Currently, HPC takes a generic MLModel.
This is fine, except a user may pass in the wrong model (i.e they may train an image model and not a hpc model).
The framework should check to make sure that a model attempting to be used has the right inputs and outputs, and if not, it should provide some sort of error warning.
Currently, HPC takes a generic MLModel.
This is fine, except a user may pass in the wrong model (i.e they may train an image model and not a hpc model).
The framework should check to make sure that a model attempting to be used has the right inputs and outputs, and if not, it should provide some sort of error warning.