【Hackathon 5th No.15】 为 Paddle 新增 Tensor.to() 以及 Layer.astype() API -part (#58244)

YibinLiu666 · sunzhongkai588 · web-flow · commit 2927ed9aba12 · 2023-11-09T11:28:39.000+08:00
* support tensor.to and layer.astype

* add UT and comments

* update

* update com

* update dtype

* fix example test

* update example

* add some ut to fix ci-coverage

* fix codestyle

* fix codestyle

* update

* add ut to test layer params' and buffers' type

* update test

* fix doc

* Update python/paddle/base/dygraph/tensor_patch_methods.py

Co-authored-by: zachary sun &lt;70642955+sunzhongkai588@users.noreply.github.com&gt;

* update doc

---------

Co-authored-by: zachary sun &lt;70642955+sunzhongkai588@users.noreply.github.com&gt;
diff --git a/python/paddle/base/dygraph/tensor_patch_methods.py b/python/paddle/base/dygraph/tensor_patch_methods.py
@@ -512,6 +512,129 @@ def transform(t, device, dtype, blocking):
             warnings.filterwarnings("ignore", category=UserWarning)
             return transform(self, device, dtype, blocking)
 
+    @framework.dygraph_only
+    def to(self, *args, **kwargs):
+        """
+        Performs Tensor dtype and/or device conversion. A paddle.dtype and place
+        are inferred from the arguments of ``self.to(*args, **kwargs)``.There are
+        three ways to call `to`:
+
+            1. to(dtype, blocking=True)
+            2. to(device, dtype=None, blocking=True)
+            3. to(other, blocking=True)
+
+        Returns:
+            Tensor: self
+
+        Examples:
+            .. code-block:: python
+
+                >>> import paddle
+                >>> tensorx = paddle.to_tensor([1,2,3])
+                >>> print(tensorx)
+                Tensor(shape=[3], dtype=int64, place=Place(gpu:0), stop_gradient=True,
+                    [1, 2, 3])
+
+                >>> tensorx = tensorx.to("cpu")
+                >>> print(tensorx.place)
+                Place(cpu)
+
+                >>> tensorx = tensorx.to("float32")
+                >>> print(tensorx.dtype)
+                paddle.float32
+
+                >>> tensorx = tensorx.to("gpu", "int16")
+                >>> print(tensorx)
+                Tensor(shape=[3], dtype=int16, place=Place(gpu:0), stop_gradient=True,
+                    [1, 2, 3])
+                >>> tensor2 = paddle.to_tensor([4,5,6])
+                >>> tensor2
+                Tensor(shape=[3], dtype=int64, place=Place(gpu:0), stop_gradient=True,
+                    [4, 5, 6])
+                >>> tensor2 = tensor2.to(tensorx)
+                >>> print(tensor2)
+                Tensor(shape=[3], dtype=int16, place=Place(gpu:0), stop_gradient=True,
+                    [4, 5, 6])
+        """
+        device = None
+        dtype = None
+        blocking = None
+        size_args = len(args)
+        size_kwargs = len(kwargs)
+
+        def get_device_dtype_from_tensor(other):
+            if other is not None:
+                device = str(other.place)[6:-1]
+                dtype = other.dtype
+                return device, dtype
+            else:
+                return None, None
+
+        if size_args + size_kwargs > 3 or size_args + size_kwargs == 0:
+            raise TypeError(
+                "to() received too mant arguments - expected one of:\n  \
+                * (Union[str, paddle.CPUPlace(), paddle.CUDAPlace(), paddle.CUDAPinnedPlace(), paddle.XPUPlace(), paddle.CustomPlace()] \
+                device, Union[str, paddle.dtype, numpy.dtype] dtype, bool blocking)\n \
+                * (Union[str, paddle.dtype, numpy.dtype] dtype, bool blocking)\n \
+                * (paddle.Tensor other, bool blocking) "
+            )
+        valid_keys = {"device", "dtype", "blocking", "other"}
+        valid_dtypes = [
+            "bfloat16",
+            "float16",
+            "float32",
+            "float64",
+            "int8",
+            "int16",
+            "int32",
+            "int64",
+            "uint8",
+            "complex64",
+            "complex128",
+            "bool",
+        ]
+        invalid_keys = set(kwargs.keys()) - valid_keys
+        if len(invalid_keys) != 0:
+            raise TypeError(
+                "to() got an unexpected keyword argument "
+                + list(invalid_keys)[0]
+            )
+        if size_args > 0:
+            if isinstance(args[0], paddle.Tensor):
+                device, dtype = get_device_dtype_from_tensor(args[0])
+                if size_args == 2:
+                    blocking = args[1]
+                else:
+                    blocking = kwargs.get("blocking", None)
+            elif (
+                isinstance(args[0], (paddle.dtype, np.dtype))
+                or isinstance(args[0], str)
+                and args[0].lower() in valid_dtypes
+            ):
+                dtype = args[0]
+                if size_args == 2:
+                    blocking = args[1]
+                else:
+                    blocking = kwargs.get("blocking", None)
+            else:
+                device = args[0]
+                if size_args == 2:
+                    dtype = args[1]
+                elif size_args == 3:
+                    dtype, blocking = args[1], args[2]
+                else:
+                    dtype = kwargs.get("dtype", None)
+                    blocking = kwargs.get("blocking", None)
+        else:
+            device = kwargs.get("device", None)
+            dtype = kwargs.get("dtype", None)
+            blocking = kwargs.get("blocking", None)
+            if device is None and dtype is None:
+                device, dtype = get_device_dtype_from_tensor(
+                    kwargs.get("other", None)
+                )
+        return self._to(device, dtype, blocking)
+
     @property
     def grad(self):
         """
@@ -1020,6 +1143,7 @@ def coalesce(self, name=None):
         ("item", item),
         ("__setitem__", __setitem__),
         ("_to", _to),
+        ("to", to),
         ("values", values),
         ("to_dense", to_dense),
         ("to_sparse_coo", to_sparse_coo),
diff --git a/python/paddle/nn/layer/layers.py b/python/paddle/nn/layer/layers.py
@@ -919,6 +919,86 @@ def parameters(self, include_sublayers=True):
         ]
         return ret
 
