Data supporting: "CM-Net: A general machine learning architecture for predicting mechanical properties of cellular materials"

Our project aims to construct​ a general machine learning architecture to predict the mechanical properties of cellular materials, which helps researchers design specialized configurations tailored to their specific demands. In this project, we focus on the out-of-plane mechanical performances​ of thin-walled structures—such as tubes and honeycombs—that exhibit significant nonlinear behavior due to plasticity, buckling, and frictional contact. The dataset comprises over 30,000 thin-walled structures constituted​ from various metal materials.Train and validation set (.npy format). Complete architectural database of thin-walled structures for training​ and validation of CM-Net, including basic geometric units, adjacency matrices, and labels.Test set (.npy format). Complete architectural database of thin-walled structures for testing the interpolation and extrapolation capabilities​ of CM-Net.The dataset is intended for reviews of potential publications and should not be cited or used before public publication.