OpenAI: error inference result

Sarah Kuhn 10

i'm trying to make an inference request with my prox. I.ex i deployed a model on azure openAI services and now want to test and communicate with it.

But i get the following error while trying to do a inference request:

Can someone maybe help me on what i'm doing wrong ? Its my first project using Azure OpenAI Service...

Inference result: {'error': 'Error communicating with OpenAI: HTTPSConnectionPool(host=\'agiopenai.openai.azure.com\', port=443): Max retries exceeded with url: //openai/deployments/%3Cai-for-proxy%3E/chat/completions?api-version=2023-05-15 (Caused by NameResolutionError("<urllib3.connection.HTTPSConnection object at 0x7f7010033ca0>: Failed to resolve \'agiopenai.openai.azure.com\' ([Errno -3] Temporary failure in name resolution)"))'}
{'_content': b'{"error":"Error communicating with OpenAI: HTTPSConnectionPool(h'
             b"ost='agiopenai.openai.azure.com', port=443): Max retries exceede"
             b'd with url: //openai/deployments/%3Cai-for-proxy%3E/chat/complet'
             b'ions?api-version=2023-05-15 (Caused by NameResolutionError(\\'
             b'"<urllib3.connection.HTTPSConnection object at 0x7f7010033ca0>: '
             b"Failed to resolve 'agiopenai.openai.azure.com' ([Errno -3] Tempo"
             b'rary failure in name resolution)\\"))"}\n',

Saurabh Sharma 23,816 Reputation points Microsoft Employee

2023-09-08T19:15:40.06+00:00

Hi Sarah Kuhn,

Thanks for using Microsoft Q&A!!

I believe you are not passing the correct openai endpoint to the call and thus are you getting this error.

Please validate if you are passing the correct openai endpoint. (Please check the screenshot below)-

If that's not the case, then can you provide the code (after removing any sensitive information) which you using for OpenAI?

If you are using rest endpoint can you provide the endpoint and the request details (body, headers etc.) ?

Please let me know if you have any questions.

Thanks

Saurabh

Sarah Kuhn 10

hey Saurabh Sharma, thanks for the help. I double checked and I'm pretty sure I'm using the right endpoint as showed in the picture.

Here's the code (my goal is to do an inference request with my ai proxy)

endpoint = f"{deployment.deployment_url}/v2/envs"
headers = {"Authorization": ai_core_client.rest_client.get_token(),
           "ai-resource-group": resource_group_id,
           "Content-Type": "application/json"}
response = requests.get(endpoint, headers=headers)

legacy_davinci = False # set True if you have a davinci model deployment on Azure OpenAI Services
if legacy_davinci:
    body = {
        "engine": "<ai-for-proxy>", # The deployment name you chose when you deployed the ChatGPT or GPT-4 model.
                                   #   For information of deployment creation and name Refer article https://learn.microsoft.com/en-us/azure/cognitive-services/openai/how-to/create-resource?pivots=web-portal
        "prompt": "Classify the following news article into 1 of the following categories: categories: [Business, Tech, Politics, Sport, Entertainment]\n\nnews article: Donna Steffensen Is Cooking Up a New Kind of Perfection. The Internet’s most beloved cooking guru has a buzzy new book and a fresh new perspective:\n\nClassified category:",
        "max_tokens": 60,
        "temperature": 0,
        "frequency_penalty": 0,
        "presence_penalty": 0,
        "top_p": 1,
        "best_of": 1,
        "stop": "null"
    }
    endpoint = f"{deployment.deployment_url}/v2/completion"
else:
    body = {
        "engine": "<ai-for-proxy>", # include your engine from a deployment of an Azure OpenAI services model
        "prompt": "Classify the following news article into 1 of the following categories: categories: [Business, Tech, Politics, Sport, Entertainment]\n\nnews article: Donna Steffensen Is Cooking Up a New Kind of Perfection. The Internet’s most beloved cooking guru has a buzzy new book and a fresh new perspective:\n\nClassified category:",
        "max_tokens": 60,
        "temperature": 0,
        "frequency_penalty": 0,
        "presence_penalty": 0,
        "stop": "null"
    }
    endpoint = f"{deployment.deployment_url}/v2/chat-completion"

headers = {"Authorization": ai_core_client.rest_client.get_token(),
           "ai-resource-group": resource_group_id,
           "Content-Type": "application/json"}
response = requests.post(endpoint, headers=headers, json=body)

print("Inference result:", response.json())
pprint(vars(response))

