【Hackathon 9th Sprint No.1】add api compatibility for paddle.unique - part

python/paddle/compat/__init__.py

@@ -47,6 +47,7 @@
     'split',
     'min',
     'max',
+    'unique',
     'median',
     'nanmedian',
     'seed',
@@ -795,6 +796,75 @@ def sort(
     return SortRetType(values=outputs, indices=indices)
 
 
+@ForbidKeywordsDecorator(
+    illegal_keys={"x", "axis"},
+    func_name="paddle.compat.unique",
+    correct_name="paddle.unique",
+)
+def unique(
+    input: Tensor,
+    sorted: bool = True,
+    return_inverse: bool = False,
+    return_counts: bool = False,
+    dim=None,
+) -> Tensor | tuple[Tensor, ...]:
+    r"""
+    Returns the unique elements of `input` in ascending order.
+
+    Args:
+        input(Tensor): The input tensor, it's data type should be float32, float64, int32, int64.
+        sorted(bool, optional): Does not affect the return result, same as PyTorch.
+        return_inverse(bool, optional): If True, also return the indices for where elements in
+            the original input ended up in the returned unique tensor.
+        return_counts(bool, optional): If True, also return the counts for each unique element.
+        dim(int, optional): The axis to apply unique. If None, the input will be flattened.
+            Default: None.
+
+    Returns:
+        tuple (output, inverse_indices, counts). `output` is the unique tensor for `input`. \
+            `inverse_indices` is provided only if `return_inverse` \
+            is True. `counts` is provided only if `return_counts` is True.
+
+    Examples:
+        .. code-block:: python
+
+            >>> import paddle
+
+            >>> x = paddle.to_tensor([2, 3, 3, 1, 5, 3])
+            >>> unique = paddle.compat.unique(x)
+            >>> print(unique)
+            Tensor(shape=[4], dtype=int64, place=Place(cpu), stop_gradient=True,
+            [1, 2, 3, 5])
+
+            >>> _, inverse_indices, counts = paddle.compat.unique(x, return_inverse=True, return_counts=True)
+            >>> print(inverse_indices)
+            Tensor(shape=[6], dtype=int64, place=Place(cpu), stop_gradient=True,
+            [1, 2, 2, 0, 3, 2])
+            >>> print(counts)
+            Tensor(shape=[4], dtype=int64, place=Place(cpu), stop_gradient=True,
+            [1, 1, 3, 1])
+
+            >>> x = paddle.to_tensor([[2, 1, 3], [3, 0, 1], [2, 1, 3]])
+            >>> unique = paddle.compat.unique(x)
+            >>> print(unique)
+            Tensor(shape=[4], dtype=int64, place=Place(cpu), stop_gradient=True,
+            [0, 1, 2, 3])
+
+            >>> unique = paddle.compat.unique(x, dim=0)
+            >>> print(unique)
+            Tensor(shape=[2, 3], dtype=int64, place=Place(cpu), stop_gradient=True,
+            [[2, 1, 3],
+             [3, 0, 1]])
+    """
+    return paddle.unique(
+        input,
+        return_inverse=return_inverse,
+        return_counts=return_counts,
+        axis=dim,
+        sorted=sorted,
+    )
+
+
 @ForbidKeywordsDecorator(
     illegal_keys={"x", "num_or_sections", "axis", "name"},
     func_name="paddle.compat.split",

python/paddle/tensor/manipulation.py

@@ -3672,6 +3672,7 @@ def unique(
     return_counts: Literal[True] = ...,
     axis: int | None = ...,
     dtype: DTypeLike = ...,
+    sorted: bool = ...,
     name: str | None = ...,
 ) -> tuple[Tensor, Tensor, Tensor, Tensor]: ...
 
@@ -3684,6 +3685,7 @@ def unique(
     return_counts: Literal[True] = ...,
     axis: int | None = ...,
     dtype: DTypeLike = ...,
+    sorted: bool = ...,
     name: str | None = ...,
 ) -> tuple[Tensor, Tensor, Tensor]: ...
 
