fsmn

Documentation

Index

• type Model
  • func New(in, out, order int) *Model
  • func (m *Model) NewProc(ctx nn.Context) nn.Processor
• type Processor
  • func (p *Processor) Forward(xs ...ag.Node) []ag.Node
  • func (p *Processor) SetInitialState(state *State)
• type State

Types

type Model

type Model struct {
	W    *nn.Param   `type:"weights"`
	WRec *nn.Param   `type:"weights"`
	WS   []*nn.Param `type:"weights"` // coefficient vectors for scaling
	B    *nn.Param   `type:"biases"`
	// contains filtered or unexported fields
}

This is variant of the Feedforward Sequential Memory Networks (https://arxiv.org/pdf/1512.08301.pdf) where the neurons in the same hidden layer are independent of each other and they are connected across layers as in the IndRNN.

func New

func New(in, out, order int) *Model

New returns a new model with parameters initialized to zeros.

func (*Model) NewProc

func (m *Model) NewProc(ctx nn.Context) nn.Processor

NewProc returns a new processor to execute the forward step.

type Processor

type Processor struct {
	nn.BaseProcessor

	States []*State
	// contains filtered or unexported fields
}

Model contains the serializable parameters.

func (*Processor) Forward

func (p *Processor) Forward(xs ...ag.Node) []ag.Node

Forward performs the forward step for each input and returns the result.

func (*Processor) SetInitialState

func (p *Processor) SetInitialState(state *State)

type State

type State struct {
	Y ag.Node
}