Hello World, We Are Team Chainforge 👋

Much Has Changed For The Better! The Team Has Grown!

We’re a lively round-up of AI researchers, enthusiasts, and community members looking to save AI practitioners headaches. We got tired of seeing arbitrary benchmarks for LLMs put out eval results that were properly too obscure.

How we got here

It’s funny how our first organic chats quickly turned into much more than a group chat. Our shared frustration quickly morphed into a bigger ambition—growing the seed of Chainforge’s research roots as a product, into the start of our greater purpose. (A big shoutout to the Cosmos Institute’s paper, The Philosopher Builder, for inspiring our initial conversations.)

Our Builder’s Philosophy

• When it comes to AI, there is no free lunch. (Yes, we do mean that theorem)

  • We believe that as we develop more performant technologies, the inherent strengths and weaknesses of various models, architectures, and datasets will always be present. There will be no single “oracle model” that rules them all.

• Dream big, share big.

  • We envision building a source where developers and engineers can trust KNOWING which AI model is the best out-of-the-box fit for the use case needed.

  • We see this path forming through scale - thus through community; One that we can provide for, and where privacy is handled along the way.

• Truth should be accessible.

  • We believe knowledge channels should be democratized.

    • Therefore: Practitioners should benefit from the insights of leading research institutions and industry players.

      • Therefore: Users and stakes relying on AI models for veracity of outputs will be fundamentally more truthful.

  • This means we are here to improve topic-based AI accuracy challenges plaguing the industry. We want to make high performance with efficient tuning AIOps cycles commonplace.

So…Where are we going with this?

If you’ve read this far, welcome to the Chainforge Community <3 With the sweat of MILA researchers, and industry practitioners we’re building our first means to solve for truth in AI.

We posit that with the right mechanisms, truly comparative eval leaderboards should be publicly transparent.

There’s something good coming, so give us a follow on any of our socials, and strap in 🚀

Look forward to hearing more from us in the coming weeks!

- Team Chainforge ⛓🔥