Gaussian#

class sinabs.activation.Gaussian(mu: float = 0.0, sigma: float = 0.5, grad_scale: float = 1.0)[source]#

Gaussian surrogate gradient function.

Parameters

mu: The mean of the Gaussian. sigma: The standard deviation of the Gaussian. grad_scale: Scale the gradients arbitrarily.

Parameters:
  • mu (float) –

  • sigma (float) –

  • grad_scale (float) –