Do large language vision models understand 3D shapes?

3DV
Re-assign community