Explore a new open source paradigm for AI benchmarking called DynaTask in this 48-minute conference talk by Sam Lightstone, Software Engineer at Facebook. Discover how this novel benchmark suite utilizes dynamic adversarial data collection to evaluate AI models and assess their vulnerability to human deception. Learn about the importance of robust AI evaluation methods and gain insights into the latest advancements in AI benchmarking techniques. Understand the potential implications of DynaTask for improving AI model performance and resilience against adversarial attacks.
DynaTask: A New Open Source Approach for AI Benchmarking
Toronto Machine Learning Series (TMLS) via YouTube
Overview
Syllabus
DynaTask, A New Open Source Approach for AI Benchmarking
Taught by
Toronto Machine Learning Series (TMLS)
