Undue idle experiment warning

[Jae-Won Chung]

Hi,

I have a GPU node experiment running DNN training. The experiment Id is fee69f29-644e-11ec-b318-e4434b2381fc.

I just got an automated message warning that the experiment is idle for a long period, and it might be terminated if it stays idle. However the experiment has never been idle; I've been grinding all GPUs with DNN training ever since I first created the experiment. I suspect that the idle experiment detection logic does not take into account GPU utilization. Could you look into this issue?

Thanks a lot.

Best,

Jae-Won