Resource Deployment and Environment Variables

Required Resources

To use the rag-experiment-accelerator, create the following resources:

Azure AI Search Service
- Turning on Semantic Ranking is optional
Azure OpenAI
- Create new deployments using models gpt-35-turbo and text-embedding-ada-002
Azure ML Workspace
Azure AI Service for Language is optional

Below are the required environment variables, to be added to a local .env file at the root of the repo.

Variable	Value	Note
AZURE_SEARCH_SERVICE_ENDPOINT		Azure AI Search API Endpoint
AZURE_SEARCH_ADMIN_KEY		Azure AI Search Key
AZURE_SEARCH_USE_SEMANTIC_SEARCH	True / False	Whether to use semantic search (not all regions support this feature)
OPENAI_API_KEY		OpenAI API key
OPENAI_API_TYPE	azure, open_ai	Must be `azure` for Azure OpenAI or `open_ai` for OpenAI.
OPENAI_ENDPOINT		Azure OpenAI API endpoint.
OPENAI_API_VERSION	2023-03-15-preview	Azure OpenAI API version. See https://learn.microsoft.com/en-us/azure/ai-services/openai/reference.
AML_SUBSCRIPTION_ID		Azure Machine Learning subscription ID
AML_WORKSPACE_NAME		Name of deployed Azure Machine Learning Workspace
AML_RESOURCE_GROUP_NAME		Azure Machine Learning resource group name
LOGGING_LEVEL	NOTSET, DEBUG, INFO, WARN, ERROR, CRITICAL	LOGGING_LEVEL is INFO by default

Below are optional environment variables, to be added to a local .env file at the root of the repo.

Variable	Value	Note
AZURE_LANGUAGE_SERVICE_ENDPOINT		Azure AI Service for Language API Endpoint
AZURE_LANGUAGE_SERVICE_KEY		Azure AI Service for Language Key