Search papers, labs, and topics across Lattice.
University of Science and Technology of China
3
0
6
0
Maintaining a consistent identity in sequential-action video generation is possible without sacrificing performance, as demonstrated by a novel three-stage pipeline that ranked third in a competitive challenge.
Smaller models can achieve empathetic dialogue performance that rivals larger counterparts by leveraging privileged information during training.
Unified multimodal models suffer from internal conflict, but this work shows how to turn that interference into a surprisingly effective source of performance gains.
