GPT3-NGen-WebApp

GPT-3 powered Web Application for generating corrective action. WebApp uses a finetuned model and is built on top of flask.

Running the webapp using the key information (without config file)

1. Update the key information from GPT3 in app.py file at openai.api_key= <GPT-3-KEY-HERE>
2. Move to the location of app.py file and run python app.py to trigger run for the flask app

Running app via Docker container (without config file)

1. Clone the repo and update you GPT-3 key in app.py file at openai.api_key= <GPT-3-KEY-HERE>
2. Move to the location where `app.py' file exists in the cloned folder
3. Build docker using command: docker build -f Dockerfile -t app_nga_2:latest .
4. Run docker run -p 8100:8100 -ti app_nga_2 /bin/bash -c "cd src && source activate nga_scratch && python api.py"
5. If using linux, use link to directly navigate to the app, else use Docker App to launch the web-interface

Running the webapp on local device (with config file)

1. Create a config file that contains GPT-3 key information
2. Modify code in app.py to load config
3. Exporting of key is required before web-app is run: export OPENAI_CONFIG=/path/to/config/openai.cfg
4. Run python file for starting the webapp

Running app via Docker container (with config file): Present Working Version

1. Clone the repo and place file named openai.configin nga folder. It should contain one line of config information in format: OPENAI_KEY= '<KEY>'
2. Move to the nga folder where Dockerfile is present for this project (nga folder)
3. Build docker using command: docker build -f Dockerfile -t app_nga_3:latest .
4. Run docker run -p 8100:8100 -ti app_nga_3
5. If using linux, use link to directly navigate to the app, else use Docker App to launch the web-interface

Heroku Deployed App GIF