CaMLAB

This 4th iteration of CaMLAB (Cambridge Machine Learning for Alignment Bootcamp) will run as a two-week full-time program between April 8-21 in Cambridge, UK, and we have space for 20-25 participants.

The program aims to provide the machine learning skills which are necessary to work on a range of technical problems in AI alignment, such as mechanistic interpretability. We aim to build up the ML engineering fundamentals you need for technical AI Safety research, like:

  • The basics of PyTorch

  • Building, training and tuning models like GPTs and ResNets

  • Interpreting language models with TransformerLens

The deadline for applying is Sunday 25 February, 23:59 GMT+0.

Please contact Dron at dron@cambridgeaisafety.org if you have any questions.

If you’re an experienced in ML, apply to be a TA.

FAQ

The program is designed for participants with strong quantitative skills and familiarity with Python programming, and a demonstrated interest in AI Safety; no prior ML experience is required. We consider applicants with less experience if we expect them to be able to get up to speed in time for the program. If you're unsure whether you'd be a good fit for this camp, we encourage you to err on the side of applying!

The program is free to attend and travel/accommodation support will be available for participants.

  • We expect applicants to have a demonstrated interest in AI Safety, either by doing our AI Safety Intro Fellowship, an equivalent fellowship elsewhere, or demonstrating equivalent knowledge and engagement.

    We think you’d be a particularly a good fit if:

    • You’re keen to work on technical AI safety, but are limited by a lack of ML experience

    • You’re interested in AI governance and want a deeper technical understanding of ML

  • We expect applicants to be:

    • Confident programmers (preferably in Python)

    • Comfortable with first-year undergraduate maths, especially linear algebra

  • Yes! Our cohorts include both graduate and undergraduate students.

  • Apply here, and we’ll be in touch!

  • Yes, we plan to have a remote stream. Please select this option in your application if you can only attend remotely.