Models in the cognitive sciences and AI are typically based on heuristic combinations of soft computing methods - including fuzzy approaches, neural networks and others. It is often difficult, if not completely intractable to apply operations between such models, as they are usually given in different mathematical representations or using different frameworks that may or may not be suitable for their unification. This paper focuses on the TP model transformation, which plays an important role in transforming various model representations to a unified form that fits well with formalised mathematical design concepts. The novelty of the paper is a new extension of the TP model transformation that is capable of transforming a set of models with a different number of inputs. This is in contrast to previous solutions, in which the requirement for all models to have the same number of inputs was a strong limitation.