Due to the interest in AI and embedding storage options, we assembled this service based on the opensource postgresql module PGVector. PostgreSQL, also known as Postgres, is a free and open-source relational database management system emphasizing extensibility and SQL compliance. 

[PostgreSQL Official Documentation](https://www.postgresql.org/docs/)
[PGVector Official Github](https://github.com/pgvector/pgvector)

(WE G)ot PGVector 🔗

• Workflow Enablement Group (WEG) is proud to offer PGVector hosted as a Docker container in our LC Kubernetes/OpenShift cluster.
• We currently support version 16 of Postgresql with the corresponding version of pgvector.

Working with PGVector and embeddings

PGVector is an opensource vector storage module that can be added to postgresql. Users can request their own pgvector instance via LaunchIT, just as they can with any of our other services.

We maintain a gitlab repository with an example of one way to make use of the service below:

https://lc.llnl.gov/gitlab/weg/examples/pgvector-example

This provides an easy introduction for using pgvector to store data intended for retrieval augmented generation (RAG) queries with a large language model. In this case we are making use of LC llamame, which gives users access to several different models across CZ, RZ, and even SCF. 

LlamaMe details can be found here:

 https://hpc.llnl.gov/services/cloud-services/ai-ml-services/llamame-llm-api-service

LaunchIT is detailed here:

https://hpc.llnl.gov/services/cloud-services/launchit