Adapts pre-trained vision models using prompts. Enables efficient fine-tuning for new tasks without extensive retraining.
