LLM Workflow Engine (LWE) Azure OpenAI Chat Provider plugin

Azure OpenAI Chat Provider plugin for LLM Workflow Engine

Access to Azure OpenAI Chat models.

Installation

Azure setup

You'll need to create a service resource and deploy models see here.

You'll also need a key and endpoint for the resource, see here

From packages

Install the latest version of this software directly from github with pip:

pip install git+https://github.com/llm-workflow-engine/lwe-plugin-provider-azure-openai-chat

From source (recommended for development)

Install the latest version of this software directly from git:

git clone https://github.com/llm-workflow-engine/lwe-plugin-provider-azure-openai-chat.git

Install the development package:

cd lwe-plugin-provider-azure-openai-chat
pip install -e .

Configuration

The following provider variables/environment variables need to be set:

• export AZURE_OPENAI_API_KEY=[key]
• export AZURE_ENDPOINT=[endpoint]
• export AZURE_OPENAI_API_VERSION=2024-02-01

Add the following to config.yaml in your profile:

plugins:
  enabled:
    - provider_azure_openai_chat
    # Any other plugins you want enabled...

Usage

From a running LWE shell:

/provider azure_openai_chat
/model deployment_name gpt-35-turbo
# Instead of environment variables, these values can also be set directly on the model:
/model openai_api_key [key]
/model openai_endpoint [endpoint]
/model openai_api_version 2023-05-15