I am trying to build a network with 2 layers of a custom LIF neuron type. For the connection from the first layer to the second I want to apply a modified ReLu activation function to the weighted outputs (instead of f(x) = x if x > 0, i want to be able to change the threshold value/have a particular value for each neuron), and then only allow the inputs that are top inputs (for example, only the highest input, or the top 5 inputs) to be used in updating the voltage values of the next layer. While I am not sure on how to do alot of this, my first question is simply how do I look at the inputs to a given layer? I want to see how the inputs to the second layer change as I am building this network. I saw that ‘inputs’ is probeable for ensembles, but when I attempt to probe the inputs for an ensemble and then plot the results, I get simply a blank plot.
Hello @n299, can you please share your code to reproduce your issue?
After further testing on other neuron types, I found that the ensemble input probe works for both types. However, is there a way to see numerical values for the input (in matrix/array format)? I understand that ‘input’ is what is being given to the network to decode and reproduce, but is there a way to access the ‘input’ in terms of what is directly influencing the voltages of the layer, for example, in the LIF update rules for voltage in the step function?