onnx 1.16.2__cp311-cp311-win32.whl → 1.18.0__cp311-cp311-win32.whl

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

@@ -1,6 +1,7 @@
 # Copyright (c) ONNX Project Contributors
 #
 # SPDX-License-Identifier: Apache-2.0
+from __future__ import annotations
 
 import numpy as np
 

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

@@ -1,6 +1,7 @@
 # Copyright (c) ONNX Project Contributors
 #
 # SPDX-License-Identifier: Apache-2.0
+from __future__ import annotations
 
 import numpy as np
 

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

@@ -1,6 +1,7 @@
 # Copyright (c) ONNX Project Contributors
 #
 # SPDX-License-Identifier: Apache-2.0
+from __future__ import annotations
 
 import numpy as np
 
@@ -10,7 +11,7 @@ from onnx.backend.test.case.node import expect
 from onnx.defs import AI_ONNX_PREVIEW_TRAINING_DOMAIN
 
 
-def apply_momentum(r, t, x, g, v, norm_coefficient, alpha, beta):  # type: ignore
+def apply_momentum(r, t, x, g, v, norm_coefficient, alpha, beta):
     # Add gradient of regularization term.
     g_regularized = norm_coefficient * x + g
     # Coefficient of gradient should be 1 at the first iteration.
@@ -22,7 +23,7 @@ def apply_momentum(r, t, x, g, v, norm_coefficient, alpha, beta):  # type: ignor
     return x_new, v_new
 
 
-def apply_nesterov(r, t, x, g, v, norm_coefficient, alpha, beta):  # type: ignore
+def apply_nesterov(r, t, x, g, v, norm_coefficient, alpha, beta):
     # Add gradient of regularization term.
     g_regularized = norm_coefficient * x + g
     # Coefficient of gradient should be 1 at the first iteration.

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

@@ -1,6 +1,7 @@
 # Copyright (c) ONNX Project Contributors
 #
 # SPDX-License-Identifier: Apache-2.0
+from __future__ import annotations
 
 import numpy as np
 
@@ -28,11 +29,36 @@ class Mul(Base):
         z = x * y
         expect(node, inputs=[x, y], outputs=[z], name="test_mul")
 
+        x = np.random.randint(4, size=(3, 4, 5), dtype=np.int8)
+        y = np.random.randint(24, size=(3, 4, 5), dtype=np.int8)
+        z = x * y
+        expect(node, inputs=[x, y], outputs=[z], name="test_mul_int8")
+
+        x = np.random.randint(4, size=(3, 4, 5), dtype=np.int16)
+        y = np.random.randint(24, size=(3, 4, 5), dtype=np.int16)
+        z = x * y
+        expect(node, inputs=[x, y], outputs=[z], name="test_mul_int16")
+
         x = np.random.randint(4, size=(3, 4, 5), dtype=np.uint8)
         y = np.random.randint(24, size=(3, 4, 5), dtype=np.uint8)
         z = x * y
         expect(node, inputs=[x, y], outputs=[z], name="test_mul_uint8")
 
+        x = np.random.randint(4, size=(3, 4, 5), dtype=np.uint16)
+        y = np.random.randint(24, size=(3, 4, 5), dtype=np.uint16)
+        z = x * y
+        expect(node, inputs=[x, y], outputs=[z], name="test_mul_uint16")
+
+        x = np.random.randint(4, size=(3, 4, 5), dtype=np.uint32)
+        y = np.random.randint(24, size=(3, 4, 5), dtype=np.uint32)
+        z = x * y
+        expect(node, inputs=[x, y], outputs=[z], name="test_mul_uint32")
+
+        x = np.random.randint(4, size=(3, 4, 5), dtype=np.uint64)
+        y = np.random.randint(24, size=(3, 4, 5), dtype=np.uint64)
+        z = x * y
+        expect(node, inputs=[x, y], outputs=[z], name="test_mul_uint64")
+
     @staticmethod
     def export_mul_broadcast() -> None:
         node = onnx.helper.make_node(

