hxtorch.spiking.functional.superspike.SuperSpike
-
class
hxtorch.spiking.functional.superspike.SuperSpike(*args, **kwargs) Bases:
torch.autograd.function.FunctionDefine Surrogate Gradient ‘SuperSpike’ (negative side of Fast Sigmoid) See: https://arxiv.org/abs/1705.11146
-
__init__(*args, **kwargs) Initialize self. See help(type(self)) for accurate signature.
Methods
backward(ctx, grad_output)Implements surrogate gradient ‘SuperSpike’ for backward.
forward(ctx, input, alpha)Forward function, generating spikes at positions > 0.
Attributes
-
static
backward(ctx, grad_output: torch.Tensor) → Tuple[Optional[torch.Tensor], …] Implements surrogate gradient ‘SuperSpike’ for backward.
- Parameters
grad_output – Back-propagated gradient.
- Returns
SuperSpike gradient multiplied to back-propagated gradient.
-
static
forward(ctx, input: torch.Tensor, alpha: float) → torch.Tensor Forward function, generating spikes at positions > 0.
- Parameters
input – Tensor holding v_mem - v_th, of > 0 a spike at the corresponding entry is generated. Saved for backward.
alpha – Parameter controlling the slope of the surrogate derivative. Saved for backward.
- Returns
A boolean tensor holding True at the entries for neurons emitting a spike.
-