This Stanford CME295 course provides a comprehensive introduction to Transformers and Large Language Models (LLMs), designed for students and professionals interested in modern artificial intelligence and deep learning systems.
The course begins with the fundamentals of Transformer architecture, explaining how attention mechanisms work and how they form the backbone of today’s AI models. It then expands into Transformer-based models and practical techniques used to improve performance in real-world applications.
Students will explore how LLMs are trained, fine-tuned, and optimized, including key concepts such as training pipelines, model scaling, and adaptation strategies. The course also covers advanced reasoning capabilities in LLMs, helping learners understand how AI systems process complex tasks and generate meaningful outputs.
Further lectures introduce agentic LLMs, where models are designed to perform tasks autonomously, as well as evaluation techniques used to measure model performance and reliability. The final lectures provide a recap of modern trends in AI research and the future direction of large-scale language models.
By the end of this course, learners will have a strong theoretical foundation in Transformers and LLMs, along with a clear understanding of how modern AI systems are built, evaluated, and improved in real-world environments.
