dddraxxx/docvqa_aug_val_1k

该数据集包含了图像和相关的文本信息，每个样本都包含一个图像字段以及数据来源、奖励模型等相关信息。奖励模型中包含了一个地面真实字段，用于表示真实的答案。额外信息字段中包含了所有可能的答案、OCR识别结果以及一些文档相关的索引信息。数据集分为测试集，测试集包含了1000个示例。

The dataset consists of images and associated textual information, with each sample including an image field and related information such as data source and reward model. The reward model contains a ground truth field representing the correct answer. The extra information field includes all possible answers, OCR recognition results, and some document-related index information. The dataset is split into a test set, which contains 1000 examples.