Stanford CME295: Transformers and Large Language Models | Autumn 2025
Afshine Amidi and Shervine Amidi
Stanford’s Autumn 2025 course on Transformers and large language models, taught by Afshine and Shervine Amidi in nine lectures of almost two hours each. It begins with tokenization, embeddings and the attention mechanism, then covers how LLMs are pretrained and tuned, how they reason, how they act as agents and how they are evaluated. The last lecture reviews the course and current trends in the field.
What you'll learn
- Tokenization, embeddings and self-attention
- The Transformer architecture and the models built on it
- Pretraining and fine-tuning large language models
- Reasoning models and agentic LLMs
- Evaluating LLMs
