Open
Description
we should have an integraiton test based on this for using nvidia ml workloads
and we should check the machine for GPU before running the test and if it doesnt have nvidia gpu we should skip the test, so we dont have a failing test
#19345 (comment)