Lecture series

This is the second edition of a popular 16 lecture series.

We will explain how language-model systems are built in order to understand and predict their behaviour. Frontier language models are now being used as the foundation for agentic systems, which can carry out tasks that require extended reasoning and long-horizon planning. We investigate the potential safety and security risks associated with such systems, and present current research directions that aim to mitigate them.

The series is designed to be accessible for a broad audience across academia and industry, requiring knowledge from an introductory course in machine learning or statistics (e.g. backpropagation). We emphasise conceptual understanding of such systems, but will discuss technical details where necessary.

We hope that the course will empower researchers to make better use of language model systems and inform deployment across academia and industry. We also hope to stimulate engagement with the serious risks associated with intelligent systems, and encourage further work to address them.
 

Sign up and stay in touch

If you would like to be informed of any updates regarding the course, please sign up to the course mailing list via the form below.

Time and location

All lectures will be held on Tuesdays and Thursdays from 17:00-18:00 in LR4, Cambridge University Engineering Department.

Programme

Part I. What is a Language Model? 
1. Introduction to Language Models (Tuesday October 14th)
2. The Transformer Architecture (Thursday October 16th)
3. Scaling Laws (Tuesday October 21st)

Part II. Crafting Agentic Systems
4. Post-Training (Thursday October 23rd)
5. Reinforcement Learning for Language Models (Tuesday October 28th)
6. Reward Modelling  (Thursday October 30th)
7. Agents and Agent architectures (Tuesday November 4th)

Part III. Agentic Behaviour
8. Optimisation and Reasoning (Thursday November 6th)
9. Reward Hacking and Goal Misgeneralisation (Tuesday November 11th)
10. Out-of-Context Reasoning and Situational Awareness (Thursday November 13th)
11. Deceptive Alignment and Alignment Faking (Tuesday November 18th)

Part IV. Frontiers
12. Threat modelling, Safety Cases, and Systemic Risk (Thursday November 20th)
13. Evaluations (Tuesday November 25th)
14. AI Control (Thursday November 27th)
15. AGI labs and the AI Safety Landscape (Tuesday December 2nd)
16. The Future of Language Models (Thursday December 4th)

More information

This series is run by Meridian Cambridge, in association with C2D3. More information about the course content and organisers can be found here: https://www.meridiancambridge.org/language-models-course-in-person

Recordings from the first series

This second series will be delivered live and will not be recorded. We encourage live participation to get the most out of this interactive series. However you can catch-up on missed lectures through the playlist from the first lecture series. Recordings can be found here.

First series, first lecture