npx.npx_numpy_tensors#
- class onnx_array_api.npx.npx_numpy_tensors.EagerNumpyTensor(tensor: ndarray)[source]#
Defines a value for a specific backend.
- class onnx_array_api.npx.npx_numpy_tensors.JitNumpyTensor(tensor: ndarray)[source]#
Defines a value for a specific backend.
- class onnx_array_api.npx.npx_numpy_tensors.NumpyTensor(tensor: ndarray)[source]#
Default backend based on
onnx.reference.ReferenceEvaluator()
.- Parameters:
input_names – input names
onx – onnx model
- class Evaluator(tensor_class: type, input_names: List[str], onx: ModelProto)[source]#
Wraps class
onnx.reference.ReferenceEvaluator
to have a signature closer to python function.- run(*inputs: List[NumpyTensor]) List[NumpyTensor] [source]#
Executes the function.
- Parameters:
inputs – function inputs
- Returns:
outputs
- const_cast(to: Any | None = None) EagerTensor [source]#
Casts a constant without any ONNX conversion.
- classmethod create_function(input_names: List[str], onx: ModelProto) Callable [source]#
Creates a python function calling the onnx backend used by this class.
- Parameters:
onx – onnx model
- Returns:
python function
- property dims#
Returns the dimensions of the tensor. First dimension is the batch dimension if the tensor has more than one dimension.
- classmethod get_ir_version(ir_version)[source]#
Updates the IR version. This method should be overloaded. By default, it returns ir_version.
- classmethod get_opsets(opsets)[source]#
Updates the opsets for a given backend. This method should be overloaded. By default, it returns opsets.
- property ndim#
Returns the number of dimensions (rank).
- property tensor_type: TensorType#
Returns the tensor type of this tensor.
- property tensor_type_dims: TensorType#
Returns the tensor type of this tensor. This property is used to define a key used to cache a jitted function. Same keys keys means same ONNX graph. Different keys usually means same ONNX graph but different input shapes.