Projects are used to organize deployed clusters and the experiments run on those clusters. When creating a project you will want to choose a name that identifies the type of experiments you are going to run as well as the size of the cluster you want to run those experiments on.
The advanced dropdown requires further explanation to understand all of your choices presented here.
Worker Nodes is the count of workers to deploy your Instance Type to.
Instance Type describes the AWS EC2 Instance type you will be deploying. It is a combination of CPU and Memory. When choosing an instance type take into account the size of your dataset and memory requirements that will be needed.
Cluster Settings allows you to run stable or experimental. Stable is fully tested and official release of Auger software stack. Experimental is bleeding edge changes and generally lightly tested. Stable is recommended.
Automatic Termination is the number of hours a cluster can sit idle before Auger terminates the cluster. Idle in this case is defined as no training runs being observed by the system. This is useful if you forget to stop your cluster.