metadata
datasets:
- breadlicker45/gender-bluesky-classification-v4
base_model:
- Lajavaness/bilingual-embedding-base
It's 48% accurate at guessing male, female, and non-binary. (random guessing is 33.33% accurate)
But I am aware it is biased towards males because of a data diversity issue.
The diversity issues being I collected different post amounts based on each user's account gender.
I plan to fix this in v5.