kevinpro/R-PRM
收藏Hugging Face2025-03-28 更新2025-04-12 收录
下载链接：
https://hf-mirror.com/datasets/kevinpro/R-PRM
下载链接
链接失效反馈官方服务：
资源简介：
R-PRM数据集是一个为训练推理驱动的过程奖励模型而开发的数据集，包括监督微调(SFT)和直接偏好优化(DPO)两个阶段。该数据集用于训练生成奖励模型，能够对数学推理过程进行逐步分析和判断，提高策略模型的评估质量和指导能力。
The R-PRM dataset is developed for training Reasoning-Driven Process Reward Models, including two stages: Supervised Fine-Tuning (SFT) and Direct Preference Optimization (DPO). This dataset is used to train a generative reward model that performs step-by-step analysis and judgment of mathematical reasoning processes, improving the evaluation quality and guidance capabilities for policy models.
提供机构：
kevinpro
