MoPD: Mixture-of-Prompts Distillation for Vision-Language Models
This addresses a generalization issue in adapting vision-language models to downstream tasks, but it is incremental as it builds on existing soft prompt learning methods.
The paper tackles the problem of soft prompt learning methods overfitting to seen classes and performing poorly on unseen classes in vision-language models, proposing Mixture-of-Prompts Distillation (MoPD) which outperforms state-of-the-art baselines, particularly on unseen classes.
Soft prompt learning methods are effective for adapting vision-language models (VLMs) to downstream tasks. Nevertheless, empirical evidence reveals a tendency of existing methods that they overfit seen classes and exhibit degraded performance on unseen classes. This limitation is due to the inherent bias in the training data towards the seen classes. To address this issue, we propose a novel soft prompt learning method, named Mixture-of-Prompts Distillation (MoPD), which can effectively transfer useful knowledge from hard prompts manually hand-crafted (a.k.a. teacher prompts) to the learnable soft prompt (a.k.a. student prompt), thereby enhancing the generalization ability of soft prompts on unseen classes. Moreover, the proposed MoPD method utilizes a gating network that learns to select hard prompts used for prompt distillation. Extensive experiments demonstrate that the proposed MoPD method outperforms state-of-the-art baselines especially on on unseen classes.