module `npy.numpy_onnx_impl_body`#

Short summary#

module mlprodict.npy.numpy_onnx_impl_body

Design to implement graph as parameter.

class	truncated documentation
`AttributeGraph`	Class wrapping a function to make it simple as a parameter.
`if_then_else`	Overloads class `OnnxVarGraph`.
`OnnxVarGraph`	Overloads `OnnxVar` to handle graph attribute.

method	truncated documentation
`__init__`
`__init__`
`__init__`
`__repr__`	usual
`__repr__`	usual
`_graph_guess_dtype`	Guesses the graph inputs.
`_graph_guess_dtype`	Guesses the graph inputs.
`to_algebra`	Converts the variable into an operator.
`to_algebra`	Converts the variable into an operator.
`to_algebra`	Converts the variable into an operator.

Design to implement graph as parameter.

New in version 0.8.

class mlprodict.npy.numpy_onnx_impl_body.AttributeGraph(fct, *inputs)#

Class wrapping a function to make it simple as a parameter.

Parameters:

fct – function taking the list of inputs defined as OnnxVar, the function returns an OnnxVar
inputs – list of input as OnnxVar

New in version 0.8.

_graph_guess_dtype(i, var)#

Guesses the graph inputs.

Parameters:

Returns:

input type

to_algebra(op_version=None)#

Converts the variable into an operator.

class mlprodict.npy.numpy_onnx_impl_body.OnnxVarGraph(*inputs, op=None, select_output=None, dtype=None, **kwargs)#

Overloads OnnxVar to handle graph attribute.

Parameters:

inputs – variable name or object
op – ONNX operator
select_output – if multiple output are returned by ONNX operator op, it takes only one specifed by this argument
dtype – specifies the type of the variable held by this class (op is None) in that case
fields – list of attributes with the graph type
kwargs – addition argument to give operator op