Custom number of outputs from spiking layer

[RFalowo]

I understand that the number of neurons in a given layer is equal to the input but i wanted to know if there is a way to set the output feature number?

[fangwei123456]

if there is a way to set the output feature number?

I think you can add a layer to implement a shape transform on the outputs of neurons?

[RFalowo]

And by this you mean a torch layer, not a spikingjelly container? If so this wouldnt work for me as I need my network to be completely SNN based. Also many thanks for your help!

[fangwei123456]

And by this you mean a torch layer, not a spikingjelly container?

You can also add SpikingJelly's modules after a spiking neuron layer.

[RFalowo]

Brilliant! and in regards to the using activation_based.layer.linear before neuron.LIFNode. Does this create a hybrid network and add normal ANN nodes to the network even in the case of using a container? for instance:

net = layer.LinearRecurrentContainer( nn.Sequential( layer.Linear(in_features, out_features), neuron.LIFNode(), ), in_features, out_features)

[RFalowo]

Apologies i think i just need an example of what you mean.

[fangwei123456]

Does this create a hybrid network

Yes, but whether it is a problem depends on your purpose of using SNNs. If you seek for biological plausibility rather than deployment on neuromorphic chips, then the float values will not be a problem.