onnx 1.15.0__cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl → 1.16.1__cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl

onnx/__init__.py

@@ -16,6 +16,7 @@ __all__ = [
     "IR_VERSION_2019_9_19",
     "IR_VERSION_2020_5_8",
     "IR_VERSION_2021_7_30",
+    "IR_VERSION_2023_5_5",
     "EXPERIMENTAL",
     "STABLE",
     # Modules
@@ -93,6 +94,7 @@ from onnx.onnx_pb import (
     IR_VERSION_2019_9_19,
     IR_VERSION_2020_5_8,
     IR_VERSION_2021_7_30,
+    IR_VERSION_2023_5_5,
     ModelProto,
     NodeProto,
     OperatorSetIdProto,
@@ -186,7 +188,7 @@ def _get_serializer(
 
 def load_model(
     f: IO[bytes] | str | os.PathLike,
-    format: _SupportedFormat | None = None,
+    format: _SupportedFormat | None = None,  # noqa: A002
     load_external_data: bool = True,
 ) -> ModelProto:
     """Loads a serialized ModelProto into memory.
@@ -218,7 +220,7 @@ def load_model(
 
 def load_tensor(
     f: IO[bytes] | str | os.PathLike,
-    format: _SupportedFormat | None = None,
+    format: _SupportedFormat | None = None,  # noqa: A002
 ) -> TensorProto:
     """Loads a serialized TensorProto into memory.
 
@@ -237,7 +239,7 @@ def load_tensor(
 
 def load_model_from_string(
     s: bytes | str,
-    format: _SupportedFormat = _DEFAULT_FORMAT,
+    format: _SupportedFormat = _DEFAULT_FORMAT,  # noqa: A002
 ) -> ModelProto:
     """Loads a binary string (bytes) that contains serialized ModelProto.
 
@@ -256,7 +258,7 @@ def load_model_from_string(
 
 def load_tensor_from_string(
     s: bytes,
-    format: _SupportedFormat = _DEFAULT_FORMAT,
+    format: _SupportedFormat = _DEFAULT_FORMAT,  # noqa: A002
 ) -> TensorProto:
     """Loads a binary string (bytes) that contains serialized TensorProto.
 
@@ -276,7 +278,7 @@ def load_tensor_from_string(
 def save_model(
     proto: ModelProto | bytes,
     f: IO[bytes] | str | os.PathLike,
-    format: _SupportedFormat | None = None,
+    format: _SupportedFormat | None = None,  # noqa: A002
     *,
     save_as_external_data: bool = False,
     all_tensors_to_one_file: bool = True,
@@ -284,8 +286,7 @@ def save_model(
     size_threshold: int = 1024,
     convert_attribute: bool = False,
 ) -> None:
-    """
-    Saves the ModelProto to the specified path and optionally, serialize tensors with raw data as external data before saving.
+    """Saves the ModelProto to the specified path and optionally, serialize tensors with raw data as external data before saving.
 
     Args:
         proto: should be a in-memory ModelProto
@@ -330,10 +331,9 @@ def save_model(
 def save_tensor(
     proto: TensorProto,
     f: IO[bytes] | str | os.PathLike,
-    format: _SupportedFormat | None = None,
+    format: _SupportedFormat | None = None,  # noqa: A002
 ) -> None:
-    """
-    Saves the TensorProto to the specified path.
+    """Saves the TensorProto to the specified path.
 
     Args:
         proto: should be a in-memory TensorProto

onnx/backend/base.py

@@ -14,9 +14,7 @@ from onnx import IR_VERSION, ModelProto, NodeProto
 
 
 class DeviceType:
-    """
-    Describes device type.
-    """
+    """Describes device type."""
 
     _Type = NewType("_Type", int)
     CPU: _Type = _Type(0)
@@ -24,8 +22,7 @@ class DeviceType:
 
 
 class Device:
-    """
-    Describes device type and device id
+    """Describes device type and device id
     syntax: device_type:device_id(optional)
     example: 'CPU', 'CUDA', 'CUDA:1'
     """
@@ -56,8 +53,7 @@ def namedtupledict(
 
 
 class BackendRep:
-    """
-    BackendRep is the handle that a Backend returns after preparing to execute
+    """BackendRep is the handle that a Backend returns after preparing to execute
     a model repeatedly. Users will then pass inputs to the run function of
     BackendRep to retrieve the corresponding results.
     """
@@ -68,8 +64,7 @@ class BackendRep:
 
 
 class Backend:
-    """
-    Backend is the entity that will take an ONNX model with inputs,
+    """Backend is the entity that will take an ONNX model with inputs,
     perform a computation, and then return the output.
 
