when NNI runs on k8s with frameworkcontroller, is there any way to share one gpu resource between trials(pod)?

[mf-giwoong-lee]

https://github.com/microsoft/nni/issues/4378

I uploaded question already in issue channel, but I didn't get exact answer..

The typescript nni codes get maxTrialNumberPerGpu as key value.

However, I cannot find running nnictl create command with typescript codes.

Or I find the another solution for this problem.