PictoViLT/merged_SC_L1_T_final

该数据集是一个包含文本和图像特征的多模态数据集，适用于如掩码语言建模等NLP任务。数据集包含input_ids, attention_mask, token_type_ids等文本处理相关的特征，以及pixel_values, masked_indices等图像处理相关的特征。同时，metadata中提供了额外的上下文信息。数据集分为训练集和测试集，分别包含2048和512个样本。

This dataset is a multimodal dataset containing both text and image features, suitable for NLP tasks such as masked language modeling. The dataset includes features related to text processing such as input_ids, attention_mask, token_type_ids, and features related to image processing such as pixel_values, masked_indices. Additionally, the metadata provides extra contextual information. The dataset is split into a training set and a test set, containing 2048 and 512 samples respectively.