     For one-off execution, users can use run_node and run_model to obtain results quickly.
@@ -112,11 +107,16 @@ class Backend:
         **kwargs: Dict[str, Any],
     ) -> Optional[Tuple[Any, ...]]:
         """Simple run one operator and return the results.
+
         Args:
+            node: The node proto.
+            inputs: Inputs to the node.
+            device: The device to run on.
             outputs_info: a list of tuples, which contains the element type and
-            shape of each output. First element of the tuple is the dtype, and
-            the second element is the shape. More use case can be found in
-            https://github.com/onnx/onnx/blob/main/onnx/backend/test/runner/__init__.py
+                shape of each output. First element of the tuple is the dtype, and
+                the second element is the shape. More use case can be found in
+                https://github.com/onnx/onnx/blob/main/onnx/backend/test/runner/__init__.py
+            kwargs: Other keyword arguments.
         """
         # TODO Remove Optional from return type
         if "opset_version" in kwargs:
@@ -131,8 +131,7 @@ class Backend:
 
     @classmethod
     def supports_device(cls, device: str) -> bool:
-        """
-        Checks whether the backend is compiled with particular device support.
+        """Checks whether the backend is compiled with particular device support.
         In particular it's used in the testing suite.
         """
         return True

onnx/backend/sample/ops/abs.py

@@ -3,5 +3,5 @@
 import numpy as np
 
 
-def abs(input: np.ndarray) -> np.ndarray:
+def abs(input: np.ndarray) -> np.ndarray:  # noqa: A001
     return np.abs(input)  # type: ignore[no-any-return]

onnx/backend/test/case/model/__init__.py

@@ -42,7 +42,6 @@ BASE_URL = "onnx/backend/test/data/light/light_%s.onnx"
 
 def collect_testcases() -> List[TestCase]:
     """Collect model test cases defined in python/numpy code."""
-
     real_model_testcases = []
 
