1 paper
Xianghong Fang, Litao Guo, Hengchao Chen +8
The effectiveness of modern visual representation learning and autoregressive models critically depends on vector quantization (VQ), which discretizes continuous feature representa…