The issue of receiving "I don't know" as a response when using the REST API for your prompt flow could be related to several factors. Here are some steps to follow.
- Ensure that the endpoint identity has the necessary permissions to access the Azure AI Search index. If your flow contains an Index Look Up tool, the endpoint needs to access the workspace datastore to read the necessary files.
- Check if the model used in your flow is powerful enough. If the model is not suitable for the queries being made, it might return a generic response. Consider using a more powerful model if available.
- If the index query is not optimized, it may lead to poor performance or generic responses. Review the index and ensure that it is set up correctly to handle the queries you are sending.
- If your flow has many steps or is overly complex, it might lead to delays or failures in processing the query. Simplifying the flow could help improve response accuracy.
- Before deploying, ensure that you test the flow thoroughly in the Azure AI Foundry portal to confirm that it behaves as expected.
- kindly refer these https://learn.microsoft.com/en-us/azure/machine-learning/prompt-flow/how-to-deploy-for-real-time-inference?view=azureml-api-2#troubleshoot-endpoints-deployed-from-prompt-flow
Hope this helps. Do let us know if you any further queries.
If this answers your query, do click Accept Answer
and Yes
for was this answer helpful. And, if you have any further query do let us know.