RHC
Menu

← All projects

ML evaluation

labelbudget

An active-learning benchmark that reports labels needed to reach a target accuracy, with variance across many seeds, next to a prominent random-sampling baseline. A comparison guard refuses claims the evidence does not support.

labelbudget — Concept render
Concept render

Drawing

Chart: active learning reaches the target accuracy with fewer labels than the random baseline

Highlights

  • Target accuracy is 97% of the full-pool score; the random baseline is always run.
  • Paired bootstrap CI and exact sign test, with no comparative claim below 10 seeds.
  • Pure Python standard library with 25 tests.