Azure-Samples / azure-search-openai-demo-java

This repo is the Java version of Microsoft's sample app for ChatGPT + Enterprise data.
MIT License
67 stars 66 forks source link

Get same reply when enable or disable "Use semantic ranker for retrieval" in the settings tab #64

Closed faynef closed 1 month ago

faynef commented 7 months ago

Describe the issue: Get same reply when enable or disable Use semantic ranker for retrieval in the settings tab, expecting a different reply. image image

Environment:

Repro Steps: 1.Run azd auth login. 2.Run azd up. 3.Select or enter "What is my deductible" in the chat text box. 4.From the answer, select one of the citations. 5.In the browser, select the Developer Settings tab. 6.Check the Suggest follow-up questions checkbox and ask the same question again. 7.In the Settings tab, deselect Use semantic ranker for retrieval. 8.Ask the same question again, get the same reply.

Expected behavior: Get diffrient reply when enable or disable "Use semantic ranker for retrieval" in the settings tab

@charris-msft for notification.

github-actions[bot] commented 5 months ago

This issue is stale because it has been open 60 days with no activity. Remove stale label or comment or this issue will be closed.

dantelmomsft commented 1 month ago

can't reproduce as it's showing different answers. However there might be cases where same facts are retrieved ( maybe in different orders) using different search options hence LLM generate the same response +/-. So I'm not sure the above steps and expected behavior can be qualified as a bug. My tests below:

image

image

image