dddraxxx/curriculum_1_docvqa_augmented

这是一个包含图像和对应数据源信息的训练数据集，其中每个样本都包含了地面真实情况的奖励模型，以及一些额外信息，如所有可能的答案、正确答案、增强索引、增强类型、训练阶段、数据分割方式、文档ID、是否有原始OCR结果、索引、OCR结果、原始问题、问题ID、提示内容和角色、能力类型以及环境名称等。数据集分为训练集，共有3999个样本。

This is a training dataset containing images and corresponding data source information. Each sample includes a reward model with ground truth, as well as additional information such as all possible answers, correct answer, augmentation index, augmentation type, training stage, data split method, document ID, whether there is original OCR result, index, OCR result, original question, question ID, prompt content and role, ability type, and environment name. The dataset is split into a training set with a total of 3999 samples.