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

@@ -1,6 +1,7 @@
 # Copyright (c) ONNX Project Contributors
 #
 # SPDX-License-Identifier: Apache-2.0
+from __future__ import annotations
 
 import numpy as np
 

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

@@ -1,6 +1,7 @@
 # Copyright (c) ONNX Project Contributors
 #
 # SPDX-License-Identifier: Apache-2.0
+from __future__ import annotations
 
 import numpy as np
 
@@ -9,7 +10,9 @@ from onnx.backend.test.case.base import Base
 from onnx.backend.test.case.node import expect
 
 
-def compute_negative_log_likelihood_loss(input, target, weight=None, reduction="mean", ignore_index=None):  # type: ignore
+def compute_negative_log_likelihood_loss(
+    input, target, weight=None, reduction="mean", ignore_index=None
+):
     input_shape = input.shape
     if len(input_shape) == 1:
         raise RuntimeError("Unsupported shape")

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

@@ -1,6 +1,7 @@
 # Copyright (c) ONNX Project Contributors
 #
 # SPDX-License-Identifier: Apache-2.0
+from __future__ import annotations
 
 import numpy as np
 

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

@@ -1,6 +1,7 @@
 # Copyright (c) ONNX Project Contributors
 #
 # SPDX-License-Identifier: Apache-2.0
+from __future__ import annotations
 
 import numpy as np
 

onnx/backend/test/case/node/not.py

@@ -1,6 +1,7 @@
 # Copyright (c) ONNX Project Contributors
 #
 # SPDX-License-Identifier: Apache-2.0
+from __future__ import annotations
 
 import numpy as np
 

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

@@ -1,6 +1,7 @@
 # Copyright (c) ONNX Project Contributors
 #
 # SPDX-License-Identifier: Apache-2.0
+from __future__ import annotations
 
 import numpy as np
 
@@ -9,7 +10,7 @@ from onnx.backend.test.case.base import Base
 from onnx.backend.test.case.node import expect
 
 
-def one_hot(indices, depth, axis=-1, dtype=np.float32):  # type: ignore
+def one_hot(indices, depth, axis=-1, dtype=np.float32):
     """Compute one hot from indices at a specific axis"""
     values = np.asarray(indices)
     rank = len(values.shape)

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

@@ -1,8 +1,9 @@
 # Copyright (c) ONNX Project Contributors
 #
 # SPDX-License-Identifier: Apache-2.0
+from __future__ import annotations
 
-from typing import Any, Optional
+from typing import Any
 
 import numpy as np
 
@@ -11,7 +12,7 @@ from onnx.backend.test.case.base import Base
 from onnx.backend.test.case.node import expect
 
 
-def optional_get_element_reference_implementation(optional: Optional[Any]) -> Any:
+def optional_get_element_reference_implementation(optional: Any | None) -> Any:
     assert optional is not None
     return optional
 

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

@@ -1,8 +1,7 @@
 # Copyright (c) ONNX Project Contributors
 #
 # SPDX-License-Identifier: Apache-2.0
-
-from typing import Optional
+from __future__ import annotations
 
 import numpy as np
 
@@ -12,7 +11,7 @@ from onnx.backend.test.case.node import expect
 
 
 def optional_has_element_reference_implementation(
-    optional: Optional[np.ndarray],
+    optional: np.ndarray | None,
 ) -> np.ndarray:
     if optional is None:
         return np.array(False)

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

@@ -1,6 +1,7 @@
 # Copyright (c) ONNX Project Contributors
 #
 # SPDX-License-Identifier: Apache-2.0
+from __future__ import annotations
 
 import numpy as np
 

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

@@ -1,6 +1,7 @@
 # Copyright (c) ONNX Project Contributors
 #
 # SPDX-License-Identifier: Apache-2.0
+from __future__ import annotations
 
 import numpy as np
 
@@ -9,7 +10,7 @@ from onnx.backend.test.case.base import Base
 from onnx.backend.test.case.node import expect
 
 
-def pad_impl(data, raw_pads, mode, constant_values=0.0, axes=None):  # type: ignore
+def pad_impl(data, raw_pads, mode, constant_values=0.0, axes=None):
     input_rank = data.ndim
     if axes is None:
         axes = list(range(input_rank))
@@ -18,7 +19,7 @@ def pad_impl(data, raw_pads, mode, constant_values=0.0, axes=None):  # type: ign
     num_axes = len(axes)
 
     if num_axes * 2 != raw_pads.size:
-        raise Exception("The number of elements in raw_pads should be 2 * num_axes")
+        raise ValueError("The number of elements in raw_pads should be 2 * num_axes")
 
     pad_width = []
     for _ in range(input_rank):

