Azure ML prompt flow QnA model works in worskspace but the deployed endpoint does not work

Oussama Chiboub 35 Reputation points
2024-09-27T13:47:57.5466667+00:00

I have been working on a prompt flow QnA chatbot for a while now and it has been working well. Recently i redeployed my model with minor changes and the endpoint does not work anymore: below is the error that i have : "Execution failure in 'lookup': (TypeError) unsupported operand type(s) for +: 'NoneType' and 'int' azure". Keep in mind that i haven't changed anything in the "lookup" node, and i couldn't find where the + operand is being done on the flow.

User's image

User's image

The prompt flow model works well in azure ML studio but deploying somehow makes this error.

Thank you for your help ! It's kind of an urgent matter

Azure Machine Learning
Azure Machine Learning
An Azure machine learning service for building and deploying models.
2,887 questions
Azure AI Search
Azure AI Search
An Azure search service with built-in artificial intelligence capabilities that enrich information to help identify and explore relevant content at scale.
991 questions
Microsoft Deployment Toolkit
Microsoft Deployment Toolkit
A collection of Microsoft tools and documentation for automating desktop and server deployment. Previously known as Microsoft Solution Accelerator for Business Desktop Deployment (BDD).
892 questions
{count} votes

4 answers

Sort by: Most helpful
  1. YutongTie-MSFT 51,501 Reputation points
    2024-09-29T09:11:38.1133333+00:00

    Hello everyone,

    Thanks for reporting this issue, product team has acknowledged this issue and working on a fix. Will update here again if the issue has any progress.

    Regards,

    Yutong

    -Please kindly accept the answer if you feel helpful to support the community, thanks a lot.

    2 people found this answer helpful.

  2. Erik 10 Reputation points
    2024-09-30T10:28:44.38+00:00

    A temporary fix for me was to turn of tracing by setting an env variable:

    os.environ['PF_DISABLE_TRACING'] = 'true'
    
    2 people found this answer helpful.

  3. suwit 0 Reputation points
    2024-09-30T04:46:40.9233333+00:00

    error:

    
    
    The above exception was the direct cause of the following exception:
    
    Traceback (most recent call last):
      File "/azureml-envs/prompt-flow/runtime/lib/python3.9/site-packages/promptflow/_core/flow_execution_context.py", line 90, in invoke_tool
        result = self._invoke_tool_inner(node, f, kwargs)
      File "/azureml-envs/prompt-flow/runtime/lib/python3.9/site-packages/promptflow/_core/flow_execution_context.py", line 206, in _invoke_tool_inner
        raise ToolExecutionError(node_name=node_name, module=module) from e
    promptflow._core._errors.ToolExecutionError: Execution failure in 'opportunities': (TypeError) unsupported operand type(s) for +: 'NoneType' and 'NoneType'
    2024-09-30 04:39:01 +0000      22 execution.flow     WARNING  Failed to calculate metrics due to exception: unsupported operand type(s) for +: 'int' and 'NoneType'.
    

  4. Oussama Chiboub 35 Reputation points
    2024-09-30T13:25:01.3033333+00:00

    Hi everyone ! I found a fix pour this error : Apparently latest open ai versions cause a problem with prompt flow tracing, so reverting to = < 1.44.1 openai python library in the requirements file fixed the issue for me

    https://pypi.org/project/openai/#history

    0 comments No comments

Your answer

Answers can be marked as Accepted Answers by the question author, which helps users to know the answer solved the author's problem.