     model_tests = [

onnx/backend/test/case/node/affinegrid.py

@@ -1,209 +1,209 @@
-# Copyright (c) ONNX Project Contributors
-#
-# SPDX-License-Identifier: Apache-2.0
-
-import numpy as np
-
-import onnx
-from onnx.backend.test.case.base import Base
-from onnx.backend.test.case.node import expect
-from onnx.reference.ops.op_affine_grid import (
-    apply_affine_transform,
-    construct_original_grid,
-)
-
-
-def create_affine_matrix_3d(
-    angle1,
-    angle2,
-    offset_x,
-    offset_y,
-    offset_z,
-    shear_x,
-    shear_y,
-    shear_z,
-    scale_x,
-    scale_y,
-    scale_z,
-):
-    rot_x = np.stack(
-        [
-            np.ones_like(angle1),
-            np.zeros_like(angle1),
-            np.zeros_like(angle1),
-            np.zeros_like(angle1),
-            np.cos(angle1),
-            -np.sin(angle1),
-            np.zeros_like(angle1),
-            np.sin(angle1),
-            np.cos(angle1),
-        ],
-        axis=-1,
-    ).reshape(-1, 3, 3)
-    rot_y = np.stack(
-        [
-            np.cos(angle2),
-            np.zeros_like(angle2),
-            np.sin(angle2),
-            np.zeros_like(angle2),
-            np.ones_like(angle2),
-            np.zeros_like(angle2),
-            -np.sin(angle2),
-            np.zeros_like(angle2),
-            np.cos(angle2),
-        ],
-        axis=-1,
-    ).reshape(-1, 3, 3)
-    shear = np.stack(
-        [
-            np.ones_like(shear_x),
-            shear_x,
-            shear_y,
-            shear_z,
-            np.ones_like(shear_x),
-            shear_x,
-            shear_y,
-            shear_x,
-            np.ones_like(shear_x),
-        ],
-        axis=-1,
-    ).reshape(-1, 3, 3)
-    scale = np.stack(
-        [
-            scale_x,
-            np.zeros_like(scale_x),
-            np.zeros_like(scale_x),
-            np.zeros_like(scale_x),
-            scale_y,
-            np.zeros_like(scale_x),
-            np.zeros_like(scale_x),
-            np.zeros_like(scale_x),
-            scale_z,
-        ],
-        axis=-1,
-    ).reshape(-1, 3, 3)
-    translation = np.transpose(np.array([offset_x, offset_y, offset_z])).reshape(
-        -1, 1, 3
-    )
-    rotation_matrix = rot_y @ rot_x @ shear @ scale  # (N, 3, 3)
-    rotation_matrix = np.transpose(rotation_matrix, (0, 2, 1))
-    affine_matrix = np.hstack((rotation_matrix, translation))
-    affine_matrix = np.transpose(affine_matrix, (0, 2, 1))
-    return affine_matrix.astype(np.float32)
-
-
-def create_affine_matrix_2d(
-    angle1, offset_x, offset_y, shear_x, shear_y, scale_x, scale_y
-):
-    rot = np.stack(
-        [np.cos(angle1), -np.sin(angle1), np.sin(angle1), np.cos(angle1)], axis=-1
-    ).reshape(-1, 2, 2)
-    shear = np.stack(
-        [np.ones_like(shear_x), shear_x, shear_y, np.ones_like(shear_x)], axis=-1
-    ).reshape(-1, 2, 2)
-    scale = np.stack(
-        [scale_x, np.zeros_like(scale_x), np.zeros_like(scale_x), scale_y], axis=-1
-    ).reshape(-1, 2, 2)
-    translation = np.transpose(np.array([offset_x, offset_y])).reshape(-1, 1, 2)
-    rotation_matrix = rot @ shear @ scale  # (N, 3, 3)
-    rotation_matrix = np.transpose(rotation_matrix, (0, 2, 1))
-    affine_matrix = np.hstack((rotation_matrix, translation))
-    affine_matrix = np.transpose(affine_matrix, (0, 2, 1))
-    return affine_matrix.astype(np.float32)
-
-
-def create_theta_2d():
-    angle = np.array([np.pi / 4, np.pi / 3])
-    offset_x = np.array([5.0, 2.5])
-    offset_y = np.array([-3.3, 1.1])
-    shear_x = np.array([-0.5, 0.5])
-    shear_y = np.array([0.3, -0.3])
-    scale_x = np.array([2.2, 1.1])
-    scale_y = np.array([3.1, 0.9])
-    theta_2d = create_affine_matrix_2d(
-        angle, offset_x, offset_y, shear_x, shear_y, scale_x, scale_y
-    )
-    return theta_2d
-
-
-def create_theta_3d():
-    angle1 = np.array([np.pi / 4, np.pi / 3])
-    angle2 = np.array([np.pi / 6, np.pi / 2])
-    offset_x = np.array([5.0, 2.5])
-    offset_y = np.array([-3.3, 1.1])
-    offset_z = np.array([-1.1, 2.2])
-    shear_x = np.array([-0.5, 0.5])
-    shear_y = np.array([0.3, -0.3])
-    shear_z = np.array([0.7, -0.2])
-    scale_x = np.array([2.2, 1.1])
-    scale_y = np.array([3.1, 0.9])
-    scale_z = np.array([0.5, 1.5])
