Is your feature request related to a problem?
The converter classes heavily overlap with the dataset classes.
Desired solution
Instead of specifying converters when instantiating NNs, let them learn a converter when fit is called. The converter is still a useful abstraction for the checks to apply to the data passed to predict.
Possible alternatives (optional)
No response
Screenshots (optional)
No response
Additional Context (optional)
No response
Is your feature request related to a problem?
The converter classes heavily overlap with the dataset classes.
Desired solution
Instead of specifying converters when instantiating NNs, let them learn a converter when
fitis called. The converter is still a useful abstraction for the checks to apply to the data passed topredict.Possible alternatives (optional)
No response
Screenshots (optional)
No response
Additional Context (optional)
No response