Hi,
I am using T4 GPU machines (Spots) for deep learning purposes in West Europe datacenter.
The problem is that I am constantly experiencing changes in instances availability. The training speed changes drastically during night/day hours, weekdays and weekends.
I want to understand if it's caused by the fact that we are using Spots and not standard tier machines.