+    def astype(self, dtype=None):
+        """
+
+        Casts all parameters and buffers to dtype and then return the Layer.
+
+        Parameters:
+            dtype(str|paddle.dtype|numpy.dtype): target data type of layer.
+                If set str, it can be "bool", "bfloat16", "float16", "float32", "float64",
+                "int8", "int16", "int32", "int64", "uint8", "complex64", "complex128".
+                Default: None
+
+        Returns:
+            Layer, self
+
+        Examples:
+            .. code-block:: python
+
+                >>> import paddle
+                >>> import paddle.nn as nn
+                >>> weight_attr = paddle.ParamAttr(name="weight",initializer=paddle.nn.initializer.Constant(value=1.5))
+                >>> bias_attr = paddle.ParamAttr(name="bias",initializer=paddle.nn.initializer.Constant(value=2.5))
+
+                >>> linear = paddle.nn.Linear(2, 2, weight_attr=weight_attr, bias_attr=bias_attr).to(device="cpu",dtype="float32")
+                >>> print(linear)
+                Linear(in_features=2, out_features=2, dtype=float32)
+                >>> print(linear.parameters())
+                [Parameter containing:
+                Tensor(shape=[2, 2], dtype=float32, place=Place(cpu), stop_gradient=False,
+                    [[1.50000000, 1.50000000],
+                        [1.50000000, 1.50000000]]), Parameter containing:
+                Tensor(shape=[2], dtype=float32, place=Place(cpu), stop_gradient=False,
+                    [2.50000000, 2.50000000])]
+
+                >>> linear=linear.astype("int8")
+                >>> print(linear)
+                Linear(in_features=2, out_features=2, dtype=paddle.int8)
+                >>> print(linear.parameters())
+                [Parameter containing:
+                Tensor(shape=[2, 2], dtype=int8, place=Place(cpu), stop_gradient=False,
+                    [[1, 1],
+                        [1, 1]]), Parameter containing:
+                Tensor(shape=[2], dtype=int8, place=Place(cpu), stop_gradient=False,
+                    [2, 2])]
+
+        """
+        valid_dtypes = [
+            "bfloat16",
+            "float16",
+            "float32",
+            "float64",
+            "int8",
+            "int16",
+            "int32",
+            "int64",
+            "uint8",
+            "complex64",
+            "complex128",
+            "bool",
+        ]
+        if (
+            isinstance(dtype, (paddle.dtype, np.dtype))
+            or type(dtype) is str
+            and dtype in valid_dtypes
+        ):
+            if isinstance(dtype, (str, np.dtype)):
+                dtype = framework.convert_np_dtype_to_dtype_(dtype)
+            self._dtype = dtype
+            for layer in self.sublayers():
+                layer._dtype = dtype
+            for _, param in self.named_parameters(include_sublayers=True):
+                param._to(None, dtype)
+            for _, buffer in self.named_buffers(include_sublayers=True):
+                buffer.to(None, dtype)
+            return self
+        else:
+            raise ValueError(
+                "dtype value error, must be 'bfloat16', 'float16', 'float32', 'float64', 'int8', 'int16', 'int32', 'int64', 'uint8', 'complex64', 'complex128', 'bool', or paddle.dtype, numpy.dtype, but recieve "
+                + str(dtype)
+            )
+
     def children(self):
         """
 
