* Add gpu workflow and a transformers API inference test * Set device-specific env variables in script instead of workflow * Fix status message --------- Co-authored-by: sgwhat <ge.song@intel.com>
