"The Cognitive Revolution" | AI Builders, Researchers, and Live Player Analysis cover image

GELU, MMLU, & X-Risk Defense in Depth, with the Great Dan Hendrycks

"The Cognitive Revolution" | AI Builders, Researchers, and Live Player Analysis

00:00

Enhancing Resilience in Machine Learning Models

This chapter explores advanced techniques for safeguarding machine learning models from adversarial attacks, particularly through the use of circuit breakers and tamper resistance. It emphasizes the importance of robust safeguards and the challenges of implementing fine-tuning resistance against tampering. By examining current models and strategies, the discussion aims to strike a balance between model performance, utility, and risk management.

Transcript
Play full episode

The AI-powered Podcast Player

Save insights by tapping your headphones, chat with episodes, discover the best highlights - and more!
App store bannerPlay store banner
Get the app