replace fluid.mean to paddle.mean (#43907)

* change fluid.mean to paddle.mean

* reverse some old code examples

python/paddle/fluid/contrib/slim/tests/imperative_test_utils.py

@@ -68,7 +68,7 @@ def train_lenet(lenet, reader, optimizer):
 
         out = lenet(img)
         loss = fluid.layers.cross_entropy(out, label)
-        avg_loss = fluid.layers.mean(loss)
+        avg_loss = paddle.mean(loss)
         avg_loss.backward()
 
         optimizer.minimize(avg_loss)

python/paddle/fluid/contrib/slim/tests/test_graph.py

@@ -46,7 +46,7 @@ def conv_block():
                                                   act="relu")
     prediction = fluid.layers.fc(input=conv_pool_2, size=10, act='softmax')
     loss = fluid.layers.cross_entropy(input=prediction, label=label)
-    avg_loss = fluid.layers.mean(loss)
+    avg_loss = paddle.mean(loss)
     return [img, label], avg_loss
 
 

python/paddle/fluid/contrib/slim/tests/test_imperative_qat.py

@@ -118,7 +118,7 @@ class TestImperativeQat(unittest.TestCase):
                     out = lenet(img)
                     acc = fluid.layers.accuracy(out, label)
                     loss = fluid.layers.cross_entropy(out, label)
-                    avg_loss = fluid.layers.mean(loss)
+                    avg_loss = paddle.mean(loss)
                     avg_loss.backward()
                     adam.minimize(avg_loss)
                     lenet.clear_gradients()

python/paddle/fluid/contrib/slim/tests/test_imperative_qat_amp.py

@@ -115,7 +115,7 @@ class TestImperativeQatAmp(unittest.TestCase):
                     out = model(img)
                     acc = fluid.layers.accuracy(out, label)
                     loss = fluid.layers.cross_entropy(out, label)
-                    avg_loss = fluid.layers.mean(loss)
+                    avg_loss = paddle.mean(loss)
                 scaled_loss = scaler.scale(avg_loss)
                 scaled_loss.backward()
 
@@ -125,7 +125,7 @@ class TestImperativeQatAmp(unittest.TestCase):
                 out = model(img)
                 acc = fluid.layers.accuracy(out, label)
                 loss = fluid.layers.cross_entropy(out, label)
-                avg_loss = fluid.layers.mean(loss)
+                avg_loss = paddle.mean(loss)
                 avg_loss.backward()
 
                 adam.minimize(avg_loss)

python/paddle/fluid/contrib/slim/tests/test_quantization_mkldnn_pass.py

@@ -45,7 +45,7 @@ def conv_net(img, label):
                                                   act="relu")
     prediction = fluid.layers.fc(input=conv_pool_2, size=10, act='softmax')
     loss = fluid.layers.cross_entropy(input=prediction, label=label)
-    avg_loss = fluid.layers.mean(loss)
+    avg_loss = paddle.mean(loss)
     return avg_loss
 
 

python/paddle/fluid/contrib/slim/tests/test_quantization_pass.py

@@ -41,7 +41,7 @@ def linear_fc(num):
     for _ in six.moves.xrange(num):
         hidden = fluid.layers.fc(hidden, size=128, act='relu')
     loss = fluid.layers.cross_entropy(input=hidden, label=label)
-    loss = fluid.layers.mean(loss)
+    loss = paddle.mean(loss)
     return loss
 
 
@@ -92,7 +92,7 @@ def residual_block(num, quant_skip_pattern=None):
                                    pool_stride=2)
     fc = fluid.layers.fc(input=pool, size=10)
     loss = fluid.layers.cross_entropy(input=fc, label=label)
-    loss = fluid.layers.mean(loss)
+    loss = paddle.mean(loss)
     return loss
 
 
@@ -116,7 +116,7 @@ def conv_net(img, label, quant_skip_pattern):
     with fluid.name_scope(quant_skip_pattern):
         prediction = fluid.layers.fc(input=hidden, size=10, act='softmax')
     loss = fluid.layers.cross_entropy(input=prediction, label=label)
-    avg_loss = fluid.layers.mean(loss)
+    avg_loss = paddle.mean(loss)
     return avg_loss
 
 
@@ -620,7 +620,7 @@ def quant_dequant_residual_block(num, quant_skip_pattern=None):
         pool_add = fluid.layers.elementwise_add(x=pool1, y=pool2, act='relu')
     fc = fluid.layers.fc(input=pool_add, size=10)
     loss = fluid.layers.cross_entropy(input=fc, label=label)
-    loss = fluid.layers.mean(loss)
+    loss = paddle.mean(loss)
     return loss
 
 

python/paddle/fluid/contrib/slim/tests/test_quantization_scale_pass.py

@@ -53,7 +53,7 @@ def conv_net(img, label):
     hidden = fluid.layers.fc(input=conv_pool_2, size=100, act='relu')
     prediction = fluid.layers.fc(input=hidden, size=10, act='softmax')
     loss = fluid.layers.cross_entropy(input=prediction, label=label)
-    avg_loss = fluid.layers.mean(loss)
+    avg_loss = paddle.mean(loss)
     return avg_loss
 
 

python/paddle/fluid/contrib/slim/tests/test_quantize_transpiler_v2.py

@@ -48,7 +48,7 @@ def conv_net(img, label):
         hidden = fluid.layers.fc(input=conv_pool_1, size=100, act='relu')
     prediction = fluid.layers.fc(input=hidden, size=10, act='softmax')
     loss = fluid.layers.cross_entropy(input=prediction, label=label)
-    avg_loss = fluid.layers.mean(loss)
+    avg_loss = paddle.mean(loss)
     return avg_loss
 
 

python/paddle/fluid/contrib/slim/tests/test_user_defined_quantization.py

@@ -55,7 +55,7 @@ def conv_net(img, label):
     hidden = fluid.layers.fc(input=conv_pool_2, size=100, act='relu')
     prediction = fluid.layers.fc(input=hidden, size=10, act='softmax')
     loss = fluid.layers.cross_entropy(input=prediction, label=label)
-    avg_loss = fluid.layers.mean(loss)
+    avg_loss = paddle.mean(loss)
     return avg_loss
 
 

python/paddle/fluid/contrib/tests/test_image_classification_fp16.py

@@ -136,7 +136,7 @@ def train(net_type, use_cuda, save_dirname, is_local):
         logits = fluid.layers.fc(input=net, size=classdim, act="softmax")
         cost, predict = fluid.layers.softmax_with_cross_entropy(
             logits, label, return_softmax=True)
-        avg_cost = fluid.layers.mean(cost)
+        avg_cost = paddle.mean(cost)
         acc = fluid.layers.accuracy(input=predict, label=label)
 
         # Test program
@@ -460,7 +460,7 @@ class TestAmpWithNonIterableDataLoader(unittest.TestCase):
                 logits = fluid.layers.fc(input=net, size=10, act="softmax")
                 cost, predict = fluid.layers.softmax_with_cross_entropy(
                     logits, label, return_softmax=True)
-                avg_cost = fluid.layers.mean(cost)
+                avg_cost = paddle.mean(cost)
 
                 optimizer = fluid.optimizer.Lamb(learning_rate=0.001)
                 amp_lists = fluid.contrib.mixed_precision.AutoMixedPrecisionLists(

python/paddle/fluid/contrib/tests/test_quantize_transpiler.py

@@ -32,7 +32,7 @@ def linear_fc(num):
     for _ in six.moves.xrange(num):
         hidden = fluid.layers.fc(hidden, size=128, act='relu')
     loss = fluid.layers.cross_entropy(input=hidden, label=label)
-    loss = fluid.layers.mean(loss)
+    loss = paddle.mean(loss)
     return loss
 
 
@@ -63,7 +63,7 @@ def residual_block(num):
         hidden = fluid.layers.elementwise_add(x=conv, y=short, act='relu')
     fc = fluid.layers.fc(input=hidden, size=10)
     loss = fluid.layers.cross_entropy(input=fc, label=label)
-    loss = fluid.layers.mean(loss)
+    loss = paddle.mean(loss)
     return loss
 
 
@@ -83,7 +83,7 @@ def conv_net(img, label):
                                                   act="relu")
     prediction = fluid.layers.fc(input=conv_pool_2, size=10, act='softmax')
     loss = fluid.layers.cross_entropy(input=prediction, label=label)
-    avg_loss = fluid.layers.mean(loss)
+    avg_loss = paddle.mean(loss)
     return avg_loss
 
 

python/paddle/fluid/contrib/tests/test_weight_decay_extend.py

@@ -87,7 +87,7 @@ def bow_net(data,
     fc_2 = fluid.layers.fc(input=fc_1, size=hid_dim2, act="tanh")
     prediction = fluid.layers.fc(input=[fc_2], size=class_dim, act="softmax")
     cost = fluid.layers.cross_entropy(input=prediction, label=label)
-    avg_cost = fluid.layers.mean(x=cost)
+    avg_cost = paddle.mean(x=cost)
 
     return avg_cost
 

python/paddle/fluid/dygraph/jit.py

@@ -103,7 +103,7 @@ def _dygraph_to_static_func_(dygraph_func):
 
