Vision-language model

Rectify representation bias in vision-language models for long-tailed recognition