Stanford Publishes CME295 Transformers Course Materials

Stanford’s CME295 Transformers and Large Language Models course from Autumn 2025 is published openly, providing complete curriculum materials including nine recorded on-campus lectures, slides, and midterm and final exams with solutions. The nine-lecture course covers tokenization, attention, decoding, mixture-of-experts, scaling laws, fine-tuning methods like LoRA and RLHF, retrieval-augmented generation, agentic LLMs, evaluation, quantization and optimization, and is recommended for learners with basic linear algebra, ML and Python.