top of page

Introducing RAGO

Jan 31, 2024
RAGO logo

Welcome to the unveiling of RAGO, our latest innovation from Neural Bridge. RAGO isn't just any Large Language Model; it's our answer to the specific needs of Retrieval Augmented Generation (RAG). If you've followed our journey at Neural Bridge, you know we're no strangers to RAG. In fact, we've brought several RAG solutions to life, proving their worth across various applications. Despite the success, we've hit a snag when it comes to open-source models. While they're fantastic for on-prem deployments or when you're trying to scale up without breaking the bank, their performance in RAG tasks often leaves much to be desired.

So, we rolled up our sleeves and decided it was time to craft a solution tailored specifically for RAG. Enter RAGO. This isn't a one-size-fits-all model; we're talking about a suite of versions, each meticulously tuned for different facets of RAG. And to make things even better, we're laying it all out for you:

  1. RAGO v1 - 1B

  2. RAGO v1 - 7B

  3. RAGO v1 - 40B

  4. RAGO v2 - 7B

  5. RAGO v2 - 13B

Here, you'll find a detailed breakdown of the different versions of RAGO we're releasing, along with the weights for these models available on Hugging Face.

We're not just stopping at the model, either. Recognizing the importance of quality data, we've curated a specific RAG dataset, complete with some intentionally included hallucination data, to push the boundaries of what our models can handle.​

For those who love to dive into the details, both our models and datasets will be accessible on Hugging Face (link). We're talking easy access to the weights and data you need to revolutionize your RAG applications.

But what's innovation without proof, right? We're in the process of compiling our findings into a paper that'll showcase just how RAGO stands up against both open-sourced and closed-source models in handling RAG tasks. This paper will be your go-to for understanding the depths of our research and the breakthroughs we've achieved with RAGO.

 

Are you as excited as we are? Make sure to subscribe to our newsletter. It's your golden ticket to being among the first to know when our paper drops. Trust us, you won't want to miss this deep dive into the world of RAG, powered by RAGO.

 

Stay tuned, and welcome to the next chapter in Large Language Models, brought to you by the innovators at Neural Bridge.

bottom of page