Skip to content

BigSalmon2/detecting-fake-text

 
 

Repository files navigation

This one was forked but then unforked. Most up to date is detecting-fake-text-2

GLTR: Giant Language Model Test Room

Detecting text that was generated from large language models (e.g. GPT-2).

webpage: http://gltr.io
online-demo: http://gltr.io/dist/index.html
paper: https://arxiv.org/abs/1906.04043

A project by Hendrik Strobelt, Sebastian Gehrmann, Alexander M. Rush.

collaboration of MIT-IBM Watson AI Lab and HarvardNLP

Quickstart

Install dependencies for Python >3.6 :

pip install -r requirements.txt

run server for gpt-2-small:

python server.py

the demo instance runs now at http://localhost:5001/client/index.html

Run the BERT server

start the server for BERT:

python server.py --model BERT

the instance runs now at http://localhost:5001/client/index.html?nodemo. HINT: we only provide demo texts for gpt2-small.

server.py options

usage: server.py [-h] [--model MODEL] [--nodebug NODEBUG] [--address ADDRESS]
                 [--port PORT] [--nocache NOCACHE] [--dir DIR] [--no_cors]

optional arguments:
  -h, --help         show this help message and exit
  --model MODEL		 choose either 'gpt-2-small' (default) or 'BERT' or your own
  --nodebug NODEBUG  server in non-debugging mode
  --port PORT	     port to launch UI and API (default:5001)
  --no_cors          launch API without CORS support (default: False)

Extend backend

The backend defines a number of model api's that can be invoked by the server by starting it with the parameter --model NAME. To add a custom model, you need to write your own api in backend/api.py and add the decorator @register_api(name=NAME).

Each api needs to be a class that inherits from AbstractLanguageChecker, which defines two functions check_probabilities and postprocess. Please follow the documentation within api.py when implementing the class and the functions.

Extend frontend

the source code for the front-end is in client/src.

To modify, installing of node dependencies is necessary:

cd client/src; npm install; cd ../..

re-compilation of front-end:

> rm -rf client/dist;cd client/src/; npm run build; cd ../..

License

Apache 2

(c) 2019 by Hendrik Strobelt, Sebastian Gehrmann, Alexander M. Rush

About

Giant Language Model Test Room, outdated

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • TypeScript 42.8%
  • Python 31.8%
  • HTML 13.2%
  • CSS 7.4%
  • JavaScript 4.8%