ONNXModel¶

class mlfmu.types.onnx_model.ONNXModel(onnx_path: str | PathLike[str], *, time_input: bool = False)¶

Bases: object

ONNX Metadata class.

Represents an ONNX model and provides methods to load inputs and outputs. Allows to import the ONNX file and figure out the input/output sizes.

Raises:

ValueError – If the ml model has 3 inputs, but the usesTime flag is set to false in the json interface.
ValueError – If the number of inputs to the ml model is larger than 3.
ValueError – If the number of outputs from the ml model is not exactly 1.

__init__(onnx_path: str | PathLike[str], *, time_input: bool = False) → None¶

Initialize the ONNXModel object by loading the ONNX file and assigning model parameters.

Parameters:

onnx_path (Union[str, os.PathLike[str]]) – The path to the ONNX file.
time_input (bool, optional) – Flag indicating whether the model uses time input. Defaults to False.

Methods

`__init__`(onnx_path, *[, time_input])	Initialize the ONNXModel object by loading the ONNX file and assigning model parameters.
`load_inputs`()	Load the inputs from the ONNX file and assign the input name and size.
`load_outputs`()	Load the outputs from the ONNX file and assign the output name and size.

Attributes

`filename`	The name of the ONNX file.
`input_name`	The name of the main input.
`input_size`	The size of the main input.
`output_name`	The name of the output.
`output_size`	The size of the output.
`state_size`	The size of the internal states input.
`states_name`	The name of the internal states input.
`time_input`	Flag indicating whether the model uses time input.
`time_input_name`	The name of the time input.

load_inputs() → None¶: Load the inputs from the ONNX file and assign the input name and size.

load_outputs() → None¶: Load the outputs from the ONNX file and assign the output name and size.

time_input: bool = False¶: Flag indicating whether the model uses time input.