CC-BY-NC blocks commercial fine-tuning — is a commercial variant planned?
#6
by hardik90 - opened
Hi @ai4bharat — this is one of the most-downloaded Indic datasets (4,600+ downloads) and the cc-by-nc-4.0 tag is correctly set, so thank you for the clean metadata.
The practical issue: the NC restriction silently disqualifies this dataset for commercial fine-tuning or RAG corpora, and my license scan (https://huggingface.co/datasets/hardik90/indic-dataset-license-matrix) shows it as the top non-commercial trap — teams often fine-tune on it unaware of the NC clause.
Questions for the team:
- Is a commercial-use variant or dual license under consideration?
- Would a permissive re-release (e.g. of a filtered subset) be possible?
- If not, is there a recommended permissively-licensed substitute for paraphrasing/QA augmentation in Indic languages?
This info would be very valuable for the community — a clear statement in the card either way helps everyone plan their stack.