Joseph Young

I'm eager to help work on RAG sources. Many of my requirements for RAG and cybersecurity involve cited sources from the RAG context. I found three significant factors controlling the...

Modifying the `RAG_TEMPLATE` to include all my cited source requirements has significantly improved responses and relevant cited sources. Some helpful hints are to describe how you want the LLM to...

> I may have misunderstood the assignment. > > What i want is to show the document chunks used by the RAG query in the LLM response.  ![RAG...

> adding it to the RAG template would be good. seems that the parameter to pull context/quotations may vary by LLM (e.g. command-r has their own special approach) this may...

Excellent points! I have been thinking a lot about this topic and found this medium article, which I added to our discord for discussion there, too. https://medium.com/@codegpt/evaluation-of-an-rag-system-the-saga-5cea918fbc47 The article is...

> There is no guide for integration with openwebui nor should there be?....both flowise and langflow allow you to drop quite a number of different LLMs into the flows. One...

Thanks! I intend to create the file "Dockerfile-cuda" in your fork. I will make the file with a combination of your changes and mine. We can rename this file once...

Today, I'm working to reduce the attack surface and minimize the container image size. The primary objective is to use a Chainguard image as the base layer because [Chainguard](https://images.chainguard.dev/directory?category=all) images...

So far, my effort to use Chainguard's [pytorch-cuda12](https://images.chainguard.dev/directory/image/pytorch-cuda12/versions) image has run into a problem due to conflicts with PyTorch, which is included in the base image and required in the...

> Thinking about some alternative ways we could do things, just some ideas: > > * Have the "default" Dockerfile and backend code setup in such a way that CPU-only...