fix: Separate out config for AZURE_OPENAI_MODEL_NAME and AZURE_OPENAI_DEPLOYMENT_NAME - Fixes #35

[timothymeyers]

Purpose

The following PR addresses an issue I found will troubleshooting #35.

Originally, it was assumed that the Azure OpenAI Model Name and Deployment Name were the same because this is how the accelerator deploys its resources. However, when doing a local deployment, it may be desirable to use an existing Azure Open AI resource and model deployment due to resource and quota constraints.

In my case, my Model name (gpt-4o) was not the same as my model deployment name (gpt4o, no dash). This was resulting in 400 errors when trying to connect to the model.

This PR introduces a new optional configuration value, AZURE_OPENAI_MODEL_NAME to config.py. It will default to using AZURE_OPENAI_DEPLOYMENT_NAME if not present.

Does this introduce a breaking change?

• Yes
• No

How to Test

• Get the code

git clone [repo-address]
cd [repo-name]
git checkout [branch-name]

• Test the code

  • Deploy another model within your Azure OpenAI instance where the deployment name (e.g. gpt-4o-testing) and model name (e.g., gpt-4o) do not match.
  • Update .env to include AZURE_OPENAI_MODEL_NAME
  • Start/Restart app.py server

What to Check

Verify that the following are valid

• In the original code, you will see 400 ERRORs due to model not being found, as it is looking for gpt-4o-testing rather than gpt-4o.
• In my updated code, it should work normally.