           @dygraph_to_static_func
           def func(x):
-              if fluid.layers.mean(x) < 0:
+              if paddle.mean(x) < 0:
                   x_v = x - 1
               else:
                   x_v = x + 1

python/paddle/fluid/dygraph/learning_rate_scheduler.py

@@ -897,7 +897,7 @@ class ReduceLROnPlateau(LearningRateDecay):
         check_type(loss, 'loss', Variable, 'ReduceLROnPlateau.step')
         assert len(loss.shape) == 1 and loss.shape[0] == 1, "the loss.shape " \
             "should be (1L,), but the current loss.shape is {}. Maybe that "  \
-            "you should call fluid.layers.mean to process it first.".format(loss.shape)
+            "you should call paddle.mean to process it first.".format(loss.shape)
 
         self.epoch_num += 1
         if self.cooldown_counter > 0:

python/paddle/fluid/incubate/fleet/tests/fleet_deep_ctr.py

@@ -16,6 +16,7 @@ import argparse
 import logging
 import time
 
+import paddle
 import paddle.fluid as fluid
 import paddle.fluid.incubate.fleet.base.role_maker as role_maker
 from paddle.fluid.incubate.fleet.parameter_server.distribute_transpiler import fleet
@@ -123,7 +124,7 @@ def model():
     auc_var, batch_auc_var, auc_states = fluid.layers.auc(input=predict,
                                                           label=label)
     cost = fluid.layers.cross_entropy(input=predict, label=label)
-    avg_cost = fluid.layers.mean(x=cost)
+    avg_cost = paddle.mean(x=cost)
 
     return datas, avg_cost, predict, train_file_path
 

python/paddle/fluid/install_check.py

@@ -102,7 +102,7 @@ def run_check():
                     exe = executor.Executor(
                         core.CUDAPlace(0) if core.is_compiled_with_cuda() and
                         (core.get_cuda_device_count() > 0) else core.CPUPlace())
-                    loss = layers.mean(out)
+                    loss = paddle.mean(out)
                     loss.persistable = True
                     optimizer.SGD(learning_rate=0.01).minimize(loss)
                     startup_prog.random_seed = 1

python/paddle/fluid/io.py

@@ -493,7 +493,7 @@ def save_params(executor, dirname, main_program=None, filename=None):
             predict = fluid.layers.fc(input=image, size=10, act='softmax')
 
             loss = fluid.layers.cross_entropy(input=predict, label=label)
-            avg_loss = fluid.layers.mean(loss)
+            avg_loss = paddle.mean(loss)
 
             exe = fluid.Executor(fluid.CPUPlace())
             exe.run(fluid.default_startup_program())
@@ -719,7 +719,7 @@ def save_persistables(executor, dirname, main_program=None, filename=None):
 
             predict = fluid.layers.fc(input=image, size=10, act='softmax')
             loss = fluid.layers.cross_entropy(input=predict, label=label)
-            avg_loss = fluid.layers.mean(loss)
+            avg_loss = paddle.mean(loss)
             exe = fluid.Executor(fluid.CPUPlace())
             exe.run(fluid.default_startup_program())
             fluid.io.save_persistables(executor=exe, dirname=dir_path, filename=file_name)
@@ -1315,7 +1315,7 @@ def save_inference_model(dirname,
             predict = fluid.layers.fc(input=image, size=10, act='softmax')
 
             loss = fluid.layers.cross_entropy(input=predict, label=label)
-            avg_loss = fluid.layers.mean(loss)
+            avg_loss = paddle.mean(loss)
 
             exe = fluid.Executor(fluid.CPUPlace())
             exe.run(fluid.default_startup_program())

python/paddle/fluid/layers/nn.py

@@ -13089,7 +13089,7 @@ def mean(x, name=None):
 
             input = fluid.layers.data(
                 name='data', shape=[2, 3], dtype='float32')
-            mean = fluid.layers.mean(input)
+            mean = paddle.mean(input)
     """
 
     if _in_legacy_dygraph():

python/paddle/fluid/optimizer.py

@@ -913,7 +913,7 @@ class Optimizer(object):
             program = loss.block.program
             assert len(loss.shape) == 1 and loss.shape[0] == 1, \
                 "The loss.shape should be (1L,), but the current loss.shape is {}. " \
-                "Maybe that you should call fluid.layers.mean to process the current loss.".format(
+                "Maybe that you should call paddle.mean to process the current loss.".format(
                     loss.shape)
             parameter_list = parameter_list if parameter_list \
                 else self._parameter_list
@@ -6834,7 +6834,7 @@ class LookaheadOptimizer(object):
             label = fluid.layers.data(name="label", shape=[1], dtype="int64")
             y = fluid.layers.fc(input=[x], size=2, act="softmax")
             loss = fluid.layers.cross_entropy(input=y, label=label)
-            loss = fluid.layers.mean(x=loss)
+            loss = paddle.mean(x=loss)
             sgd = fluid.optimizer.SGD(learning_rate=0.01)
             optimizer = fluid.optimizer.LookaheadOptimizer(sgd,
                                                 alpha=0.5,

python/paddle/fluid/tests/book/notest_understand_sentiment.py

@@ -48,7 +48,7 @@ def convolution_net(data,
                                  size=class_dim,
                                  act="softmax")
     cost = fluid.layers.cross_entropy(input=prediction, label=label)
-    avg_cost = fluid.layers.mean(cost)
+    avg_cost = paddle.mean(cost)
     accuracy = fluid.layers.accuracy(input=prediction, label=label)
     return avg_cost, accuracy, prediction
 
@@ -93,7 +93,7 @@ def dyn_rnn_lstm(data,
     last = fluid.layers.sequence_last_step(rnn())
     prediction = fluid.layers.fc(input=last, size=class_dim, act="softmax")
     cost = fluid.layers.cross_entropy(input=prediction, label=label)
-    avg_cost = fluid.layers.mean(cost)
+    avg_cost = paddle.mean(cost)
     accuracy = fluid.layers.accuracy(input=prediction, label=label)
     return avg_cost, accuracy, prediction
 
@@ -132,7 +132,7 @@ def stacked_lstm_net(data,
                                  size=class_dim,
                                  act='softmax')
     cost = fluid.layers.cross_entropy(input=prediction, label=label)
-    avg_cost = fluid.layers.mean(cost)
+    avg_cost = paddle.mean(cost)
     accuracy = fluid.layers.accuracy(input=prediction, label=label)
     return avg_cost, accuracy, prediction
 

python/paddle/fluid/tests/book/test_fit_a_line.py

@@ -56,16 +56,16 @@ def train(use_cuda, save_dirname, is_local, use_bf16, pure_bf16):
             with amp.bf16.bf16_guard():
                 y_predict = fluid.layers.fc(input=x, size=1, act=None)
             cost = fluid.layers.square_error_cost(input=y_predict, label=y)
-            avg_cost = fluid.layers.mean(cost)
+            avg_cost = paddle.mean(cost)
         else:
             y_predict = fluid.layers.fc(input=x, size=1, act=None)
             with amp.bf16.bf16_guard():
                 cost = fluid.layers.square_error_cost(input=y_predict, label=y)
-                avg_cost = fluid.layers.mean(cost)
+                avg_cost = paddle.mean(cost)
     else:
         y_predict = fluid.layers.fc(input=x, size=1, act=None)
         cost = fluid.layers.square_error_cost(input=y_predict, label=y)
-        avg_cost = fluid.layers.mean(cost)
+        avg_cost = paddle.mean(cost)
 
     lr = 5e-3 if use_bf16 else 1e-3
     sgd_optimizer = fluid.optimizer.SGD(learning_rate=lr)

python/paddle/fluid/tests/book/test_image_classification.py

@@ -126,7 +126,7 @@ def train(net_type, use_cuda, save_dirname, is_local):
 
     predict = fluid.layers.fc(input=net, size=classdim, act='softmax')
     cost = fluid.layers.cross_entropy(input=predict, label=label)
-    avg_cost = fluid.layers.mean(cost)
+    avg_cost = paddle.mean(cost)
     acc = fluid.layers.accuracy(input=predict, label=label)
 
     # Test program

python/paddle/fluid/tests/book/test_label_semantic_roles.py

@@ -160,7 +160,7 @@ def train(use_cuda, save_dirname=None, is_local=True):
                                              param_attr=fluid.ParamAttr(
                                                  name='crfw',
                                                  learning_rate=mix_hidden_lr))
-    avg_cost = fluid.layers.mean(crf_cost)
+    avg_cost = paddle.mean(crf_cost)
 
