用配对比较提升大模型推理，让多个答案竞争选优。
OpenDeepThink: Parallel Reasoning via Bradley-Terry Aggregation
- 通过两两对比投票生成全局排名，避免单个判断偏差。
- 在代码竞赛中使模型有效评分提升405点，8轮后完成。
- 无需调参即可跨模型通用，适合需要可靠推理的场景。
测试时计算扩展是提升大模型推理能力的关键途径。现有方法主要通过延长单一推理路径来增加深度，而并行生成多个候选答案虽简单，却面临选择瓶颈：缺乏真实验证时，大模型逐个评判候选答案易受噪声和偏见影响。为此，我们提出 OpenDeepThink，一种基于种群的测试时计算框架，采用两两比较的 Bradley-Terry 聚合机制进行选择。每轮中，大模型随机比较候选答案对，通过 Bradley-Terry 汇总投票得到全局排名；保留排名靠前的四分之三，并用对比过程中生成的自然语言评述对其变异；淘汰最差四分之一。该方法在八轮连续大模型调用（约27分钟实际时间）内，使 Gemini 3.1 Pro 在 Codeforces Elo 上有效提升 +405 点。该流程无需重新调整即可迁移至更弱或更强模型，在多领域 HLE 基准上，增益集中在可客观验证的领域，而在主观领域则出现逆转。我们发布了 CF-73，一个经专家评级的 73 道 Codeforces 题目集合，包含国际大师标注及 99% 本地评估一致性与官方判题结果一致。
原文摘要 · Abstract (English)
Test-time compute scaling is a primary axis for improving LLM reasoning. Existing methods primarily scale depth by extending a single reasoning trace. Scaling breadth by sampling multiple candidates in parallel is straightforward, but introduces a selection bottleneck: choosing the best candidate without a ground-truth verifier, since pointwise LLM judging is noisy and biased. To address this, we introduce OpenDeepThink, a population-based test-time compute framework that selects via pairwise Bradley-Terry comparison. Each generation, the LLM judges random pairs of candidates and aggregates votes via Bradley-Terry into a global ranking; top-ranked candidates are preserved and the top three quarters are mutated using the natural-language critiques produced during comparison; the bottom quarter is discarded. OpenDeepThink raises Gemini 3.1 Pro's effective Codeforces Elo by +405 points in eight sequential LLM-call rounds (~27 minutes wall-clock). The pipeline transfers across weaker and stronger models without retuning, and on the multi-domain HLE benchmark, gains appear concentrated in objectively verifiable domains and reverse in subjective ones. We release CF-73, a curated set of 73 expert-rated Codeforces problems with International Grandmaster annotation and 99% local-evaluation agreement against the official verdict.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。