GovBench Hackathon 🇺🇸
GovBench Hackathon 🇺🇸
Over the next year, the US Government will spend $1 billion+ on LLM applications. But how well do LLMs perform on USG domains: Homeland Security, Military, Health & Human Services, etc.?
GovBench is series of government-specific LLM benchmarks. We recently released JointStaffBench: an LLM benchmark focused on the US Military.
Join us in Washington D.C. for an intimate, high-impact hackathon where AI engineers and subject matter experts come together to create LLM benchmarks tailored to the real-world needs of the U.S. Government. These benchmarks will likely have a large influence in the USG's adoption of LLM systems.
We're looking for:
Subject Matter Experts (SMEs) from each of the 15 Executive Branch Departments: Agriculture, Commerce, Defense, Education, Energy, Health and Human Services, Homeland Security, Housing and Urban Development, Interior, Justice, Labor, State, Transportation, Treasury, and Veterans Affairs.
Coding experience not required.
4+ years experience in Department preferred but not required.
Hackers/Coders with LLM & python experience.
The goal is for a benchmark to be created for each Executive Branch Department. Skeleton code will be provided to help streamline the benchmark creation & evaluation process. Ideally, each Department will have 2-3 participants spanning SMEs & Coders.
This event is only in-person in Washington D.C. with no virtual component.
Agenda
10:00 am - 10:15 am: Check-in
10:15 am - 10:30 am: Introduction
10:30 am - 2:30 pm: Building time
2:30 pm: Submit benchmark
2:45 pm - 3:00 pm: Wrap up
Photos of the venue!