ARBITER: Reasoning Trajectory Basins and Majority Vote Failures in Test-Time Sampling

• URL: http://arxiv.org/abs/2605.26172v1
• Date: Mon, 25 May 2026 04:07:44 GMT
• Title: ARBITER: Reasoning Trajectory Basins and Majority Vote Failures in Test-Time Sampling
• Abstract summary: ARBITER is a model-agnostic approach that models interactions between basins using only the base model's own sampled outputs, hidden states, and derived evidence.<n>It improves accuracy from the mid-78% range to the mid-82% range, recovering about 22% of the available oracle headroom.
• Score: 3.2669068153591687
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: When language models use test-time sampling, they generate multiple reasoning trajectories and select an answer by majority vote. We show that these trajectories are not independent: for a given question, they concentrate into a small number of clusters, or reasoning basins, each defined by a normalized final answer and the solutions that reach it. A majority vote therefore selects the most stable basin rather than the most accurate one, which creates wrong-majority failures where the correct answer is present but outvoted. We introduce ARBITER, a model-agnostic approach that models interactions between basins using only the base model's own sampled outputs, hidden states, and derived evidence. Most direct correction strategies fail; ARBITER instead uses conservative additive evidence on top of consensus. In its simplest parameter-free form, ARBITER-Δ adds same-model evidence to the majority prior, while ARBITER-Enc augments this with bounded residual signals from hidden states over complete solutions. On GSM8K with Qwen3-4B, consensus over K=24 samples achieves around the mid-94% range, while a same-pool top-2 oracle reaches around the mid-96% range. ARBITER recovers a subset of these cases using zero external information. Across three model families and three math benchmarks, it yields consistent gains with no net-negative cases; for example, on Llama-3.1-8B MMLU-HS-Math, it improves accuracy from the mid-78% range to the mid-82% range, recovering about 22% of the available oracle headroom, indicating that this headroom can be partially recovered from the sample pool itself.