-
-    theta_3d = create_affine_matrix_3d(
-        angle1,
-        angle2,
-        offset_x,
-        offset_y,
-        offset_z,
-        shear_x,
-        shear_y,
-        shear_z,
-        scale_x,
-        scale_y,
-        scale_z,
-    )
-    return theta_3d
-
-
-class AffineGrid(Base):
-    @staticmethod
-    def export_2d_no_reference_evaluator() -> None:
-        theta_2d = create_theta_2d()
-        N, C, H, W = len(theta_2d), 3, 5, 6
-        data_size = (H, W)
-        for align_corners in (0, 1):
-            node = onnx.helper.make_node(
-                "AffineGrid",
-                inputs=["theta", "size"],
-                outputs=["grid"],
-                align_corners=align_corners,
-            )
-
-            original_grid = construct_original_grid(data_size, align_corners)
-            grid = apply_affine_transform(theta_2d, original_grid)
-
-            test_name = "test_affine_grid_2d"
-            if align_corners == 1:
-                test_name += "_align_corners"
-            expect(
-                node,
-                inputs=[theta_2d, np.array([N, C, H, W], dtype=np.int64)],
-                outputs=[grid],
-                name=test_name,
-            )
-
-    @staticmethod
-    def export_3d_no_reference_evaluator() -> None:
-        theta_3d = create_theta_3d()
-        N, C, D, H, W = len(theta_3d), 3, 4, 5, 6
-        data_size = (D, H, W)
-        for align_corners in (0, 1):
-            node = onnx.helper.make_node(
-                "AffineGrid",
-                inputs=["theta", "size"],
-                outputs=["grid"],
-                align_corners=align_corners,
-            )
-
-            original_grid = construct_original_grid(data_size, align_corners)
-            grid = apply_affine_transform(theta_3d, original_grid)
-
-            test_name = "test_affine_grid_3d"
-            if align_corners == 1:
-                test_name += "_align_corners"
-            expect(
-                node,
-                inputs=[theta_3d, np.array([N, C, D, H, W], dtype=np.int64)],
-                outputs=[grid],
-                name=test_name,
-            )
+# Copyright (c) ONNX Project Contributors
+#
+# SPDX-License-Identifier: Apache-2.0
+
+import numpy as np
+
+import onnx
+from onnx.backend.test.case.base import Base
+from onnx.backend.test.case.node import expect
+from onnx.reference.ops.op_affine_grid import (
+    apply_affine_transform,
+    construct_original_grid,
+)
+
+
+def create_affine_matrix_3d(
+    angle1,
+    angle2,
+    offset_x,
+    offset_y,
+    offset_z,
+    shear_x,
+    shear_y,
+    shear_z,
+    scale_x,
+    scale_y,
+    scale_z,
+):
+    rot_x = np.stack(
+        [
+            np.ones_like(angle1),
+            np.zeros_like(angle1),
+            np.zeros_like(angle1),
+            np.zeros_like(angle1),
+            np.cos(angle1),
+            -np.sin(angle1),
+            np.zeros_like(angle1),
+            np.sin(angle1),
+            np.cos(angle1),
+        ],
+        axis=-1,
+    ).reshape(-1, 3, 3)
+    rot_y = np.stack(
+        [
+            np.cos(angle2),
+            np.zeros_like(angle2),
+            np.sin(angle2),
+            np.zeros_like(angle2),
+            np.ones_like(angle2),
+            np.zeros_like(angle2),
+            -np.sin(angle2),
+            np.zeros_like(angle2),
+            np.cos(angle2),
+        ],
+        axis=-1,
+    ).reshape(-1, 3, 3)
+    shear = np.stack(
+        [
+            np.ones_like(shear_x),
+            shear_x,
+            shear_y,
+            shear_z,
+            np.ones_like(shear_x),
+            shear_x,
+            shear_y,
+            shear_x,
+            np.ones_like(shear_x),
+        ],
+        axis=-1,
+    ).reshape(-1, 3, 3)
+    scale = np.stack(
+        [
+            scale_x,
+            np.zeros_like(scale_x),
+            np.zeros_like(scale_x),
+            np.zeros_like(scale_x),
+            scale_y,
+            np.zeros_like(scale_x),
+            np.zeros_like(scale_x),
+            np.zeros_like(scale_x),
+            scale_z,
+        ],
+        axis=-1,
+    ).reshape(-1, 3, 3)
+    translation = np.transpose(np.array([offset_x, offset_y, offset_z])).reshape(
