Jourloy / ruGPT3-Django

Django server with ruGPT3 models

ruGPT3 server

Getting start

Change settings

Look into src/gpt3/settings.py and change SECRET_KEY and ALLOWED_HOSTS!

About ENV

Project have two env variables, look into .env.temlpate. If you are want to use Docker-Compose way, so you should create .env file. HOST_IP is allowed IPs for you backend (0.0.0.0 = Allow all). HOST_PORT is a port for your backend

If you are want to use Docker way, so you should pass this variables in command

Starting

Docker-Compose

You can just start docker-compose

$ docker-compose up -d

Done, now you can use endpoints

Warning

You should have too many memory for start docker. If docker build was killed use Install packages way.

Docker

You can build a docker container

$ docker build --build-arg HOST_IP=0.0.0.0 --build-arg HOST_PORT=3000 -t rugpt3 .

And run it

$ docker run -d --name ruGP3-Apache rugpt3

Warning

You should have too many memory for start docker. If docker build was killed use Python way

Python

Install packages

$ pip install requirements.txt

Start

In example below 0.0.0.0 is a IP. If you don't want share access and use local only, just remove this IP. 10000 is a port for app

MacOS / Linux

$ python manage.py runserver 0.0.0.0:10000

Windows

$ py manage.py runserver 0.0.0.0:10000

Endpoints

You can change it in src/rpute/urls.py

ruGPT3

Generate text on Russian

Endpoint: /gpt3/text?text=&maxLength=

• text: string. Text for start.
• maxLength: number. Recomend max length for text.

ruTitle

Generate title for text on Russian

Endpoint: /gpt3/title?text=

• text: string. Title will be generated for this text.

ruDolph

Generate image by a text on Russian

Endpoint: /gpt3/image?text=

• text: string. Short description for image

Links

ru-promts

ruGPT3

ruDolph