PhD student, Institute of Computing Technology, Chinese Academy of Sciences
1 paper at NeurIPS 2025
This work improves CLIP’s visual detail capturing ability by inverting the unCLIP generative model, which we find suitable for achieving this goal.
