Ask-picturize-it

This uses OpenAI API Whisper(whisper-1), DALL-E, GPT(gpt-3.5-turbo), also Azure OpenAI and Google PaLM (https://developers.generativeai.google)

Azure Container Registry and Docker notes

export required environment variables

export AZURE_REGISTRY_NAME = "Set Azure Container registry name here"
export P_MONGODB_DATABASE = "Mongo database"
export P_MONGODB_URI = "Mongo connection string"

build a docker image locally or use GitHub Workflow action

docker build --rm --pull \
  --file "Dockerfile" \
  --label "com.$AZURE_REGISTRY_NAME.ask-picturize-it" \
  --tag "ask-picturize-it:latest" \
  .

Run and test

docker run -e P_MONGODB_DATABASE -e P_MONGODB_URI -it --publish 80:80 --publish 27017:27017 ask-picturize-it:latest

Login to Azure Container registry

az login --use-device-code
az acr login --name $AZURE_REGISTRY_NAME.azurecr.io

Tag and Push the docker image to the Azure Container registry

docker tag ask-picturize-it $AZURE_REGISTRY_NAME.azurecr.io/ask-picturize-it
docker push $AZURE_REGISTRY_NAME.azurecr.io/ask-picturize-it

JupyterLab or Azure ML Workspace Notes

• Use https://jupyter.org (locally, or on the Cloud or in GitHub Codespaces) or Login to https://ml.azure.com and use Notebooks

• Clone repo and rename env.example to .env and set values to variables

• uncomment following lines

    #from dotenv import load_dotenv
    #load_dotenv()
  

• Install

    pip install python-dotenv
  

• Run to install dependencies

  pip install -r requirements.txt 
  

• Install pip install p2j

  pip install p2j
  

• Run to convert py to ipynb

  p2j app.py