Generative AI

This course focuses on the practical aspects of LLMs to enable students to be effective and responsible users of generative AI technologies. The course has three parts. Introduction section covers the historical aspects, key technical ideas and learnings all the way to Transformer architectures and various LLM training aspects. The Practical Aspects and Techniques section, students learn how to train, deploy, and use LLMs; and discuss core concepts like prompt tuning, quantization, and parameter efficient fine-tuning, and explore use case patterns. Finally, a discussion of challenges & opportunities offered by Generative AI, which includes highlighting critical issues like bias and inclusivity, fake information, safety, and some IP issue

Student Learning Outcomes

• Appreciate the history of the path towards Large Language Models (LLMs) and Generative AI approaches.
• Be able to understand key use case patterns of Generative AI approaches and know how to think about incorporating them into applications.
• Become aware of critical issues such as bias, inclusivity problems, hallucinations, and IP questions.
• Become conversant in PyTorch and key neural net coding strategies.
• Know how to approach improving the results obtained from LLMs through prompt-tuning, instruction-based fine-tuning, and reinforcement learning with human feedback.
• Understand the foundations of LLMs, how they are trained, and how to deploy and use them, for and beyond text-focused problems.