dddraxxx/visual_probe_hard_106

该数据集是一个包含图像和其他相关信息的训练集，具体包括图像字段、数据来源、奖励模型等信息。奖励模型中包含了地面真实情况，而额外信息中包含了答案、文档ID、原始数据来源和问题等。此外，数据集还包括了提示信息，这些提示信息包含了内容和角色两个部分。能力字段和环境名称也作为特征的一部分。整个数据集分为训练集，共有106个示例，总大小为295,884,098字节。

This dataset is a training set that includes images and other related information, specifically featuring image fields, data sources, reward models, etc. The reward model includes ground truth, while the extra information includes answer, document ID, original data source, and question. Moreover, the dataset contains prompt information, which consists of content and role parts. The ability field and environment name are also part of the features. The entire dataset is split into a training set with a total of 106 examples and a size of 295,884,098 bytes.