evaluator

Documentation

Index

• type EvaluatorMeta
  • func MakeEvaluatorMeta(id string, semaphore *semaphore.Weighted, maxEvaluatorWait time.Duration, ...) EvaluatorMeta
  • func (e EvaluatorMeta) TFMetric() *gmetric.Operation
• type Service
  • func NewEvaluator(signature *domain.Signature, session *tf.Session, meta EvaluatorMeta) *Service
  • func (s *Service) Close() error
  • func (s *Service) Evaluate(ctx context.Context, params []interface{}) ([]interface{}, error)

Types

type EvaluatorMeta

type EvaluatorMeta struct {
	// contains filtered or unexported fields
}

func MakeEvaluatorMeta

func MakeEvaluatorMeta(id string, semaphore *semaphore.Weighted, maxEvaluatorWait time.Duration,
	semaMetric, tfMetric *gmetric.Operation) EvaluatorMeta

func (EvaluatorMeta) TFMetric

func (e EvaluatorMeta) TFMetric() *gmetric.Operation

type Service

type Service struct {
	EvaluatorMeta
	// contains filtered or unexported fields
}

Represents a TF session for running predictions.

func NewEvaluator

func NewEvaluator(signature *domain.Signature, session *tf.Session, meta EvaluatorMeta) *Service

func (*Service) Close

func (s *Service) Close() error

Close will block until all goroutines waiting on the semaphore are released, and complete the Tensorflow session run, or their context errors. Will then close the session.

func (*Service) Evaluate

func (s *Service) Evaluate(ctx context.Context, params []interface{}) ([]interface{}, error)

Evaluate runs the primary model prediction via Cgo Tensorflow. params is expected to be [inputs][1][batch]T - see service/request.Request.Feeds and related methods.