1 paper
Xianghong Fang, Wenlong Mou, Yuan Yuan +2
Discrete visual tokenization, predominantly driven by vector, scalar, and product quantization, lacks a unified conceptual framework for understanding quantization tradeoffs. In th…