diff --git a/test/legacy_test/test_Tensor_to.py b/test/legacy_test/test_Tensor_to.py
@@ -0,0 +1,147 @@
+# Copyright (c) 2019 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
+#
+#     https://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 paddle
+from paddle import base
+
+
+class TensorToTest(unittest.TestCase):
+    def test_Tensor_to_dtype(self):
+        tensorx = paddle.to_tensor([1, 2, 3])
+        valid_dtypes = [
+            "bfloat16",
+            "float16",
+            "float32",
+            "float64",
+            "int8",
+            "int16",
+            "int32",
+            "int64",
+            "uint8",
+            "complex64",
+            "complex128",
+            "bool",
+        ]
+        for dtype in valid_dtypes:
+            tensorx = tensorx.to(dtype)
+            typex_str = str(tensorx.dtype)
+            self.assertTrue(typex_str, "paddle." + dtype)
+
+    def test_Tensor_to_device(self):
+        tensorx = paddle.to_tensor([1, 2, 3])
+        places = ["cpu"]
+        if base.core.is_compiled_with_cuda():
+            places.append("gpu:0")
+            places.append("gpu")
+
+        for place in places:
+            tensorx = tensorx.to(place)
+            placex_str = str(tensorx.place)
+            if place == "gpu":
+                self.assertTrue(placex_str, "Place(" + place + ":0)")
+            else:
+                self.assertTrue(placex_str, "Place(" + place + ")")
+
+    def test_Tensor_to_device_dtype(self):
+        tensorx = paddle.to_tensor([1, 2, 3])
+        places = ["cpu"]
+        if base.core.is_compiled_with_cuda():
+            places.append("gpu:0")
+            places.append("gpu")
+        valid_dtypes = [
+            "bfloat16",
+            "float16",
+            "float32",
+            "float64",
+            "int8",
+            "int16",
+            "int32",
+            "int64",
+            "uint8",
+            "complex64",
+            "complex128",
+            "bool",
+        ]
+        for dtype in valid_dtypes:
+            for place in places:
+                tensorx = tensorx.to(place, dtype)
+                placex_str = str(tensorx.place)
+                if place == "gpu":
+                    self.assertTrue(placex_str, "Place(" + place + ":0)")
+                else:
+                    self.assertTrue(placex_str, "Place(" + place + ")")
+                typex_str = str(tensorx.dtype)
+                self.assertTrue(typex_str, "paddle." + dtype)
+
+    def test_Tensor_to_blocking(self):
+        tensorx = paddle.to_tensor([1, 2, 3])
+        tensorx = tensorx.to("cpu", "int32", False)
+        placex_str = str(tensorx.place)
+        self.assertTrue(placex_str, "Place(cpu)")
+        typex_str = str(tensorx.dtype)
+        self.assertTrue(typex_str, "paddle.int32")
+        tensor2 = paddle.to_tensor([4, 5, 6])
+        tensor2 = tensor2.to(tensorx, False)
+        place2_str = str(tensor2.place)
+        self.assertTrue(place2_str, "Place(cpu)")
+        type2_str = str(tensor2.dtype)
+        self.assertTrue(type2_str, "paddle.int32")
+        tensor2 = tensor2.to("float16", False)
+        type2_str = str(tensor2.dtype)
+        self.assertTrue(type2_str, "paddle.float16")
+
+    def test_Tensor_to_other(self):
+        tensor1 = paddle.to_tensor([1, 2, 3], dtype="int8", place="cpu")
+        tensor2 = paddle.to_tensor([1, 2, 3])
+        tensor2 = tensor2.to(tensor1)
+        self.assertTrue(tensor2.dtype, tensor1.dtype)
+        self.assertTrue(type(tensor2.place), type(tensor1.place))
+
+    def test_kwargs(self):
+        tensorx = paddle.to_tensor([1, 2, 3])
+        tensorx = tensorx.to(device="cpu", dtype="int8", blocking=True)
+        placex_str = str(tensorx.place)
+        self.assertTrue(placex_str, "Place(cpu)")
+        typex_str = str(tensorx.dtype)
+        self.assertTrue(typex_str, "paddle.int8")
+        tensor2 = paddle.to_tensor([4, 5, 6])
+        tensor2 = tensor2.to(other=tensorx)
+        place2_str = str(tensor2.place)
+        self.assertTrue(place2_str, "Place(cpu)")
+        type2_str = str(tensor2.dtype)
+        self.assertTrue(type2_str, "paddle.int8")
+
+    def test_error(self):
+        tensorx = paddle.to_tensor([1, 2, 3])
+        # device value error
+        try:
+            tensorx = tensorx.to("error_device")
+        except Exception as error:
+            self.assertIsInstance(error, ValueError)
+        # to many augments
+        try:
+            tensorx = tensorx.to("cpu", "int32", False, "test_aug")
+        except Exception as error:
+            self.assertIsInstance(error, TypeError)
+        # invalid key
+        try:
+            tensorx = tensorx.to("cpu", "int32", test_key=False)
+        except Exception as error:
+            self.assertIsInstance(error, TypeError)
+
+
+if __name__ == '__main__':
+    unittest.main()
diff --git a/test/legacy_test/test_layer_astype.py b/test/legacy_test/test_layer_astype.py
