I’ve been looking around for ways to finetune or provide context to LLMs using private data. Here’s a great open source project I found:
https://python.langchain.com/en/latest/index.html
The best examples integrate w/ openAI, which means you pay for each query to your bot or agent, but you can use an open source LLM (like LLAMA or Alpaca) if you want. For private data they have data loaders for most common file formats.
My initial testing suggests the indexing of private data isn’t plug and play just yet, and having to pay 3 cents per test query to openAI can add up if you aren’t paying attention.
Anyone interested in diving into the capabilities or discussing further comment below.