clear fluid apis: square_error_cost (#48029)

* clear fluid apis in fleet and passes

* fix model.py

* fix model.py

* fix cpp_pass.py

* clear loss.py

* change test file

* fix some test_*.py

* fix adaround.py

* fix evaluator.py

* fix CI bug

* fix CI bug

* fix decode.py

* fix detection.py

* fix ci bug

* rm test_sigmoid_cross_entropy_with_logits_op_ipu.py and fix __init__.py

* fix ci bug

* fix ci BUG

python/paddle/fluid/contrib/optimizer.py

@@ -78,7 +78,6 @@ class Momentum(Optimizer):
             import numpy as np
 
             paddle.enable_static()
-
             place = fluid.CPUPlace()
             main = fluid.Program()
             with fluid.program_guard(main):

python/paddle/fluid/contrib/slim/quantization/adaround.py

@@ -16,6 +16,7 @@ import numpy as np
 import time
 import sys
 import logging
+import paddle
 
 import paddle
 import paddle.fluid as fluid
@@ -61,7 +62,7 @@ class AdaRoundLoss:
         self.default_beta_range = default_beta_range
 
     def compute_recon_loss(self, ada_quantized_output, orig_output):
-        square_cost = fluid.layers.square_error_cost(
+        square_cost = paddle.nn.functional.square_error_cost(
             ada_quantized_output, orig_output
         )
         recon_loss = paddle.mean(paddle.sum(square_cost, axis=-1))

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

@@ -50,7 +50,7 @@ class TestMovingAverageAbsMaxScaleOp(unittest.TestCase):
                 name=fc_tmp.name, dtype=fc_tmp.dtype
             )
             fc_tmp_1 = out_scale(fc_tmp)
-            cross_entropy = fluid.layers.softmax_with_cross_entropy(
+            cross_entropy = paddle.nn.functional.softmax_with_cross_entropy(
                 fc_tmp, label
             )
             loss = paddle.mean(cross_entropy)

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

@@ -127,7 +127,7 @@ def train(net_type, use_cuda, save_dirname, is_local):
             raise ValueError("%s network is not supported" % net_type)
 
         logits = fluid.layers.fc(input=net, size=classdim, act="softmax")
-        cost, predict = fluid.layers.softmax_with_cross_entropy(
+        cost, predict = paddle.nn.functional.softmax_with_cross_entropy(
             logits, label, return_softmax=True
         )
         avg_cost = paddle.mean(cost)
@@ -509,7 +509,7 @@ class TestAmpWithNonIterableDataLoader(unittest.TestCase):
 
                 net = vgg16_bn_drop(image)
                 logits = fluid.layers.fc(input=net, size=10, act="softmax")
-                cost, predict = fluid.layers.softmax_with_cross_entropy(
+                cost, predict = paddle.nn.functional.softmax_with_cross_entropy(
                     logits, label, return_softmax=True
                 )
                 avg_cost = paddle.mean(cost)

python/paddle/fluid/contrib/tests/test_multi_precision_fp16_train.py

@@ -110,7 +110,7 @@ def train(use_pure_fp16=True, use_nesterov=False, optimizer=""):
         label = fluid.layers.data(name='label', shape=[1], dtype='int64')
         net = resnet_cifar10(images)
         logits = fluid.layers.fc(input=net, size=classdim, act="softmax")
-        cost = fluid.layers.softmax_with_cross_entropy(
+        cost = paddle.nn.functional.softmax_with_cross_entropy(
             logits, label, return_softmax=False
         )
         sum_cost = paddle.sum(cost)

python/paddle/fluid/layers/detection.py

@@ -21,7 +21,7 @@ from .layer_function_generator import templatedoc
 from ..layer_helper import LayerHelper
 from ..framework import Variable, _non_static_mode, static_only, in_dygraph_mode
 from .. import core
-from .loss import softmax_with_cross_entropy
+from paddle.fluid.layers import softmax_with_cross_entropy
 from . import tensor
 from . import nn
 from ..data_feeder import check_variable_and_dtype, check_type, check_dtype

python/paddle/fluid/layers/loss.py

@@ -36,7 +36,6 @@ from paddle import _C_ops, _legacy_C_ops
 
 __all__ = [
     'cross_entropy',
-    'square_error_cost',
     'softmax_with_cross_entropy',
 ]
 
@@ -144,41 +143,6 @@ def cross_entropy2(input, label, ignore_index=kIgnoreIndex):
     return out
 
