Hackathon Showcase
LuchaLibre
Building small red‑teaming agents to probe a larger model, producing playful commentary on each agent's attack actions.
YouTube Video
Project Description
A group of agents powered by small models that collaborate and share information to persuade a large SOTA model.
Prior Work
N/A
Team
Technologies
Products & Tools
AI Tinkerers
Google
Weights & Biases
Additional Links
Social Posts
Summarizing URL...