Package org.apache.beam.sdk.ml.inference.remote

Class RemoteInference.Invoke<InputT extends BaseInput,OutputT extends BaseResponse>

java.lang.Object

org.apache.beam.sdk.transforms.PTransform<PCollection<InputT>,PCollection<Iterable<PredictionResult<InputT,OutputT>>>>

org.apache.beam.sdk.ml.inference.remote.RemoteInference.Invoke<InputT,OutputT>

All Implemented Interfaces:

Serializable, HasDisplayData

Enclosing class:

RemoteInference

public abstract static class RemoteInference.Invoke<InputT extends BaseInput,OutputT extends BaseResponse>
extends PTransform<PCollection<InputT>,PCollection<Iterable<PredictionResult<InputT,OutputT>>>>

See Also:

• Serialized Form

Field Summary

Fields inherited from class org.apache.beam.sdk.transforms.PTransform

annotations, displayData, name, resourceHints

Constructor Summary

Constructors

Constructor	Description
Invoke()

Method Summary

Modifier and Type	Method	Description
PCollection<Iterable<PredictionResult<InputT,OutputT>>>	expand(PCollection<InputT> input)	Override this method to specify how this PTransform should be expanded on the given InputT.
RemoteInference.Invoke<InputT,OutputT>	handler(Class<? extends BaseModelHandler<?,InputT,OutputT>> modelHandler)	Model handler class for inference.
RemoteInference.Invoke<InputT,OutputT>	withBatchConfig(BatchElements.BatchConfig batchConfig)	Configures the batching behavior for the inputs.
RemoteInference.Invoke<InputT,OutputT>	withOutputCoder(Coder<OutputT> outputCoder)	Configures the coder for the output of the model.
RemoteInference.Invoke<InputT,OutputT>	withOverloadRatio(double overloadRatio)	Configures the target ratio between requests sent and successful requests.
RemoteInference.Invoke<InputT,OutputT>	withParameters(BaseModelParameters modelParameters)	Configures the parameters for model initialization.
RemoteInference.Invoke<InputT,OutputT>	withSamplePeriodMs(long samplePeriodMs)	Configures the length of history to consider when setting throttling probability.
RemoteInference.Invoke<InputT,OutputT>	withSampleUpdateMs(long sampleUpdateMs)	Configures the granularity of time buckets that we store data in for throttling.
RemoteInference.Invoke<InputT,OutputT>	withThrottleDelaySecs(int throttleDelaySecs)	Configures the throttling delay when the client is preemptively throttled.

Methods inherited from class org.apache.beam.sdk.transforms.PTransform

addAnnotation, compose, compose, getAdditionalInputs, getAnnotations, getDefaultOutputCoder, getDefaultOutputCoder, getDefaultOutputCoder, getKindString, getName, getResourceHints, populateDisplayData, setDisplayData, setResourceHints, toString, validate, validate

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait

Constructor Details

Invoke

public Invoke()

Method Details

handler

public RemoteInference.Invoke<InputT,OutputT> handler(Class<? extends BaseModelHandler<?,InputT,OutputT>> modelHandler)

Model handler class for inference.

withParameters

public RemoteInference.Invoke<InputT,OutputT> withParameters(BaseModelParameters modelParameters)

Configures the parameters for model initialization.

withBatchConfig

public RemoteInference.Invoke<InputT,OutputT> withBatchConfig(BatchElements.BatchConfig batchConfig)

Configures the batching behavior for the inputs.

withThrottleDelaySecs

public RemoteInference.Invoke<InputT,OutputT> withThrottleDelaySecs(int throttleDelaySecs)

Configures the throttling delay when the client is preemptively throttled. Defaults to 5 seconds. A value of 0 disables throttling. For more context, see ReactiveThrottler

withSamplePeriodMs

public RemoteInference.Invoke<InputT,OutputT> withSamplePeriodMs(long samplePeriodMs)

Configures the length of history to consider when setting throttling probability. Defaults to a sample period of 1000ms. For more context, see

invalid reference

AdaptiveThrottler

withSampleUpdateMs

public RemoteInference.Invoke<InputT,OutputT> withSampleUpdateMs(long sampleUpdateMs)

Configures the granularity of time buckets that we store data in for throttling. Defaults to a sample period of 1000ms. For more context, see

invalid reference

AdaptiveThrottler

withOverloadRatio

public RemoteInference.Invoke<InputT,OutputT> withOverloadRatio(double overloadRatio)

Configures the target ratio between requests sent and successful requests. Defaults to an overload ratio of 2.0. For more context, see

invalid reference

AdaptiveThrottler

withOutputCoder

public RemoteInference.Invoke<InputT,OutputT> withOutputCoder(Coder<OutputT> outputCoder)

Configures the coder for the output of the model. If not provided, it will fallback to using standard Java serialization for the output element.

expand

public PCollection<Iterable<PredictionResult<InputT,OutputT>>> expand(PCollection<InputT> input)

Description copied from class: PTransform

Override this method to specify how this PTransform should be expanded on the given InputT.

NOTE: This method should not be called directly. Instead apply the PTransform should be applied to the InputT using the apply method.

Composite transforms, which are defined in terms of other transforms, should return the output of one of the composed transforms. Non-composite transforms, which do not apply any transforms internally, should return a new unbound output and register evaluators (via backend-specific registration methods).

Specified by:

expand in class PTransform<PCollection<InputT extends BaseInput>,PCollection<Iterable<PredictionResult<InputT extends BaseInput,OutputT extends BaseResponse>>>>