square_weights = get_square_weights(
input_exc_weights.view(784, n_neurons), n_sqrt, 28)
square_assignments = get_square_assignments(assignments, n_sqrt)
spikes_ = {
layer: spikes[layer].get("s")[:, 0].contiguous()
for layer in spikes
}
voltages = {"Ae": exc_voltages, "Ai": inh_voltages}
inpt_axes, inpt_ims = plot_input(image,
inpt,
label=labels[step],
axes=inpt_axes,
ims=inpt_ims)
spike_ims, spike_axes = plot_spikes(spikes_,
ims=spike_ims,
axes=spike_axes)
weights_im = plot_weights(square_weights, im=weights_im)
assigns_im = plot_assignments(square_assignments, im=assigns_im)
perf_ax = plot_performance(accuracy, ax=perf_ax)
voltage_ims, voltage_axes = plot_voltages(voltages,
ims=voltage_ims,
axes=voltage_axes,
plot_type="line")
plt.pause(1e-8)
network.reset_state_variables() # Reset state variables.
print("Progress: %d / %d (%.4f seconds)" % (epoch + 1, n_epochs, t() - start))
print("Training complete.\n")
input_exc_weights = network.connections[("X", "Ae")].w
square_weights = get_square_weights(
input_exc_weights.view(784, n_neurons), n_sqrt, 28)
voltages = {"Ae": exc_voltages, "Ai": inh_voltages}
inpt_axes, inpt_ims, inpt_fig = plot_input(image.sum(1).view(28, 28),
inpt,
label=label,
axes=inpt_axes,
ims=inpt_ims,
fig=inpt_fig)
save_plot(fig=inpt_fig, directory=directory, name="input", n=i)
spike_ims, spike_axes, spike_fig = plot_spikes(
{
layer: spikes[layer].get("s").view(time, 1, -1)
for layer in spikes
},
ims=spike_ims,
axes=spike_axes,
fig=spike_fig)
save_plot(fig=spike_fig, directory=directory, name="spikes", n=i)
weights_im, weights_fig = plot_weights(square_weights,
im=weights_im,
fig=weights_fig)
save_plot(fig=weights_fig, directory=directory, name="weights", n=i)
voltage_ims, voltage_axes, voltage_fig = plot_voltages(
voltages, ims=voltage_ims, axes=voltage_axes, fig=voltage_fig)
save_plot(fig=voltage_fig, directory=directory, name="voltage", n=i)
if i % update_interval == 0:
square_assignments = get_square_assignments(assignments, n_sqrt)
assigns_im, assigns_fig = plot_assignments(square_assignments,
im=assigns_im,
if plot:
image = batch["image"].view(28, 28)
inpt = inputs["X"].view(time, 784).sum(0).view(28, 28)
weights1 = conv_conn.w
_spikes = {
"X": spikes["X"].get("s").view(time, -1),
"Y": spikes["Y"].get("s").view(time, -1),
}
_voltages = {"Y": voltages["Y"].get("v").view(time, -1)}
inpt_axes, inpt_ims = plot_input(image,
inpt,
label=label,
axes=inpt_axes,
ims=inpt_ims)
spike_ims, spike_axes = plot_spikes(_spikes,
ims=spike_ims,
axes=spike_axes)
weights1_im = plot_conv2d_weights(weights1, im=weights1_im)
voltage_ims, voltage_axes = plot_voltages(_voltages,
ims=voltage_ims,
axes=voltage_axes)
plt.pause(1)
network.reset_state_variables() # Reset state variables.
print("Progress: %d / %d (%.4f seconds)\n" % (n_epochs, n_epochs, t() - start))
print("Training complete.\n")
network.run(inputs={"I": datum}, time=time, input_time_dim=1)
training_pairs.append([spikes["O"].get("s").sum(0), label])
if plot:
inpt_axes, inpt_ims = plot_input(
dataPoint["image"].view(28, 28),
datum.view(time, 784).sum(0).view(28, 28),
label=label,
axes=inpt_axes,
ims=inpt_ims,
)
spike_ims, spike_axes = plot_spikes(
{layer: spikes[layer].get("s").view(-1, time)
for layer in spikes},
axes=spike_axes,
ims=spike_ims,
)
voltage_ims, voltage_axes = plot_voltages(
{
layer: voltages[layer].get("v").view(-1, time)
for layer in voltages
},
ims=voltage_ims,
axes=voltage_axes,
)
weights_im = plot_weights(get_square_weights(C1.w, 23, 28),
im=weights_im,
wmin=-2,
wmax=2)
weights_im2 = plot_weights(C2.w, im=weights_im2, wmin=-2, wmax=2)