遇见数据集

R-I-S-E/RISE-Judge-SFT-20K

收藏
Hugging Face2025-02-18 更新2025-04-12 收录
官方服务:

资源简介:

RISE-Judge-SFT-20k是一个针对LLM-as-a-judge的偏好数据集,由MATH-PRM800K、Ultrafeedback和Skywork-Reward-Preference-80K-v0.2构建而成,用于训练评判模型。数据集包含了16k条有效数据,经过严格的筛选和构建过程,确保评判结果的正确性。

RISE-Judge-SFT-20k is a preference dataset for LLM-as-a-judge, constructed based on MATH-PRM800K, Ultrafeedback, and Skywork-Reward-Preference-80K-v0.2, used for training judge models. The dataset includes 16k valid entries, which have undergone a rigorous selection and construction process to ensure the correctness of the judgments.

提供机构:
R-I-S-E
二维码
社区交流群
二维码
科研交流群
商业服务