neoneye/simon-arc-rle-image-v16

该数据集是一个用于图像到文本和文本到图像任务的英语数据集，包含10K到100K之间的数据项。数据集的每个版本都针对不同的图像处理任务和图像尺寸进行了调整，包括旋转图像、颜色计数、直方图、k近邻等。图像尺寸在不同版本中也有所变化，从1像素到30像素不等。

The simons ARC (abstraction & reasoning corpus) RLE (run-length-encoding) image version 16 dataset is primarily used for image-to-text and text-to-image tasks. The dataset includes images of various sizes, ranging from 1 to 30 pixels, with a focus on image processing techniques such as rotation, color counting, and neighboring pixel counting, which have evolved with each version update. The purpose of this dataset is to assist large language models (LLM) in learning to handle these image-related tasks, particularly in dealing with rotated images and counting tasks.