ModelParallelRunner

View Source On Gitee

import com.mindspore.config.RunnerConfig;

ModelParallelRunner defines MindSpore Lite concurrent inference.

Public Member Functions

function

long getModelParallelRunnerPtr()

boolean init()

boolean predict()

boolean getInputs()

boolean getOutputs()

void free()

getModelParallelRunnerPtr

public long getModelParallelRunnerPtr()

Get the underlying concurrent inference class pointer.

  • Returns

    Low-level concurrent inference class pointer.

init

public boolean init(String modelPath, RunnerConfig runnerConfig)

Read and load models according to the path, generate one or more models, and compile all models to a state that can be run on the Device.

  • Parameters

    • modelPath: model file path.

    • runnerConfig: A RunnerConfig structure. Defines configuration parameters for the concurrent inference model.

  • Returns

    Whether the initialization is successful.

public boolean init(String modelPath)

Read and load models according to the path, generate one or more models, and compile all models to a state that can be run on the Device.

  • Parameters

    • modelPath: model file path.

  • Returns

    Whether the initialization is successful.

predict

public boolean predict(List<MSTensor> inputs, List<MSTensor> outputs)

Concurrent inference model.

  • Parameters

    • inputs: model input.

    • outputs: model output.

  • Returns

    Whether the inference is successful.

getInputs

public List<MSTensor> getInputs()

Get all input tensors of the model.

  • Returns

    A list of input tensors for the model.

getOutputs

public List<MSTensor> getOutputs()

Get all input tensors of the model.

  • Returns

    A list of output tensors for the model.

free

public void free()

Free concurrent inference class memory.