LLMs Unlocked: Evaluate AI Models with Bluelight AI
Registration
Approval Required
Your registration is subject to approval by the host.
About Event
BluelightAI is tackling one of the biggest challenges in AI today: helping teams adapt and trust cutting-edge models before they go live. The team will walk through their unique approach to mechanistic interpretability, using topological data analysis to reveal what models are really doing under the hood.
We’ll use Cobalt, our AI Evaluation platform, to analyze models - demonstrating, in real-time, how you can uncover hidden risks, reliability gaps, and unexpected strengths before your model goes public.
BluelightAI co-founders Sachin Khanna and Dr. Gunnar Carlsson (former Stanford professor) will lead a hands-on session exploring how complex AI models really work.