class

ML::Autograd::AccumulateGrad

Inherits ML::Autograd::GradFn < Reference < Object

Accumulate gradient (no-op backward, just stores gradient)

Constructors

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