Elaborated from #18 . The neural network that we train has certain characteristics which must be known for integrating into a climate model. These characteristics are explained in Arthur's 2021 paper, though with a statistical focus. Providing an overview of model input and output shape with a coding focus (e.g. variable/coord names used, how the forcing scaling works, an example of using the predicted probability distribution) would make it easier to adapt GZ21.
Elaborated from #18 . The neural network that we train has certain characteristics which must be known for integrating into a climate model. These characteristics are explained in Arthur's 2021 paper, though with a statistical focus. Providing an overview of model input and output shape with a coding focus (e.g. variable/coord names used, how the forcing scaling works, an example of using the predicted probability distribution) would make it easier to adapt GZ21.