Scaling Large Reasoning Models beyond Human Supervision: A Path toward Superintelligence· 大型推理模型超越人类监督：迈向超智能的路径
Recent advances in large reasoning models (LRMs) have shown that reinforcement learning with verifiable rewards (RLVR) can substantially improve reasoning in mathematics and code, where outcomes can be checked automatically. Extending this progress to open-ended and agentic tasks remains difficult because reliable rewards are harder to obtain and direct human supervision cannot keep pace with the scale and complexity of model-generated experience. This paper studies how LRMs can continue to improve as human supervision gradually recedes from the learning loop. We examine two connected dimensions of this problem. The reward axis traces the development from per-instance human judgments to reusable verifiers and rewards that operate even without human feedback. The experience axis examines ho
研究大型推理模型在减少人类监督情况下的自适应学习。
- 核心方法
- 论文探索了两个维度：奖励维度，从单个实例的人类评价转向可重复使用的验证器和奖励；经验维度，研究模型如何在没有人类反馈的情况下生成有效经验。
- 适合谁读
- 研究者
- 要解决的问题
- 如何在人类监督逐渐减少的情况下持续提升大型推理模型（LRMs）的性能，特别是在开放性和代理任务中。
- 关键实验
- 未提供
- 主要贡献
- 提出了减少人类监督下大型推理模型持续优化的方法论；推动了向超智能发展的路径。
- 意义与局限
- 为实现超智能提供了理论基础，但实际应用中仍需克服可靠奖励获取和复杂任务处理的挑战。