meituan-longcat/R-HORIZON-AIME25
收藏Hugging Face2025-10-21 更新2025-10-25 收录
下载链接：
https://hf-mirror.com/datasets/meituan-longcat/R-HORIZON-AIME25
下载链接
链接失效反馈官方服务：
资源简介：
R-HORIZON是一个通过查询组合来刺激大型推理模型（LRMs）进行长跨度推理行为的新方法。它将孤立的问题转化为复杂的多步推理场景，并建立了R-HORIZON基准测试，包括6个代表性数据集。这些数据集涵盖了数学、代码生成和代理应用。R-HORIZON还支持使用长跨度推理数据进行强化学习。数据集可以在Hugging Face Hub上获取，同时提供了安装、基准评估和训练的快速入门指南。
R-HORIZON is a novel method designed to stimulate long-horizon reasoning behaviors in Large Reasoning Models (LRMs) through query composition. It transforms isolated problems into complex multi-step reasoning scenarios and establishes the R-HORIZON Benchmark with 6 representative datasets covering mathematics, code generation, and agent applications. R-HORIZON also supports reinforcement learning using long-horizon reasoning data. The datasets are available on Hugging Face Hub, and there are quick start guides for installation, benchmark evaluation, and training.
提供机构：
meituan-longcat
