Explore the groundbreaking research presented by Marvin Li on non-asymptotic theory for feature emergence in diffusion models. Delve into the concept of critical windows and their significance in understanding the behavior of diffusion models. Gain insights into the theoretical foundations and practical implications of this innovative approach, which challenges traditional asymptotic analyses. Learn how this research contributes to advancing our understanding of feature emergence in machine learning and artificial intelligence.
Critical Windows: Non-Asymptotic Theory for Feature Emergence in Diffusion Models
Generative Memory Lab via YouTube
Overview
Syllabus
Critical windows: non-asymptotic theory for feature emergence in diffusion models
Taught by
Generative Memory Lab
Related Courses
-
Reflected Diffusion Models
-
Blackout Diffusion: Generative Diffusion Models in Discrete-State Spaces
-
Deep Generative Models - Diffusion Models for Discrete Data - Lecture 18
-
The Variational Book - Exploring Generative AI and Diffusion Models
-
A Theory for Emergence of Complex Skills in Language Models
-
Stanford CS236: Deep Generative Models - Score Based Diffusion Models - Lecture 16
