consistency from Dynamic Distributional Alignment Perspective on Answer Aggregation
- 2026Selection, Recombination, or a Fresh Solve? A Candidate-Free Control for Single-Pass Test-Time Aggregation
- 2026Divergent-Convergent Reasoning: Scaling Test-Time Compute through Structured Solution Synthesis
- 2025The Majority is not always right: RL training for solution aggregation
- 2026ARBITER: Reasoning Trajectory Basins and Majority Vote Failures in Test-Time Sampling