Basically, Graphormer inherits the extending usage of fairseq, which means it could easily support user-defined plug-ins.
For example, the Graphormer-base model could be defined through ~graphormer.models.GraphormerModel, which inherits the ~fairseq.models.FairseqModel class.
It's also easy to extend the Graphormer-base model, which means you could define your own model and criterion, and then use them in Graphormer.
Also, development of new model is easy. We provide a tutorial of how to implement a simple MLP model on graph in Tutorials.