add type_as (#74459)

Author: Difers

python/paddle/base/dygraph/math_op_patch.py

@@ -172,6 +172,9 @@ def conversion_method(self: Tensor) -> Tensor:
 
         return methods
 
+    def type_as(self: Tensor, other: Tensor) -> Tensor:
+        return self.astype(other.dtype)
+
     def _scalar_elementwise_op_(
         var: Tensor, scale: float, bias: float
     ) -> Tensor:
@@ -295,6 +298,7 @@ def _mT_(var: Tensor) -> Tensor:
         ('astype', astype),
         ('byte', byte),
         ('uint8', byte),
+        ('type_as', type_as),
         ('dim', dim),
         ('ndimension', ndimension),
         ('ndim', _ndim),

python/paddle/base/layers/math_op_patch.py

@@ -382,6 +382,9 @@ def astype(self, dtype):
         out.stop_gradient = self.stop_gradient
         return out
 
+    def type_as(self, other):
+        return self.astype(other.dtype)
+
     @static_only
     def append(self, var):
         """
@@ -799,6 +802,7 @@ def to_dense(var):
         ('__neg__', _neg_),
         ('__abs__', _abs_),
         ('astype', astype),
+        ('type_as', type_as),
         ('cpu', cpu),
         ('cuda', cuda),
         ('place', place),

python/paddle/pir/math_op_patch.py

@@ -434,6 +434,9 @@ def conversion_method(self):
             methods.append((method_name, method_impl))
         return methods
 
+    def type_as(self, other):
+        return self.astype(other.dtype)
+
     def _scalar_add_(var, value):
         return paddle.scale(var, 1.0, value)
 
@@ -1175,6 +1178,7 @@ def register_hook(self, hook):
         ('astype', astype),
         ('byte', byte),
         ('uint8', byte),
+        ('type_as', type_as),
         ('size', _size_),
         ('T', _T_),
         ('mT', _mT_),

test/legacy_test/test_type_as.py

@@ -0,0 +1,153 @@
+# 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
+
+
+def api_warpprt(x, y):
+    return x.type_as(y)
+
+
+class TestTypeAsBase(unittest.TestCase):
+    def setUp(self):
+        self.input_dtype_1 = "float32"
+        self.input_dtype_2 = "float16"
+        self.input_shape = (2, 3)
+
+        self.input_np_1 = self.generate_data(
+            self.input_dtype_1, self.input_shape
+        )
+        self.input_np_2 = self.generate_data(
+            self.input_dtype_2, self.input_shape
+        )
+
+        self.input_shape_1 = self.input_np_1.shape
+        self.input_shape_2 = self.input_np_2.shape
+
+        self.op_static = api_warpprt
+        self.op_dygraph = api_warpprt
+        self.places = [None, paddle.CPUPlace()]
+
+    def generate_data(self, dtype, shape):
+        if "int" in dtype:
+            data = np.arange(1, np.prod(shape) + 1).reshape(shape)
+        else:
+            data = np.arange(1, np.prod(shape) + 1, dtype='float32').reshape(
+                shape
+            )
+        return data.astype(dtype)
+
+    def check_static_result(self, place):
+        paddle.enable_static()
+        main_prog = paddle.static.Program()
+        startup_prog = paddle.static.Program()
+        with paddle.static.program_guard(main_prog, startup_prog):
+            input_name_1 = 'input_1'
+            input_name_2 = 'input_2'
+            input_var_1 = paddle.static.data(
+                name=input_name_1,
+                shape=self.input_shape_1,
+                dtype=self.input_dtype_1,
+            )
+            input_var_2 = paddle.static.data(
+                name=input_name_2,
+                shape=self.input_shape_2,
+                dtype=self.input_dtype_2,
+            )
+            res = self.op_static(input_var_1, input_var_2)
+            exe = base.Executor(place)
+            fetches = exe.run(
+                main_prog,
+                feed={
+                    input_name_1: self.input_np_1,
+                    input_name_2: self.input_np_2,
+                },
+                fetch_list=[res],
+            )
+            self.assertEqual(fetches[0].dtype, np.dtype(self.input_dtype_2))
+
+    def test_static(self):
+        for place in self.places:
+            self.check_static_result(place=place)
+
+    def check_dygraph_result(self, place):
+        with base.dygraph.guard(place):
+            input_1 = paddle.to_tensor(self.input_np_1)
+            input_2 = paddle.to_tensor(self.input_np_2)
+            result = self.op_dygraph(input_1, input_2)
+            self.assertEqual(result.dtype, input_2.dtype)
+
+    def test_dygraph(self):
+        for place in self.places:
+            self.check_dygraph_result(place=place)
+
+
+class TestTypeAsFloat32ToFloat16(TestTypeAsBase):
+    def setUp(self):
+        self.input_dtype_1 = "float32"
+        self.input_dtype_2 = "float16"
+        super().setUp()
+
+
+class TestTypeAsFloat64ToFloat32(TestTypeAsBase):
+    def setUp(self):
+        self.input_dtype_1 = "float64"
+        self.input_dtype_2 = "float32"
+        super().setUp()
+
+
+class TestTypeAsInt32ToInt64(TestTypeAsBase):
+    def setUp(self):
+        self.input_dtype_1 = "int32"
+        self.input_dtype_2 = "int64"
+        super().setUp()
+
+
+class TestTypeAsInt32ToFloat32(TestTypeAsBase):
+    def setUp(self):
+        self.input_dtype_1 = "int32"
+        self.input_dtype_2 = "float32"
+        super().setUp()
+
+
+class TestTypeAsFloat32ToInt64(TestTypeAsBase):
+    def setUp(self):
+        self.input_dtype_1 = "float32"
+        self.input_dtype_2 = "int64"
+        super().setUp()
+
+
+class TestTypeAsInt8ToFloat64(TestTypeAsBase):
+    def setUp(self):
+        self.input_dtype_1 = "int8"
+        self.input_dtype_2 = "float64"
+        self.input_shape = (4, 2)
+        super().setUp()
+
+
+class TestTypeAsUInt8ToInt32(TestTypeAsBase):
+    def setUp(self):
+        self.input_dtype_1 = "uint8"
+        self.input_dtype_2 = "int32"
+        self.input_shape = (3, 3)
+        super().setUp()
+
+
+if __name__ == "__main__":
+    unittest.main()