Awesome

CheckGPT

ATTENTION: For datasets, scripts, and more experiment code, refer to our newer version and artifact.

Description

The official repository of paper: "Check Me If You Can: Detecting ChatGPT-Generated Academic Writing using CheckGPT".

Table of Contents

• Data
• Features
• Installation
• Usage

Data

Data structure: { {KEY}:{TEXT} }. You may use your own data following this structure.

Features

To turn text into features, use features.py.

Installation

Run

pip install -r requirements.txt

Usage

For training, testing and transfer learning, use dnn.py using this format:

python dnn.py {SUBJECT} {TASK} {EXP_ID} 

Examples:

1. To train a model from scratch on CS and task 1:

python dnn.py CS 1 0001 

2. To test a model on HSS and task 3:

python dnn.py HSS 3 0001 --test 1

3. To evaluate any text, run and follow instructions:

python test.py