Search papers, labs, and topics across Lattice.
University of Electronic Science and Technology of China
1
0
3
5
Most video generation models fail to render text accurately, with even top performers exhibiting a 25% word error rate on a new benchmark designed for this critical oversight.
