Self-Supervised Pre-Training With Masked Shape Prediction for 3D Scene Understanding | Read Paper on Bytez