Openly-licensed, consented parallel corpora for under-served languages — with a clear schema, consent-first collection, and per-pair provenance — to help build translation for languages the big datasets ignore.
# Low Resource Parallel Corpora
Openly-licensed, consented parallel corpora for under-served languages — with a clear schema, consent-first collection, and per-pair provenance — to help build translation for languages the big datasets ignore.
---
This is a **Hee-Lee Oss** good-deed project. Contributors pull a task, do it with their own coding agent, and open a PR. Get started:
github.com
See `PLAN.md` for the project plan and `tasks/` for open tasks.