MGTEVAL: An Interactive Platform for Systemtic Evaluation of Machine-Generated Text Detectors

TL;DR

MGTEVAL is an extensible platform designed to systematically evaluate machine-generated text detectors across various datasets, attacks, and metrics, enhancing reproducibility and comparability.

Contribution

It introduces a unified, modular platform that streamlines the evaluation process of MGT detectors with customizable benchmarks and attack simulations.

Findings

Supports constructing custom benchmarks with configurable LLMs.

Applies 12 text attacks to test sets for robustness evaluation.

Provides comprehensive performance reports including effectiveness, robustness, and efficiency.

Abstract

We present MGTEVAL, an extensible platform for systematic evaluation of Machine-Generated Text (MGT) detectors. Despite rapid progress in MGT detection, existing evaluations are often fragmented across datasets, preprocessing, attacks, and metrics, making results hard to compare and reproduce. MGTEVAL organizes the workflow into four components: Dataset Building, Dataset Attack, Detector Training, and Performance Evaluation. It supports constructing custom benchmarks by generating MGT with configurable LLMs, applying 12 text attacks to test sets, training detectors via a unified interface, and reporting effectiveness, robustness, and efficiency. The platform provides both command-line and Web-based interfaces for user-friendly experimentation without code rewriting.