Closed jduerholt closed 1 year ago
@jduerholt Changes and additions look good to me.
One comment about model building when we have the different input types. As it is now, we would stick to using MixedSingleTaskGP, for example, when we have CategoricalMolecularInput with them treated as categoricals? i.e. Not using an additional hypothetical "molecular_kernel" argument.
This PR adds the CategoricalMolecularInput feature.