diff --git a/python/paddle/hapi/model.py b/python/paddle/hapi/model.py
index 5aa689ca324c099f239a29e2ee21b8283e378341..bba94d56cca8e77735d8921d007248b2e388a5f6 100644
--- a/python/paddle/hapi/model.py
+++ b/python/paddle/hapi/model.py
@@ -731,8 +731,8 @@ class DynamicGraphAdapter(object):
         if not self.model._optimizer or not optim_state:
             return
 
-        # If optimizer performs set_dict when state vars haven't been created,
-        # which would happen when set_dict before minimize, the state would be
+        # If optimizer performs set_state_dict when state vars haven't been created,
+        # which would happen when set_state_dict before minimize, the state would be
         # stored in optimizer._accumulators_holder and loaded lazily.
         # To contrive this when loading from static-graph saved states, extend
         # state dict to include keys named accoring to dygraph naming rules.
@@ -776,7 +776,13 @@ class DynamicGraphAdapter(object):
                                      accum_name + "_0")
                     converted_state[dy_state_name] = state_var
 
-        self.model._optimizer.set_dict(converted_state)
+        if not hasattr(self.model._optimizer, 'set_state_dict'):
+            warnings.warn(
+                "paddle.fluid.optimizer is deprecated in API 2.0, please use paddle.optimizer instead"
+            )
+            self.model._optimizer.set_dict(converted_state)
+        else:
+            self.model._optimizer.set_state_dict(converted_state)
 
 
 class Model(object):
diff --git a/python/paddle/tests/test_model.py b/python/paddle/tests/test_model.py
index 7b79b25cbc3e98b802bad87386ad0572ec6ab8d7..e078595dc9551763f2c4fc1b17f5b4220e3b1f6d 100644
--- a/python/paddle/tests/test_model.py
+++ b/python/paddle/tests/test_model.py
@@ -416,6 +416,29 @@ class TestModelFunction(unittest.TestCase):
             shutil.rmtree(path)
             fluid.disable_dygraph() if dynamic else None
 
+    def test_dynamic_load(self):
+        mnist_data = MnistDataset(mode='train')
+        for new_optimizer in [True, False]:
+            path = tempfile.mkdtemp()
+            paddle.disable_static()
+            net = LeNet()
+            inputs = [InputSpec([None, 1, 28, 28], 'float32', 'x')]
+            labels = [InputSpec([None, 1], 'int64', 'label')]
+            if new_optimizer:
+                optim = paddle.optimizer.Adam(
+                    learning_rate=0.001, parameters=net.parameters())
+            else:
+                optim = fluid.optimizer.Adam(
+                    learning_rate=0.001, parameter_list=net.parameters())
+            model = Model(net, inputs, labels)
+            model.prepare(
+                optimizer=optim, loss=CrossEntropyLoss(reduction="sum"))
+            model.fit(mnist_data, batch_size=64, verbose=0)
+            model.save(path + '/test')
+            model.load(path + '/test')
+            shutil.rmtree(path)
+            paddle.enable_static()
+
     def test_dynamic_save_static_load(self):
         path = tempfile.mkdtemp()
         # dynamic saving