xla
Add options in benchmark script to keep model on CUDA
#7193
Merged

Loading