Guided Prompt Evolution for Vision-Language Models Adaptation
EvoPrompt is a novel framework designed to explicitly steer the prompt trajectory for knowledge-preserving fine-tuning, and achieves state-of-the-art performance in few-shot learning while robustly preserving the original zero-shot capabilities of pre-trained VLMs.
E. Zhang, Jia-Yun Li, Yanlong Wang et al.
· 0 citations