With the Amazon Titan Multimodal Embeddings G1
model, you can create embeddings for multimodal data, specifically text and image data. These embeddings can then be used for multimodal search and Retrieval Augmented Generation (RAG) use-cases, for example searching images by text only, images only or a combination of text and images.
- Multimodal RAG - Multimodal RAG using the Amazon Berkley Objects dataset.
We welcome community contributions! Please ensure your sample aligns with AWS best practices, and please update the Contents section of this README file with a link to your sample, along with a description.