Abstract: Recent progress has shown great potential of visual prompt tuning (VPT) when adapting pre-trained vision transformers to various downstream tasks. However, most existing solutions ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results