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

@@ -1,6 +1,7 @@
 # Copyright (c) ONNX Project Contributors
 #
 # SPDX-License-Identifier: Apache-2.0
+from __future__ import annotations
 
 import numpy as np
 

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

@@ -1,6 +1,7 @@
 # Copyright (c) ONNX Project Contributors
 #
 # SPDX-License-Identifier: Apache-2.0
+from __future__ import annotations
 
 import numpy as np
 

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

@@ -1,6 +1,7 @@
 # Copyright (c) ONNX Project Contributors
 #
 # SPDX-License-Identifier: Apache-2.0
+from __future__ import annotations
 
 import numpy as np
 

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

@@ -1,6 +1,7 @@
 # Copyright (c) ONNX Project Contributors
 #
 # SPDX-License-Identifier: Apache-2.0
+from __future__ import annotations
 
 import numpy as np
 

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

@@ -1,6 +1,7 @@
 # Copyright (c) ONNX Project Contributors
 #
 # SPDX-License-Identifier: Apache-2.0
+from __future__ import annotations
 
 import numpy as np
 
@@ -275,6 +276,44 @@ class QuantizeLinear(Base):
             name="test_quantizelinear_int4",
         )
 
+    @staticmethod
+    def export_float4e2m1() -> None:
+        node = onnx.helper.make_node(
+            "QuantizeLinear",
+            inputs=["x", "y_scale", "y_zero_point"],
+            outputs=["y"],
+            axis=0,
+        )
+
+        x = np.array(
+            [
+                [0.0, 2.5, 4.8, 8.6],
+                [-30, -20, 6, 9],
+                [-0.0, -2.5, -4.8, -8.6],
+            ]
+        ).astype(np.float32)
+
+        y_scale = np.asarray([2.0, 3.0, 4.0], dtype=np.float32)
+        y_zero_point = make_tensor(
+            "y_zero_point",
+            TensorProto.FLOAT4E2M1,
+            y_scale.shape,
+            np.zeros_like(y_scale),
+        )
+        y = make_tensor(
+            "y",
+            TensorProto.FLOAT4E2M1,
+            x.shape,
+            [0, 1, 2, 4, -6, -6, 2, 3, 0, -0.5, -1, -2],
+        )
+
+        expect(
+            node,
+            inputs=[x, y_scale, y_zero_point],
+            outputs=[y],
+            name="test_quantizelinear_float4e2m1",
+        )
+
     @staticmethod
     def export_blocked_asymmetric() -> None:
         node = onnx.helper.make_node(

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

@@ -1,6 +1,7 @@
 # Copyright (c) ONNX Project Contributors
 #
 # SPDX-License-Identifier: Apache-2.0
+from __future__ import annotations
 
 import numpy as np
 

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

@@ -1,6 +1,7 @@
 # Copyright (c) ONNX Project Contributors
 #
 # SPDX-License-Identifier: Apache-2.0
+from __future__ import annotations
 
 import numpy as np
 

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

@@ -1,6 +1,7 @@
 # Copyright (c) ONNX Project Contributors
 #
 # SPDX-License-Identifier: Apache-2.0
+from __future__ import annotations
 
 import numpy as np
 

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

@@ -1,6 +1,7 @@
 # Copyright (c) ONNX Project Contributors
 #
 # SPDX-License-Identifier: Apache-2.0
+from __future__ import annotations
 
 import numpy as np
 

onnx/backend/test/case/node/reducel1.py

@@ -1,6 +1,7 @@
 # Copyright (c) ONNX Project Contributors
 #
 # SPDX-License-Identifier: Apache-2.0
+from __future__ import annotations
 
 import numpy as np
 

onnx/backend/test/case/node/reducel2.py

@@ -1,6 +1,7 @@
 # Copyright (c) ONNX Project Contributors
 #
 # SPDX-License-Identifier: Apache-2.0
+from __future__ import annotations
 
 import numpy as np
 

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

@@ -1,6 +1,7 @@
 # Copyright (c) ONNX Project Contributors
 #
 # SPDX-License-Identifier: Apache-2.0
+from __future__ import annotations
 
 import numpy as np
 
@@ -226,5 +227,5 @@ class ReduceMax(Base):
             node,
             inputs=[data, axes],
             outputs=[reduced],
-            name="test_reduce_min_empty_set",
+            name="test_reduce_max_empty_set",
         )

