flytekitplugins.onnxtensorflow.TensorFlow2ONNXConfig

class flytekitplugins.onnxtensorflow.TensorFlow2ONNXConfig(input_signature, custom_ops=None, target=None, custom_op_handlers=None, custom_rewriter=None, opset=None, extra_opset=None, shape_override=None, inputs_as_nchw=None, large_model=False)

TensorFlow2ONNXConfig is the config used during the tensorflow to ONNX conversion.

Parameters

• input_signature (Union[tensorflow.python.framework.tensor.TensorSpec, numpy.ndarray]) – The shape/dtype of the inputs to the model.

• custom_ops (Optional[Dict[str, Any]]) – If a model contains ops not recognized by ONNX runtime, you can tag these ops with a custom op domain so that the runtime can still open the model.

• target (Optional[List[Any]]) – List of workarounds applied to help certain platforms.

• custom_op_handlers (Optional[Dict[Any, Tuple]]) – A dictionary of custom op handlers.

• custom_rewriter (Optional[List[Any]]) – A list of custom graph rewriters.

• opset (Optional[int]) – The ONNX opset number.

• extra_opset (Optional[List[int]]) – The extra ONNX opset numbers to be used by, say, custom ops.

• shape_override (Optional[Dict[str, List[Any]]]) – Dict with inputs that override the shapes given by tensorflow.

• inputs_as_nchw (Optional[List[str]]) – Transpose inputs in list from nhwc to nchw.

• large_model (bool) – Whether to use the ONNX external tensor storage format.

Return type

None

Methods

classmethod from_dict(kvs, *, infer_missing=False)

Parameters

kvs (Optional[Union[dict, list, str, int, float, bool]]) –

Return type

dataclasses_json.api.A

classmethod from_json(s, *, parse_float=None, parse_int=None, parse_constant=None, infer_missing=False, **kw)

Parameters

s (Union[str, bytes, bytearray]) –

Return type

dataclasses_json.api.A

classmethod schema(*, infer_missing=False, only=None, exclude=(), many=False, context=None, load_only=(), dump_only=(), partial=False, unknown=None)

Parameters

• infer_missing (bool) –

• many (bool) –

• partial (bool) –

Return type

dataclasses_json.mm.SchemaF[dataclasses_json.mm.A]

to_dict(encode_json=False)

Return type

Dict[str, Optional[Union[dict, list, str, int, float, bool]]]

to_json(*, skipkeys=False, ensure_ascii=True, check_circular=True, allow_nan=True, indent=None, separators=None, default=None, sort_keys=False, **kw)

Parameters

• skipkeys (bool) –

• ensure_ascii (bool) –

• check_circular (bool) –

• allow_nan (bool) –

• indent (Optional[Union[int, str]]) –

• separators (Optional[Tuple[str, str]]) –

• default (Optional[Callable]) –

• sort_keys (bool) –

Return type

str

Attributes

custom_op_handlers: Optional[Dict[Any, Tuple]] = None

custom_ops: Optional[Dict[str, Any]] = None

custom_rewriter: Optional[List[Any]] = None

dataclass_json_config = None

extra_opset: Optional[List[int]] = None

inputs_as_nchw: Optional[List[str]] = None

large_model: bool = False

opset: Optional[int] = None

shape_override: Optional[Dict[str, List[Any]]] = None

target: Optional[List[Any]] = None

input_signature: Union[tensorflow.python.framework.tensor.TensorSpec, numpy.ndarray]