Score value range shift in Azure Cognitive Search Vector Search

[seddamani]

Hello!

I was browsing the Azure Cognitive Search Vector Search example, and I noticed a shift in the range of score values between vector search with one vector: score~0.8

and when involving multiple vectors (cross-field or multi-vector): score~0.03

Do you have any explanation or resources concerning the score value shift?

In my use-case, I am thinking about leveraging score values, for instance filtering based on some score value. To do so, I need to grasp the

Thanks for your help!