1 paper
Jiajun Dong, Chengkun Wang, Wenzhao Zheng +3
Effective image tokenization is crucial for both multi-modal understanding and generation tasks due to the necessity of the alignment with discrete text data. To this end, existing…