Data Source Scout

What it does

Turns your criteria into several Hugging Face Hub searches and pulls each candidate's license, size, modality and update date from the Hub. It then ranks the best fits and explains what it ruled out.

What you provide

Any of these: task, domain, modality, size, license needs, exclusions. Partial requests are fine, and it states its assumptions.

What you get back

The searches it ran, up to 8 ranked datasets (each with a link, why it fits, and its license, size, modality, downloads and last update), and notes on excluded near-misses.

Rules it follows

• Searches the Hugging Face Hub only. License and size come from Hub tags, and missing details show as "not specified."
• License info is a starting point, so check it yourself before commercial use.
• It doesn't download the data, so it can't judge label quality.
• If the Hub can't be reached, the job fails with a notice instead of reporting "no matches."