     # TODO(qiao)
     # check other optimizers and check why out will be NAN

python/paddle/fluid/tests/book/test_recognize_digits.py

@@ -34,7 +34,7 @@ BATCH_SIZE = 64
 def loss_net(hidden, label):
     prediction = fluid.layers.fc(input=hidden, size=10, act='softmax')
     loss = fluid.layers.cross_entropy(input=prediction, label=label)
-    avg_loss = fluid.layers.mean(loss)
+    avg_loss = paddle.mean(loss)
     acc = fluid.layers.accuracy(input=prediction, label=label)
     return prediction, avg_loss, acc
 

python/paddle/fluid/tests/book/test_recommender_system.py

@@ -153,7 +153,7 @@ def model():
 
     label = layers.data(name='score', shape=[1], dtype='float32')
     square_cost = layers.square_error_cost(input=scale_infer, label=label)
-    avg_cost = layers.mean(square_cost)
+    avg_cost = paddle.mean(square_cost)
 
     return scale_infer, avg_cost
 

python/paddle/fluid/tests/book/test_rnn_encoder_decoder.py

@@ -158,7 +158,7 @@ def seq_to_seq_net():
                               dtype='int64',
                               lod_level=1)
     cost = fluid.layers.cross_entropy(input=prediction, label=label)
-    avg_cost = fluid.layers.mean(cost)
+    avg_cost = paddle.mean(cost)
 
     return avg_cost, prediction
 

python/paddle/fluid/tests/book/test_word2vec_book.py

@@ -85,7 +85,7 @@ def train(target,
                                        size=dict_size,
                                        act='softmax')
         cost = fluid.layers.cross_entropy(input=predict_word, label=words[4])
-        avg_cost = fluid.layers.mean(cost)
+        avg_cost = paddle.mean(cost)
         return avg_cost, predict_word
 
     word_dict = paddle.dataset.imikolov.build_dict()

python/paddle/fluid/tests/test_beam_search_decoder.py

@@ -145,7 +145,7 @@ def train_main(use_cuda):
                         dtype='int64',
                         lod_level=1)
     cost = layers.cross_entropy(input=rnn_out, label=label)
-    avg_cost = layers.mean(x=cost)
+    avg_cost = paddle.mean(x=cost)
 
     optimizer = fluid.optimizer.Adagrad(learning_rate=1e-3)
     optimizer.minimize(avg_cost)

python/paddle/fluid/tests/test_error_clip.py

@@ -35,7 +35,7 @@ with fluid.program_guard(main_program=prog):
     label = fluid.layers.data(name='y', shape=[1], dtype='int64')
 
     cost = fluid.layers.cross_entropy(input=predict, label=label)
-    avg_cost = fluid.layers.mean(cost)
+    avg_cost = paddle.mean(cost)
 
 prog_clip = prog.clone()
 prog_clip.block(0).var(hidden1.name)._set_error_clip(

python/paddle/fluid/tests/test_if_else_op.py

@@ -66,7 +66,7 @@ class TestMNISTIfElseOp(unittest.TestCase):
                                     mask=cond,
                                     x=image)
             loss = layers.cross_entropy(input=prob, label=label)
-            avg_loss = layers.mean(loss)
+            avg_loss = paddle.mean(loss)
 
             optimizer = MomentumOptimizer(learning_rate=0.001, momentum=0.9)
             optimizer.minimize(avg_loss, startup_prog)
@@ -124,7 +124,7 @@ class TestMNISTIfElseOp(unittest.TestCase):
 
             prob = ie()
             loss = layers.cross_entropy(input=prob[0], label=label)
-            avg_loss = layers.mean(loss)
+            avg_loss = paddle.mean(loss)
 
             optimizer = MomentumOptimizer(learning_rate=0.001, momentum=0.9)
             optimizer.minimize(avg_loss, startup_prog)

python/paddle/fluid/tests/unittests/asp/asp_pruning_base.py

@@ -61,7 +61,7 @@ class TestASPHelperPruningBase(unittest.TestCase):
 
     def run_training_pruning_test(self, get_mask_gen_func, get_mask_check_func):
         with fluid.program_guard(self.main_program, self.startup_program):
-            loss = fluid.layers.mean(
+            loss = paddle.mean(
                 fluid.layers.cross_entropy(input=self.predict,
                                            label=self.label))
             optimizer = paddle.incubate.asp.decorate(

python/paddle/fluid/tests/unittests/asp/test_asp_customized_pruning.py

@@ -242,7 +242,7 @@ class TestASPStaticCustomerizedPruneFunc(unittest.TestCase):
 
     def test_training_pruning(self):
         with fluid.program_guard(self.main_program, self.startup_program):
-            loss = fluid.layers.mean(
+            loss = paddle.mean(
                 fluid.layers.cross_entropy(input=self.predict,
                                            label=self.label))
             optimizer = sparsity.decorate(

python/paddle/fluid/tests/unittests/asp/test_asp_optimize_static.py

@@ -48,7 +48,7 @@ class TestASPStaticOptimize(unittest.TestCase):
 
         with fluid.program_guard(self.main_program, self.startup_program):
             self.img, self.label, predict = build_model()
-            self.loss = fluid.layers.mean(
+            self.loss = paddle.mean(
                 fluid.layers.cross_entropy(input=predict, label=self.label))
             self.optimizer = fluid.optimizer.SGD(learning_rate=0.01)
 

python/paddle/fluid/tests/unittests/asp/test_asp_pruning_static.py

@@ -66,7 +66,7 @@ class TestASPStaticPruningBase(unittest.TestCase):
 
     def test_training_pruning(self):
         with fluid.program_guard(self.main_program, self.startup_program):
-            loss = fluid.layers.mean(
+            loss = paddle.mean(
                 fluid.layers.cross_entropy(input=self.predict,
                                            label=self.label))
             optimizer = paddle.incubate.asp.decorate(

python/paddle/fluid/tests/unittests/asp/test_asp_save_load.py

@@ -141,7 +141,7 @@ class TestASPStaticOptimize(unittest.TestCase):
 
         with fluid.program_guard(self.main_program, self.startup_program):
             self.img, self.label, predict = build_model()
-            self.loss = fluid.layers.mean(
+            self.loss = paddle.mean(
                 fluid.layers.cross_entropy(input=predict, label=self.label))
             self.optimizer = fluid.optimizer.SGD(learning_rate=0.01)
             self.optimizer = paddle.incubate.asp.decorate(self.optimizer)

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

@@ -96,7 +96,7 @@ def mlp_pretrain_forward(train_program, start_program):
         predict = mlp(input)
 
         cost = layers.cross_entropy(input=predict, label=label)
-        avg_cost = layers.mean(x=cost)
+        avg_cost = paddle.mean(x=cost)
 
     return avg_cost, train_program, start_program
 

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

@@ -70,7 +70,7 @@ def net():
     cost, y_predict = fluid.layers.softmax_with_cross_entropy(
         hidden, y, return_softmax=True)
     acc_top1 = fluid.layers.accuracy(input=y_predict, label=y, k=1)
-    avg_cost = fluid.layers.mean(cost)
+    avg_cost = paddle.mean(cost)
 
     sgd_optimizer = fluid.optimizer.SGD(learning_rate=0.05)
     sgd_optimizer.minimize(avg_cost)

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

@@ -84,7 +84,7 @@ class TestDistMnist2x2(TestDistRunnerBase):
         # Train program
         predict = cnn_model(images)
         cost = fluid.layers.cross_entropy(input=predict, label=label)
-        avg_cost = fluid.layers.mean(x=cost)
+        avg_cost = paddle.mean(x=cost)
 
         # Evaluator
         batch_size_tensor = fluid.layers.create_tensor(dtype='int64')

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

@@ -92,7 +92,7 @@ class TestDistCTR2x2(TestDistRunnerBase):
         auc_var, batch_auc_var, auc_states = fluid.layers.auc(input=predict,
                                                               label=label)
         cost = fluid.layers.cross_entropy(input=predict, label=label)
-        avg_cost = fluid.layers.mean(x=cost)
+        avg_cost = paddle.mean(x=cost)
 
         inference_program = paddle.fluid.default_main_program().clone()
 

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

@@ -143,7 +143,7 @@ class TestDistCTR2x2(FleetDistRunnerBase):
                                                               label=label)
 
         cost = fluid.layers.cross_entropy(input=predict, label=label)
-        avg_cost = fluid.layers.mean(x=cost)
+        avg_cost = paddle.mean(x=cost)
 
         self.feeds = datas
         self.train_file_path = ["fake1", "fake2"]

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

@@ -116,7 +116,7 @@ class TestHeterPipelinePsCTR2x2(FleetDistHeterRunnerBase):
             predict = fluid.layers.fc(input=merge_layer, size=2, act='softmax')
 
             cost = fluid.layers.cross_entropy(input=predict, label=label)
-            avg_cost = fluid.layers.mean(x=cost)
+            avg_cost = paddle.mean(x=cost)
             fluid.layers.Print(avg_cost, message="avg_cost")
 
         self.feeds = datas

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

@@ -76,7 +76,7 @@ class TestFleetMetaOptimizerPrecision(TestDistRunnerBase):
         # Train program
         predict = cnn_model(images)
         cost = fluid.layers.cross_entropy(input=predict, label=label)
-        avg_cost = fluid.layers.mean(x=cost)
+        avg_cost = paddle.mean(x=cost)
 