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

@@ -1,6 +1,7 @@
 # Copyright (c) ONNX Project Contributors
 #
 # SPDX-License-Identifier: Apache-2.0
+from __future__ import annotations
 
 import numpy as np
 

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

@@ -1,6 +1,7 @@
 # Copyright (c) ONNX Project Contributors
 #
 # SPDX-License-Identifier: Apache-2.0
+from __future__ import annotations
 
 import numpy as np
 

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

@@ -1,6 +1,7 @@
 # Copyright (c) ONNX Project Contributors
 #
 # SPDX-License-Identifier: Apache-2.0
+from __future__ import annotations
 
 import numpy as np
 

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

@@ -1,6 +1,7 @@
 # Copyright (c) ONNX Project Contributors
 #
 # SPDX-License-Identifier: Apache-2.0
+from __future__ import annotations
 
 import numpy as np
 
@@ -192,7 +193,7 @@ class ReduceSum(Base):
             node,
             inputs=[data, axes],
             outputs=[reduced],
-            name="test_reduce_sum_negative_axes_keepdims_random",
+            name="test_reduce_sum_empty_axes_input_noop",
         )
 
     @staticmethod

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

@@ -1,6 +1,7 @@
 # Copyright (c) ONNX Project Contributors
 #
 # SPDX-License-Identifier: Apache-2.0
+from __future__ import annotations
 
 import numpy as np
 

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

@@ -1,6 +1,7 @@
 # Copyright (c) ONNX Project Contributors
 #
 # SPDX-License-Identifier: Apache-2.0
+from __future__ import annotations
 
 import numpy as np
 

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

@@ -1,6 +1,7 @@
 # Copyright (c) ONNX Project Contributors
 #
 # SPDX-License-Identifier: Apache-2.0
+from __future__ import annotations
 
 import numpy as np
 

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

@@ -1,6 +1,7 @@
 # Copyright (c) ONNX Project Contributors
 #
 # SPDX-License-Identifier: Apache-2.0
+from __future__ import annotations
 
 import numpy as np
 

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

@@ -1,6 +1,7 @@
 # Copyright (c) ONNX Project Contributors
 #
 # SPDX-License-Identifier: Apache-2.0
+from __future__ import annotations
 
 import numpy as np
 
@@ -856,7 +857,7 @@ class Resize(Base):
             node,
             inputs=[data, roi, sizes],
             outputs=[output],
-            name="test_resize_tf_crop_and_resize",
+            name="test_resize_tf_crop_and_resize_extrapolation_value",
         )
 
     @staticmethod
@@ -1563,7 +1564,7 @@ class Resize(Base):
             node,
             inputs=[data, sizes],
             outputs=[output],
-            name="test_resize_upsample_sizes_nearest_not_larger",
+            name="test_resize_upsample_sizes_nearest_not_smaller",
         )
 
     @staticmethod

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

@@ -1,6 +1,7 @@
 # Copyright (c) ONNX Project Contributors
 #
 # SPDX-License-Identifier: Apache-2.0
+from __future__ import annotations
 
