Large Language Models (LLMs) such as ChatGPT and Gemini are widely deployed in the real world, powering applications that range from conversational assistants, coding assistants, autonomous agents, creative content generation, and scientific research support. This has fundamentally transformed how people access information, automate tasks, and innovate across domains such as education, healthcare, finance, and entertainment.
This course will involve two parts.
- In the first part, we will learn the foundation of LLMs, including model architecture, pre-training, and post-training.
- In the second part, we will discuss diverse topics on trustworthy LLMs when deploying them in practice, including safety, prompt injection attacks/defenses, watermarking AI-generated content, privacy, knowledge corruption attacks/defenses, explanation/interpretation, and so on.
