Safe Haskell	Safe-Inferred
Language	Haskell2010

Torch.Typed.Autograd

Synopsis

Documentation

class HasGrad a b | a -> b Source #

Minimal complete definition

grad, toDependent

Instances details

(HasGrad a b, HasGrad (HList as) (HList bs), Castable (HList (b ': bs)) [ATenTensor]) => HasGrad (HList (a ': as)) (HList (b ': bs)) Source #
Instance details Defined in Torch.Typed.Autograd Methods grad :: forall (dtype :: DType) (device :: (DeviceType, Nat)). Tensor device dtype '[] -> HList (a ': as) -> HList (b ': bs) Source # toDependent :: HList (a ': as) -> HList (b ': bs)
HasGrad (HList ('[] :: [Type])) (HList ('[] :: [Type])) Source #
Instance details Defined in Torch.Typed.Autograd Methods grad :: forall (dtype :: DType) (device :: (DeviceType, Nat)). Tensor device dtype '[] -> HList '[] -> HList '[] Source # toDependent :: HList '[] -> HList '[]
HasGrad (Parameter device dtype shape) (Tensor device dtype shape) Source #
Instance details Defined in Torch.Typed.Autograd Methods grad :: forall (dtype0 :: DType) (device0 :: (DeviceType, Nat)). Tensor device0 dtype0 '[] -> Parameter device dtype shape -> Tensor device dtype shape Source # toDependent :: Parameter device dtype shape -> Tensor device dtype shape

grad :: forall dtype device. HasGrad a b => Tensor device dtype '[] -> a -> b Source #

calculate gradients of a zero-dimensional tensor with respect to a list of parameters