AI编程模型通过后训练与迭代优化，首次超越人类最高分完成国际信息学奥赛。
Post-Training Language Models for Gold-Medal Performance in Coding Competitions
- 用2.2万道题目和合成推理路径，对300亿/5500亿参数模型进行微调与强化学习。
- 经优化后模型在IOI 2025中得分从130提升至468，超金牌线；2026年预演达535.4分。
- 引入生成-验证-修正反馈机制，实现测试时动态计算，适合竞赛级代码生成任务。
编程竞赛已成为检验大语言模型推理能力的关键场景，国际信息学奥林匹克（IOI）和国际大学生程序设计竞赛（ICPC）代表了最严苛的挑战。本文提出端到端专业化训练流程，融合大规模题库筛选、合成推理轨迹、监督微调（SFT）与强化学习（RL）。基于22,000道精选题目，我们对Nemotron-3-Nano-CC（30B-A3B）进行SFT与RL训练，对Nemotron-3-Ultra-CC（550B-A55B）仅使用SFT。进一步提出GenCorrect——一种反馈驱动的测试时计算策略，可迭代生成、评估并优化多样解法。在IOI 2025中，Nano-CC得分由130提升至291（后训练），再达468（结合GenCorrect），超过金牌线438.3。Ultra-CC得分为502。据此，我们构建面向竞赛的Ultra-CC系统，并在IOI 2026中前瞻性评估：在与人类选手相同的时限、网络访问及提交限制下，该系统获得535.4分（满分600），高于金牌线361.12与最高人类得分498.27。据我们所知，这是首个在完整IOI题集上超越最高人类得分的AI系统。
原文摘要 · Abstract (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.
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。