Get raw predictions from CompiledInferenceNetwork

[neworderofjamie]

As ever, have taken inspiration from Keras for this but to obtain raw predictions rather than evaluate metrics from an inference network you do:

predictions, _ = compiled_net.predict({input: testing_images * 0.01},
                                      output)

rather than:

metrics, _ = compiled_net.evaluate({input: testing_images * 0.01},
                                   {output: testing_labels})

The result will be a dictionary mapping output populations to (dataset_size, num_output_neurons) numpy arrays containing the raw readout values i.e. spike counts or average var value