neoneye/simon-arc-combine-v212

该数据集是一个用于图像到文本和文本到图像任务的组合数据集，包含多个子数据集，涉及颜色、旋转、平移、细胞自动机、形状、图像、质量、直方图、对称性、压缩、分形、轮廓、网格、Z索引等任务。数据集的规模在1K到10K之间，每次更新都会使用不同的随机种子。数据集的主要目的是通过不同的任务和参数调整来提高模型的抽象和推理能力。

This dataset is a combined dataset for image-to-text and text-to-image tasks, containing multiple sub-datasets involving tasks such as color, rotation, translation, cellular automata, shape, image, mass, histogram, symmetry, compression, fractal, outline, grid, Z-index, etc. The dataset size ranges from 1K to 10K, and each update uses a different random seed. The main purpose of the dataset is to improve the models abstraction and reasoning capabilities through different tasks and parameter adjustments.