Band of Agents Hackathon

Next Hackathon

Band of Agents Hackathon

Starts Jun 12, 2026

Join this hackathon →
Online

Unhallucinate Challenge: 24 Hours Fight

Build an advanced RAG based app with reduced hallucinations by using the newest Vectara’s HHEM model and win up to $1500!

  • • 🕒 24 hours to complete the challenge
  • • 💡 Implement RAG based app by utilizing HHEM model
  • • 🤝 Take part alone or form a team with other participants
  • • 🚀 3 finalists will share the Vectara’s credits worth of $3000
Unhallucinate Challenge: 24 Hours Fight event thumbnail
a financial AI application

In late 2023, Vectara introduced the Hughes Hallucination Evaluation Model (HHEM). It's a tool that helps measure how much a language model (LLM) generates text that doesn't make sense or strays from the original content. We looked at various models, both open source like Llama 2 or Mistral 7B, and commercial ones like OpenAI’s GPT-4, Anthropic Claude, or Google’s Gemini. The model showed that there are big differences between these models in how likely they are to produce text that doesn't make sense

Source: https://www.nytimes.com/2023/11/06/technology/chatbots-hallucination-rates.html

Hackathon Challenge

We want you to enhance the use of HHEM in our RAG application, ensuring maximum benefits for GenAI. Your participation is genuinely appreciated. Let's work together to boost GenAI applications. Your innovative touch is what we're looking for—apply now!

Prizes

Top performers will be rewarded with Vectara credits for the best uses of HHEM:

  • 🥇 First Place:💰 $1500 worth of Vectara's credits
  • 🥈 Second Place:💰 $1000 worth of Vectara's credits
  • 🥉 Third Place:💰 $500 worth of Vectara's credits

Seize the opportunity to win these valuable credits!

Here’s what is truly helpful from Vectara’s HHEM Model

The Hughes Hallucination Evaluation Model (HHEM) Leaderboard assesses how frequently Large Language Models (LLMs) like GPT-4, Google Gemini, or Meta’s Llama 2 generate hallucinations in document summaries. Vectara has open-sourced HHEM to democratize LLM hallucination evaluation, shedding light on performance differences. Initially a Huggingface model with a Github repository, HHEM Leaderboard now accommodates the evaluation of new models. The LLM community is encouraged to submit models for assessment.

HHEM model features:

  • Open-Source Nature: HHEM is open source, making the model and its evaluation accessible to the broader community.
  • Versatility: The model is adaptable for assessing different types of LLMs, allowing for the evaluation of both open-source models and commercial ones.
  • Leaderboard Functionality: The creation of a HHEM Leaderboard suggests a platform for comparing and showcasing the hallucination rates of various LLMs, providing valuable insights to the community.

This Hackathon is for you if

You are AI enthusiast and developer

Dive into cutting-edge AI technology, showcase your coding skills, and contribute to refining language models' performance.

You have an startup idea you want to test

Discover how HHEM can elevate your AI applications. Leverage the hackathon to optimize language models for more reliable and accurate outputs.

You want to build great LLM

Use HHEM in your RAG application to reduce overall hallucinations, or by picking the best LLM for your app.
Hackathon Community Winner

The community can judge!

The community judges will review hackathon projects and vote for the best use case, carefully following the rules. For the micro hackathons, the community feedback decides the winner! 🏆

StartUp Opportunities

Slingshot 🚀 Slingshot Incubator accelerates AI prototypes into MVPs. To qualify, have a functional AI prototype, a basic business plan, and a team of 2-5. Get mentorship, partner credits, and legal support.

👉 Apply through the Slingshot Application link.

GAIA 🌱 GAIA Accelerator fuels generative AI startups. To qualify, have an AI MVP, be between MVP and seed stage, and relocate to Riyadh, KSA. Receive a $100K investment, expert mentorships, and access to investors.

👉 Apply through the GAIA Application Link and unlock your project's potential!

Event Schedule

  • To be announced

Submitted concepts, prototypes and pitches

Submissions from the teams participating in the Unhallucinate Challenge: 24 Hours Fight event and making it to the end 👊

Band of Agents Hackathon

Next Hackathon

Band of Agents Hackathon

Starts Jun 12, 2026

Join this hackathon →