Azure OpenAI Embeddings and GPT models for document summarization and QnA

1) Overview of the application

This is a sample application written in Angular/PrimeNG as the frontend and Python/Flask as the backend services which use Azure OpenAI embeddings and GPT-3 models for text summarication and Q&A.

User Interface

Architecture (Referenced from azure-open-ai-embeddings-qna repo)

2) Prerequisite to run the application

To run the application, you need to have the following Azure services and tools:

Once you have these services and tools, you can run the application on your local machine using the docker-compose command.

Make sure to add Azure API keys and OpenAI models to your .env file and have it in the same directory where you run the docker command.

Sample .env file:

OPENAI_API_KEY=XXXxxxxxxXXXXXxxxxxxx
OPENAI_API_BASE=https://youraoaiservice.openai.azure.com/
OPENAI_API_TYPE=azure
OPENAI_API_VERSION=2022-12-01
DOCUMENT_MODEL_NAME=curie-search-doc
QUERY_MODEL_NAME=curie-query-doc
DEPLOYMENT_NAME=text-davinci-003
AZURE_FORM_RECOGNIZER_ENDPOINT=https://formrecognizer-sf.cognitiveservices.azure.com/
AZURE_FORM_RECOGNIZER_KEY=XXXxxxxxxXXXXXxxxxxxx
REDIS_HOST=redis
REDIS_PORT=6379
REDIS_USERNAME=test
REDIS_PASSWORD=
REDIS_SSL=False

docker-compose.yml file:

version: '3'
services:
  flask:
    image: "sinafakhraee/aoai-backend-service:latest"
    env_file: .env
    ports:
      - "5000:5000"
  redis:
    image: "redislabs/redismod:latest"
  ng-app:
    image: "sinafakhraee/ng-app:latest"
    ports:
      - "8080:8080"

docker-compose command:

docker-compose up

sinafakhraee/aoai-angular-app