Class ModelExecutor

• Defined in File multi_model_runner.h

Class Documentation

class ModelExecutor

Public Functions

ModelExecutor() = default

Constructor of ModelExecutor.

inline ModelExecutor(const std::vector<std::shared_ptr<ModelImpl>> &models, const std::vector<std::string> &executor_input_names, const std::vector<std::string> &executor_output_names, const std::vector<std::vector<std::string>> &subgraph_input_names)

Constructor of ModelExecutor.

Parameters

• models – [in] Which is a vector of ModelImplPtr, used to inference in ModelExecutor.

• executor_input_names – [in] Which is a vector of string, name of ModelExecutor's inputs.

• executor_output_names – [in] Which is a vector of string, name of ModelExecutor's outputs.

• subgraph_input_names – [in] Which is a vector of vector of string, name of every model's inputs in ModelExecutor.

~ModelExecutor() = default

Destructor of ModelExecutor.

Status Predict(const std::vector<MSTensor> &inputs, std::vector<MSTensor> *outputs)

Inference ModelExecutor API.

Parameters

• inputs – [in] A vector where ModelExecutor inputs are arranged in sequence.

• outputs – [out] Which is a pointer to a vector. The ModelExecutor outputs are filled in the container in sequence.

Returns

Status.

std::vector<MSTensor> GetInputs() const

Obtains all input tensors of the ModelExecutor.

Returns

The vector that includes all input tensors.

std::vector<MSTensor> GetOutputs() const

Obtains all output tensors of the ModelExecutor.

Returns

The vector that includes all output tensors.

Status Initialize(const std::shared_ptr<Context> &model_context)

Inference ModelExecutor API. If use this API in train mode, it's equal to RunStep API.

Parameters

model_context – [in] Define the context used to store options during execution.

Returns

Status.