In general, the idea of the TensorSpace "Model Preprocessing" is a serious of processes which convert the classic deep learning models to the format that could be used by TensorSpace. These process should be completed before start to build any further structure which could be based on the given structure.

The success or not of the Model Preprocessing is the precondition of applying existed deep learning models by TensorSpace. The result of the Model Preprocessing should be valid, reachable and consistent.

For more details, please check Introduction: Model Preprocessing.