A: Thank you so much for registering! We just got a huge room (Scaife Hall 105) that can fit everyone who is interested in this class. However, due to teaching staff and compute resource constraints, we still cannot officially admit everyone who's on the waitlist and we will admit students based on their positions on the waitlist, their home departments, and their programs. Graduate students in MLD and SCS will be given some reserved seats. That being said, we are fully open to any form of auditing. Details can be found in the self-study guides and below.

A: Yes, however, we do recommend refreshing the pre-req knowledge before the class starts.

A: Yes! This class is open to any form of auditing, please refer to the self-study guides for more details.

A: As a CMU auditing student, you'll have full access to in-person lectures, Discord discussions, and office hours just like registered students! The main differences are that (1) you won't receive 6 credits, (2) your work won't be graded, and (3) you'll need to find your own computational resources (see the compute resources guide). Check out the full access comparison table for details.

A: Absolutely! This class is completely open-sourced. All lecture recordings, notes, homework descriptions, code, and resources will be publicly available online. While you won't have access to in-person lectures, Discord, or office hours, you can still learn all the material and email the instructor with questions. Please refer to the self-study guides for more details.

A: Many questions in the homework and quizzes are meant to be open-ended so there won't be any standardized solutions. However, we will post both reference code solutions and the written solutions for the questions that have standardized answers online after they have taken place in the class.

A: The final presentation is primarily for registered students who need grades. However, if you are at CMU and you've completed a project you're proud of, talk to Kelly about potentially showcasing it!

A: Discord is currently limited to CMU students only. Invite links will be shared via email to waitlisted students and posted in the first lecture. If you're a CMU student attending the class and didn't receive an invite, please reach out to Kelly.

A: Unfortunately no. All students in the class will have to work on the same dataset.

A: The final presentation will be in-class poster presentation during the last week of class. More details to be posted.

A: You can feel free to additional contents besides the required ones. However, the final presentation needs to contain your efforts made in the class and throughout the homework in order to receive credits. If you only present your research outside of the class it will not be counted as a valid final presentation.

A: You should complete each assignment independently. Discussion with classmates is encouraged, but don't copy code directly and all collaboration should be disclosed. All submitted code must be your own work. You can feel free to work with any AI systems :).

A: You have 6 late days total for the semester. Each assignment can use up to 2 late days.

A: You'll primarily use Python with PyTorch or JAX. We'll provide starter code for your homework.

A: Absolutely! This course is ChatGPT-friendly and open-internet. You're encouraged to use pre-trained models, open-source GitHub repositories, and AI coding assistants. Just make sure to provide proper citations and ensure your final implementation demonstrates originality and understanding.

A: There will be seven short (≤10 minutes) in-class quizzes. The lowest two scores will be dropped, and each quiz counts for 3% of your total grade. Quizzes are closed-book and closed-internet but should be easy with no trick questions.

A: Generally no, but since the lowest two scores are dropped, you have some flexibility. Medical emergencies with documentation may be exceptions.

A: Nope, we only keep track of the top 5 quiz scores so if you are happy with your previous quizzes, feel free to skip the last two.

A: Attendance is required for in-class quizzes and the final presentation week. Otherwise, feel free to watch recordings at your own pace if that works better for you.

A: No! There are no traditional exams. Assessment consists of homework assignments (70%), in-class quizzes (15%), and a final presentation (15%).

A: Since this is the first iteration of the course, we will have to decide whether/how to curve the final grades based on the class performance when it takes place. However, we believe that the goal of education is to facilitate learning rather than discrimination. Therefore, this class will never be curved down (i.e. we will only bump up the letter grades if we ever decide to curve).

A: Yes! We will sprinkle opportunities to earn extra credits throughout the class (details still TBD for now). Please check out the syllabus and announcements for updates later.

A: While personal laptops can handle some work, you'll need GPU access for training diffusion models. Registered students will receive course-provided compute resources from our sponsors (thank you [Modal](https://modal.com/)). Auditing students and students outside of CMU should check out the comprehensive compute resources guide which includes free options, grants that you can apply for, and paid options.

A: Feel free to use any computational resources you'd like! If you have access to compute on your own, go for it!

A: No required textbook! Check out the resources page for curated monographs, tutorials, and key papers on diffusion models and flow matching.

A: Multiple support channels available! Office hours (check the schedule for times and locations), Discord server for course discussions (CMU students only), and email Kelly at yutonghe[at]andrew.cmu.edu for private matters or if you're outside CMU.

A: Yes! All lecture recordings, slides, homework descriptions, code, and resources will be publicly available online. This makes the course accessible to anyone worldwide, though only CMU students get access to Discord and office hours.