Let's say someone wants to train a neural network model on 50GB of data, he/she can just use an AWS EC2 instances with a large number of CPUs and a large memory. The largest AWS EC2 instance provides 448 vCPUs and more 12TB of memory, would do some people still use SageMaker to train the model on multiple instances as 448 vCPUs and 12TB of memory can even train a large model like GPT?
Checking the versions of EC2 instances and reviewed the distributed training process on AWS
