Masked Scene Contrast: A Scalable Framework for Unsupervised 3D Representation Learning | Read Paper on Bytez