RLAIF-V/RLPR-Benchmarks
收藏Hugging Face2025-06-22 更新2025-11-01 收录
下载链接：
https://hf-mirror.com/datasets/RLAIF-V/RLPR-Benchmarks
下载链接
链接失效反馈官方服务：
资源简介：
RLPR评估基准是一个包含7个不同基准测试的数据集，用于评估RLPR的推理能力。这些基准测试涵盖了数学推理和一般领域推理，包括MATH-500、Minerva、AIME24、MMLU-Pro、GPQA、TheoremQA和WebInstruct验证集。数据集支持问题回答和文本生成任务，语言为英语。
The RLPR Evaluation Benchmarks is a dataset consisting of seven different benchmarks designed for evaluating the reasoning capabilities of RLPR. These benchmarks cover mathematical reasoning and general domain reasoning, including MATH-500, Minerva, AIME24, MMLU-Pro, GPQA, TheoremQA, and the WebInstruct validation set. The dataset supports question answering and text generation tasks, with English as the language.
提供机构：
RLAIF-V
