Unable to converse with gpt model through Azure OpenAI Chat Assistant

[gujju62]

Please provide us with the following information: While trying to chat with Azure Open AI gpt-4 model as service using the chat assistant feature, I am getting error message "Failed to create thread Error Code 404. I have done following :

• Have followed the installation guidelines
• Have validated the access to gpt-4 service endpoint using nslookup
• Tried to access the services using the URL provided by Azure OpenAI from Postman and I am getting response.

• Have used deployment names in "Settings" --> system assistant and while creating new chat assistant

This issue is for a: (mark with an x)

- [X ] bug report -> please search issues before submitting
- [ ] feature request
- [ ] documentation issue or request
- [ ] regression (a behavior that used to work and stopped in a new release)

Minimal steps to reproduce

• Create New Chat Assistant
• Start the conversation

Any log messages given by the failure

Expected/desired behavior

Should be able to have the chat conversation

OS and Version?

Windows 7, 8 or 10. Linux (which distribution). macOS (Yosemite? El Capitan? Sierra?) Linux ubuntu 20.04

Versions

0.4.5 version for the Assistant

Mention any other details that might be useful

---

Thanks! We'll be in touch soon.

[jhakulin]

@gujju62 Thanks for the report. In the first error message, I see "Resource not found", thus I wonder if the endpoint or the model deployment name is set correctly for assistant.

Did you have following format in AZURE_OPENAI_ENDPOINT="your_endpoint_data.openai.azure.com/" ? Then did you use the model deployment name from Azure AI Studio, ie, not the model name (as that may be different).

[gujju62]

@jhakulin - Thank you . Yes, I am using the endpoints and key provided by Azure OpenAI service.

I am also using using the deployment name that i have provided while creating the service.

cheers.

[jhakulin]

@gujju62 Need to figure out what is wrong as I cannot reproduce.

Could you please enable OpenAI logging check box in Debug view and send new trace to me to jhakulin@microsoft.com? NOTE: When you enable the OpenAI logging your api-key seems get exposed, so be careful and remove that before sharing the information.

Also could you do pip list for the components you have installed.

[gujju62]

@jhakulin - Please find attached the list of packages installed [pip install]. Under Menu Item "Diagnostics --> Debug View", I don't see any option for enabling OpenAI logging. packageslist.txt

[jhakulin]

@gujju62 You do not see check box like below?

[gujju62]

@jhakulin - I have emailed you sample trace . Thanks.

[jhakulin]

@gujju62 One problem could be that you have 2023-03-15-preview api version.

For Azure OpenAI Assistants, please use

set AZURE_OPENAI_API_VERSION=2024-05-01-preview

[gujju62]

@jhakulin - With the API version change, I was able to access the tool and able to perform chat with the assistant. Thank you for your help.

I am also trying out speech capability of the tool, is there any documentation that you could point to me?

Thank you.

[jhakulin]

@gujju62 Great, please see instructions here how to get Azure Speech resource

https://learn.microsoft.com/en-us/azure/ai-services/speech-service/get-started-speech-to-text?tabs=linux%2Cterminal&pivots=programming-language-python

Once you get key and region, please update environment variables

set AZURE_AI_SPEECH_KEY=Your_Speech_Key set AZURE_AI_SPEECH_REGION=Your_Speech_Region

The same resource should work for both speech to text and text to speech

[jhakulin]

closing the issue, please let me know if need more help.