class

GS::Autograd::SigmoidBackward

Inherits GS::Autograd::GradFn < Reference < Object

Gradient for Sigmoid: out = 1 / (1 + exp(-x)) d(sigmoid)/dx = out * (1 - out)

Constructors

new(output_data : Tensor)
Source

Instance methods

backward(grad_output : Tensor) : Array(Tensor | Nil)

Compute gradients w.r.t. inputs given gradient of output Returns array of gradients, one per input (or nil if input doesn't require grad)

Source