Paper

FloMo: 3D Scene Flow as a World Action Model Intermediate for Learning from Human Video · ByteTrail