dddraxxx/reason_synth_extreme_simple_dfs

这是一个包含图像和与其相关的描述性表达等信息的图像理解数据集。数据集中的每个样本都包含了图像路径、网格大小、参照表达式、表达式类型、推理级别、目标要求、主要目标索引、匹配对象和图像本身等特征。此外，数据集分为训练集，共有7652个样本。

This is an image understanding dataset that includes images and their associated descriptive expressions. Each sample in the dataset contains features such as image path, grid size, referring expression, expression type, reasoning level, target requirements, primary target index, matching objects, and the image itself. Moreover, the dataset is split into a training set with a total of 7652 samples.