Spaces:
No application file
Apply for a GPU community grant: Academic project
Dear Hugging Face Team,
I am writing on behalf of "Open Kurdish AI," a non-profit initiative dedicated to democratizing AI for the Kurdish language (spoken by 40+ million people globally but severely under-resourced in the digital space).
We have successfully aggregated the world's largest open-source Kurdish data collection, including:
The Grand Corpus (2B+ tokens).
Mozilla Common Voice & AsoSoft Speech datasets.
Aya Collection (Instruction tuning).
You can verify our organized collection here: https://huggingface.co/collections/OpenKurdishAI/the-grand-corpus-of-kurdish
We are now ready to fine-tune a foundation model (Llama-3-8B or Mistral) to create the first robust "Instruct-Kurdish" model. However, as a community-driven project, we lack the compute resources.
We kindly request a Community GPU grant (A10G or A100) to train this model. This project aligns perfectly with Hugging Face's mission of open science and preserving linguistic diversity.
Thank you for empowering us to build our digital future.
Best regards,
The Open Kurdish AI Team