         # Evaluator
         batch_size_tensor = fluid.layers.create_tensor(dtype='int64')

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

@@ -76,7 +76,7 @@ class TestFleetMetaOptimizerFuseAllReducePrecision(TestDistRunnerBase):
         # Train program
         predict = cnn_model(images)
         cost = fluid.layers.cross_entropy(input=predict, label=label)
-        avg_cost = fluid.layers.mean(x=cost)
+        avg_cost = paddle.mean(x=cost)
 
         # Evaluator
         batch_size_tensor = fluid.layers.create_tensor(dtype='int64')

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

@@ -91,7 +91,7 @@ def get_loss(cos_q_pt, cos_q_nt):
                                                    shape=[-1, 1],
                                                    value=0.0,
                                                    dtype='float32'), loss_op2)
-    avg_cost = fluid.layers.mean(loss_op3)
+    avg_cost = paddle.mean(loss_op3)
     return avg_cost
 
 

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

@@ -133,7 +133,7 @@ class TestDistCTR2x2(FleetDistRunnerBase):
         acc = fluid.layers.accuracy(input=predict, label=label)
         auc_var, _, _ = fluid.layers.auc(input=predict, label=label)
         cost = fluid.layers.cross_entropy(input=predict, label=label)
-        avg_cost = fluid.layers.mean(x=cost)
+        avg_cost = paddle.mean(x=cost)
 
         self.feeds = datas
         self.train_file_path = ["fake1", "fake2"]

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

@@ -85,7 +85,7 @@ class TestDistMnist2x2(TestDistRunnerBase):
         # Train program
         predict = cnn_model(images)
         cost = fluid.layers.cross_entropy(input=predict, label=label)
-        avg_cost = fluid.layers.mean(x=cost)
+        avg_cost = paddle.mean(x=cost)
 
         # Evaluator
         batch_size_tensor = fluid.layers.create_tensor(dtype='int64')

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

@@ -58,7 +58,7 @@ class TestDistMnist2x2(TestDistRunnerBase):
         # Train program
         predict = cnn_model(images)
         cost = fluid.layers.cross_entropy(input=predict, label=label)
-        avg_cost = fluid.layers.mean(x=cost)
+        avg_cost = paddle.mean(x=cost)
 
         # Evaluator
         batch_size_tensor = fluid.layers.create_tensor(dtype='int64')

python/paddle/fluid/tests/unittests/dist_mnist_fp16_allreduce.py

@@ -38,7 +38,7 @@ class TestDistMnist2x2(TestDistRunnerBase):
         # Train program
         predict = cnn_model(images)
         cost = fluid.layers.cross_entropy(input=predict, label=label)
-        avg_cost = fluid.layers.mean(x=cost)
+        avg_cost = paddle.mean(x=cost)
 
         # Evaluator
         batch_size_tensor = fluid.layers.create_tensor(dtype='int64')

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

@@ -37,7 +37,7 @@ class TestDistMnist2x2(TestDistRunnerBase):
         # Train program
         predict = cnn_model(images)
         cost = fluid.layers.cross_entropy(input=predict, label=label)
-        avg_cost = fluid.layers.mean(x=cost)
+        avg_cost = paddle.mean(x=cost)
 
         # Evaluator
         batch_size_tensor = fluid.layers.create_tensor(dtype='int64')

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

@@ -49,7 +49,7 @@ class TestDistMnist2x2(TestDistRunnerBase):
         # Train program
         predict = cnn_model(images)
         cost = fluid.layers.cross_entropy(input=predict, label=label)
-        avg_cost = fluid.layers.mean(x=cost)
+        avg_cost = paddle.mean(x=cost)
 
         # Evaluator
         batch_size_tensor = fluid.layers.create_tensor(dtype='int64')

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

@@ -229,7 +229,7 @@ class DistSeResneXt2x2(TestDistRunnerBase):
         out = model.net(input=image, class_dim=102)
         cost = fluid.layers.cross_entropy(input=out, label=label)
 
-        avg_cost = fluid.layers.mean(x=cost)
+        avg_cost = paddle.mean(x=cost)
         acc_top1 = fluid.layers.accuracy(input=out, label=label, k=1)
         acc_top5 = fluid.layers.accuracy(input=out, label=label, k=5)
 

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

@@ -56,7 +56,7 @@ def runtime_main():
                                                 act='softmax')
             cost = paddle.fluid.layers.cross_entropy(input=prediction,
                                                      label=input_y)
-            avg_cost = paddle.fluid.layers.mean(x=cost)
+            avg_cost = paddle.mean(x=cost)
 
             strategy = paddle.distributed.fleet.DistributedStrategy()
             strategy.sharding = True

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

@@ -137,7 +137,7 @@ class TestDistTextClassification2x2(TestDistRunnerBase):
         # Train program
         predict = conv_net(data, dict_dim)
         cost = fluid.layers.cross_entropy(input=predict, label=label)
-        avg_cost = fluid.layers.mean(x=cost)
+        avg_cost = paddle.mean(x=cost)
         acc = fluid.layers.accuracy(input=predict, label=label)
         inference_program = fluid.default_main_program().clone()
 

python/paddle/fluid/tests/unittests/dist_word2vec.py

@@ -94,7 +94,7 @@ class TestDistWord2vec2x2(TestDistRunnerBase):
                     initializer=fluid.initializer.Constant(value=0.1)))
             cost = fluid.layers.cross_entropy(input=predict_word,
                                               label=words[4])
-            avg_cost = fluid.layers.mean(cost)
+            avg_cost = paddle.mean(cost)
             return avg_cost, predict_word
 
         word_dict = paddle.dataset.imikolov.build_dict()

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

@@ -13,7 +13,7 @@
 # limitations under the License.
 
 from __future__ import absolute_import, division, print_function
-
+import paddle
 import paddle.fluid as fluid
 from paddle.fluid.dygraph import Embedding, Layer, Linear
 from paddle.fluid.dygraph.jit import declarative
@@ -357,7 +357,7 @@ class PretrainModelLayer(Layer):
 
         mask_lm_loss = fluid.layers.softmax_with_cross_entropy(logits=fc_out,
                                                                label=mask_label)
-        mean_mask_lm_loss = fluid.layers.mean(mask_lm_loss)
+        mean_mask_lm_loss = paddle.mean(mask_lm_loss)
 
         next_sent_fc_out = self.next_sent_fc(next_sent_feat)
 
@@ -367,7 +367,7 @@ class PretrainModelLayer(Layer):
         next_sent_acc = fluid.layers.accuracy(input=next_sent_softmax,
                                               label=labels)
 
-        mean_next_sent_loss = fluid.layers.mean(next_sent_loss)
+        mean_next_sent_loss = paddle.mean(next_sent_loss)
 
         loss = mean_next_sent_loss + mean_mask_lm_loss
         return next_sent_acc, mean_mask_lm_loss, loss

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

@@ -41,7 +41,7 @@ def dyfunc_empty_nonlocal(x):
 
 
 def dyfunc_with_if_else(x_v, label=None):
-    if fluid.layers.mean(x_v).numpy()[0] > 5:
+    if paddle.mean(x_v).numpy()[0] > 5:
         x_v = x_v - 1
     else:
         x_v = x_v + 1
@@ -87,14 +87,14 @@ def dyfunc_with_if_else3(x):
         m = x + 2
         n = x + 3
         return q, x, y, z
-    q, x, y, z = fluid.layers.cond(fluid.layers.mean(x)[0] < 5, lambda :
+    q, x, y, z = fluid.layers.cond(paddle.mean(x)[0] < 5, lambda :
         paddle.jit.dy2static.convert_call(true_fn_0)(q, x, y),
         lambda : paddle.jit.dy2static.convert_call(false_fn_0)(q,
         x, y))
     """
     y = x + 1
     # NOTE: x_v[0] < 5 is True
-    if fluid.layers.mean(x).numpy()[0] < 5:
+    if paddle.mean(x).numpy()[0] < 5:
         x = x + 1
         z = x + 2
         q = x + 3
@@ -155,13 +155,13 @@ def nested_if_else(x_v):
         batch_size = fluid.layers.shape(x_v)[0]
 
     # if tensor.shape is [1], now support to compare with numpy.
-    if fluid.layers.mean(x_v).numpy() < 0:
+    if paddle.mean(x_v).numpy() < 0:
         y = x_v + bias
         w = fluid.layers.fill_constant([feat_size], dtype='float32', value=10)
         if y.numpy()[0] < 10:
             tmp = y * w
             y = fluid.layers.relu(tmp)
-            if fluid.layers.mean(y).numpy()[0] < batch_size:
+            if paddle.mean(y).numpy()[0] < batch_size:
                 y = fluid.layers.abs(y)
             else:
                 tmp = fluid.layers.fill_constant([feat_size],
@@ -257,7 +257,7 @@ class NetWithControlFlowIf(fluid.dygraph.Layer):
                                                                 value=1)
         # Control flow `if` statement
         fc_out = self.fc(input)
-        if fluid.layers.mean(fc_out).numpy()[0] < 0:
+        if paddle.mean(fc_out).numpy()[0] < 0:
             y = fc_out + self.constant_vars['bias']
             self.constant_vars['w'] = fluid.layers.fill_constant(
                 [5], dtype='float32', value=10)
@@ -280,13 +280,13 @@ class NetWithControlFlowIf(fluid.dygraph.Layer):
         else:
             y = fc_out - self.constant_vars['bias']
 
