I’d like to announce the JuliaGenAI organization!
A few of us (@svilupp, @sethaxen, Brian Chen) have been discussing the use of Julia in a generative AI world. We’re focused on providing and supporting libraries that make it easy and interesting to work with generative models. It’s a new organization and we’re looking for ideas and community involvement!
I personally was inspired by @svilupp’s suite of incredible packages, like PromptingTools.jl, LLMTextAnalysis.jl, and AIHelpMe.jl. Many of this is also built on top of the wonderful OpenAI.jl, which provides easy support for OpenAI APIs.
We want to support development of interconnects with tools in external languages, like llama.cpp, ggml, ollama, and others. I personally use Julia for many generative AI tasks and the richer the support for these, the better off the community will be. It’s a weird-ass new world and I’m looking forward to the community embracing and applying generative tools to our favorite language.
I think many of us in the Julia community are in agreement that it will be difficult to compete with other languages as far as training neural nets are concerned, and the Julia-native inference tooling for transformers are still relatively undeveloped (though improving!). I would love to see this happen and will try to provide support where I can, but that’s not the primary focus of the organization.
Our hope is that the JuliaGenAI organization can be a home for:
- Tools to develop AI agents, handle chain of thought, etc. Think something like a native LangChain. I’m still tinkering/experimenting with this at colang, but I would love to see all kinds of other tools.
- Experiment with things like structured text generation. outlines does this really well, and it may be worth our time to add a Julia wrapper around it.
- Other stuff yet to be determined (please make suggestions!)