If you are a Python person, and looking for a simple web interface to try out OpenAI ChatGPT API, then this is the repo for you.
This is a Python one-stop shop project that I built myself to try out different context/prompt engineering. I hope you will find it useful.
Tested on MacOS and Ubuntu20 LTS
Python version == python3.8
Technically, it should work on >= python3.7, but you will have to edit the module versions accordingly in install.sh file
- At project root, create and enter a
python3.8virtual environment
cd $PROJECT_ROOT
python3.8 -m venv venv
source venv/bin/activate
- One bash script installation for all
bash ./install.sh
Visit openai to obtain your API key and Organization ID
Then export to env variable
export OPENAI_API_KEY="sk-abcdef..."
export OPENAI_ORG_ID="org-zxcvbb..."
Start the server using bash script
bash ./app.sh start --host 0.0.0.0 --port 8080
You should see the below sample stdout
INFO: Waiting for application startup.
INFO: Application startup complete.
INFO: Uvicorn running on http://0.0.0.0:8080 (Press CTRL+C to quit)
chatgpt launched
--------------------------
Note that this server is detached from your terminal, you're free to close the terminal without interrupting the service.
To stop the server, execute bash ./app.sh stop
Now, open web browser and visit http://0.0.0.0:8080. Enjoy
Step 1: create your own config_lang.py, similar to the following files
./config
├── config_en.py
└── config_zh.py
Step 2: change the import in main.py:32 accordingly
from config.config_en import Args
# from config.config_zh import Args
Step 3: create your own prompts, arranging the files in the following way
./prompts
├── prompts_en
│ ├── chat-agent.txt
│ └── context-summarizer.txt
└── prompts_zh
├── chat-agent.txt
└── context-summarizer.txt
Step 4: edit the config_lang.py created in step 1, make sure its PROMPTS_DIR is assigned with the intended directory. Also edit all the message related variables (line 30:50). Replace them with your desired texts.
to be added
- This project was built upon
https://medium.com/@ahtishamshafi9906/how-to-build-a-simple-chat-application-in-fastapi-7bafad755654 - Spinning donut
https://www.a1k0n.net/2011/07/20/donut-math.html
This project is licensed under the MIT License.