OpenMathReasoning nvidia
OpenMathReasoning OpenMathReasoning is a large-scale math reasoning dataset for training large language models (LLMs). This dataset contains 306K unique mathematical problems sourced from AoPS forums with: 3.2M long chain-of-thought (CoT) solutions 1.7M long tool-integrated reasoning (TIR) solutions 566K samples that select the most promising solution out of many candidates (GenSelect) Additional 193K problems sourced from AoPS forums (problems only, no solutions) We used… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/OpenMathReasoning.
- 类型
- dataset
- 许可
- cc-by-4.0
- 语言
- en
- 下载量
- 45,655
- 点赞
- 470
- 访问
- public
- 文件
- 0
标签
- 数学推理
- 指令微调
- 合成数据
- 预训练语料
- 文本数据集
- 推理
- 大语言模型
- 数据
摘要
OpenMathReasoning是NVIDIA发布的大规模数学推理训练数据集，包含30.6万道来自AoPS论坛的数学题，配套320万条长思维链（CoT）、170万条工具集成推理（TIR）解题方案及GenSelect最佳方案选择数据，并额外提供19.3万道纯题目。该数据集用于训练OpenMath-Nemotron系列模型，在AIME等数学基准上取得SOTA表现，适合大语言模型的数学推理指令微调与强化学习。
README
--- language: - en license: cc-by-4.0 size_categories: - 1M<n<10M task_categories: - question-answering - text-generation pretty_name: OpenMathReasoning tags: - math - nvidia configs: - config_name: default data_files: - split: cot path: data/cot-* - split: tir path: data/tir-* - split: genselect p…