Regarding the embeddings.openai process in AI Studio indexing, encountering an error that says 'public access disabled'. How can I resolve this issue?

구동민/(협력사) Engineer/H&A R&D데이터인사이트Project 0

Hello, I am encountering an issue in Azure AI Studio. After uploading data and attempting to index it, the process fails during the LLM-Crack, Chunk, and Embed Data step, with the following error log:

INFO     azureml.rag.embeddings.openai - Attempt 0 to embed 16 documents. (openai.py:196)  
WARNING  azureml.rag.embeddings.openai - Error embedding: Error code: 403 -  
{'error': {'code': '403', 'message': 'Public access is disabled. Please configure private endpoint.'}} (openai.py:210)

For reference, a private endpoint has already been created and connected to the OpenAI service in the Azure AI Studio Hub. I am curious why this error occurs. Furthermore, there are no issues when uploading to the vector store in OpenAI Studio, but the same OpenAI service connection results in this error only in AI Studio. I would like to understand why this discrepancy occurs.

AshokPeddakotla-MSFT 32,946 Reputation points

2024-09-02T08:00:20.5633333+00:00
구동민/(협력사) Engineer/H&A R&D데이터인사이트Project Greetings & Welcome to Microsoft Q&A forum!

For reference, a private endpoint has already been created and connected to the OpenAI service in the Azure AI Studio Hub.

Did you follow any documentation for the setup?

If not, see How to configure a private link for Azure AI Studio hubs for more details.

Furthermore, there are no issues when uploading to the vector store in OpenAI Studio, but the same OpenAI service connection results in this error only in AI Studio. I would like to understand why this discrepancy occurs.

Please see the below limitations.

Private Azure AI Services and Azure AI Search aren't supported.

The "Add your data" feature in the Azure AI Studio playground doesn't support private storage account.

You might encounter problems trying to access the private endpoint for your hub if you're using Mozilla Firefox. This problem might be related to DNS over HTTPS in Mozilla Firefox. We recommend using Microsoft Edge or Google Chrome.

Do let me know if you are there any further queries.
구동민/(협력사) Engineer/H&A R&D데이터인사이트Project 0 Reputation points

2024-09-03T05:46:11.3766667+00:00
Thank you for your response. The AI Studio Hub is configured as Public, and the AI Search Service connected to the Hub is also configured as Public. The OpenAI Service connected to the Hub is configured as Private and has a private endpoint set up. In this state, when I upload data to the blob storage of the Hub's project and index that data, I encounter the following error:

[2024-08-30 07:58:32] INFO azureml.rag.embeddings.openai - Attempt 0 to embed 16 documents. (openai.py:196) [2024-08-30 07:58:32] WARNING azureml.rag.embeddings.openai - Error embedding: Error code: 403 - {'error': {'code': '403', 'message': 'Public access is disabled. Please configure private endpoint.'}} (openai.py:210)

Based on the error code, it seems that the issue occurs when Machine Learning attempts to connect to the OpenAI resource. However, the same OpenAI resource works perfectly fine in the Hub's project chat playground, where it successfully loads and operates the 4 omni models. I am curious why this error occurs when accessing the same OpenAI Service from Machine Learning, despite it working fine in the chat playground of the Hub's project.

Share via

Regarding the embeddings.openai process in AI Studio indexing, encountering an error that says 'public access disabled'. How can I resolve this issue?

Your answer