microsoft

/

Florence-2-base-ft

like 78

on my gtx 1080
on start model in gpu use 1.2G
after first inference use 4.5G
is it normal? can model use only 1.2G?