Prompt-Based Continual Compositional Zero-Shot Learning

This article likely discusses a novel approach to zero-shot learning, focusing on continual learning and compositional generalization using prompts. The research probably explores how to enable models to learn new tasks and concepts sequentially without forgetting previously learned information, while also allowing them to combine existing knowledge to solve unseen tasks. The use of prompts suggests an investigation into how to effectively guide large language models (LLMs) or similar architectures to achieve these goals.