【Hackathon 5th No.27】为 Paddle 新增 select_scatter API -part

python/paddle/__init__.py

@@ -242,6 +242,7 @@
     tolist,
     take_along_axis,
     put_along_axis,
+    select_scatter,
     tensordot,
     as_complex,
     as_real,
@@ -893,6 +894,7 @@
     'renorm_',
     'take_along_axis',
     'put_along_axis',
+    'select_scatter',
     'multigammaln',
     'multigammaln_',
     'nan_to_num',

python/paddle/tensor/__init__.py

@@ -60,6 +60,7 @@
     zeros,
     zeros_like,
 )
+
 from .einsum import einsum  # noqa: F401
 from .linalg import (  # noqa: F401
     bincount,
@@ -173,6 +174,7 @@
     scatter_,
     scatter_nd,
     scatter_nd_add,
+    select_scatter,
     shard_index,
     slice,
     split,
@@ -693,6 +695,7 @@
     'repeat_interleave',
     'take_along_axis',
     'put_along_axis',
+    'select_scatter',
     'put_along_axis_',
     'exponential_',
     'heaviside',

python/paddle/tensor/manipulation.py

@@ -5921,3 +5921,109 @@ def diagonal_scatter(x, y, offset=0, axis1=0, axis2=1, name=None):
 
     """
     return fill_diagonal_tensor(x, y, offset, axis1, axis2, name)
+
+
+def select_scatter(x, values, axis, index, name=None):
+    """
+    Embeds the values of the values tensor into x at the given index of axis.
+
+    Args:
+        x (Tensor) : The Destination Tensor. Supported data types are `bool`, `float16`, `float32`, `float64`, `uint8`, `int8`, `int16`, `int32`, `int64`, `bfloat16`, `complex64`, `complex64`.
+        values (Tensor) : The tensor to embed into x. Supported data types are `bool`, `float16`, `float32`, `float64`, `uint8`, `int8`, `int16`, `int32`, `int64`, `bfloat16`, `complex64`, `complex64`.
+        axis (int) : the dimension to insert the slice into.
+        index (int) : the index to select with.
+        name (str, optional): Name for the operation (optional, default is None).
+
+    Returns:
+        Tensor, same dtype and shape with x
+
+    Examples:
+        .. code-block:: python
+
+            >>> import paddle
+
+            >>> x = paddle.zeros((2,3,4)).astype("float32")
+            >>> values = paddle.ones((2,4)).astype("float32")
+            >>> res = paddle.select_scatter(x,values,1,1)
+            >>> print(res)
+            Tensor(shape=[2, 3, 4], dtype=float32, place=Place(cpu), stop_gradient=True,
+                   [[[0., 0., 0., 0.],
+                     [1., 1., 1., 1.],
+                     [0., 0., 0., 0.]],
+                    [[0., 0., 0., 0.],
+                     [1., 1., 1., 1.],
+                     [0., 0., 0., 0.]]])
+
+    """
+    x_shape = x.shape
+    value_shape = values.shape
+    if not isinstance(x_shape, list):
+        x_shape = list(x_shape)
+    if index < 0:
+        index += x_shape[axis]
+    if axis < 0:
+        axis += len(x_shape)
+    del x_shape[axis]
+    if len(x_shape) != len(value_shape):
+        raise RuntimeError(
+            "expected values to have a size equal to the slice of x. value size = "
+            + str(value_shape)
+            + " slice size = "
+            + str(x_shape)
+        )
+    for i in range(len(x_shape)):
+        if x_shape[i] != value_shape[i]:
+            raise RuntimeError(
+                "expected values to have a size equal to the slice of x. value size = "
+                + str(value_shape)
+                + " slice size = "
+                + str(x_shape)
+            )
+    from ..base.framework import default_main_program
+
+    starts = [index]
+    ends = [index + 1]
+    steps = [1]
+    axes = [axis]
+    none_axes = []
+    decrease_axes = [axis]
+    inputs = {'Input': x}
+    attrs = {
+        'axes': axes,
+        'starts': starts,
+        'ends': ends,
+        'steps': steps,
+        'decrease_axes': decrease_axes,
+        'none_axes': none_axes,
+    }
+
+    dtype = x.dtype
+    attrs['dtype'] = dtype
+
+    values = values.astype(dtype)
+    inputs["ValueTensor"] = values
+
+    if in_dynamic_or_pir_mode():
+        return _C_ops.set_value_with_tensor(
+            x,
+            values,
+            starts,
+            ends,
+            steps,
+            axes,
+            decrease_axes,
+            none_axes,
+        )
+    else:
+        helper = LayerHelper('select_scatter', **locals())
+        output = helper.create_variable_for_type_inference(dtype=x.dtype)
+        cur_block = default_main_program().current_block()
+        cur_block.append_op(
+            type="set_value",
+            inputs=inputs,
+            outputs={'Out': output},
+            attrs=attrs,
+            inplace_map={"Input": "Out"},
+        )
+
+        return output