 
-def square_error_cost(input, label):
-    r"""
-
-    Accept input predictions and target label and returns the
-    squared error cost.
-
-    For predictions label, and target label, the equation is:
-
-    .. math::
-
-        Out = (input - label)^2
-
-    Parameters:
-        input (Tensor): Input tensor, the data type should be float32.
-        label (Tensor): Label tensor, the data type should be float32.
-
-    Returns:
-        Tensor, The tensor storing the element-wise squared
-        error difference between input and label.
-
-    Examples:
-
-        .. code-block:: python
-
-            import paddle
-            input = paddle.to_tensor([1.1, 1.9])
-            label = paddle.to_tensor([1.0, 2.0])
-            output = paddle.nn.functional.square_error_cost(input, label)
-            print(output)
-            # [0.01, 0.01]
-
-    """
-    return paddle.nn.functional.square_error_cost(input, label)
-
-
 def softmax_with_cross_entropy(
     logits,
     label,
@@ -189,49 +153,32 @@ def softmax_with_cross_entropy(
     axis=-1,
 ):
     r"""
-
     This operator implements the cross entropy loss function with softmax. This function
     combines the calculation of the softmax operation and the cross entropy loss function
     to provide a more numerically stable gradient.
-
     Because this operator performs a softmax on logits internally, it expects
     unscaled logits. This operator should not be used with the output of
     softmax operator since that would produce incorrect results.
-
     When the attribute :attr:`soft_label` is set :attr:`False`, this operators
     expects mutually exclusive hard labels, each sample in a batch is in exactly
     one class with a probability of 1.0. Each sample in the batch will have a
     single label.
-
     The equation is as follows:
-
     1) Hard label (one-hot label, so every sample has exactly one class)
-
     .. math::
-
         loss_j =  -\\text{logits}_{label_j} +
         \\log\\left(\\sum_{i=0}^{K}\\exp(\\text{logits}_i)\\right), j = 1,..., K
-
     2) Soft label (each sample can have a distribution over all classes)
-
     .. math::
-
         loss_j =  -\\sum_{i=0}^{K}\\text{label}_i
         \\left(\\text{logits}_i - \\log\\left(\\sum_{i=0}^{K}
         \\exp(\\text{logits}_i)\\right)\\right), j = 1,...,K
-
     3) If :attr:`numeric_stable_mode` is :attr:`True`, softmax is calculated first by:
-
     .. math::
-
         max_j &= \\max_{i=0}^{K}{\\text{logits}_i}
-
         log\\_max\\_sum_j &= \\log\\sum_{i=0}^{K}\\exp(logits_i - max_j)
-
         softmax_j &= \\exp(logits_j - max_j - {log\\_max\\_sum}_j)
-
     and then cross entropy loss is calculated by softmax and label.
-
     Args:
         logits (Tensor): A multi-dimension ``Tensor`` , and the data type is float32 or float64. The input tensor of unscaled log probabilities.
         label (Tensor): The ground truth  ``Tensor`` , data type is the same
@@ -258,7 +205,6 @@ def softmax_with_cross_entropy(
         axis (int, optional): The index of dimension to perform softmax calculations. It
                               should be in range :math:`[-1, rank - 1]`, while :math:`rank`
                               is the rank of input :attr:`logits`. Default: -1.
-
     Returns:
         ``Tensor`` or Tuple of two ``Tensor`` : Return the cross entropy loss if \
                                                     `return_softmax` is False, otherwise the tuple \
@@ -266,13 +212,10 @@ def softmax_with_cross_entropy(
                                                     with input logits and cross entropy loss is in \
                                                     the same shape with input logits except shape \
                                                     in dimension :attr:`axis` as 1.
-
     Examples:
         .. code-block:: python
-
             import paddle
             import numpy as np
-
             data = np.random.rand(128).astype("float32")
             label = np.random.rand(1).astype("int64")
             data = paddle.to_tensor(data)

