When we have current model == None, we cannot train models / display visualizations, etc. Currently it gives 500 internal error as current model points to nothing. We will have to do an explicit check and throw and error.
We need to update UI, backend and corresponding test cases to reflect this.