@@ -3696,6 +3698,7 @@ def unique(
     return_counts: Literal[True] = ...,
     axis: int | None = ...,
     dtype: DTypeLike = ...,
+    sorted: bool = ...,
     name: str | None = ...,
 ) -> tuple[Tensor, Tensor, Tensor]: ...
 
@@ -3708,6 +3711,7 @@ def unique(
     return_counts: Literal[False] = ...,
     axis: int | None = ...,
     dtype: DTypeLike = ...,
+    sorted: bool = ...,
     name: str | None = ...,
 ) -> tuple[Tensor, Tensor, Tensor]: ...
 
@@ -3720,6 +3724,7 @@ def unique(
     return_counts: Literal[True] = ...,
     axis: int | None = ...,
     dtype: DTypeLike = ...,
+    sorted: bool = ...,
     name: str | None = ...,
 ) -> tuple[Tensor, Tensor]: ...
 
@@ -3732,6 +3737,7 @@ def unique(
     return_counts: Literal[False] = ...,
     axis: int | None = ...,
     dtype: DTypeLike = ...,
+    sorted: bool = ...,
     name: str | None = ...,
 ) -> tuple[Tensor, Tensor]: ...
 
@@ -3744,6 +3750,7 @@ def unique(
     return_counts: Literal[False] = ...,
     axis: int | None = ...,
     dtype: DTypeLike = ...,
+    sorted: bool = ...,
     name: str | None = ...,
 ) -> tuple[Tensor, Tensor]: ...
 
@@ -3756,6 +3763,7 @@ def unique(
     return_counts: Literal[False] = ...,
     axis: int | None = ...,
     dtype: DTypeLike = ...,
+    sorted: bool = ...,
     name: str | None = ...,
 ) -> Tensor: ...
 
@@ -3768,6 +3776,7 @@ def unique(
     return_counts: bool = False,
     axis: int | None = ...,
     dtype: DTypeLike = ...,
+    sorted: bool = ...,
     name: str | None = ...,
 ) -> Tensor | tuple[Tensor, ...]: ...
 
@@ -3779,6 +3788,7 @@ def unique(
     return_counts=False,
     axis=None,
     dtype="int64",
+    sorted=True,
     name=None,
 ):
     r"""
@@ -3795,6 +3805,7 @@ def unique(
             Default: None.
         dtype(str|paddle.dtype|np.dtype, optional): The date type of `indices` or `inverse` tensor: int32 or int64.
             Default: int64.
+        sorted(bool, optional): Does not affect the return result, same as PyTorch.
         name(str|None, optional): Name for the operation. For more information, please refer to
             :ref:`api_guide_Name`. Default: None.
 

