The number of elastic inference accelerators to attach to the instance.
The type of elastic inference accelerator.