Devlog

The phenotype catalog goes open on Hugging Face

All 484 groups, each with an original profile of 300 to 450 words, published as an open CC BY 4.0 dataset one load_dataset() call away.

  • Data
  • Research
  • Catalog

On May 5 the phenotype catalog was published as an open dataset on Hugging Face, EthnicErotic/phenotype-catalog, under CC BY 4.0. It covers all 484 groups in the catalog plus the atlas reference, in CSV and JSONL.

The same day every one of the 484 group pages gained an original profile of 300 to 450 words, and the full set went into the dataset. The dataset is built as a structured complement to Wikipedia: Wikipedia carries the long prose, the dataset carries machine-readable fields.

Under the hood

  • Two configs at launch: ethnicities (484 rows) and atlas (21 trait categories), each shipped as CSV and JSONL and loadable with a single load_dataset() call.
  • Every row carries a canonical URL back to its page, so the dataset and the live catalog stay linked record by record.
  • A language model drafted each profile from the group's structured record (homeland, region, subgroup, language and religion) under an editorial style guide for present-tense reference prose, with output held to a small set of HTML elements.
groups, each with a profile
484
words per profile
300 to 450
atlas categories in v1
21

See it on the site

Sources and open data

Read the full log