Code-based evaluators show `score` as `Error` in Azure Foundry evaluations

[bastbu]

• Package Name: azure-ai-projects
• Package Version: 2.0.0
• Operating System: Ubuntu
• Python Version: 3.10

Describe the bug

There are a couple of issues with the ai-projects-sdk when running custom evaluators.

• When running the sample for a custom code-based evaluator here, the UI in Foundry creates two metrics instead of one. The score is shown as "Error".
• Logging output is not shown in the user logs, which makes it hard to debug custom scoring logic

To Reproduce

Steps to reproduce the behavior:

1. Run the sample code

The my_custom_evaluator_code: score column shows an error:

Note: the current sample code in the repository uses somewhat questionable queries, but I did not modify those for the purpose of running the sample in an unmodified manner.

Expected behavior

• There should be an explanation of why this creates two metrics
• The score should not display as "Error" when using the SDK
• The sample should use neutral questions

[github-actions]

Thanks for the feedback! We are routing this to the appropriate team for follow-up. cc @dargilco @glharper @howieleung @kingernupur @nick863 @trangevi @trrwilson.

[kashifkhan]

Thank you for the feedback @bastbu . Ill forward this to the right team who can help you further