python/paddle/fluid/optimizer.py

@@ -1441,13 +1441,14 @@ class SGDOptimizer(Optimizer):
             import paddle.fluid as fluid
             import numpy as np
 
+            paddle.enable_static()
             place = fluid.CPUPlace()
             main = fluid.Program()
             with fluid.program_guard(main):
                 x = fluid.layers.data(name='x', shape=[13], dtype='float32')
                 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)
+                cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
                 avg_cost = fluid.layers.mean(cost)
 
                 sgd_optimizer = fluid.optimizer.SGD(learning_rate=0.001)
@@ -1642,13 +1643,14 @@ class MomentumOptimizer(Optimizer):
             import paddle.fluid as fluid
             import numpy as np
 
+            paddle.enable_static()
             place = fluid.CPUPlace()
             main = fluid.Program()
             with fluid.program_guard(main):
                 x = fluid.layers.data(name='x', shape=[13], dtype='float32')
                 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)
+                cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
                 avg_cost = fluid.layers.mean(cost)
 
                 moment_optimizer = fluid.optimizer.MomentumOptimizer(learning_rate=0.001, momentum=0.9)
@@ -2219,13 +2221,14 @@ class AdamOptimizer(Optimizer):
             import paddle
             import paddle.fluid as fluid
 
+            paddle.enable_static()
             place = fluid.CPUPlace()
             main = fluid.Program()
             with fluid.program_guard(main):
                 x = fluid.data(name='x', shape=[None, 13], dtype='float32')
                 y = fluid.data(name='y', shape=[None, 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)
+                cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
                 avg_cost = fluid.layers.mean(cost)
 
                 adam_optimizer = fluid.optimizer.AdamOptimizer(0.01)
@@ -2247,13 +2250,14 @@ class AdamOptimizer(Optimizer):
             import paddle.fluid as fluid
             import paddle.fluid.layers.learning_rate_scheduler as lr_scheduler
 
+            paddle.enable_static()
             place = fluid.CPUPlace()
             main = fluid.Program()
             with fluid.program_guard(main):
                 x = fluid.data(name='x', shape=[None, 13], dtype='float32')
                 y = fluid.data(name='y', shape=[None, 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)
+                cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
                 avg_cost = fluid.layers.mean(cost)
 
                 # define beta decay variable
@@ -3276,13 +3280,14 @@ class RMSPropOptimizer(Optimizer):
             import paddle.fluid as fluid
             import numpy as np
 
+            paddle.enable_static()
             place = fluid.CPUPlace()
             main = fluid.Program()
             with fluid.program_guard(main):
                 x = fluid.layers.data(name='x', shape=[13], dtype='float32')
                 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)
+                cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
                 avg_cost = fluid.layers.mean(cost)
 
                 rms_optimizer = fluid.optimizer.RMSProp(learning_rate=0.1)
@@ -3493,13 +3498,15 @@ class FtrlOptimizer(Optimizer):
             import paddle.fluid as fluid
             import numpy as np
 
+            paddle.enable_static()
+
             place = fluid.CPUPlace()
             main = fluid.Program()
             with fluid.program_guard(main):
                 x = fluid.layers.data(name='x', shape=[13], dtype='float32')
                 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)
+                cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
                 avg_cost = fluid.layers.mean(cost)
 
                 ftrl_optimizer = fluid.optimizer.Ftrl(learning_rate=0.1)

