Add fp16 support for dygraph (#19828)

* Add fp16 support for dygraph
test=develop

* Add unit test
test=develop

python/paddle/fluid/dygraph/base.py

@@ -196,6 +196,8 @@ def to_variable(value, block=None, name=None):
             stop_gradient=True)
         var = py_var._ivar.value()
         tensor = var.get_tensor()
+        if value.dtype == np.float16:
+            value = value.view(np.uint16)
         tensor.set(value, framework._current_expected_place())
         return py_var
     elif isinstance(value, framework.Variable):

python/paddle/fluid/tests/unittests/test_dygraph_mnist_fp16.py

@@ -0,0 +1,132 @@
+# 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
+#
+#     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.
+
+from __future__ import print_function
+
+import unittest
+import numpy as np
+
+import paddle.fluid as fluid
+from paddle.fluid.dygraph.nn import Conv2D, Pool2D, FC
+
+
+class SimpleImgConvPool(fluid.dygraph.Layer):
+    def __init__(self,
+                 name_scope,
+                 num_filters,
+                 filter_size,
+                 pool_size,
+                 pool_stride,
+                 pool_padding=0,
+                 pool_type='max',
+                 global_pooling=False,
+                 conv_stride=1,
+                 conv_padding=0,
+                 conv_dilation=1,
+                 conv_groups=1,
+                 act=None,
+                 use_cudnn=False,
+                 dtype='float32',
+                 param_attr=None,
+                 bias_attr=None):
+        super(SimpleImgConvPool, self).__init__(name_scope)
+
+        self._conv2d = Conv2D(
+            self.full_name(),
+            num_filters=num_filters,
+            filter_size=filter_size,
+            stride=conv_stride,
+            padding=conv_padding,
+            dilation=conv_dilation,
+            groups=conv_groups,
+            param_attr=param_attr,
+            bias_attr=bias_attr,
+            use_cudnn=use_cudnn,
+            dtype=dtype,
+            act=act)
+
+        self._pool2d = Pool2D(
+            self.full_name(),
+            pool_size=pool_size,
+            pool_type=pool_type,
+            pool_stride=pool_stride,
+            pool_padding=pool_padding,
+            global_pooling=global_pooling,
+            use_cudnn=use_cudnn)
+
+    def forward(self, inputs):
+        x = self._conv2d(inputs)
+        x = self._pool2d(x)
+        return x
+
+
+class MNIST(fluid.dygraph.Layer):
+    def __init__(self, name_scope, dtype="float32"):
+        super(MNIST, self).__init__(name_scope)
+
+        self._simple_img_conv_pool_1 = SimpleImgConvPool(
+            self.full_name(),
+            num_filters=20,
+            filter_size=5,
+            pool_size=2,
+            pool_stride=2,
+            act="relu",
+            dtype=dtype,
+            use_cudnn=True)
+
+        self._simple_img_conv_pool_2 = SimpleImgConvPool(
+            self.full_name(),
+            num_filters=50,
+            filter_size=5,
+            pool_size=2,
+            pool_stride=2,
+            act="relu",
+            dtype=dtype,
+            use_cudnn=True)
+
+        pool_2_shape = 50 * 4 * 4
+        SIZE = 10
+        scale = (2.0 / (pool_2_shape**2 * SIZE))**0.5
+        self._fc = FC(self.full_name(),
+                      10,
+                      param_attr=fluid.param_attr.ParamAttr(
+                          initializer=fluid.initializer.NormalInitializer(
+                              loc=0.0, scale=scale)),
+                      act="softmax",
+                      dtype=dtype)
+
+    def forward(self, inputs, label):
+        x = self._simple_img_conv_pool_1(inputs)
+        x = self._simple_img_conv_pool_2(x)
+        cost = self._fc(x)
+        loss = fluid.layers.cross_entropy(cost, label)
+        avg_loss = fluid.layers.mean(loss)
+        return avg_loss
+
+
+class TestMnist(unittest.TestCase):
+    def test_mnist_fp16(self):
+        if not fluid.is_compiled_with_cuda():
+            return
+        x = np.random.randn(1, 3, 224, 224).astype("float16")
+        y = np.random.randn(1, 1).astype("int64")
+        with fluid.dygraph.guard(fluid.CUDAPlace(0)):
+            model = MNIST("mnist", dtype="float16")
+            x = fluid.dygraph.to_variable(x)
+            y = fluid.dygraph.to_variable(y)
+            print(model(x, y))
+
+
+if __name__ == "__main__":
+    unittest.main()