Can earthformer be used for multivariate prediction of univariate? For example, my input variable is [T,H,W,5], and my output variable is some other quantity [T,H,W,1]?
Sure. You may want to have a try in our simplest test case to set different values for the input channels and target channels via input_shape and target_shape.
Can earthformer be used for multivariate prediction of univariate? For example, my input variable is [T,H,W,5], and my output variable is some other quantity [T,H,W,1]?