Leveraging MLLM Embeddings and Attribute Smoothing for Compositional Zero-Shot Learning

Open in new window