I'm doing the following tutorial from gitHub and have allocated all the resources but however whenever I want to make a request to my proxy I get this error from OpenAi.

(Here's the link to the git tutorial: https://github.com/SAP-samples/azure-openai-aicore-cap-api/blob/main/01-ai-core-azure-openai-proxy/proxy.ipynb

and I'm currently executing the steps from the proxy.ipynb i.ex step 8 in the file.)

Thank you so much for your help because I really don't understand where this error is coming from.

here's the error I still get:

Inference result: {'error': 'Error communicating with OpenAI: HTTPSConnectionPool(host=\'agiopenai.openai.azure.com\', port=443): Max retries exceeded with url: //openai/deployments/%3Cai-for-proxy%3E/chat/completions?api-version=2023-05-15 (Caused by NameResolutionError("<urllib3.connection.HTTPSConnection object at 0x7f0487ff38b0>: Failed to resolve \'agiopenai.openai.azure.com\' ([Errno -3] Temporary failure in name resolution)"))'}
{'_content': b'{"error":"Error communicating with OpenAI: HTTPSConnectionPool(h'
             b"ost='agiopenai.openai.azure.com', port=443): Max retries exceede"
             b'd with url: //openai/deployments/%3Cai-for-proxy%3E/chat/complet'
             b'ions?api-version=2023-05-15 (Caused by NameResolutionError(\\'
             b'"<urllib3.connection.HTTPSConnection object at 0x7f0487ff38b0>: '
             b"Failed to resolve 'agiopenai.openai.azure.com' ([Errno -3] Tempo"
             b'rary failure in name resolution)\\"))"}\n',

Balur, Harsha 15 Reputation points

2023-09-12T11:14:51.1266667+00:00

I am facing similar error, any resolution for the same issue?
Saurabh Sharma 23,816 Reputation points Microsoft Employee

2023-09-12T17:58:44.1866667+00:00

@Sarah Kuhn Thanks for sharing details. I will check and get back to you on this one.
Färber Melanie 0 Reputation points

2023-09-22T08:30:47.9266667+00:00

I have the same error. Has anyone found a solution?
Balur, Harsha 15 Reputation points

2023-09-25T09:10:24.9333333+00:00

any help would be appreciated here
Saurabh Sharma 23,816 Reputation points Microsoft Employee

2023-09-26T23:48:03.1033333+00:00

@Sarah Kuhn Which OpenAI model(GPT 3.5 Turbo/GPT 4) you are trying to use for the chat completions. I believe the model doesn't support what you are trying to use it for but wanted to confirm.
Balur, Harsha 15 Reputation points

2023-09-27T10:33:00.6433333+00:00

Hi Saurabh,

I was using GPT 3.5 Turbo for similar error case
Bir Bahadur 20 Reputation points

2023-10-18T09:26:09.1666667+00:00

I am facing the same issues, please help @Balur, Harsha , its urgent, Have you resolved the issue,
Rahul Nyati 0 Reputation points

2023-10-18T13:23:13.48+00:00

I am facing the same issues, please help.
Gowni, Amaranatha Reddy 5 Reputation points

2023-10-25T06:20:36.5833333+00:00

I am facing the same issues, please help does any have solution,
@Balur, Harsha have resolve this issue ? If you have any reference contact please share me.
Balur, Harsha 15 Reputation points

2023-10-25T06:55:03.53+00:00

SAP confirmed that it is an issue from AI Core side, following up with them for a confirmation on this fix.

Share via

OpenAI: error inference result

Your answer