test/legacy_test/test_compat_unique.py

@@ -0,0 +1,85 @@
+# Copyright (c) 2025 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import unittest
+
+import numpy as np
+
+import paddle
+from paddle import base
+
+
+class TestCompatUniqueAPI(unittest.TestCase):
+    def test_basic(self):
+        paddle.disable_static()
+        x = paddle.to_tensor([2, 3, 3, 1, 5, 3])
+        result = paddle.compat.unique(x)
+        expected = paddle.to_tensor([1, 2, 3, 5], dtype='int64')
+        np.testing.assert_allclose(result.numpy(), expected.numpy())
+
+        _, inverse_indices, counts = paddle.compat.unique(
+            x, return_inverse=True, return_counts=True
+        )
+        expected_indices = paddle.to_tensor([1, 2, 2, 0, 3, 2], dtype='int64')
+        expected_counts = paddle.to_tensor([1, 1, 3, 1], dtype='int64')
+        np.testing.assert_allclose(
+            inverse_indices.numpy(), expected_indices.numpy()
+        )
+        np.testing.assert_allclose(counts.numpy(), expected_counts.numpy())
+
+        x = paddle.to_tensor([[2, 1, 3], [3, 0, 1], [2, 1, 3]])
+        result = paddle.compat.unique(x)
+        expected = paddle.to_tensor([0, 1, 2, 3], dtype='int64')
+        np.testing.assert_allclose(result.numpy(), expected.numpy())
+        paddle.enable_static()
+
+    def test_static(self):
+        paddle.enable_static()
+
+        with paddle.static.program_guard(
+            paddle.static.Program(), paddle.static.Program()
+        ):
+            x = paddle.static.data(name='input', shape=[6], dtype='int64')
+            out, inverse_indices, counts = paddle.compat.unique(
+                x, return_inverse=True, return_counts=True
+            )
+
+            exe = base.Executor(base.CPUPlace())
+            x_data = np.array([2, 3, 3, 1, 5, 3], dtype='int64')
+            result = exe.run(
+                feed={'input': x_data},
+                fetch_list=[out, inverse_indices, counts],
+            )
+
+            np.testing.assert_allclose(result[1], [1, 2, 2, 0, 3, 2])
+            np.testing.assert_allclose(result[2], [1, 1, 3, 1])
+
+        with paddle.static.program_guard(
+            paddle.static.Program(), paddle.static.Program()
+        ):
+            x = paddle.static.data(name='input', shape=[3, 3], dtype='int64')
+            out = paddle.compat.unique(x)
+
+            exe = base.Executor(base.CPUPlace())
+            x_data = np.array([[2, 1, 3], [3, 0, 1], [2, 1, 3]], dtype='int64')
+            result = exe.run(feed={'input': x_data}, fetch_list=[out])
+
+            expected = np.array([0, 1, 2, 3], dtype='int64')
+            np.testing.assert_allclose(result[0], expected)
+
+        paddle.disable_static()
+
+
+if __name__ == '__main__':
+    unittest.main()

test/legacy_test/test_unique.py

@@ -500,5 +500,50 @@ def test_dygraph_api_out(self):
         np.testing.assert_allclose(out.numpy(), expected_out)
 
 
+class TestUniqueAPI_Compatibility(unittest.TestCase):
+    def setUp(self):
+        self.x_np = np.random.random(size=[3, 5]).astype("float32")
+        self.place = (
+            core.CUDAPlace(0)
+            if core.is_compiled_with_cuda()
+            else core.CPUPlace()
+        )
+
+    def test_dygraph(self):
+        paddle.disable_static()
+        out = paddle.unique(paddle.to_tensor(self.x_np))
+        expected_out = np.unique(self.x_np)
+        np.testing.assert_allclose(out.numpy(), expected_out)
+
+    def test_static(self):
+        paddle.enable_static()
+        x = paddle.static.data(name='x1', shape=[-1, 5], dtype='float32')
+        out1 = paddle.unique(x)
+        out2 = paddle.unique(x=x)
+        exe = paddle.static.Executor(self.place)
+        res = exe.run(
+            feed={
+                'x1': self.x_np.reshape(3, 5),
+            },
+            fetch_list=[out1, out2],
+        )
+        expected_out = np.unique(self.x_np)
+        for result in res:
+            np.testing.assert_array_equal(result, expected_out)
+        paddle.disable_static()
+
+    def test_dygraph_sorted(self):
+        paddle.disable_static()
+        out = paddle.unique(paddle.to_tensor(self.x_np), sorted=True)
+        expected_out = np.unique(self.x_np)
+        np.testing.assert_allclose(out.numpy(), expected_out)
+
+    def test_dygraph_axis(self):
+        paddle.disable_static()
+        out = paddle.unique(paddle.to_tensor(self.x_np), sorted=True, axis=1)
+        expected_out = np.unique(self.x_np, axis=1)
+        np.testing.assert_allclose(out.numpy(), expected_out)
+
+
 if __name__ == "__main__":
     unittest.main()