Error while connecting data to Azure Open AI service Preview option

Question

Hi Team,

I am connecting my data with OpenAI service using the Preview feature. While doing so, I have enabled vector search and added an embedding model. We have also selected the semantic search standard plan and enabled Hybrid + Vector Search. This setup was working previously but after the recent deployment it keeps on giving me the error:

I have an embedding model already in place and I am able to interact with the data in the AI Studio. However once the application is deployed onto the web app this error is showing up.

Any suggestions?

Answer 1

Checking to see if you're still experiencing this problem. Have you attempted to configure the environment variables by accessing your web app through App Services in the Azure portal? To resolve this, set two environment variables with the API version "2023-08-01-preview."

AZURE_OPENAI_EMBEDDING_ENDPOINT = https://{your-resource-name}/openai/deployments/{deployment-id}/extensions/chat/completions?api-version={api-version}
AZURE_OPENAI_EMBEDDING_KEY = your-api-key

https://learn.microsoft.com/en-us/azure/ai-services/openai/tutorials/embeddings?tabs=python%2Ccommand-line&pivots=programming-language-python

Answer 2

Hi,

I had the same problem and solved it by setting two environment variables:

AZURE_OPENAI_EMBEDDING_ENDPOINT = https://{your-resource-name}/openai/deployments/{deployment-id}/extensions/chat/completions?api-version={api-version}
AZURE_OPENAI_EMBEDDING_KEY = your-api-key

You can set the environment variables in by navigating to your web app through App-Services in the azure portal. (As API version, "2023-08-01-preview" worked for me.)

I hope that helps.

Answer 3

Hi Suraj, thanks for responding to my query but the problem is that this application has been deployed from Azure Open AI Studio and I just selected the embedding model from the default drop down menu. Add Data Source> Blob Storage > Selected the container and search service > Clicked the vector embedding option.

Answer 4

The error message you’re seeing, 'embeddingEndpoint' is a required parameter for the AzureCognitiveSearch data source type, indicates that you need to include the ‘embeddingEndpoint’ parameter when configuring the Azure Cognitive Search data source for your application.

This parameter is required when you’re using the vectorSimpleHybrid query type, which combines both semantic and vector search. The ‘embeddingEndpoint’ should point to the endpoint of your embedding model.

Please replace 'YOUR_EMBEDDING_MODEL_ENDPOINT' with the actual endpoint of your embedding model2.

Here’s an example of how you can include this in your configuration:

"dataSources": [
  {
    "type": "AzureCognitiveSearch",
    "parameters": {
      "endpoint": "'YOUR_AZURE_COGNITIVE_SEARCH_ENDPOINT'",
      "key": "'YOUR_AZURE_COGNITIVE_SEARCH_KEY'",
      "indexName": "'YOUR_AZURE_COGNITIVE_SEARCH_INDEX_NAME'",
      "embeddingEndpoint": "'YOUR_EMBEDDING_MODEL_ENDPOINT'"
    }
  }
]