-        loss = fluid.layers.mean(y)
+        loss = paddle.mean(y)
         return loss
 
 
 def if_with_and_or(x_v, label=None):
     batch_size = fluid.layers.shape(x_v)
-    if x_v is not None and (fluid.layers.mean(x_v).numpy()[0] > 0 or label
+    if x_v is not None and (paddle.mean(x_v).numpy()[0] > 0 or label
                             is not None) and batch_size[0] > 1 and True:
         x_v = x_v - 1
     else:
@@ -318,7 +318,7 @@ def if_with_and_or_2(x, y=None):
 
 def if_with_and_or_3(x, y=None):
     batch_size = fluid.layers.shape(x)
-    mean_res = fluid.layers.mean(x)
+    mean_res = paddle.mean(x)
     if x is not None and batch_size[0] > 1 and y is not None and mean_res.numpy(
     )[0] > 0:
         x = x + 1
@@ -329,7 +329,7 @@ def if_with_and_or_3(x, y=None):
 
 def if_with_and_or_4(x, y=None):
     batch_size = fluid.layers.shape(x)
-    mean_res = fluid.layers.mean(x)
+    mean_res = paddle.mean(x)
     if (x is not None and batch_size[0] > 1) or (y is not None
                                                  and mean_res.numpy()[0] > 0):
         x = x + 1
@@ -349,7 +349,7 @@ def if_with_class_var(x, y=None):
 
     foo = Foo()
     batch_size = fluid.layers.shape(x)
-    mean_res = fluid.layers.mean(x)
+    mean_res = paddle.mean(x)
 
     if batch_size[0] > foo.a:
         x = x + foo.b
@@ -361,7 +361,7 @@ def if_with_class_var(x, y=None):
 def if_tensor_case(x):
     x = fluid.dygraph.to_variable(x)
 
-    mean = fluid.layers.mean(x)
+    mean = paddle.mean(x)
     # It is equivalent to `if mean != 0`
     if mean:
         for i in range(0, 10):
@@ -376,7 +376,7 @@ def if_tensor_case(x):
             x += i
 
     # join `and`/`or`
-    if fluid.layers.mean(x) + 1 and mean > 1 and x is not None or 2 > 1:
+    if paddle.mean(x) + 1 and mean > 1 and x is not None or 2 > 1:
         x -= 1
 
     # `not` statement

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

@@ -19,6 +19,7 @@ import textwrap
 from paddle.utils import gast
 import inspect
 import numpy as np
+import paddle
 import paddle.fluid as fluid
 from paddle.fluid.dygraph.dygraph_to_static.utils import ast_to_func
 
@@ -59,7 +60,7 @@ class TestAST2Func(unittest.TestCase):
 
         def func(x):
             y = fluid.layers.relu(x)
-            loss = fluid.layers.mean(y)
+            loss = paddle.mean(y)
             return loss
 
         x_data = np.random.random([10, 16]).astype('float32')

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

@@ -590,7 +590,7 @@ def val_bmn(model, args):
 
         loss, tem_loss, pem_reg_loss, pem_cls_loss = bmn_loss_func(
             pred_bm, pred_start, pred_end, gt_iou_map, gt_start, gt_end, args)
-        avg_loss = fluid.layers.mean(loss)
+        avg_loss = paddle.mean(loss)
 
         loss_data += [
             avg_loss.numpy()[0],
@@ -665,7 +665,7 @@ class TestTrain(unittest.TestCase):
                     loss, tem_loss, pem_reg_loss, pem_cls_loss = bmn_loss_func(
                         pred_bm, pred_start, pred_end, gt_iou_map, gt_start,
                         gt_end, args)
-                    avg_loss = fluid.layers.mean(loss)
+                    avg_loss = paddle.mean(loss)
 
                     avg_loss.backward()
                     adam.minimize(avg_loss)

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

@@ -17,7 +17,7 @@ from __future__ import print_function
 import unittest
 import numpy as np
 from collections import Counter
-
+import paddle
 import paddle.fluid as fluid
 
 from paddle.fluid.dygraph.jit import declarative
@@ -113,7 +113,7 @@ class TestCacheProgramWithOptimizer(unittest.TestCase):
 
 def simple_func(x):
     inputs = fluid.dygraph.to_variable(x)
-    mean = fluid.layers.mean(inputs)
+    mean = paddle.mean(inputs)
     return mean
 
 

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

@@ -37,7 +37,7 @@ np.random.seed(SEED)
 # Use a decorator to test exception
 @paddle.jit.to_static
 def dyfunc_with_if(x_v):
-    if fluid.layers.mean(x_v).numpy()[0] > 5:
+    if paddle.mean(x_v).numpy()[0] > 5:
         x_v = x_v - 1
     else:
         x_v = x_v + 1
@@ -58,7 +58,7 @@ def nested_func(x_v):
 @paddle.jit.to_static
 def dyfunc_with_third_library_logging(x_v):
     logging.info('test dyfunc_with_third_library_logging')
-    if fluid.layers.mean(x_v).numpy()[0] > 5:
+    if paddle.mean(x_v).numpy()[0] > 5:
         x_v = x_v - 1
     else:
         x_v = x_v + 1

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

@@ -33,7 +33,7 @@ def inner_func():
 def func_error_in_compile_time(x):
     x = fluid.dygraph.to_variable(x)
     inner_func()
-    if fluid.layers.mean(x) < 0:
+    if paddle.mean(x) < 0:
         x_v = x - 1
     else:
         x_v = x + 1
@@ -78,7 +78,7 @@ class LayerErrorInCompiletime(fluid.dygraph.Layer):
     def forward(self, x):
         y = self._linear(x)
         z = fluid.layers.fill_constant(shape=[1, 2], value=9, dtype="int")
-        out = fluid.layers.mean(y[z])
+        out = paddle.mean(y[z])
         return out
 
 
@@ -386,7 +386,7 @@ class TestJitSaveInCompiletime(TestErrorBase):
              'y = self._linear(x)',
              'z = fluid.layers.fill_constant(shape=[1, 2], value=9, dtype="int")',
              '<--- HERE',
-             'out = fluid.layers.mean(y[z])',
+             'out = paddle.mean(y[z])',
              'return out'
              ]
 

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

@@ -16,7 +16,7 @@ from __future__ import print_function
 
 import numpy as np
 import unittest
-
+import paddle
 import paddle.fluid as fluid
 from paddle.fluid.dygraph.jit import declarative
 from paddle.fluid.dygraph.dygraph_to_static import ProgramTranslator
@@ -59,7 +59,7 @@ class Linear(fluid.dygraph.Layer):
     @declarative
     def forward(self, x):
         pre = self.fc(x)
-        loss = fluid.layers.mean(pre)
+        loss = paddle.mean(pre)
         return pre, loss
 
 

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

@@ -15,6 +15,7 @@
 from __future__ import print_function
 
 import numpy as np
+import paddle
 import paddle.fluid as fluid
 import unittest
 from paddle.fluid.dygraph import declarative
@@ -23,7 +24,7 @@ from paddle.fluid.dygraph import declarative
 @fluid.dygraph.declarative
 def dygraph_decorated_func(x):
     x = fluid.dygraph.to_variable(x)
-    if fluid.layers.mean(x) > 0:
+    if paddle.mean(x) > 0:
         x_v = x - 1
     else:
         x_v = x + 1
@@ -33,7 +34,7 @@ def dygraph_decorated_func(x):
 @fluid.dygraph.declarative
 def jit_decorated_func(x):
     x = fluid.dygraph.to_variable(x)
-    if fluid.layers.mean(x) > 0:
+    if paddle.mean(x) > 0:
         x_v = x - 1
     else:
         x_v = x + 1

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

@@ -251,7 +251,7 @@ def relu(x):
 
 
 def call_external_func(x, label=None):
-    if fluid.layers.mean(x) < 0:
+    if paddle.mean(x) < 0:
         x_v = x - 1
     else:
         x_v = add_fn(x)
@@ -274,7 +274,7 @@ class NetWithExternalFunc(fluid.dygraph.Layer):
 
     @declarative
     def forward(self, x, label=None):
-        if fluid.layers.mean(x) < 0:
+        if paddle.mean(x) < 0:
             x_v = x - 1
         else:
             x_v = add_fn(x)

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

@@ -403,7 +403,7 @@ class LexNet(fluid.dygraph.Layer):
         crf_cost = self.linear_chain_crf(input=emission,
                                          label=target,
                                          length=length)
-        avg_cost = fluid.layers.mean(x=crf_cost)
+        avg_cost = paddle.mean(x=crf_cost)
         crf_decode = self.crf_decoding(input=emission, length=length)
         return avg_cost, crf_decode
 

