Search papers, labs, and topics across Lattice.
This work was supported in part by the National Natural Science Fund for Distinguished Young Scholars (62025205), in part by the National Natural Science Foundation of China under Grant 62541327, Grant 62532009, Grant U25B2042, and Grant 62302396, in part by the Natural Science Foundation of Shaanxi Province (Grant No.2024JC-YBQN-0665). (Corresponding author: Yao Zhang) Jing Zhang and Ke Huang are with the School of Computer Science and Technology, Xi鈥檃n University of Science and Technology, Xi鈥檃n 710600, China. Yao Zhang and Bin Guo are with the School of Computer Science, Northwestern Polytechnical University, Xi鈥檃n 710021, China. Z. Yu is with Harbin Engineering University, Harbin, Heilongjiang, China, and also with the School of Computer Science, Northwestern Polytechnical University, Xi鈥檃n, Shaanxi, China
1
0
3
0
Squeeze 3D vision transformers: XStreamVGGT slashes memory consumption by 4.42x and accelerates inference by 5.48x via pruning and quantization of the KV cache, all with negligible performance loss.
