dbtool
@dbtool.bsky.social
Open models & data for meta-research.
genderize (gender + country from a name): open weights, 98% accuracy, CPU only.
Now building in public: sample size + PICO from clinical abstracts.
EU-hosted API · free research keys · dbtool.it
genderize (gender + country from a name): open weights, 98% accuracy, CPU only.
Now building in public: sample size + PICO from clinical abstracts.
EU-hosted API · free research keys · dbtool.it
Build log, week 1. What's next at dbtool: one API call that turns a clinical abstract into sample size + PICO.
One disk, one consumer GPU, one box. Every number ships with its test set; the public endpoint waits until it clears 80% on hand-annotated abstracts.
Follow along → https://dbtool.it
One disk, one consumer GPU, one box. Every number ships with its test set; the public endpoint waits until it clears 80% on hand-annotated abstracts.
Follow along → https://dbtool.it
September 13, 2026 at 2:30 PM
Build log, week 1. What's next at dbtool: one API call that turns a clinical abstract into sample size + PICO.
One disk, one consumer GPU, one box. Every number ships with its test set; the public endpoint waits until it clears 80% on hand-annotated abstracts.
Follow along → https://dbtool.it
One disk, one consumer GPU, one box. Every number ships with its test set; the public endpoint waits until it clears 80% on hand-annotated abstracts.
Follow along → https://dbtool.it
genderize cora + ultra are out as open weights: byte-level, CPU-only gender + country (226 ISO codes) from a name. Gender 97.9%/98.2%, country top-1 82.6%/83.7% on 25k unseen names, network alone. CC BY-NC 4.0.
September 12, 2026 at 5:10 PM
genderize cora + ultra are out as open weights: byte-level, CPU-only gender + country (226 ISO codes) from a name. Gender 97.9%/98.2%, country top-1 82.6%/83.7% on 25k unseen names, network alone. CC BY-NC 4.0.
genderize cora & ultra are now open weights: a personal name → probable gender + a ranking of 226 countries, from 48 UTF-8 bytes. No tokenizer, no dictionary, CPU only. 98.2% gender accuracy on 25,000 held-out names. Weights CC BY-NC 4.0, script MIT.
https://huggingface.co/textpie/genderize
https://huggingface.co/textpie/genderize
September 12, 2026 at 5:07 PM
genderize cora & ultra are now open weights: a personal name → probable gender + a ranking of 226 countries, from 48 UTF-8 bytes. No tokenizer, no dictionary, CPU only. 98.2% gender accuracy on 25,000 held-out names. Weights CC BY-NC 4.0, script MIT.
https://huggingface.co/textpie/genderize
https://huggingface.co/textpie/genderize