 import numpy as np
 

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

@@ -0,0 +1,126 @@
+# Copyright (c) ONNX Project Contributors
+#
+# SPDX-License-Identifier: Apache-2.0
+from __future__ import annotations
+
+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_rms_normalization import _rms_normalization
+
+
+def calculate_normalized_shape(x_shape, axis):
+    rank = len(x_shape)
+    if axis < 0:
+        axis = axis + rank
+    return x_shape[axis:]
+
+
+class RMSNormalization(Base):
+    @staticmethod
+    def export() -> None:
+        X = np.random.randn(2, 3, 4, 5).astype(np.float32)
+
+        def case(axis: int) -> None:
+            normalized_shape = calculate_normalized_shape(X.shape, axis)
+            W = np.random.randn(*normalized_shape).astype(np.float32)
+            Y = _rms_normalization(X, W, axis=axis)
+
+            node = onnx.helper.make_node(
+                "RMSNormalization",
+                inputs=["X", "W"],
+                outputs=["Y"],
+                axis=axis,
+            )
+
+            if axis < 0:
+                name = f"test_rms_normalization_4d_axis_negative_{-axis}"
+            else:
+                name = f"test_rms_normalization_4d_axis{axis}"
+
+            expect(node, inputs=[X, W], outputs=[Y], name=name)
+
+        for i in range(len(X.shape)):
+            case(i)
+            case(i - len(X.shape))
+
+    @staticmethod
+    def export_default_axis() -> None:
+        X = np.random.randn(2, 3, 4, 5).astype(np.float32)
+
+        # Default axis in RMSNormalization is -1.
+        normalized_shape = calculate_normalized_shape(X.shape, -1)
+        W = np.random.randn(*normalized_shape).astype(np.float32)
+        # Axis is default to -1 in the reference implementation.
+        Y = _rms_normalization(X, W)
+
+        # Not specifying axis attribute means -1.
+        node = onnx.helper.make_node(
+            "RMSNormalization",
+            inputs=["X", "W"],
+            outputs=["Y"],
+        )
+
+        expect(
+            node,
+            inputs=[X, W],
+            outputs=[Y],
+            name="test_rms_normalization_default_axis",
+        )
+
+    @staticmethod
+    def export2d() -> None:
+        X = np.random.randn(3, 4).astype(np.float32)
+
+        def case(axis: int) -> None:
+            normalized_shape = calculate_normalized_shape(X.shape, axis)
+            W = np.random.randn(*normalized_shape).astype(np.float32)
+            Y = _rms_normalization(X, W, axis=axis)
+
+            node = onnx.helper.make_node(
+                "RMSNormalization",
+                inputs=["X", "W"],
+                outputs=["Y"],
+                axis=axis,
+            )
+
+            if axis < 0:
+                name = f"test_rms_normalization_2d_axis_negative_{-axis}"
+            else:
+                name = f"test_rms_normalization_2d_axis{axis}"
+
+            expect(node, inputs=[X, W], outputs=[Y], name=name)
+
+        for i in range(len(X.shape)):
+            case(i)
+            case(i - len(X.shape))
+
+    @staticmethod
+    def export3d_epsilon() -> None:
+        epsilon = 1e-1
+        X = np.random.randn(2, 3, 5).astype(np.float32)
+
+        def case(axis: int) -> None:
+            normalized_shape = calculate_normalized_shape(X.shape, axis)
+            W = np.random.randn(*normalized_shape).astype(np.float32)
+            Y = _rms_normalization(X, W, axis=axis, epsilon=epsilon)
+            node = onnx.helper.make_node(
+                "RMSNormalization",
+                inputs=["X", "W"],
+                outputs=["Y"],
+                axis=axis,
+                epsilon=epsilon,
+            )
+
+            if axis < 0:
+                name = f"test_rms_normalization_3d_axis_negative_{-axis}_epsilon"
+            else:
+                name = f"test_rms_normalization_3d_axis{axis}_epsilon"
+
+            expect(node, inputs=[X, W], outputs=[Y], name=name)
+
+        for i in range(len(X.shape)):
+            case(i)
+            case(i - len(X.shape))

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

@@ -1,8 +1,9 @@
 # Copyright (c) ONNX Project Contributors
 #
 # SPDX-License-Identifier: Apache-2.0
+from __future__ import annotations
 
-from typing import Any, Tuple
+from typing import Any
 
 import numpy as np
 
@@ -28,9 +29,9 @@ class RNNHelper:
         self.num_directions = params[str(W)].shape[0]
 
         if self.num_directions == 1:
-            for k in params:
+            for k, v in params.items():
                 if k != X:
-                    params[k] = np.squeeze(params[k], axis=0)
+                    params[k] = np.squeeze(v, axis=0)
 
             hidden_size = params[R].shape[-1]
             batch_size = params[X].shape[1]
@@ -62,7 +63,7 @@ class RNNHelper:
     def f(self, x: np.ndarray) -> np.ndarray:
         return np.tanh(x)
 
-    def step(self) -> Tuple[np.ndarray, np.ndarray]:
+    def step(self) -> tuple[np.ndarray, np.ndarray]:
         seq_length = self.X.shape[0]
         hidden_size = self.H_0.shape[-1]
         batch_size = self.X.shape[1]