Overview

Large language models are widely used but often poorly understood. This course explains the core mechanisms that connect text, vector representations, attention, transformer architectures and next-token prediction.

The course separates the statistical objective used during training from the behaviours that emerge at deployment. Learners examine pretraining, fine-tuning, retrieval, context windows, prompting, hallucination and evaluation.

The aim is not to train a frontier model. It is to build enough technical understanding to reason about what an LLM can do, why it sometimes appears to reason and where its outputs remain unreliable.

Learning objectives

By the end of the course, learners should be able to:

• explain tokenization, embeddings and attention at a conceptual and mathematical level;
• describe how transformer models perform next-token prediction;
• distinguish pretraining, fine-tuning, retrieval and inference;
• explain why fluent outputs can still be false or unsupported;
• evaluate the effect of context, prompting and tool use;
• identify tasks for which LLMs require external evidence, computation or human review.

Intended participants and prerequisites

Course structure

The course is organised into six units. Each unit combines concise explanation, worked examples and short analytical exercises.

Unit	Subject	Principal questions
1	Text as tokens	How is language converted into numerical objects?
2	Embeddings and representation	How are semantic relationships encoded in vector space?
3	Attention and transformers	How does a model combine information across a sequence?
4	Training and scaling	What objective is optimized, and what changes as models scale?
5	Inference and tool use	How do prompting, retrieval and external tools change model behaviour?
6	Failure and evaluation	Why do hallucination, brittleness and weak verification persist?

Teaching and assessment

Teaching is delivered through concise lectures, technical notes, worked examples and guided exercises. AI-supported interaction may be used to explain concepts, generate additional practice and challenge a learner's reasoning, but assessment remains based on the learner's demonstrated understanding.

Component	Purpose	Weight
Short exercises	Check understanding of the principal concepts	40%
Applied investigation	Use the course framework on a new problem or dataset	30%
Final assessment	Integrated reasoning across the course	30%

A certificate of completion is issued only when the required assessments have been completed at the stated standard. Participation alone does not constitute completion.

Faculty

Materials and technical requirements

Learning environment

Learners need a current web browser and access to the Gordon learning platform. Where a course uses code, spreadsheets, datasets or external readings, the relevant requirements are identified inside the unit.

Course materials

Required readings, datasets and practice materials are identified within each unit. Public courses are designed so that required core materials can be accessed without a separate tuition charge.