Large language models can enhance the interpretability of embeddings, bridging the gap between rich data representations and expressive capabilities.
