API Summary¶

Device¶

The package is compiled for a specific device, GPU or CPU. The CPU implementation includes optimizations such as MKL (Math Kernel Libary). The following function indicates the chosen option:

onnxruntime.get_device() → str¶

Return the device used to compute the prediction (CPU, MKL, …)

Examples and datasets¶

The package contains a few models stored in ONNX format used in the documentation. These don’t need to be downloaded as they are installed with the package.

onnxruntime.datasets.get_example(name)[source]¶

Retrieves the absolute file name of an example.

Load and run a model¶

ONNX Runtime reads a model saved in ONNX format. The main class InferenceSession wraps these functionalities in a single place.

class onnxruntime.ModelMetadata¶

Pre-defined and custom metadata about the model. It is usually used to identify the model used to run the prediction and facilitate the comparison.

custom_metadata_map¶

additional metadata

description¶

description of the model

domain¶

ONNX domain

graph_name¶

graph name

producer_name¶

producer name

version¶

version of the model

class onnxruntime.InferenceSession(path_or_bytes, sess_options=None, providers=[])[source]¶

This is the main class used to run a model.

disable_fallback()[source]¶

Disable session.run() fallback mechanism.

enable_fallback()[source]¶

Enable session.Run() fallback mechanism. If session.Run() fails due to an internal Execution Provider failure, reset the Execution Providers enabled for this session. If GPU is enabled, fall back to CUDAExecutionProvider. otherwise fall back to CPUExecutionProvider.

end_profiling()[source]¶

End profiling and return results in a file.

The results are stored in a filename if the option onnxruntime.SessionOptions.enable_profiling().

get_inputs()[source]¶

Return the inputs metadata as a list of onnxruntime.NodeArg.

get_modelmeta()[source]¶

Return the metadata. See onnxruntime.ModelMetadata.

get_outputs()[source]¶

Return the outputs metadata as a list of onnxruntime.NodeArg.

get_overridable_initializers()[source]¶

Return the inputs (including initializers) metadata as a list of onnxruntime.NodeArg.

get_providers()[source]¶

Return list of registered execution providers.

get_session_options()[source]¶

Return the session options. See onnxruntime.SessionOptions.

run(output_names, input_feed, run_options=None)[source]¶

Compute the predictions.

Parameters:	output_names – name of the outputs input_feed – dictionary { input_name: input_value } run_options – See onnxruntime.RunOptions.

sess.run([output_name], {input_name: x})

set_providers(providers)[source]¶

Register the input list of execution providers. The underlying session is re-created.

Parameters:	providers – list of execution providers

The list of providers is ordered by Priority. For example [‘CUDAExecutionProvider’, ‘CPUExecutionProvider’] means execute a node using CUDAExecutionProvider if capable, otherwise execute using CPUExecutionProvider.

class onnxruntime.NodeArg¶

Node argument definition, for both input and output, including arg name, arg type (contains both type and shape).

name¶

node name

shape¶

node shape (assuming the node holds a tensor)

type¶

node type

class onnxruntime.RunOptions¶

Configuration information for a single Run.

log_severity_level¶

Log severity level for a particular Run() invocation. 0:Verbose, 1:Info, 2:Warning. 3:Error, 4:Fatal. Default is 2.

log_verbosity_level¶

VLOG level if DEBUG build and run_log_severity_level is 0. Applies to a particular Run() invocation. Default is 0.

logid¶

To identify logs generated by a particular Run() invocation.

terminate¶

Set to True to terminate any currently executing calls that are using this RunOptions instance. The individual calls will exit gracefully and return an error status.

class onnxruntime.SessionOptions¶

Configuration information for a session.

enable_cpu_mem_arena¶

Enables the memory arena on CPU. Arena may pre-allocate memory for future usage. Set this option to false if you don’t want it. Default is True.

enable_mem_pattern¶

Enable the memory pattern optimization. Default is true.

enable_profiling¶

Enable profiling for this session. Default is false.

execution_mode¶

Sets the execution mode. Default is sequential.

graph_optimization_level¶

Graph optimization level for this session.

inter_op_num_threads¶

Sets the number of threads used to parallelize the execution of the graph (across nodes). Default is 0 to let onnxruntime choose.

intra_op_num_threads¶

Sets the number of threads used to parallelize the execution within nodes. Default is 0 to let onnxruntime choose.

log_severity_level¶

Log severity level. Applies to session load, initialization, etc. 0:Verbose, 1:Info, 2:Warning. 3:Error, 4:Fatal. Default is 2.

log_verbosity_level¶

VLOG level if DEBUG build and session_log_verbosity_level is 0. Applies to session load, initialization, etc. Default is 0.

logid¶

Logger id to use for session output.

optimized_model_filepath¶

File path to serialize optimized model. By default, optimized model is not serialized if optimized_model_filepath is not provided.

Backend¶

In addition to the regular API which is optimized for performance and usability, ONNX Runtime also implements the ONNX backend API for verification of ONNX specification conformance. The following functions are supported:

onnxruntime.backend.is_compatible(model, device=None, **kwargs)¶

Return whether the model is compatible with the backend.

Parameters:	model – unused device – None to use the default device or a string (ex: ‘CPU’)
Returns:	boolean

onnxruntime.backend.prepare(model, device=None, **kwargs)¶

Load the model and creates a onnxruntime.InferenceSession ready to be used as a backend.

Parameters:	model – ModelProto (returned by onnx.load), string for a filename or bytes for a serialized model device – requested device for the computation, None means the default one which depends on the compilation settings kwargs – see onnxruntime.SessionOptions
Returns:	onnxruntime.InferenceSession

onnxruntime.backend.run(model, inputs, device=None, **kwargs)¶

Compute the prediction.

Parameters:	model – onnxruntime.InferenceSession returned by function prepare inputs – inputs device – requested device for the computation, None means the default one which depends on the compilation settings kwargs – see onnxruntime.RunOptions
Returns:	predictions

onnxruntime.backend.supports_device(device)¶

Check whether the backend is compiled with particular device support. In particular it’s used in the testing suite.