test/legacy_test/test_select_scatter_op.py

@@ -0,0 +1,167 @@
+#   Copyright (c) 2021 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 copy
+import unittest
+
+import numpy as np
+
+import paddle
+from paddle.framework import core
+from paddle.pir_utils import test_with_pir_api
+
+paddle.enable_static()
+
+
+class TestSelectScatterAPI(unittest.TestCase):
+    def setUp(self):
+        np.random.seed(0)
+        self.shape = [2, 3, 4]
+        self.type = np.float32
+        self.x_np = np.random.random(self.shape).astype(self.type)
+        self.place = [paddle.CPUPlace()]
+        self.axis = 1
+        self.index = 1
+        self.value_shape = [2, 4]
+        self.value_np = np.random.random(self.value_shape).astype(self.type)
+        self.x_feed = copy.deepcopy(self.x_np)
+        if core.is_compiled_with_cuda():
+            self.place.append(paddle.CUDAPlace(0))
+
+    def get_out_ref(self, out_ref, index, value_np):
+        for i in range(2):
+            for j in range(4):
+                out_ref[i, index, j] = value_np[i, j]
+
+    @test_with_pir_api
+    def test_api_static(self):
+        paddle.enable_static()
+
+        def run(place):
+            with paddle.static.program_guard(paddle.static.Program()):
+                x = paddle.static.data('Src', self.shape, self.type)
+                value = paddle.static.data(
+                    'Values', self.value_shape, self.type
+                )
+                out = paddle.select_scatter(x, value, self.axis, self.index)
+                exe = paddle.static.Executor(place)
+                res = exe.run(
+                    feed={
+                        'Src': self.x_feed,
+                        'Values': self.value_np,
+                    },
+                    fetch_list=[out],
+                )
+
+            out_ref = copy.deepcopy(self.x_np)
+            self.get_out_ref(out_ref, self.index, self.value_np)
+            for out in res:
+                np.testing.assert_allclose(out, out_ref, rtol=0.001)
+
+        for place in self.place:
+            run(place)
+
+    def test_api_dygraph(self):
+        def run(place):
+            paddle.disable_static(place)
+            x_tensor = paddle.to_tensor(self.x_np)
+            value_tensor = paddle.to_tensor(self.value_np)
+            out = paddle.select_scatter(
+                x_tensor, value_tensor, self.axis, self.index
+            )
+            out_ref = copy.deepcopy(self.x_np)
+            self.get_out_ref(out_ref, self.index, self.value_np)
+            np.testing.assert_allclose(out.numpy(), out_ref, rtol=0.001)
+
+            paddle.enable_static()
+
+        for place in self.place:
+            run(place)
+
+
+class TestSelectScatterAPICase2(TestSelectScatterAPI):
+    def setUp(self):
+        np.random.seed(0)
+        self.shape = [2, 3, 4, 5]
+        self.type = np.float64
+        self.x_np = np.random.random(self.shape).astype(self.type)
+        self.place = [paddle.CPUPlace()]
+        self.axis = 2
+        self.index = 1
+        self.value_shape = [2, 3, 5]
+        self.value_np = np.random.random(self.value_shape).astype(self.type)
+        self.x_feed = copy.deepcopy(self.x_np)
+        if core.is_compiled_with_cuda():
+            self.place.append(paddle.CUDAPlace(0))
+
+    def get_out_ref(self, out_ref, index, value_np):
+        for i in range(2):
+            for j in range(3):
+                for k in range(5):
+                    out_ref[i, j, index, k] = value_np[i, j, k]
+
+
+class TestSelectScatterAPICase3(TestSelectScatterAPI):
+    def setUp(self):
+        np.random.seed(0)
+        self.shape = [2, 3, 4, 5, 6]
+        self.type = np.int32
+        self.x_np = np.random.random(self.shape).astype(self.type)
+        self.place = [paddle.CPUPlace()]
+        self.axis = 2
+        self.index = 1
+        self.value_shape = [2, 3, 5, 6]
+        self.value_np = np.random.random(self.value_shape).astype(self.type)
+        self.x_feed = copy.deepcopy(self.x_np)
+        if core.is_compiled_with_cuda():
+            self.place.append(paddle.CUDAPlace(0))
+
+    def get_out_ref(self, out_ref, index, value_np):
+        for i in range(2):
+            for j in range(3):
+                for k in range(5):
+                    for w in range(6):
+                        out_ref[i, j, index, k, w] = value_np[i, j, k, w]
+
+
+class TestSelectScatterAPIError(unittest.TestCase):
+    def setUp(self):
+        np.random.seed(0)
+        self.shape = [2, 3, 4]
+        self.x_np = np.random.random(self.shape).astype(np.float32)
+        self.place = [paddle.CPUPlace()]
+        self.axis = 1
+        self.index = 1
+        self.value_shape = [2, 4]
+        self.value_np = np.random.random(self.value_shape).astype(np.float32)
+        self.x_feed = copy.deepcopy(self.x_np)
+        if core.is_compiled_with_cuda():
+            self.place.append(paddle.CUDAPlace(0))
+
+    def test_len_of_shape_not_equal_error(self):
+        with self.assertRaises(RuntimeError):
+            x_tensor = paddle.to_tensor(self.x_np)
+            value_tensor = paddle.to_tensor(self.value_np).reshape((2, 2, 2))
+            res = paddle.select_scatter(x_tensor, value_tensor, 1, 1)
+
+    def test_one_of_size_not_equal_error(self):
+        with self.assertRaises(RuntimeError):
+            x_tensor = paddle.to_tensor(self.x_np)
+            value_tensor = paddle.to_tensor([[2, 2], [2, 2]]).astype(np.float32)
+            res = paddle.select_scatter(x_tensor, value_tensor, 1, 1)
+
+
+if __name__ == "__main__":
+    paddle.enable_static()
+    unittest.main()