- Students interested in the course should first try enrolling in the course in CalCentral. The class number for CS194-196 is 32397. The class number for CS294-196 is 32392. Please join the waitlist if the class is full. Please fill in the signup form if and only if you can neither enroll in the class nor join the waitlist, and we will try to get back to you as soon as we can.
- For general course content related questions, please join our edstem.
- Do not email the course staff or TAs. For private matters, post a private question on edstem and make sure it is visible to all teaching staff.
Course Staff
 |
 |
| Prof. Dawn Song | Prof. Matei Zaharia |
| UC Berkeley | UC Berkeley |
Guest lecturer & project mentor: John Whaley
GSI: Yu Gai
Reader: Shiladitya Dutta and Bill Zheng
Course Description
Generative AI and foundational models including ChatGPT have ushered the world into a new era with rich new capabilities for wide-ranging application domains. With these new capabilities also comes unprecedented challenges such as privacy, safety, ethics, alignment, and decentralization. In this class, we will introduce foundations of Large Language Models (LLMs), discuss the infrastructure, tooling, and best practices for building and running applications with LLMs, and explore the risks and challenges with these technologies and how we can build towards a responsible and democratized and decentralized future with AI. In particular, this class will cover a wide-ranging topics including:
- Foundations of LLMs
- Tech stack for building LLM applications
- Sample applications domains including code generation, data science, computer security, etc.
- Autonomous agents; personalized virtual assistants
- Trustworthy LLM, including trustworthiness evaluation & benchmarks, privacy issues in LLM, attacks on LLMs such as prompt injection
- Open source & decentralization of LLM stack, including decentralized training/inference, federated learning, open source models, open source data, edge inference
- Decentralized decision making and democratized AI governance
Syllabus (subject to change)
| Date | Topic |
|---|---|
| Aug 27 | Join The Future of Decentralization, AI, and Computing Summit! |
| Aug 29 | No class |
| Sep 5 | Intro (slides) & Foundations of LLM (slides) Guest speaker: Eric Wallace, UC Berkeley |
| Sep 12 | Infrastructure Layer I: Training and Inference, Performance Optimization, Scalability (slides) Guest speaker: Jonathan Frankle, MosaicML |
| Sep 19 | Infrastructure Layer II: Retrieval, Vector Databases, Search (slides) Guest speaker: Jerry Liu, LlamaIndex (slides) |
| Sep 26 | App Development Layer: Prompt Engineering, Chains, Tools |
| Oct 3 | Application Domains I: Software Engineering/Code Generation, Data Science |
| Oct 10 | Application Domains II: Security, Education |
| Oct 17 | Agents: RPA, Virtual Assistants |
| Oct 24 | Decentralized Training and Inference, Open-Source Models Guest speaker: Ce Zhang, Together |
| Oct 31 | Trustworthiness: Privacy, Hallucinations, Adversarial Attacks Guest speaker: Nicholas Carlini, Google DeepMind |
| Nov 7 | Decentralized Decision Making |
| Nov 14 | Ethics and Fairness, Safety, Alignment |
| Nov 21 | Edge compute; Federated learning; Open source data Guest speaker: Tianqi Chen, CMU |
| Nov 28 | Project Demos |
Enrollment and Grading
Students are strongly encouraged to have had experience and basic understanding of Machine Learning and Deep Learning before taking this class, e.g., have taken courses such as CS182, CS188, and CS189. Graduate students should enroll in CS294-196. Undergraduate students should enroll in CS194-196.
If you are not able to enroll in the class, please fill out the signup form and we will get back to you as soon as possible.
This is a variable-unit course. All enrolled students are expected to participate in lectures in person. Students enrolling in one unit are expected to submit an article that summarizes either one of the lectures or the Future of Decentralization, AI, and Computing Summit. Students enrolling in more than one units are expected to submit a lab assignment and a project instead of the article. The project of students enrolling in 2 units should have a written report, which can be a survey in a certain area related to LLMs. The project of students enrolling in 3 units should also have an implementation (coding) component that programmatically interacts with LLMs, and the project of students enrolling in 4 units should have a very significant implementation component with the potential for either real world impacts or intellectual contributions. The grade breakdowns for students enrolled in different units are the following:
| 1 unit | 2 units | 3/4 units | |
|---|---|---|---|
| Participation | 50% | 20% | 10% |
| Article | 50% | ||
| Lab | 20% | 10% | |
| Project | |||
| Proposal | 10% | 10% | |
| Milestone | 10% | 10% | |
| Presentation | 25% | 25% | |
| Report | 15% | 15% | |
| Implementation | 20% |
Lab and Project Timeline
| Released | Due | |
|---|---|---|
| Project group formation | Sep 5 | Sep 19 |
| Project proposal | Sep 12 | Oct 3 |
| Lab | Sep 19 | Oct 17 |
| Project milestone | Oct 3 | Oct 31 |
| Project presentation | Oct 31 | Nov 28 |
| Project final report | Oct 31 | Dec 12 |
Logistics
Lecture: 4-6pm PT Tuesday in Physics Building 1. (Click here if you cannot find the classroom)
- For general course content related questions, please join our edstem.
- Do not email the course staff or TAs. For private matters, post a private question on edstem and make sure it is visible to all teaching staff.
Office Hours
Yu Gai: 2:45-3:45pm PT Tuesday. Location to be announced on edstem.