Gemini PRO returns more details than on the suggested Page/URL

[BlythAI]

I was playing around with Gemini Pro on Document extraction and for trial i gave a sample URL , say this Uber API reference page and requested an extract of all Rest points in the page. I see that the model works however it is returning much more data than what is on the page. How can one limit the scope of the model to the suggested documents alone ? I tried playing around with temperature and other modifiers with no luck. Any advice on how to solve this will be very helpful.

https://developer.uber.com/docs/riders/references/api

I see that Bard is more accurate than Gemini Pro for the same task.

[lsolatorio]

Hi @BlythAI, 

Welcome and thank you for reaching out to our community.

For now, the only workaround that I can think of to limit the scope/ response of Gemini Pro  is through precise prompting as the tool doesn't have a built-in function to achieve this use case.

Clearly specify in your prompt that the information you want to receive is only limited on the page URL you have provided. 

Example: "Get all the REST endpoints information on this Uber API reference page: [https://developer.uber.com/docs/riders/references/api]."

It's also a good attempt to experiment with Temperature adjustments as it also fits this use case.

Alternatively, you can also submit this as a feature request in our Issue tracking system and product feature requests.

Hope this helps.