Get in touch.

The lab is small and the work is public. Questions, corrections, data, and collaboration are all welcome.

01/Channels

GitHub

Source code for every project. Fork it, send a pull request, or read how the models were trained.

Issues

Bugs, model failures, and questions about training details or usage. The fastest way to get a reply.

Hugging Face

Model weights, datasets, and hosted demos you can try in the browser without installing anything.

For collaboration, research partnerships, or anything that doesn't belong in a public thread.

02/Ways to help

Annotated Khmer text, scanned documents, or transcriptions are the scarcest resource in this whole space. Even small, well-labelled sets are useful.

Real documents that the models get wrong are worth more than another benchmark. Failure cases directly shape the next training run.

Build something on top of the models and say what broke. Nothing surfaces gaps faster than someone trying to ship with them.