This is a project for setting up chatGPT for local use in Python and Docker
To be able to use the project, some configuration files are needed (a single one for now):
- .env file at the project root
The .env
file is used to store credentials for connections to databases for instance. Since passwords are sensitive,
this file should never be committed as is. Instead, a CI system will be implemented to insert the sensitive data in a
template .env
file.
To setup your local environment, create at the project root a .env
file with:
# OPEN_AI_KEY
OPEN_AI_KEY=YOUR_API_KEY
# GPT Models
CHAT_MODEL=CHAT_MODEL
TEXT_COMPLETION_MODEL=TEXT_COMPLETION_MODEL
Fill the missing fields with their appropriate values.
The application can simply be run by using this code in the terminal:
- run application:
python -m main