rl-research/multimodal-rewardbench-2
收藏Hugging Face2025-12-20 更新2025-12-20 收录
下载链接：
https://hf-mirror.com/datasets/rl-research/multimodal-rewardbench-2
下载链接
链接失效反馈官方服务：
资源简介：
多模态奖励基准2（MMRB2）数据集，处理自https://github.com/facebookresearch/MMRB2。该数据集包含四种配置（edit、interleaved、reasoning、t2i），每种配置都包含文本提示、图像响应、人类标注等信息，用于评估多模态模型的性能。
Multimodal RewardBench 2 (MMRB2) dataset, processed from https://github.com/facebookresearch/MMRB2. The dataset contains four configurations (edit, interleaved, reasoning, t2i), each including text prompts, image responses, human annotations, etc., for evaluating the performance of multimodal models.
提供机构：
rl-research
