heroes-ai-handler 0.2.12

A package for handling GPT, embeddings, and data for an AI Assistant.

python version >= 3.10

Heroes AI Handler

Description

This package will handle all elements for an AI assistant, from ingesting data to using the GPT models.

Development

python3.10 -m venv .venv
pip install -r requirements.txt

Usage of the package

For using this package, a few environment variables are necessary to use certain functionalities of the package. For example, place them all in a .env file, and load them in using package python-dotenv:

from dotenv import load_dotenv
load_dotenv()

Full list of necessary variables:

Varibale name	description
AZURE_OPENAI_ENDPOINT	Endpoint to the Azure Open AI resource.
AZURE_OPENAI_API_KEY	API-key to access the Azure Open AI resource.
AZURE_OPENAI_CHAT_DEPLOYMENT_NAME	Name of the deployment of the chat completion model.
AZURE_OPENAI_API_VERSION	API-versie azure Open AI resource.
AZURE_OPENAI_EMBEDDINGS_DEPLOYMENT_NAME	Name of the deployment of the embedding model.
WEVIATE_CLUSTER_ENDPOINT	Endpoint to the Weaviate cluster.
WEVIATE_CLUSTER_API_KEY	API-key to access the Weaviate cluster.
STORAGE_ACCOUNT_NAME	Name of the Azure Storage Account.
STORAGE_ACCOUNT_KEY	Key to the Azure Storage Account
STORAGE_ACCOUNT_CONTAINER_NAME	Container name within the Azure Storage Account
SQL_SERVER_NAME	SQL Server name.
SQL_DATABASE_NAME	Database name within SQL Server.
SQL_USERNAME	Username for accessing Database.
SQL_PASSWORD	Password for accessing Database.
SQL_DRIVER	Formulate SQL driver.

Manual actions

Create the following resources:

Azure

• Azure OpenAI
  • Deploy chat completion model: gpt-3.5-turbo / gpt4
  • Deploy text embedding model: text-embedding-ada-002
• Storage account
  • Create container
• SQL server
  • Make sure you can access the SQL Server from the network configurations.
• SQL database
  • To create tables in database, run: setup/1__create_database_tables.ipynb

Weaviate

• Weaviate account with a cluster
  • To create collection in Weaviate cluster, run: setup/2__create_weaviate_collection.ipynb

Manual deployment of package

Only run this when automatical deployment through azure devops doesn't suit the purpose.

python3.10 -m venv .venv_publish
source .venv_publish/bin/activate
pip install wheel twine
rm -rf build dist *.egg-info
python setup.py sdist bdist_wheel
twine upload -u $(twineUsername) -p $(twinePassword) dist/*
rm -rf build dist *.egg-info