flowingpurplecrane/genrm
收藏Hugging Face2025-03-06 更新2025-04-12 收录
下载链接：
https://hf-mirror.com/datasets/flowingpurplecrane/genrm
下载链接
链接失效反馈官方服务：
资源简介：
这是一个伴随匿名提交论文"生成验证器：作为下一个标记预测的奖励建模"的数据集。它包括用于微调生成验证器的验证理性。数据集分为两个配置：critiques和rm，每个配置都包含训练集和验证集。
This dataset accompanies the anonymous submission, "Generative Verifiers: Reward Modeling as Next-Token Prediction". It includes the verification rationales used to finetune the generative verifiers. The dataset consists of two configurations: critiques and rm, each with its own training and validation sets.
提供机构：
flowingpurplecrane
