Version: v0.0.0-...-1581884 Latest Latest

This package is not in the latest version of its module.

Go to latest
Published: Apr 6, 2018 License: Apache-2.0 Imports: 9 Imported by: 1


This is an implementation of a datasetter specialized in getting char based training set




This section is empty.


This section is empty.


This section is empty.


type Prediction

type Prediction struct {
	// contains filtered or unexported fields

Prediction is the based type that can be used as a training dataset

func NewPrediction

func NewPrediction(input string, runeToIdx func(r rune) (int, error), sampleSize, vocabSize int) *Prediction

NewPrediction return an object suitable for the LSTM

func (*Prediction) GetOutput

func (p *Prediction) GetOutput() [][]float32

GetOutput ...

func (*Prediction) Read

func (p *Prediction) Read() ([]float32, error)

Float32Read ...

func (*Prediction) Write

func (p *Prediction) Write(val []float32) error

Float32Write ...

type Section

type Section struct {
	// contains filtered or unexported fields

Section ...

func (*Section) GetComputedVectors

func (s *Section) GetComputedVectors() G.Nodes

GetComputedVectors ..

func (*Section) GetExpectedValue

func (s *Section) GetExpectedValue(offset int) (int, error)

GetExpectedValue returns the encoded value of the rune next to the one present at offset

func (*Section) ReadInputVector

func (s *Section) ReadInputVector(g *G.ExprGraph) (*G.Node, error)

ReadInputVector returns the input vector until it reach the penultimate rune the ultimate rune is not used as input within the current section as an input

func (*Section) WriteComputedVector

func (s *Section) WriteComputedVector(n *G.Node) error

WriteComputedVector add the computed vectors to the output

type TrainingSet

type TrainingSet struct {
	// contains filtered or unexported fields

TrainingSet ...

func NewTrainingSet

func NewTrainingSet(rs io.ReadSeeker, runeToIdx func(r rune) (int, error), vocabSize, batchsize, step int) *TrainingSet

NewTrainingSet from a ReadSeeker

func (*TrainingSet) GetTrainer

func (t *TrainingSet) GetTrainer() (datasetter.Trainer, error)

GetTrainer returns a pointer so a Section. It reads batchSize runes and add it to the returned section. The offset of the underlying io.ReadSeeker is set to the position it had when entering the function + step * runes * rune_size Any error in reading is returned

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL