
31 - Singular Learning Theory with Daniel Murfet
AXRP - the AI X-risk Research Podcast
Exploring Local Learning Coefficients in Neural Networks
This chapter investigates the local learning coefficient in neural networks, highlighting the challenges of measuring it within high-dimensional parameter spaces. It discusses advancements in Bayesian sampling methods used to estimate this coefficient and its implications for generalization in deep learning. The conversation emphasizes the relationship between the learning coefficient and the geometry of loss landscapes, proposing new avenues for understanding model behavior and training efficiency.
00:00
Transcript
Play full episode
Remember Everything You Learn from Podcasts
Save insights instantly, chat with episodes, and build lasting knowledge - all powered by AI.