Description
- script: |
pip install openai pandas
pip install azure.identity
pip install azure-ai-evaluation # Install the correct Azure AI Evaluation library
pip install azure-ai-projects # Install the missing dependency
pip install azure-ai-inference # Install the required Azure OpenAI library
displayName: "Install Dependencies"
Describe the bug
I was using azure ai evaluations sdk to evaluate model deployment and inference endpoint. Used below piece of code to publish results in ai foundry project. code ran successfully but the message in the logs confused customer/us because it gave aistudio link for results whereas customer was using ai foundry..
result = evaluate( evaluation_name="adversarial_model_evaluation", data=jsonl_path, # Replace with the actual file path target=None, # No need for a target since responses are already captured evaluators=evaluators, evaluator_config=evaluator_config, azure_ai_project=azure_ai_project
], "metrics": { "indirect_attack.xpia_manipulated_content": 0.03636363636363636, "indirect_attack.xpia_intrusion": 0.12727272727272726, "indirect_attack.xpia_information_gathering": 0.01818181818181818, "protected_material.protected_material_defect_rate": 0.0, "indirect_attack.xpia_defect_rate": 0.18 }, "studio_url": "[https://ai.azure.com/build/evaluation/b39f20ab-8846-44ca-8f9c-3dbb3ffde4a2?wsid=/subscriptions/e69c5bc8-9ce2-43da-ab1e-9b5851243390/resourceGroups/test_aifoundryrbac/providers/Microsoft.MachineLearningServices/workspaces/yoaifoundry-4251"](https://ai.azure.com/build/evaluation/b39f20ab-8846-44ca-8f9c-3dbb3ffde4a2?wsid=/subscriptions/e69c5bc8-9ce2-43da-ab1e-9b5851243390/resourceGroups/test_aifoundryrbac/providers/Microsoft.MachineLearningServices/workspaces/yoaifoundry-4251%22) }
To Reproduce
Steps to reproduce the behavior:
1.
Expected behavior
https response should not contain references to old product names it confuses customers
Screenshots
Additional context
Add any other context about the problem here.