VOvoidly
Classifier

Censorship classifier v3 metadata

Changed on

Version, training date, full feature list, and evaluation. Read evaluation.honest_evaluation.loco_mean_f1 as the declared headline; the LOCO median is inflated by small-sample perfect scores. Read honest_caveat.training_label_composition before interpreting F1 as censorship accuracy.

get/v1/classifier/info

Request

  • Base URL
    https://api.voidly.ai
  • URL
    https://api.voidly.ai/v1/classifier/info
  • Auth
    none required

Response

Classifier info + LOCO evaluation

Changes