Search papers, labs, and topics across Lattice.
Zhejiang University
1
0
2
Hallucinations in video models can be mitigated by intelligently patching visual tokens, leading to a 50.60% improvement in response accuracy.
