Skip to content

References: Core Concepts of Large Language Models

  1. Large language model - Wikipedia - Overview of LLM architecture, training, and capabilities, including how these models differ from earlier statistical language models. Grounds this chapter's central object of study.

  2. Transformer (deep learning architecture) - Wikipedia - Detailed coverage of the transformer architecture, including tokenization, embeddings, and the scaled dot-product self-attention mechanism this chapter introduces conceptually. Includes diagrams of the encoder-decoder structure and multi-head attention.

  3. Machine learning - Wikipedia - Comprehensive overview of machine learning as a field, its relationship to AI and deep learning, and its major learning paradigms. Supports the chapter's AI/ML/DL/NN family-tree discussion.

  4. Deep Learning - Ian Goodfellow, Yoshua Bengio, and Aaron Courville - MIT Press - The field's standard reference; credited for establishing much of deep learning's now-common notation and its clear layered explanation building from single neurons to deep, multi-layer architectures.

  5. Speech and Language Processing (3rd ed. draft) - Dan Jurafsky and James H. Martin - Stanford University (online draft) - Widely credited for one of the clearest published explanations of the transformer and self-attention mechanism, using concrete worked examples other NLP courses have since adopted as a teaching standard.

  6. The Illustrated Transformer - Jay Alammar - The most widely cited visual walkthrough of self-attention and the transformer architecture, using step-by-step diagrams that clarify how a model relates distant words to one another.

  7. What Are Large Language Models (LLMs)? - IBM Think - Practitioner-oriented explainer covering LLM parameter scale, training cost, and the small-vs-large-model tradeoff this chapter frames in economic rather than purely technical terms.

  8. What Is an API? - IBM Think - Explains APIs as the request/response bridge between client and server, the foundational concept behind the chapter's discussion of API endpoints and model providers.

  9. What are Large Language Models? - NVIDIA Glossary - Industry explainer with concrete figures on parameter counts and training compute cost, reinforcing the chapter's training-versus-inference cost distinction with real dollar amounts.

  10. What Is a Multimodal LLM (MLLM)? - IBM Think - Explains how multimodal models process and reason across text, images, and audio together as a single system, directly supporting the chapter's multimodal model discussion.