Search papers, labs, and topics across Lattice.
Department of Data Science and Artificial Intelligence, The Hong Kong Polytechnic University, Department of Applied Mathematics, The Hong Kong Polytechnic University
2
0
4
Geometry-aware modeling via Riemannian flow matching achieves state-of-the-art fidelity in human motion generation, outperforming Euclidean-space methods.
MDLMs can be significantly improved *without* retraining by using attention weights to guide sampling based on inter-token dependencies.
