​📄 Diffusion Model Paper Reading Group Debut - Diffusion Model Walk through & DDPM Paper Implementation

​🗓 Agenda

• ​[10 min] Welcome & community check-in

• ​[30 min] 📄 DDPM Paper (Ho et al. 2020):

  • ​– How Diffusion Model Works

  • ​– Key mathematical ideas and intuitions

• ​[20 min] 💻 Code Walkthrough:

  • ​– Hands-on with two open-source repos that implemented the DDPM paper:

  • ​– xdiffusion

  • ​– mindiffusion

• ​[5 min] Study Group Structure:

  • ​– Format, curriculum, and projects for the 5-month track

  • ​– How to join and what to expect

​📚 Pre-Class Learning

​Skim these ahead of time to get the most out of our session:

• ​Paper: Denoising Diffusion Probabilistic Models (DDPM)

• ​Codebases: xdiffusion, mindiffusion

​🧠 Why Learn Diffusion Models Now?

​With major breakthroughs in video generation and diffusion-based LLMs, diffusion models have become a foundational architecture in generative AI.

By starting with DDPM — the foundation paper — you’ll build the understanding needed to tackle advanced techniques like Stable Diffusion, Flow Matching, and Diffusion Transformers.

​✨ About the Diffusion Model Learning Program

​This Paper Reading Group is part of our Diffusion Model Study Group, a 12-person, peer-led program based on MIT’s lecture notes:

• ​📚 5-month curriculum (MIT lecture notes: link)

• ​🧪 Final project: Train your own model + Pipeline in ComfyUI + Build a GenAI app

• ​👩‍🔬 Learn with researchers, engineers, and technical artists

• ​⏱ Weekly: ~2 hrs live + ~2 hrs self-paced

• ​💬 Ongoing support + monthly research updates

• ​🎥 Past recordings: AI Scholars Notion

​👥 Who’s Organizing?

• ​AI Scholar Community – A peer-led community for learning AI deeply through study groups, built for engineers and product builders

​💸 Cost & Joining Process

• ​🆓 This debut session is free and open to all

• ​💵 $100/month during the 5-month program

  • ​– Covers TA support, Q&A, and community coordination

• ​🤝 Financial discount available for those with demonstrated need

• ​📅 The current cohort is already underway

• ​🚀 The next cohort begins in October 2025

​📣 Spots are limited — join the debut to see if it’s a good fit!

​👉 Questions? DM the organizer!