vikhyatk/docmatix-single

该数据集是Docmatix数据集的一个变体，但过滤掉了多图像样本。数据集包含图像和文本特征，其中文本部分细分为用户、助手和来源。数据集仅包含训练集，共有565,009个示例，总大小为244,951,255,658.16818字节，下载大小为145,422,811,605字节。

The dataset contains image and text data. The image data is stored in the images feature, and the text data is stored in the texts feature, which includes three string-type sub-features: user, assistant, and source. The dataset consists of only a training set with 565009 samples, totaling 244951255658.16818 bytes. The download size of the dataset is 145422811605 bytes. The dataset configuration is default, and the training data files are located at data/train-* path.