Who builds the world's best RAG? The wait is over!
After an intense and thrilling competition, we are excited to unveil the top-performing solutions and the talented developers behind them.
🗓️ WHERE AND WHEN
The challenge took place on Thursday, February 27th with an opening keynote by Stephan Gillich. Afterwards, we provided the PDF files participants needed to ingest into their RAG to tackle the challenge questions. Watch Rinat Abdullin unveiling the winners of the Challenge in this video!
Opening Keynote by Renowned AI Expert Stephan Gillich
"The Power of Context: RAG enhanced AI"
Retrieval Augmented Generation (RAG) has emerged as a transformative paradigm in AI, bridging the gap between AI models trained on public data and the vast ocean of private information available. It enables businesses to get truly intelligent applications without re-training or fine tuning AI models while keeping their core data and IP more private.
In this keynote we'll delve shortly into the challenges businesses have with integration of AI. We will cover the principles of RAG, examining how effectively retrieving and integrating relevant context can enhance the reliability, and factual grounding of generated outputs. Beyond the fundamentals, we'll discuss some forward looking ideas, and address the open challenges and exciting future directions of RAG.
Open Source for Open Minds: Learning Together!
The code for the question generator, submission results, and submission API are open source. We'll also share reports on winning architectures and encourage participants to share their solutions, fostering a collaborative learning experience.
You can take a look at the winning solution from the last round here
How the challenge works
STEP 1: FEED YOUR RAG
Prepare your RAG by uploading a set of public annual reports. To keep things fair and exciting, we’ll randomly select approximately 100 PDFs from the available reports for you to work with.
STEP 2: PROCESSING TIME
You’ll have at least 6 hours to process the documents before we release the questions your RAG needs to answer. This gives your RAG enough time to handle all the files.
STEP 3: SUBMIT YOUR RAGS' ANSWERS
We'll be accepting answers for 24-48 hours after the question announcement, to let people experiment with different architectures. Once completed, submit the answers to us via API.
IBM watsonx: Exclusive Track at the Enterprise RAG Challenge!
Are you interested in exploring the potential of IBM watsonx? As part of the Enterprise RAG Challenge, we’re introducing an exclusive IBM watsonx track designed to give you hands-on experience!
What’s in it for you?
- IBM watsonx in Action: Get the chance to work directly with IBM watsonx and leverage the platform’s powerful capabilities.
- Guidance from IBM Experts: You won’t be on your own! Experienced IBM experts will be available to help you with any questions or challenges you encounter.
- All the Resources You Need: IBM will provide everything you need to get started and succeed.
🏆 Exciting Additional Prizes:
Compete for additional prizes of €500, €350, and €200 in vouchers for the top-performing teams on this track!
ANSWERS TO SOME FREQUENTLY ASKED QUESTIONS
Does everyone receive an own set of pdfs or will everyone receive the same set, dependent on the seed generated?
It will be the same set of PDFs for everybody (it will be the same seed for everybody)
Does everyone receive an own set or will everyone receive the same set of questions?
It will be the same set of questions for everybody (it will be the same seed for everybody)
Do I push my answers to a git repo?
We will setup an API to which you can push your answers (programmatically or upload by hand).
This API will immediately save, sign and timestamp the submission, so it will be provable that is was submitted at a specific point of time.