Documentation
¶
Overview ¶
Package inference has the cross language implementation of RunInference API implemented in Python SDK. An expansion service for python external transforms can be started by running
$ python -m apache_beam.runners.portability.expansion_service_main -p $PORT_FOR_EXPANSION_SERVICE
Index ¶
Constants ¶
This section is empty.
Variables ¶
This section is empty.
Functions ¶
func SklearnModel ¶
func SklearnModel(modelURI string, opts ...sklearnConfig) sklearn
SklearnModel configures the parameters required to perform RunInference transform on Sklearn Model. It returns an sklearn object which should be used to call RunInference transform. ModelURI is the required parameter indicating the path to the sklearn model.
Example:
modelURI := "gs://storage/model"
model := inference.SklearnModel(modelURI)
prediction := model.RunInference(s, input, inference.WithExpansionAddr("localhost:9000"))
func WithArgs ¶
func WithArgs(args []string) runInferenceOption
WithArgs set arguments for the RunInference transform parameters.
func WithExpansionAddr ¶
func WithExpansionAddr(expansionAddr string) runInferenceOption
WithExpansionAddr provides URL for Python expansion service.
func WithExtraPackages ¶
func WithExtraPackages(extraPackages []string) runInferenceOption
WithExtraPackages is used to specify additional packages when using an automated expansion service. Packages required to run the required Model are included implicitly, eg: scikit-learn, pandas for Sklearn Model Handler.
Types ¶
type PredictionResult ¶
type PredictionResult struct {
Example []int64 `beam:"example"`
Inference int32 `beam:"inference"`
}
PredictionResult represents the result of a prediction obtained from Python's RunInference API.
Source Files