Enhancing Cross-Prompt Transferability in Vision-Language Models through Contextual Injection of Target Tokens

Devs

Enhancing Cross-Prompt Transferability in Vision-Language Models through Contextual Injection of Target Tokens | Read Paper on Bytez