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

@@ -16,6 +16,7 @@ from __future__ import print_function
 
 import numpy as np
 import unittest
+import paddle
 import paddle.fluid as fluid
 
 from paddle.fluid.dygraph import declarative
@@ -25,7 +26,7 @@ def call_lambda_as_func(x):
     x = fluid.dygraph.to_variable(x)
 
     add_func = lambda x, y: x + y
-    mean_func = lambda x: fluid.layers.mean(x)
+    mean_func = lambda x: paddle.mean(x)
 
     y = add_func(x, 1)
     y = add_func(y, add_func(y, -1))
@@ -38,7 +39,7 @@ def call_lambda_directly(x):
     x = fluid.dygraph.to_variable(x)
 
     y = (lambda x, y: x + y)(x, x)
-    out = (lambda x: fluid.layers.mean(x))(y)
+    out = (lambda x: paddle.mean(x))(y)
 
     return out
 
@@ -48,7 +49,7 @@ def call_lambda_in_func(x):
 
     add_func = lambda x: x + 1
 
-    y = fluid.layers.mean((lambda x: fluid.layers.relu(x))(x))
+    y = paddle.mean((lambda x: fluid.layers.relu(x))(x))
     out = add_func(y) if y > 1 and y < 2 else (lambda x: x**2)(y)
 
     return out
@@ -59,7 +60,7 @@ def call_lambda_with_ifExpr(x):
 
     add_func = lambda x: x + 1
 
-    y = fluid.layers.mean(x)
+    y = paddle.mean(x)
     out = add_func(y) if y or y < 2 else (lambda x: x**2)(y)
 
     return out
@@ -70,7 +71,7 @@ def call_lambda_with_ifExpr2(x):
 
     add_func = lambda x: x + 1
 
-    y = fluid.layers.mean(x)
+    y = paddle.mean(x)
 
     # NOTE: y is Variable, but z<2 is python bool value
     z = 0

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

@@ -119,7 +119,7 @@ class MNIST(fluid.dygraph.Layer):
         if label is not None:
             acc = fluid.layers.accuracy(input=x, label=label)
             loss = fluid.layers.cross_entropy(x, label)
-            avg_loss = fluid.layers.mean(loss)
+            avg_loss = paddle.mean(loss)
 
             return x, acc, avg_loss
         else:

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

@@ -479,7 +479,7 @@ def train_mobilenet(args, to_static):
                 softmax_out = fluid.layers.softmax(out, use_cudnn=False)
                 loss = fluid.layers.cross_entropy(input=softmax_out,
                                                   label=label)
-                avg_loss = fluid.layers.mean(x=loss)
+                avg_loss = paddle.mean(x=loss)
                 acc_top1 = fluid.layers.accuracy(input=out, label=label, k=1)
                 acc_top5 = fluid.layers.accuracy(input=out, label=label, k=5)
                 t_start_back = time.time()

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

@@ -33,7 +33,7 @@ def nested_input(x, y):
     sub_res = z_elem[0] - z_elem[1]
 
     mul_res = y[-1]['d']['da'] * y[-1]['d']['dc']
-    mean_func = fluid.layers.mean
+    mean_func = paddle.mean
     out = mean_func(sub_res) + mean_func(sum_res) + mean_func(mul_res)
     return out
 

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

@@ -42,7 +42,7 @@ def simple_func(x, weight_numpy):
     x = fluid.dygraph.to_variable(x)
     w = fluid.dygraph.to_variable(weight_numpy)
     y = fluid.layers.matmul(x, w)
-    z = fluid.layers.mean(y)
+    z = paddle.mean(y)
     return z
 
 
@@ -51,7 +51,7 @@ def decorated_simple_func(x, weight_numpy):
     x = fluid.dygraph.to_variable(x)
     w = fluid.dygraph.to_variable(weight_numpy)
     y = fluid.layers.matmul(x, w)
-    z = fluid.layers.mean(y)
+    z = paddle.mean(y)
     return z
 
 
@@ -91,7 +91,7 @@ class StaticCode1():
             return x_v
 
         _jst.IfElse(
-            fluid.layers.mean(x_v)[0] > 5, true_fn_0, false_fn_0, get_args_0,
+            paddle.mean(x_v)[0] > 5, true_fn_0, false_fn_0, get_args_0,
             set_args_0, ('x_v', ))
 
         def get_args_1():
@@ -148,7 +148,7 @@ class StaticCode2():
             return x_v
 
         _jst.IfElse(
-            fluid.layers.mean(x_v)[0] > 5, true_fn_2, false_fn_2, get_args_2,
+            paddle.mean(x_v)[0] > 5, true_fn_2, false_fn_2, get_args_2,
             set_args_2, ('x_v', ))
 
         def get_args_3():

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

@@ -262,7 +262,7 @@ class ResNetHelper:
 
                     pred = resnet(img)
                     loss = fluid.layers.cross_entropy(input=pred, label=label)
-                    avg_loss = fluid.layers.mean(x=loss)
+                    avg_loss = paddle.mean(x=loss)
                     acc_top1 = fluid.layers.accuracy(input=pred,
                                                      label=label,
                                                      k=1)

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

@@ -75,7 +75,7 @@ def train(to_static, build_strategy=None):
                     # precision problem, need to figure out the underlying reason.
                     # If we remove it, the loss between dygraph and dy2stat is exactly same.
                     loss = fluid.layers.cross_entropy(input=pred, label=label)
-                avg_loss = fluid.layers.mean(x=pred)
+                avg_loss = paddle.mean(x=pred)
                 acc_top1 = fluid.layers.accuracy(input=pred, label=label, k=1)
                 acc_top5 = fluid.layers.accuracy(input=pred, label=label, k=5)
 

python/paddle/fluid/tests/unittests/dygraph_to_static/test_resnet_pure_fp16.py

@@ -77,7 +77,7 @@ def train(to_static, build_strategy=None):
                                       level='O2'):
                 pred = resnet(img)
                 loss = fluid.layers.cross_entropy(input=pred, label=label)
-            avg_loss = fluid.layers.mean(x=pred)
+            avg_loss = paddle.mean(x=pred)
             acc_top1 = fluid.layers.accuracy(input=pred, label=label, k=1)
             acc_top5 = fluid.layers.accuracy(input=pred, label=label, k=5)
 

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

@@ -45,7 +45,7 @@ class SimpleFcLayer(fluid.dygraph.Layer):
     def forward(self, x):
         y = self._linear(x)
         z = self._linear(y)
-        out = fluid.layers.mean(z)
+        out = paddle.mean(z)
         return out, y
 
 

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

@@ -318,7 +318,7 @@ class SeResNeXt(fluid.dygraph.Layer):
 
         softmax_out = fluid.layers.softmax(out, use_cudnn=False)
         loss = fluid.layers.cross_entropy(input=softmax_out, label=label)
-        avg_loss = fluid.layers.mean(x=loss)
+        avg_loss = paddle.mean(x=loss)
 
         acc_top1 = fluid.layers.accuracy(input=softmax_out, label=label, k=1)
         acc_top5 = fluid.layers.accuracy(input=softmax_out, label=label, k=5)

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

@@ -97,7 +97,7 @@ class CNN(fluid.dygraph.Layer):
         prediction = self._fc_prediction(fc_1)
 
         cost = fluid.layers.cross_entropy(input=prediction, label=label)
-        avg_cost = fluid.layers.mean(x=cost)
+        avg_cost = paddle.mean(x=cost)
         acc = fluid.layers.accuracy(input=prediction, label=label)
         return avg_cost, prediction, acc
 
@@ -141,7 +141,7 @@ class BOW(fluid.dygraph.Layer):
         prediction = self._fc_prediction(fc_2)
 
         cost = fluid.layers.cross_entropy(input=prediction, label=label)
-        avg_cost = fluid.layers.mean(x=cost)
+        avg_cost = paddle.mean(x=cost)
         acc = fluid.layers.accuracy(input=prediction, label=label)
         return avg_cost, prediction, acc
 
@@ -189,7 +189,7 @@ class GRU(fluid.dygraph.Layer):
         prediction = self._fc_prediction(fc_2)
 
         cost = fluid.layers.cross_entropy(input=prediction, label=label)
-        avg_cost = fluid.layers.mean(x=cost)
+        avg_cost = paddle.mean(x=cost)
         acc = fluid.layers.accuracy(input=prediction, label=label)
         return avg_cost, prediction, acc
 
@@ -247,7 +247,7 @@ class BiGRU(fluid.dygraph.Layer):
         # TODO(Aurelius84): Uncomment the following codes when we support return variable-length vars.
         # if label is not None:
         cost = fluid.layers.cross_entropy(input=prediction, label=label)
-        avg_cost = fluid.layers.mean(x=cost)
+        avg_cost = paddle.mean(x=cost)
         acc = fluid.layers.accuracy(input=prediction, label=label)
         return avg_cost, prediction, acc
         # else:

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

@@ -303,7 +303,7 @@ def train(args, fake_data_reader, to_static):
                 loss = fluid.layers.cross_entropy(input=outputs,
                                                   label=labels,
                                                   ignore_index=-1)
-                avg_loss = fluid.layers.mean(loss)
+                avg_loss = paddle.mean(loss)
                 acc_top1 = fluid.layers.accuracy(input=outputs,
                                                  label=labels,
                                                  k=1)