Related papers

• When Many Answers Are Valid, Voting Fails: Symbolic Verification for Best-of-K Causal Reasoning in LLMs [42.32694162421078]
  We introduce CALVER, a training-free symbolic verifier that scores structured traces against Pearl's causal criteria.<n>On CLEAR find-one-valid queries that admit multiple graph-valid answers, CALVER reaches 42.1% where plurality, a reward model, an LLM judge, and model confidence remain near 30%.
  arXiv  Detail & Related papers  (2026-08-04T11:45:46Z)
• Oracle Gap and Signal Fidelity: A Fixed-Pool Diagnostic for Test-Time Collaboration [8.515185888009833]
  Test-time collaboration is often credited with improving LLM reasoning, yet its gains are uneven and sometimes negative.<n>For a fixed candidate pool, we decompose a selector or verifier's net gain into measurable factors.<n>This reframes collaboration as a candidate-selection problem rather than as an intrinsic property of a multi-agent topology.
  arXiv  Detail & Related papers  (2026-07-20T04:09:19Z)
• Disagreement-Based Cross-Model Routing for Implicit Video Question Answering [0.0]
  We study multiple-choice video question answering on the ImplicitQA benchmark.<n>On this benchmark a single frontier video LLM already operates near its accuracy ceiling.<n>We propose disagreement-based cross-model routing, a pure inference-time procedure that requires no labels and no training.
  arXiv  Detail & Related papers  (2026-05-31T05:56:27Z)
• Design and Evaluation of Multi-Agent AI Oracle Systems for Prediction Market Resolution [0.0]
  Prediction markets aggregate collective intelligence to forecast uncertain events.<n>Existing oracle systems tradeoff fast but brittle automation against accurate but costly human arbitration.<n>We evaluate whether multi-agent LLM architectures can improve oracle resolution accuracy over single-model baselines.
  arXiv  Detail & Related papers  (2026-05-29T03:44:19Z)
• Self-Verified Distillation: Your Language Model Is Secretly Its Own Synthetic Data Pipeline [56.53954182896384]
  We propose Self-Verified Distillation, a simple post-training refinement algorithm for large language models.<n>Self-Verified Distillation generates candidate solutions to unlabeled seed questions.<n>It filters them using prompt-based self-verification, and trains on the resulting self-curated dataset.<n>We find that sampling more candidate generations and using a larger verification budget during training data construction produces higher-quality self-curated data.
  arXiv  Detail & Related papers  (2026-05-20T17:26:10Z)
• 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)
• Beyond Pass@k: Breadth-Depth Metrics for Reasoning Boundaries [2.9807229517491827]
  We propose Cover@tau, which measures the fraction of problems that a model can solve for which at least a tau proportion of completions are correct.<n>Unlike Pass@k, Cover@tau captures reasoning under an explicit reliability threshold.<n>We evaluate several RLVR models using Cover@tau-based metrics and illustrate how the relative rankings of popular algorithms change compared to Pass@1.
  arXiv  Detail & Related papers  (2025-10-09T15:14:58Z)
• DISCO: Diversifying Sample Condensation for Efficient Model Evaluation [59.01400190971061]
  Costly evaluation reduces inclusivity, slows the cycle of innovation, and worsens environmental impact.<n>We argue that promoting diversity among samples is not essential; what matters is to select samples thatmaximise diversity in model responses.<n>Our method, $textbfDiversifying Sample Condensation (DISCO)$, selects the top-k samples with the greatest model disagreements.
  arXiv  Detail & Related papers  (2025-10-09T08:53:59Z)
• Critique to Verify: Accurate and Honest Test-Time Scaling with RL-Trained Verifiers [63.99316853136304]
  Mirror-Critique is a framework that trains a verifier with informative critiques.<n>We deploy a small instruction-tuned model to synthesize high-quality critique data.<n>The resulting Mirror-Verifier is deployed to evaluate candidate solutions by generating multiple critiques per solution.
  arXiv  Detail & Related papers  (2025-09-27T06:50:24Z)
• Internalizing Self-Consistency in Language Models: Multi-Agent Consensus Alignment [22.305033366660187]
  Language Models (LMs) are inconsistent reasoners, often generating contradictory responses to identical prompts.<n>We formalize self-consistency as an intrinsic property of well-aligned reasoning models and introduce Multi-Agent Consensus Alignment (MACA)<n>MACA enables agents to teach themselves to be more decisive and concise, and better leverage peer insights in multi-agent settings without external supervision.
  arXiv  Detail & Related papers  (2025-09-18T17:27:28Z)
• The Majority is not always right: RL training for solution aggregation [53.1050856072799]
  We train an aggregator model to review, reconcile, and synthesize a final, correct answer.<n>A key ingredient is careful balancing of easy and hard training examples.<n>We find our method, AggLM, outperforms both strong rule-based and reward-model baselines.
  arXiv  Detail & Related papers  (2025-09-08T16:39:38Z)
• Not All Votes Count! Programs as Verifiers Improve Self-Consistency of Language Models for Math Reasoning [24.386388107656334]
  We introduce Prove, a framework that leverages translated programs derived from natural language solutions as a verification mechanism.<n>Unlike vanilla majority voting, our approach filters out solutions whose corresponding program output is inconsistent with the generated solution, aggregating only those that pass verification.<n>Our results show that Prove consistently outperforms vanilla majority voting for solving mathematical reasoning tasks across all model sizes and datasets.
  arXiv  Detail & Related papers  (2024-10-16T14:24:55Z)
• DCR: Divide-and-Conquer Reasoning for Multi-choice Question Answering with LLMs [9.561022942046279]
  We propose Divide and Conquer Reasoning (DCR) to enhance the reasoning capability of large language models (LLMs) We first categorize questions into two subsets based on confidence score ($mathcalCS$), which is estimated by statistical frequency of generated answers. In particular, we first categorize questions into two subsets based on confidence score ($mathcalCS$), which is estimated by statistical frequency of generated answers.
  arXiv  Detail & Related papers  (2024-01-10T14:38:46Z)
• IBADR: an Iterative Bias-Aware Dataset Refinement Framework for Debiasing NLU models [52.03761198830643]
  We propose IBADR, an Iterative Bias-Aware dataset Refinement framework. We first train a shallow model to quantify the bias degree of samples in the pool. Then, we pair each sample with a bias indicator representing its bias degree, and use these extended samples to train a sample generator. In this way, this generator can effectively learn the correspondence relationship between bias indicators and samples.
  arXiv  Detail & Related papers  (2023-11-01T04:50:38Z)

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.