Consider https://simonwillison.net/2023/Sep/4/llm-embeddings/

[cmungall]

https://simonwillison.net/2023/Sep/4/llm-embeddings/

Note currently we let chromadb handle all things embeddings but the above may provide a better way to decouple the choice of object store (e.g. sqlite, mongodb) vs dedicated vector database

[oneilsh]

fyi, the sentence-transformers package contains a few different embedding models, SPECTER in particular might be useful as it was trained on scientific citations, https://www.sbert.net/docs/pretrained_models.html