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

@@ -101,7 +101,7 @@ def net(batch_size=4, lr=0.01):
         predict = fluid.layers.fc(input=merge_layer, size=2, act='softmax')
 
         cost = fluid.layers.cross_entropy(input=predict, label=label)
-        avg_cost = fluid.layers.mean(x=cost)
+        avg_cost = paddle.mean(x=cost)
     return datas, avg_cost
 
 

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

@@ -66,7 +66,7 @@ class TestFleetMetaOptimizer(unittest.TestCase):
                                                     act='softmax')
                 cost = paddle.fluid.layers.cross_entropy(input=prediction,
                                                          label=input_y)
-                avg_cost = paddle.fluid.layers.mean(x=cost)
+                avg_cost = paddle.mean(x=cost)
 
                 strategy = paddle.distributed.fleet.DistributedStrategy()
         return avg_cost, strategy
@@ -101,7 +101,7 @@ class TestFleetMetaOptimizer(unittest.TestCase):
                                                         act='softmax')
                     cost = paddle.fluid.layers.cross_entropy(input=prediction,
                                                              label=input_y)
-                    avg_cost = paddle.fluid.layers.mean(x=cost)
+                    avg_cost = paddle.mean(x=cost)
 
         strategy = paddle.distributed.fleet.DistributedStrategy()
         return avg_cost, strategy

python/paddle/fluid/tests/unittests/ipu/test_mean_op_ipu.py

@@ -48,7 +48,7 @@ class TestBase(IPUOpTest):
         x = paddle.static.data(name=self.feed_list[0],
                                shape=self.feed_shape[0],
                                dtype='float32')
-        out = paddle.fluid.layers.mean(x)
+        out = paddle.mean(x)
         self.fetch_list = [out.name]
 
     def run_model(self, exec_mode):

python/paddle/fluid/tests/unittests/ir/inference/test_trt_conv_quant_dequant_pass.py

@@ -18,6 +18,7 @@ import unittest
 import numpy as np
 from inference_pass_test import InferencePassTest
 from quant_dequant_test import QuantDequantTest
+import paddle
 import paddle.fluid as fluid
 import paddle.fluid.core as core
 from paddle.fluid.core import PassVersionChecker
@@ -54,7 +55,7 @@ class QuantDequantTensorRTSubgraphPassConvTest(QuantDequantTest):
                 cout = fluid.layers.reshape(conv_out, shape=[1, 1, 10816])
             result = fluid.layers.relu(cout)
             loss = fluid.layers.cross_entropy(input=result, label=label_shape)
-            avg_loss = fluid.layers.mean(loss)
+            avg_loss = paddle.mean(loss)
             return avg_loss, result
 
         self.main_program.random_seed = 2
@@ -152,7 +153,7 @@ class DynamicShapeQuantDequantTensorRTSubgraphPassConvTest(QuantDequantTest):
             cout = fluid.layers.reshape(conv_out, shape=[1, 1, 10816])
             result = fluid.layers.relu(cout)
             loss = fluid.layers.cross_entropy(input=result, label=label_shape)
-            avg_loss = fluid.layers.mean(loss)
+            avg_loss = paddle.mean(loss)
             return avg_loss, result
 
         self.main_program.random_seed = 2
@@ -245,7 +246,7 @@ class QuantDequantTensorRTSubgraphPassConvTransposeTest(QuantDequantTest):
                 cout = fluid.layers.reshape(conv_out, shape=[1, 1, 10816])
             result = fluid.layers.relu(cout)
             loss = fluid.layers.cross_entropy(input=result, label=label_shape)
-            avg_loss = fluid.layers.mean(loss)
+            avg_loss = paddle.mean(loss)
             return avg_loss, result
 
         self.main_program.random_seed = 2

python/paddle/fluid/tests/unittests/ir/inference/test_trt_fc_fuse_quant_dequant_pass.py

@@ -18,6 +18,7 @@ import unittest
 import numpy as np
 from inference_pass_test import InferencePassTest
 from quant_dequant_test import QuantDequantTest
+import paddle
 import paddle.fluid as fluid
 import paddle.fluid.core as core
 from paddle.fluid.core import AnalysisConfig
@@ -40,7 +41,7 @@ class FCQuantDequantFusePassTRTDims3Cols1Test(QuantDequantTest):
                                      act="relu")
             result = fluid.layers.relu(fc_out)
             loss = fluid.layers.cross_entropy(input=result, label=self.label)
-            avg_loss = fluid.layers.mean(loss)
+            avg_loss = paddle.mean(loss)
             return avg_loss, result
 
         self.main_program.random_seed = 2
@@ -105,7 +106,7 @@ class FCQuantDequantFusePassTRTDims3Cols2Test(QuantDequantTest):
             c_out = fluid.layers.reshape(fc_out, shape=[0, 784])
             result = fluid.layers.relu(c_out)
             loss = fluid.layers.cross_entropy(input=result, label=self.label)
-            avg_loss = fluid.layers.mean(loss)
+            avg_loss = paddle.mean(loss)
             return avg_loss, result
 
         self.main_program.random_seed = 2
@@ -172,7 +173,7 @@ class FCQuantDequantFusePassTRTDims3Cols3Test(QuantDequantTest):
             c_out = fluid.layers.reshape(fc_out, shape=[1, 1, 2744])
             result = fluid.layers.relu(c_out)
             loss = fluid.layers.cross_entropy(input=result, label=label_shape)
-            avg_loss = fluid.layers.mean(loss)
+            avg_loss = paddle.mean(loss)
             return avg_loss, result
 
         self.main_program.random_seed = 2

python/paddle/fluid/tests/unittests/ir/inference/test_trt_matmul_quant_dequant.py

@@ -16,6 +16,7 @@ import unittest
 import numpy as np
 from inference_pass_test import InferencePassTest
 from quant_dequant_test import QuantDequantTest
+import paddle
 import paddle.fluid as fluid
 import paddle.fluid.core as core
 from paddle.fluid.core import PassVersionChecker
@@ -44,7 +45,7 @@ class TensorRTMatMulQuantDequantDims3Test(QuantDequantTest):
                                      act=None)
             result = fluid.layers.relu(fc_out)
             loss = fluid.layers.cross_entropy(input=result, label=self.label)
-            avg_loss = fluid.layers.mean(loss)
+            avg_loss = paddle.mean(loss)
             return avg_loss, result
 
         self.main_program.random_seed = 2
@@ -136,7 +137,7 @@ class TensorRTMatMulQuantDequantDims4Test(QuantDequantTest):
                                      act=None)
             result = fluid.layers.relu(fc_out)
             loss = fluid.layers.cross_entropy(input=result, label=self.label)
-            avg_loss = fluid.layers.mean(loss)
+            avg_loss = paddle.mean(loss)
             return avg_loss, result
 
         self.main_program.random_seed = 2
@@ -227,7 +228,7 @@ class TensorRTMatMulQuantDequantDims3DynamicTest(QuantDequantTest):
                                      act=None)
             result = fluid.layers.relu(fc_out)
             loss = fluid.layers.cross_entropy(input=result, label=self.label)
-            avg_loss = fluid.layers.mean(loss)
+            avg_loss = paddle.mean(loss)
             return avg_loss, result
 
         self.main_program.random_seed = 2

python/paddle/fluid/tests/unittests/ir/pass_test.py

@@ -20,7 +20,7 @@ import random
 import unittest
 import warnings
 import numpy as np
-
+import paddle
 import paddle.fluid as fluid
 import paddle.fluid.core as core
 from paddle.fluid.framework import Program, Block
@@ -56,7 +56,7 @@ class PassTest(unittest.TestCase):
 
     def append_gradients(self, outs):
         with fluid.program_guard(self.main_program, self.startup_program):
-            loss = fluid.layers.mean(outs)
+            loss = paddle.mean(outs)
             fluid.backward.append_backward(loss)
 
     def check_output(self, startup_on_cpu=False, atol=1e-5):

python/paddle/fluid/tests/unittests/ir/test_ir_subgraph_python_interface.py

@@ -41,7 +41,7 @@ class TestQuantizationSubGraph(unittest.TestCase):
             for _ in six.moves.xrange(num):
                 hidden = fluid.layers.fc(hidden, size=128, act='relu')
             loss = fluid.layers.cross_entropy(input=hidden, label=label)
-            loss = fluid.layers.mean(loss)
+            loss = paddle.mean(loss)
             return loss
 
         main_program = Program()

python/paddle/fluid/tests/unittests/mlu/test_momentum_op_mlu.py

@@ -148,7 +148,7 @@ class TestMomentumV2(unittest.TestCase):
             y = fluid.layers.data(name='y', shape=[1], dtype='float32')
             y_predict = fluid.layers.fc(input=x, size=1, act=None)
             cost = fluid.layers.square_error_cost(input=y_predict, label=y)
-            avg_cost = fluid.layers.mean(cost)
+            avg_cost = paddle.mean(cost)
 
