The capacity of the model controls the scope of the mapping functions that it is able to learn.
To do this we need to Claculate MSE for example, and loop for multiple capacities to determine which capacity best fit the toy data I am using.
This improvement step would be added to the timeseries forecasting White paper, to show more realistic staging starting from a simple model, iterating through some improvements to get a nice model that predicts more accurate results.
The capacity of the model controls the scope of the mapping functions that it is able to learn. To do this we need to Claculate MSE for example, and loop for multiple capacities to determine which capacity best fit the toy data I am using.