测试时计算扩展是改进 LLM 推理的主轴。现有方法主要通过扩展单个推理轨迹来扩展深度。通过并行采样多个候选者来扩展广度很简单,，但引入了选择瓶颈:，在没有真实验证者的情况下选择最佳候选者,，因为逐点 LLM 判断存在噪音且存在偏见。为了解决这个,，我们引入了 OpenDeepThink, 一个基于群体的测试时计算框架，它通过成对的 Bradley-Terry 比较进行选择。每一代, 法学硕士都会随机对候选人进行判断，并通过 Bradley-Terry 将选票汇总为全球排名; 保留排名靠前的候选人，并使用比较过程中产生的自然语言批评来改变前四分之三的; 底部四分之一被丢弃。 OpenDeepThink 在八个连续的 LLM 通话轮次 (~27 分钟挂钟) 中将 Gemini 3.1 Pro的 的有效 Codeforces Elo 提高了 +405 点。管道在较弱和较强的模型之间传输，无需重新调整,，并且在多域 HLE 基准, 上，收益似乎集中在客观可验证的领域，而在主观领域则相反。我们发布了 CF-73,，这是一组精选的 73 个专家评级的 Codeforces 问题，带有国际大师注释和 99% 与官方裁决相反的本地评估协议。
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的 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.
科目:人工智能(cs.AI)
Subjects: Artificial Intelligence (cs.AI)