             rms_optimizer = paddle.optimizer.Momentum(learning_rate=0.1,
                                                       momentum=0.9)
@@ -271,7 +271,7 @@ class TestMomentumOpWithDecayAPI(unittest.TestCase):
             y = fluid.layers.data(name='y', shape=[1], dtype='float32')
             y_predict = fluid.layers.fc(input=x, size=1, act=None)
             cost = fluid.layers.square_error_cost(input=y_predict, label=y)
-            avg_cost = fluid.layers.mean(cost)
+            avg_cost = paddle.mean(cost)
 
             momentum_optimizer = paddle.fluid.contrib.optimizer.Momentum(
                 learning_rate=0.1, momentum=0.9)
@@ -591,7 +591,7 @@ class TestMultiTensorMomentumStatic(unittest.TestCase):
                                           name='X',
                                           dtype='float32')
             hidden = paddle.static.nn.fc(x=data, size=10)
-            loss = paddle.fluid.layers.mean(hidden)
+            loss = paddle.mean(hidden)
             optimizer.minimize(loss)
         exe.run(startup_program)
         if use_amp:

python/paddle/fluid/tests/unittests/mlu/test_where_op_mlu.py

@@ -107,7 +107,7 @@ class TestWhereAPI(unittest.TestCase):
                     x.stop_gradient = x_stop_gradient
                     y.stop_gradient = y_stop_gradient
                     result = paddle.where(cond, x, y)
-                    append_backward(layers.mean(result))
+                    append_backward(paddle.mean(result))
                     for use_mlu in [False, True]:
                         place = (paddle.device.MLUPlace(0)
                                  if use_mlu else fluid.CPUPlace())

python/paddle/fluid/tests/unittests/npu/test_momentum_op_npu.py

@@ -117,7 +117,7 @@ class TestMomentumV2(unittest.TestCase):
             y = fluid.layers.data(name='y', shape=[1], dtype='float32')
             y_predict = fluid.layers.fc(input=x, size=1, act=None)
             cost = fluid.layers.square_error_cost(input=y_predict, label=y)
-            avg_cost = fluid.layers.mean(cost)
+            avg_cost = paddle.mean(cost)
 
             rms_optimizer = paddle.optimizer.Momentum(learning_rate=0.1,
                                                       momentum=0.9)
@@ -243,7 +243,7 @@ class TestMomentumOpWithDecayAPI(unittest.TestCase):
             y = fluid.layers.data(name='y', shape=[1], dtype='float32')
             y_predict = fluid.layers.fc(input=x, size=1, act=None)
             cost = fluid.layers.square_error_cost(input=y_predict, label=y)
-            avg_cost = fluid.layers.mean(cost)
+            avg_cost = paddle.mean(cost)
 
             momentum_optimizer = paddle.fluid.contrib.optimizer.Momentum(
                 learning_rate=0.1, momentum=0.9)

python/paddle/fluid/tests/unittests/npu/test_softmax_op_npu.py

@@ -86,7 +86,7 @@ class TestSoftmaxNet(unittest.TestCase):
             prob = fluid.layers.softmax(prediction, axis=1)
 
             cost = fluid.layers.cross_entropy(input=prob, label=label)
-            loss = fluid.layers.mean(cost)
+            loss = paddle.mean(cost)
             sgd = fluid.optimizer.SGD(learning_rate=0.01)
             sgd.minimize(loss)
 

python/paddle/fluid/tests/unittests/npu/test_where_op_npu.py

@@ -105,7 +105,7 @@ class TestNPUWhereAPI(unittest.TestCase):
                     y.stop_gradient = y_stop_gradient
 
                     result = paddle.where(cond, x, y)
-                    append_backward(fluid.layers.mean(result))
+                    append_backward(paddle.mean(result))
 
                     exe = fluid.Executor(self.place)
                     exe.run(startup)

python/paddle/fluid/tests/unittests/npu/test_while_op_npu.py

@@ -90,7 +90,7 @@ class TestWhileOp(unittest.TestCase):
                 layers.array_write(result2, i=j, array=mem_array)
                 layers.less_than(x=j, y=array_len2, cond=cond2)
         sum_result = layers.array_read(array=mem_array, i=j)
-        loss = layers.mean(sum_result)
+        loss = paddle.mean(sum_result)
         return loss, sum_result
 
     def test_simple_net(self):

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

@@ -112,7 +112,7 @@ class MNIST(fluid.dygraph.Layer):
         x = fluid.layers.reshape(x, shape=[-1, self.pool_2_shape])
         cost = self._fc(x)
         loss = fluid.layers.cross_entropy(cost, label)
-        avg_loss = fluid.layers.mean(loss)
+        avg_loss = paddle.mean(loss)
         return avg_loss
 
 

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

@@ -335,7 +335,7 @@ class TestSeResNeXt(TestParallelDyGraphRunnerBase):
         out = model(img)
         softmax_out = fluid.layers.softmax(out, use_cudnn=False)
         loss = fluid.layers.cross_entropy(input=softmax_out, label=label)
-        avg_loss = fluid.layers.mean(x=loss)
+        avg_loss = paddle.mean(x=loss)
         return avg_loss
 
 

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

@@ -94,7 +94,7 @@ class TestSyncBatchNorm(TestParallelDyGraphRunnerBase):
 
         out = model(img)
 
-        out = fluid.layers.mean(out)
+        out = paddle.mean(out)
 
         return out
 

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

@@ -104,7 +104,7 @@ class TestDistMnist2x2(TestDistRunnerBase):
             predict = cnn_model(images)
         with fluid.device_guard("gpu:1"):
             cost = fluid.layers.cross_entropy(input=predict, label=label)
-            avg_cost = fluid.layers.mean(x=cost)
+            avg_cost = paddle.mean(x=cost)
 
         # Evaluator
         with fluid.device_guard("gpu:1"):

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

@@ -104,7 +104,7 @@ class TestDistMnist2x2(TestDistRunnerBase):
             predict = cnn_model(images)
         with fluid.device_guard("gpu:1"):
             cost = fluid.layers.cross_entropy(input=predict, label=label)
-            avg_cost = fluid.layers.mean(x=cost)
+            avg_cost = paddle.mean(x=cost)
 
         # Evaluator
         with fluid.device_guard("gpu:1"):

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

@@ -98,7 +98,7 @@ class TestDistMnist2x2(TestDistRunnerBase):
             predict = cnn_model(images)
         with fluid.device_guard("gpu:0"):
             cost = fluid.layers.cross_entropy(input=predict, label=label)
-            avg_cost = fluid.layers.mean(x=cost)
+            avg_cost = paddle.mean(x=cost)
 
         # Evaluator
         with fluid.device_guard("gpu:0"):

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

@@ -17,6 +17,7 @@ import paddle.fluid as fluid
 
 fluid.core._set_eager_deletion_mode(-1, -1, False)
 
+import paddle
 import paddle.fluid.layers.ops as ops
 from paddle.fluid.layers.learning_rate_scheduler import cosine_decay
 from simple_nets import init_data
@@ -172,7 +173,7 @@ def SE_ResNeXt50Small(use_feed):
     # Classifier layer:
     prediction = fluid.layers.fc(input=dropout, size=1000, act='softmax')
     loss = fluid.layers.cross_entropy(input=prediction, label=label)
-    loss = fluid.layers.mean(loss)
+    loss = paddle.mean(loss)
     return loss
 
 

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

@@ -12,6 +12,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
+import paddle
 import paddle.fluid as fluid
 import numpy as np
 
@@ -27,7 +28,7 @@ def simple_fc_net_with_inputs(img, label, class_num=10):
                 value=1.0)))
     prediction = fluid.layers.fc(hidden, size=class_num, act='softmax')
     loss = fluid.layers.cross_entropy(input=prediction, label=label)
-    loss = fluid.layers.mean(loss)
+    loss = paddle.mean(loss)
     return loss
 
 
@@ -51,7 +52,7 @@ def batchnorm_fc_with_inputs(img, label, class_num=10):
 
     prediction = fluid.layers.fc(hidden, size=class_num, act='softmax')
     loss = fluid.layers.cross_entropy(input=prediction, label=label)
-    loss = fluid.layers.mean(loss)
+    loss = paddle.mean(loss)
     return loss
 
 
@@ -87,7 +88,7 @@ def bow_net(use_feed,
     fc_2 = fluid.layers.fc(input=fc_1, size=hid_dim2, act="tanh")
     prediction = fluid.layers.fc(input=[fc_2], size=class_dim, act="softmax")
     cost = fluid.layers.cross_entropy(input=prediction, label=label)
-    avg_cost = fluid.layers.mean(x=cost)
+    avg_cost = paddle.mean(x=cost)
 
     return avg_cost
 

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

@@ -136,7 +136,7 @@ class TestAdadeltaV2(unittest.TestCase):
             y = fluid.layers.data(name='y', shape=[1], dtype='float32')
             y_predict = fluid.layers.fc(input=x, size=1, act=None)
             cost = fluid.layers.square_error_cost(input=y_predict, label=y)
-            avg_cost = fluid.layers.mean(cost)
+            avg_cost = paddle.mean(cost)
 
             rms_optimizer = paddle.optimizer.Adadelta(learning_rate=0.1)
             rms_optimizer.minimize(avg_cost)