Abstract: Large-scale multi-modal pre-training models such as CLIP [30] and PaLI [8] exhibit strong generalization on various visual domains and tasks. However, existing image classification ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results