Self-Study Guides & Unofficial Auditing Policy
This class is open to any form of unofficial auditing for CMU students! You’re welcome to:
- ✅ Attend lectures and sit in the classroom
- ✅ Join the course Discord server for discussions
- ✅ Come to office hours
- ✅ Follow along with homework assignments
- ✅ Access all course materials
For students outside CMU: All lecture recordings, slides, homework, code, and resources will be publicly available online!
If you're interested in this class, please join the waitlist now. I will be sending course information via email to all students who have expressed interest in the class soon.
Access Levels by Student Type
| Aspect | CMU Registered Students | CMU Auditing Students | Outside Students |
|---|---|---|---|
| Course Credits | ✓ Receive 6 credits | ✗ No credits | ✗ No credits |
| Grading | ✓ All work graded | ✗ Not graded | ✗ Not graded |
| Computational Resources | ✓ Provided by course | ✗ Use own resources | ✗ Use own resources |
| In-Person Lectures | ✓ Full access | ✓ Welcome to attend | ✗ Not available |
| Lecture Recordings | ✓ Full access | ✓ Full access | ✓ Publicly available online |
| Discord Server | ✓ Full access | ✓ Full access | ✗ CMU students only |
| Office Hours | ✓ In-person & virtual | ✓ In-person & virtual | ✗ Not available |
| Course Materials | ✓ Full access | ✓ Full access | ✓ Full access |
| Email Instructor | ✓ Available | ✓ Available | ✓ Available |
Study Materials
All the materials you need to follow along with the course are available on this website:
📅 Schedule
Weekly lecture topics, readings, and deliverables. Follow along with the course timeline.
View Schedule →📚 Resources
Curated collection of monographs, tutorials, key papers, and other courses on diffusion models and flow matching.
Browse Resources →📝 Syllabus
Course policies, grading breakdown, homework descriptions, and learning objectives.
Read Syllabus →🎥 Lecture Recordings
Video recordings of lectures will be made available throughout the semester.
Coming soon - check back after lectures begin!
📄 Lecture Slides
Detailed slides from each lecture covering key concepts.
Will be posted on the schedule page as they become available.
💻 Homework Assignments
Assignment descriptions and starter code. Build a complete image generation system!
Coming soon - check back after lectures begin!
Computational Resources
Since auditing students won’t receive course-provided compute resources, here are some options for running your experiments.
If you're paying for computational resources, make sure to only use those large GPUs when you absolutely need to and try to delegate small debugging, data processing, etc to free/cheaper resources (e.g. CPUs & smaller GPUs) as much as possible!
Also remember to place budget notifications and shut down the machines when you are not using them. We do not want anyone to go broke by studying diffusion!
Some of the services are preemptable, meaning that they can interrupt your training and reallocate your resources to someone else. Make sure to save your checkpoints frequently and to a persistent location!
🆓 Free Options
Modal
- Serverless compute for ML (one of our sponsors!)
- Zero cloud configuration, everything in code
- Choose your own GPU type
- \$30/month free credits
- Frequently gives out free credits for academia
- Learn More →
Google Colab (Free Tier)
- Free GPU access (T4 GPUs typically)
- 12-hour session limits
- Get Started →
Lightning AI
- Choose your own GPU type
- 15 free credits per month
- Free CPUs
- Learn More →
Kaggle Kernels
- Free GPU access (P100 GPUs)
- 30 hours/week GPU quota
- Get Started →
Azure for Students
- Azure provides \$100 free credits for all full-time university students
- Learn More →
🎁 Grants
AWS Cloud Credit for Research
- Graduate students can apply for up to \$5000 AWS credits
- Learn More →
TPU Research Cloud
- Great for JAX
- Learn More →
💰 Paid Options (as of Dec 2025)
Modal
- Serverless compute for ML (one of our sponsors!)
- Zero cloud configuration, everything in code
- Pay-as-you-go after using up the free credits
- \$1.95/h L40S and \$3.95/h H100
- Learn More →
Google Colab Pro/Pro+
- \$10-\$50/month
- Faster GPUs (A100, V100) in comparison to free tier
- Also has pay-as-you-go option
- Learn More →
Hyperbolic
- Cheap H100 access and pay-as-you-go
- \$0.30/h 4090 and \$1.50/h H100
- Learn More →
RunPod
- Various GPU options and pay-as-you-go
- \$0.33/h A6000, \$0.79/h L40S and \$1.99/h H100
- Learn More →
Lambda Labs
- Various GPU options and pay-as-you-go
- \$0.8/h A6000 and \$2.49/h H100
- Learn More →
Lightning AI
- Pay-as-you-go after using up the free credits
- Can also subscribe for \$20-\$140 per month for more monthly/annual credits
- \$1.80 (preemptable)-\$2.89/h L40S and \$1.99-\$5.37/h H100
- Learn More →
🏫 CMU-Specific Resources
SCS Resources
- If you're affiliated with SCS, you may have access to our compute clusters Babel and FLAME. Check with your advisor!
Research Lab Resources
- If you're affiliated with a research lab, check if they have compute resources you can use.
Ready to Start Learning?
Review the schedule, dive into the resources, and start exploring diffusion models and flow matching. Whether you're at CMU or learning from anywhere in the world, welcome to the community!
Questions about auditing or self-studying? Check out the FAQs email Kelly at yutonghe[at]cs.cmu.edu