Open data and code
What I release and what I do not, with the code and the models. The datasets are listed on the Datasets page.
Code
Released implementations and companion repositories.
Layered self-supervised knowledge distillation
Official implementation of the layered distillation framework for efficient multimodal learning.
2026GitHubNo licence statedMaterials for the PRISMA review of 232 studies on where responsibility for generative AI outcomes sits.
2025GitHubNo licence statedPDRP: personalised drug response prediction
The code behind the lung cancer drug response prediction method published in Scientific Reports.
2024GitHubNo licence statedImplementation of multistage self-supervised augmented consecutive knowledge distillation, from the CVPR workshop paper.
2024GitHubNo licence statedComputation of drug-disease associations from the COVID-19 era work.
2022GitHubNo licence stated
Models and releases in preparation
Not released. Listed so the record of what is coming is complete, and given no title where none is public.
Modelling development at cellular scale rather than segmenting one structure at a time. In development.
2026GitHubNo licence statedCompanion repository for the Perspective on foundation models in biomedical imaging. No public title, because the paper is in preparation.
2026GitHubNo licence stated
On reuse
Where a deposit carries no licence file, I have granted no permission beyond what copyright law allows. Several of these repositories were released as the code accompanying a paper and were never licensed for redistribution, and I would rather say so than imply otherwise.
Where a licence is stated it is printed against the record: the deepfake dataset deposit is the one that names its terms.
Two entries are works in progress rather than releases, and are marked as such. Neither has a public release, and the Perspective companion has no public title.