Search papers, labs, and topics across Lattice.
State Key Laboratory of Multimedia Information Processing, School of Computer Science, Peking University
1
0
2
Pruning 77.8% of visual tokens without losing performance could revolutionize the efficiency of multimodal large language models.