+        -1, 1, 3
+    )
+    rotation_matrix = rot_y @ rot_x @ shear @ scale  # (N, 3, 3)
+    rotation_matrix = np.transpose(rotation_matrix, (0, 2, 1))
+    affine_matrix = np.hstack((rotation_matrix, translation))
+    affine_matrix = np.transpose(affine_matrix, (0, 2, 1))
+    return affine_matrix.astype(np.float32)
+
+
+def create_affine_matrix_2d(
+    angle1, offset_x, offset_y, shear_x, shear_y, scale_x, scale_y
+):
+    rot = np.stack(
+        [np.cos(angle1), -np.sin(angle1), np.sin(angle1), np.cos(angle1)], axis=-1
+    ).reshape(-1, 2, 2)
+    shear = np.stack(
+        [np.ones_like(shear_x), shear_x, shear_y, np.ones_like(shear_x)], axis=-1
+    ).reshape(-1, 2, 2)
+    scale = np.stack(
+        [scale_x, np.zeros_like(scale_x), np.zeros_like(scale_x), scale_y], axis=-1
+    ).reshape(-1, 2, 2)
+    translation = np.transpose(np.array([offset_x, offset_y])).reshape(-1, 1, 2)
+    rotation_matrix = rot @ shear @ scale  # (N, 3, 3)
+    rotation_matrix = np.transpose(rotation_matrix, (0, 2, 1))
+    affine_matrix = np.hstack((rotation_matrix, translation))
+    affine_matrix = np.transpose(affine_matrix, (0, 2, 1))
+    return affine_matrix.astype(np.float32)
+
+
+def create_theta_2d():
+    angle = np.array([np.pi / 4, np.pi / 3])
+    offset_x = np.array([5.0, 2.5])
+    offset_y = np.array([-3.3, 1.1])
+    shear_x = np.array([-0.5, 0.5])
+    shear_y = np.array([0.3, -0.3])
+    scale_x = np.array([2.2, 1.1])
+    scale_y = np.array([3.1, 0.9])
+    theta_2d = create_affine_matrix_2d(
+        angle, offset_x, offset_y, shear_x, shear_y, scale_x, scale_y
+    )
+    return theta_2d
+
+
+def create_theta_3d():
+    angle1 = np.array([np.pi / 4, np.pi / 3])
+    angle2 = np.array([np.pi / 6, np.pi / 2])
+    offset_x = np.array([5.0, 2.5])
+    offset_y = np.array([-3.3, 1.1])
+    offset_z = np.array([-1.1, 2.2])
+    shear_x = np.array([-0.5, 0.5])
+    shear_y = np.array([0.3, -0.3])
+    shear_z = np.array([0.7, -0.2])
+    scale_x = np.array([2.2, 1.1])
+    scale_y = np.array([3.1, 0.9])
+    scale_z = np.array([0.5, 1.5])
+
+    theta_3d = create_affine_matrix_3d(
+        angle1,
+        angle2,
+        offset_x,
+        offset_y,
+        offset_z,
+        shear_x,
+        shear_y,
+        shear_z,
+        scale_x,
+        scale_y,
+        scale_z,
+    )
+    return theta_3d
+
+
+class AffineGrid(Base):
+    @staticmethod
+    def export_2d_no_reference_evaluator() -> None:
+        theta_2d = create_theta_2d()
+        N, C, H, W = len(theta_2d), 3, 5, 6
+        data_size = (H, W)
+        for align_corners in (0, 1):
+            node = onnx.helper.make_node(
+                "AffineGrid",
+                inputs=["theta", "size"],
+                outputs=["grid"],
+                align_corners=align_corners,
+            )
+
+            original_grid = construct_original_grid(data_size, align_corners)
+            grid = apply_affine_transform(theta_2d, original_grid)
+
+            test_name = "test_affine_grid_2d"
+            if align_corners == 1:
+                test_name += "_align_corners"
+            expect(
+                node,
+                inputs=[theta_2d, np.array([N, C, H, W], dtype=np.int64)],
+                outputs=[grid],
+                name=test_name,
+            )
+
+    @staticmethod
+    def export_3d_no_reference_evaluator() -> None:
+        theta_3d = create_theta_3d()
+        N, C, D, H, W = len(theta_3d), 3, 4, 5, 6
+        data_size = (D, H, W)
+        for align_corners in (0, 1):
+            node = onnx.helper.make_node(
+                "AffineGrid",
+                inputs=["theta", "size"],
+                outputs=["grid"],
+                align_corners=align_corners,
+            )
+
+            original_grid = construct_original_grid(data_size, align_corners)
+            grid = apply_affine_transform(theta_3d, original_grid)
+
+            test_name = "test_affine_grid_3d"
+            if align_corners == 1:
+                test_name += "_align_corners"
+            expect(
+                node,
+                inputs=[theta_3d, np.array([N, C, D, H, W], dtype=np.int64)],
+                outputs=[grid],
+                name=test_name,
+            )