SOTA multimodal models for fashion product embeddings -> https://github.com/marqo-ai/marqo-FashionCLIP/