CAMBRIA — Cambridge Boston Alignment Initiative
CAMBRIA (Cambridge Bootcamp for Research in Interpretability and Alignment) is a 3-week ML upskilling bootcamp for AI safety, focusing on interpretability and RL, based on the ARENA curriculum and run by the Cambridge Boston Alignment Initiative.
CAMBRIA (Cambridge Bootcamp for Research in Interpretability and Alignment) is a 3-week ML upskilling bootcamp for AI safety, focusing on interpretability and RL, based on the ARENA curriculum and run by the Cambridge Boston Alignment Initiative. Prerequisites are familiarity with Python and comfort with multivariable calculus (e.g., chain rule) and linear algebra (e.g., eigenvalues, SVD). For a more detailed list of helpful prior knowledge, see here. Participant applications for the Summer 2026 CAMBRIA cohorts are now closed. TA applications for our August 10-28 cohort in Cambridge, MA…
saved by
related reading
- Fall 2026boazbk.github.io
- Prospects for Alignment Automation: Interpretability Case Study — LessWronglesswrong.com
- The Iliad Intensive Course Materials — LessWronglesswrong.com
- Student Projects - CS 2881R AI Safetyboazbk.github.io
- ARENA - AI Safety Curriculumlearn.arena.education
- A woefully incomplete guide to technical upskillingjason.ml
- A Pragmatic Vision for Interpretability — AI Alignment Forumalignmentforum.org
- AI Alignmentcourse.aisafetyfundamentals.com
- GitHub - jacobhilton/deep_learning_curriculum: Language model alignment-focused deep learning curriculumgithub.com
- ARENA 6.0 Impact Report — LessWronglesswrong.com
- In (highly contingent!) defense of interpretability-in-the-loop ML training — AI Alignment Forumalignmentforum.org
- Tim Hua Personal Websitetimhua.me