How do you scale training as models gets increasingly large and accurate? Great insights on programmable networks for performant distributed deep learning by @elfmar in the first session of @DistributedML.