Skip to content

Draft: Script for benchmarking

Romane Gallienne requested to merge 21-script-benchmarking into main

Script for benchmarking dataset.

TODO :

  • Script to run texts in all detectors and save prediction (scores + predictions)
  • Script to evaluate (accuracy, tpr, fpr) detection : human-text vs machine-generated text
  • Script to have graphs
  • Need ramification according characterization and degradation

Closes #21

Edited by Arnaud Valladier

Merge request reports

Loading