Course notes / Autumn 2025

CME295Transformers & LLMs

Nine lectures, from the foundations of attention to reasoning and agents. Explore detailed study notes, lecture slides, and questions for review.

Stanford CME29509 lecturesEnglish notes

The lectures

01 — 09
  1. 01

    Transformers

    From tokens and embeddings to self-attention and the transformer architecture.

  2. 02

    Transformer-based models & tricks

    Position embeddings, normalization, attention variants, and efficient inference.

  3. 03

    Transformers & large language models

    How transformer architectures become large language models.

  4. 04

    LLM training

    Data, pretraining objectives, and the foundations of training at scale.

  5. 05

    LLM tuning

    Fine-tuning, human preferences, and aligning model behavior.

  6. 06

    LLM reasoning

    Reasoning methods, reinforcement learning, and inference-time computation.

  7. 07

    Agentic LLMs

    Tool use, retrieval, and language models that act across multiple steps.

  8. 08

    LLM evaluation

    Benchmarks, human judgments, and measuring model quality.

  9. 09

    Recap & current trends

    A course review, multimodal models, and emerging directions for LLMs.

The original CME295 lectures, videos, slides, and other course materials belong to their respective course creators and copyright holders. Eyegorithm does not claim ownership of those materials.

© 2026 Eyegorithm LLC. Eyegorithm reserves all rights in its original, copyrightable contributions to these recreated study notes, including its writing, commentary, and organization. This claim excludes the underlying course materials and any third-party content.

These independently prepared study notes are provided for learning and reference. They are not official course materials and are not affiliated with or endorsed by Stanford University or the course creators. Any errors in the notes are Eyegorithm’s own.