I work at the UK AI Safety Institute, on research to support AI governance and evaluations for advanced AI systems.

I am also a 3rd year PhD student at the University of Cambridge in the Cambridge Applied Machine Learning Lab, Department of Engineering with the AI4ER PhD programme.

Throughout 2022 and 2023 I was a researcher at EleutherAI & CarperAI working on large language models, including the design and development of OpenELM, a framework for the generation of diverse and high-quality synthetic data with LLMs.

My research interests are primarily related to large language models, including reinforcement learning from human feedback (RLHF), open-endedness with LLMs, interpretability, evaluations, AI governance, and more. For a list of publications see my Google Scholar or Semantic Scholar profiles.

You can follow me on Twitter, where I mostly retweet interesting research in machine learning along with the work of myself and my collaborators.

I can be contacted at mail [at] firstnamelastname.com.

About - {"twitter"=>"herbiebradley", "name"=>"Herbie Bradley"}