Stanford CME296 Diffusion & Large Vision Models | Spring 2026 | Lecture 5 - Architectures

Learn more details about this course: https://online.stanford.edu/courses/cme296-diffusion-and-large-vision-models To follow along with the course schedule and syllabus, visit: https://cme296.stanford.edu/syllabus/ Chapters: 00:00:00 Introduction 00:05:26 Objective 00:09:58 Convolutions, filters 00:14:44 Receptive field 00:17:14 Pooling 00:19:06 U-Net 00:27:52 Timestep representation 00:30:31 Class label representation 00:33:21 Timeline of U-Net models 00:35:43 Diffusion Transformer (DiT) 00:48:08 Adaptive layer normalization (adaLN) 01:02:30 DiT end-to-end example 01:12:57 Multimodal DiT (MM-DiT) 01:23:33 Qwen-Image, Z-Image, FLUX.1 01:24:27 Timeline of DiT models 01:25:25 Absolute position embeddings 01:38:48 Rotary position embeddings (RoPE) 01:39:59 2D RoPE variants For more information about Stanford’s graduate programs, visit: https://online.stanford.edu/graduate-education Afshine Amidi is an Adjunct Lecturer at Stanford University. Shervine Amidi is an Adjunct Lecturer at Stanford University. View the course playlist: https://www.youtube.com/playlist?list=PLoROMvodv4rNdy8rt2rZ4T2xM0OjADnfu