dddraxxx/sealvqa-spatial-hard-1335

该数据集是一个包含图像和多种相关信息的复合数据集，适用于视觉问答等任务。它包括图像字段、数据源、奖励模型以及额外信息等。额外信息中包含了答案、边界框格式描述、会话轮数、数据拆分方式、数据集类型、是否有目标、图像路径、索引、原始对话、问题、问题ID、搜索次数和目标实例等信息。奖励模型包括地面真实信息。数据集分为训练集，配置文件为默认配置。

The dataset is a composite dataset containing images and various related information, suitable for tasks such as visual question answering. It includes fields such as images, data source, reward model, and additional information. The additional information contains answer, bounding box format description, number of conversation turns, data splitting method, dataset type, whether it has targets, image path, index, original conversations, question, question ID, search times, and target instances, etc. The reward model includes ground truth information. The dataset is split into a training set, with the configuration file set to default.