后训练语言模型以在编程竞赛中取得金牌级表现

摘要

竞技编程已成为检验大语言模型推理能力的关键测试，其中IOI和ICPC等国际竞赛代表了最具挑战性的场景。我们提出了一条端到端的专业化流水线，结合了大规模题目筛选、合成推理轨迹、监督微调（SFT）和强化学习（RL）。利用22,000道精选题目，我们通过SFT和RL训练了Nemotron-3-Nano-CC（30B-A3B），并仅使用SFT训练了Nemotron-3-Ultra-CC（550B-A55B）。我们进一步引入了GenCorrect，一种反馈驱动的测试时计算策略，能够迭代式地生成、评估和优化多样化解决方案。在IOI 2025上，Nano-CC在训练后从130分提升至291分，结合GenCorrect后达到468分，超过了438.3分的金牌门槛，而Ultra-CC则达到502分。在这些结果的指导下，我们开发了一个面向竞赛的Ultra-CC系统，并在IOI 2026期间进行了前瞻性评估。在与人类选手相同的时间、网络访问和提交限制条件下，该系统在600分中获得了535.4分，既超过了361.12分的金牌门槛，也超过了498.27分的最高人类选手得分。据我们所知，这是首个在IOI赛题集上超越最高分人类选手的AI系统。

English

Competitive programming has become a key test of large language model reasoning, with international competitions such as IOI and ICPC representing its most challenging settings. We present an end-to-end specialization pipeline combining large-scale problem curation, synthetic reasoning traces, supervised fine-tuning (SFT), and reinforcement learning (RL). Using 22,000 curated problems, we train Nemotron-3-Nano-CC (30B-A3B) with SFT and RL and Nemotron-3-Ultra-CC (550B-A55B) with SFT alone. We further introduce GenCorrect, a feedback-driven test-time compute strategy that iteratively generates, evaluates, and refines diverse solutions. On IOI 2025, Nano-CC improves from 130 points to 291 after post-training and to 468 with GenCorrect, exceeding the gold threshold of 438.3 while Ultra-CC reaches 502. Guided by these results, we develop a competition-specific Ultra-CC system and evaluate it prospectively during IOI 2026. Under the same time, internet-access, and submission constraints as human contestants, it scores 535.4 out of 600, exceeding both the gold threshold of 361.12 and the top human score of 498.27. To our knowledge, this is the first AI system to outscore the highest-scoring human contestant on an IOI problem set.