Anti GPT checker

This project aims to provide a reliable way of distinguishing between AI generated text and human written one. It is the implementation part of my Masters's Thesis at the Warsaw University of Technology. It is also a part of ANANAS project at the Warsaw University of Technology.

Installation

• Python 3.10 or higher is required, with python-dev and pip installed.
• Start by installing stylometrix following the instructions in the stylometrix repository. This includes:
  • Installing spacy according to their instructions, select Polish and English models, and pipeline for accuracy
  • Downloading and installing pl_nask model python -m pip install <PATH_TO_MODEL/pl_nask-0.0.7.tar.gz>
  • Installing stylometrix pip install stylometrix
• Initialize Polish dictionaries:
  • SJP download the newest available
  • NKJP (1-gram)
  • Unpack the archives and move files odm.txt and 1grams into data directory inside the project
• Install the rest of the dependencies pip install -r requirements.txt
  • If install fails on pycld3 run sudo apt-get install -y libprotobuf-dev protobuf-compiler gcc g++

.env files

Create a .env file for the core of anti-gpt-checker in the root directory of the project with the following fields:

MONGODB_URI=""
MONGODB_PORT=""
MONGODB_DB_NAME=""
MONGODB_AUTH_USER=""
MONGODB_AUTH_PASS=""
MONGODB_AUTH_DB="" # usually admin
PERPLEXITY_POLISH_GPT2_MODEL_ID="sdadas/polish-gpt2-xl"
PERPLEXITY_POLISH_QRA_MODEL_ID="OPI-PG/Qra-13b"
PERPLEXITY_ENGLISH_GPT2_MODEL_ID="openai-community/gpt2-large"
RELATIVE_PATH_TO_PROJECT="" # relative path to the project directory from working directory
DICT_FILE_1GRAMS_PATH="data/1grams"
DICT_FILE_ODM_PATH="data/odm.txt"

Create a .env file for the API part of anti-gpt-checker in the api/ directory with the following fields:

API_MONGODB_URI=""
API_MONGODB_PORT=""
API_MONGODB_DB_NAME=""
API_MONGODB_AUTH_USER=""
API_MONGODB_AUTH_PASS="" 
API_MONGODB_AUTH_DB="" # usually admin
API_DEBUG="" # True or False
API_HISTOGRAMS_PATH=""

Docker

To run the project using Docker, you need to complete the Initialize Polish dictionaries step first. The docker-compose file will use the dictionaries from the data directory.