interleaved-bench/TweetBench

该数据集是一个包含文本和图像输入输出的多配置数据集，主要用于某种形式的文本和图像处理任务。它包含了提示文本（prompt）、是否修改提示、是否填充空白、输入和输出URLs、社区反馈、输入和输出图像、是否有成人内容等字段。数据集分为不同的子集，包括纯文本子集、交织子集、训练子集等。

This dataset is a multi-config dataset containing text and image inputs and outputs, primarily used for some form of text and image processing tasks. It includes fields such as prompt text, whether the prompt is modified, whether the blank is filled, input and output URLs, community feedback, input and output images, whether it contains adult content, etc. The dataset is divided into different subsets, including a pure text subset, interleaved subsets, training subsets, etc.