[Dy2stat]Allow users to switch eval/train mode when using @to_static to decorate a function (#37383) (#37432)

本PR之前使用@to_static装饰一个单独的function时，对于生成的Program无法切换train/eval模式，只能运行在train模式下。这也就导致动转静后用户多次调用function显存会一直增长。
本PR之后，使用@to_static装饰一个单独的function时，可以通过function.train()或者function.eval()的方式来切换train/eval模式。

Author: 0x45f

python/paddle/fluid/dygraph/dygraph_to_static/program_translator.py

@@ -261,6 +261,25 @@ def __init__(self, function, input_spec=None, **kwargs):
         # Note: Hold a reference to ProgramTranslator for switching `enable_to_static`.
         self._program_trans = ProgramTranslator()
         self._kwargs = kwargs
+        self._training = True
+
+    def train(self):
+        if isinstance(self._class_instance,
+                      layers.Layer) and self._class_instance.training == False:
+            raise RuntimeError(
+                "Failed to switch train mode. {} is a Layer's method, "
+                "please use Layer.train() to switch train mode.".format(
+                    self.dygraph_function))
+        self._training = True
+
+    def eval(self):
+        if isinstance(self._class_instance,
+                      layers.Layer) and self._class_instance.training == True:
+            raise RuntimeError(
+                "Failed to switch eval mode. {} is a Layer's method, "
+                "please use Layer.eval() to switch eval mode.".format(
+                    self.dygraph_function))
+        self._training = False
 
     def __get__(self, instance, owner):
         """
@@ -340,6 +359,8 @@ def __call__(self, *args, **kwargs):
             # 3. synchronize self.training attribute.
             if isinstance(self._class_instance, layers.Layer):
                 partial_program_layer.training = self._class_instance.training
+            else:
+                partial_program_layer.training = self._training
 
             # 4. return outputs.
             try:

python/paddle/fluid/tests/unittests/dygraph_to_static/test_program_translator.py

@@ -297,5 +297,52 @@ def test_raise_error(self):
             self.program_translator.get_program(net.forward, self.x)
 
 
+class SwitchModeNet(paddle.nn.Layer):
+    def __init__(self):
+        super(SwitchModeNet, self).__init__()
+
+    @paddle.jit.to_static
+    def forward(self, x):
+        return x + 1
+
+    @paddle.jit.to_static
+    def foo(self):
+        return True
+
+
+@paddle.jit.to_static
+def switch_mode_funciton():
+    return True
+
+
+class TestFunctionTrainEvalMode(unittest.TestCase):
+    def test_switch_mode(self):
+        paddle.disable_static()
+        switch_mode_funciton.eval()
+        switch_mode_funciton()
+        self.assertEqual(switch_mode_funciton._training, False)
+        _, partial_layer = switch_mode_funciton.program_cache.last()[-1]
+        self.assertEqual(partial_layer.training, False)
+
+        switch_mode_funciton.train()
+        switch_mode_funciton()
+        self.assertEqual(switch_mode_funciton._training, True)
+        _, partial_layer = switch_mode_funciton.program_cache.last()[-1]
+        self.assertEqual(partial_layer.training, True)
+
+    def test_raise_error(self):
+        paddle.disable_static()
+        net = SwitchModeNet()
+
+        self.assertEqual(net.training, True)
+        with self.assertRaises(RuntimeError):
+            net.forward.eval()
+
+        net.eval()
+        self.assertEqual(net.training, False)
+        with self.assertRaises(RuntimeError):
+            net.foo.train()
+
+
 if __name__ == '__main__':
     unittest.main()