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

Mechanistic Interpretability: Philosophy, Practice & Progress with Goodfire's Dan Balsam & Tom McGrath

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

00:00

Unpacking Machine Learning Interpretability

This chapter explores the challenges and techniques surrounding interpretability in machine learning, such as feature representation and sparsification. It highlights the gaps in understanding model features and advocates for improved methods for semantic assignment and validation, while drawing parallels with biological concepts and the nature of scientific measurement.

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