qualia_plugin_snn.learningmodel.pytorch.SMLP module

Contains the template for a spiking multi-layer perceptron.

class qualia_plugin_snn.learningmodel.pytorch.SMLP.SMLP

Bases: SNN

Spiking multi-layer perceptron template.

Similar to qualia_core.learningmodel.pytorch.MLP.MLP but with spiking neuron activation layers (e.g., IF) instead of torch.nn.ReLU.

Last torch.nn.Linear layer matching number of output classes is implicitely added.

Example TOML configuration for a 3-layer spiking MLP over 4 timesteps with soft-reset multi-step IF based on the SMLP template:

[[model]]
kind = "SMLP"
name = "smlp_128-128-10"
params.units        = [128, 128]
params.timesteps    = 4
params.neuron.kind                = 'IFNode'
params.neuron.params.v_reset      = false # Soft reset
params.neuron.params.v_threshold  = 1.0
params.neuron.params.detach_reset = true
params.neuron.params.step_mode    = 'm' # Multi-step mode, make sure to use SpikingJellyMultiStep learningframework
params.neuron.params.backend      = 'torch'

__init__(input_shape: tuple[int, ...], output_shape: tuple[int, ...], units: list[int], timesteps: int, neuron: RecursiveConfigDict | None = None) → None

Construct SMLP.

Parameters:

• input_shape (tuple[int, ...]) – Input shape

• output_shape (tuple[int, ...]) – Output shape

• units (list[int]) – List of torch.nn.Linear layer out_features to add in the network

• neuron (RecursiveConfigDict | None) – Spiking neuron configuration, see qualia_plugin_snn.learningmodel.pytorch.SNN.SNN.__init__()

• timesteps (int) – Number of timesteps

Return type:

None

forward(input: Tensor) → Tensor

Forward calls each of the MLP layers sequentially.

Parameters:

input (Tensor) – Input tensor

Returns:

Output tensor

Return type:

Tensor