GCN

lamp.nn.graph.GCN
See theGCN companion object
case class GCN[M <: Module](transform: M & Module) extends GraphModule

Attributes

Companion
object
Graph
Supertypes
trait Serializable
trait Product
trait Equals
class Object
trait Matchable
class Any
Show all

Members list

Value members

Concrete methods

override def forward[S : Sc](x: Graph): Graph

The implementation of the function.

The implementation of the function.

In addition of x it can also use all the `state to compute its value.

Attributes

Definition Classes
def state: Seq[(Constant, PTag)]

List of optimizable, or non-optimizable, but stateful parameters

List of optimizable, or non-optimizable, but stateful parameters

Stateful means that the state is carried over the repeated forward calls.

Attributes

Inherited methods

def apply[S : Sc](a: Graph): B

Alias of forward

Alias of forward

Attributes

Inherited from:
GenericModule
final def gradients(loss: Variable, zeroGrad: Boolean): Seq[Option[STen]]

Computes the gradient of loss with respect to the parameters.

Computes the gradient of loss with respect to the parameters.

Attributes

Inherited from:
GenericModule
final def learnableParameters: Long

Returns the total number of optimizable parameters.

Returns the total number of optimizable parameters.

Attributes

Inherited from:
GenericModule
final def parameters: Seq[(Constant, PTag)]

Returns the state variables which need gradient computation.

Returns the state variables which need gradient computation.

Attributes

Inherited from:
GenericModule
def productElementNames: Iterator[String]

Attributes

Inherited from:
Product
def productIterator: Iterator[Any]

Attributes

Inherited from:
Product
final def zeroGrad(): Unit

Attributes

Inherited from:
GenericModule