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

@@ -56,16 +56,20 @@ def train(use_cuda, save_dirname, is_local, use_bf16, pure_bf16):
         if not 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)
+            cost = paddle.nn.functional.square_error_cost(
+                input=y_predict, label=y
+            )
             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)
+                cost = paddle.nn.functional.square_error_cost(
+                    input=y_predict, label=y
+                )
                 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)
+        cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
         avg_cost = paddle.mean(cost)
 
     lr = 5e-3 if use_bf16 else 1e-3

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

@@ -167,7 +167,9 @@ def model():
     scale_infer = paddle.scale(x=inference, scale=5.0)
 
     label = layers.data(name='score', shape=[1], dtype='float32')
-    square_cost = layers.square_error_cost(input=scale_infer, label=label)
+    square_cost = paddle.nn.functional.square_error_cost(
+        input=scale_infer, label=label
+    )
     avg_cost = paddle.mean(square_cost)
 
     return scale_infer, avg_cost

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

@@ -69,7 +69,7 @@ class AutoCheckpointBase(unittest.TestCase):
             label = fluid.data(name='label', shape=[-1, 1], dtype='int64')
 
             fc_tmp = fluid.layers.fc(image, size=CLASS_NUM)
-            cross_entropy = fluid.layers.softmax_with_cross_entropy(
+            cross_entropy = paddle.nn.functional.softmax_with_cross_entropy(
                 fc_tmp, label
             )
             loss = paddle.mean(cross_entropy)

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

@@ -63,7 +63,7 @@ def net():
         hidden = fluid.layers.fc(input=hidden, size=400, act="sigmoid")
 
     hidden = fluid.layers.fc(input=hidden, size=3, act=None)
-    cost, y_predict = fluid.layers.softmax_with_cross_entropy(
+    cost, y_predict = paddle.nn.functional.softmax_with_cross_entropy(
         hidden, y, return_softmax=True
     )
     acc_top1 = paddle.static.accuracy(input=y_predict, label=y, k=1)

python/paddle/fluid/tests/unittests/collective/fleet/hybrid_parallel_pp_embedding.py

@@ -57,7 +57,7 @@ class SimpleNet(Layer):
         fc = fluid.layers.matmul(x_emb, self.softmax_weight)
         fc = fluid.layers.elementwise_add(fc, self.softmax_bias)
         projection = paddle.reshape(fc, shape=[-1, vocab_size])
-        loss = fluid.layers.softmax_with_cross_entropy(
+        loss = paddle.nn.functional.softmax_with_cross_entropy(
             logits=projection, label=y1, soft_label=False
         )
         return loss.mean()
@@ -106,7 +106,7 @@ class LossNet(Layer):
 
     def forward(self, args, y1):
         projection, x2 = args
-        loss = fluid.layers.softmax_with_cross_entropy(
+        loss = paddle.nn.functional.softmax_with_cross_entropy(
             logits=projection, label=y1[0], soft_label=False
         )
         return loss.mean()

python/paddle/fluid/tests/unittests/collective/fleet/hybrid_parallel_shared_weight.py

@@ -67,7 +67,7 @@ class SimpleNet(Layer):
 
         projection = paddle.matmul(projection, self.word_embeddings.weight)
 
-        loss = fluid.layers.softmax_with_cross_entropy(
+        loss = paddle.nn.functional.softmax_with_cross_entropy(
             logits=projection, label=y1, soft_label=False
         )
         return loss.mean()
@@ -120,7 +120,7 @@ class LossNet(Layer):
 
     def forward(self, args, y1):
         projection = args
-        loss = fluid.layers.softmax_with_cross_entropy(
+        loss = paddle.nn.functional.softmax_with_cross_entropy(
             logits=projection, label=y1[0], soft_label=False
         )
         return loss.mean()

python/paddle/fluid/tests/unittests/collective/fleet/parallel_dygraph_transformer.py

@@ -941,7 +941,7 @@ class TransFormer(Layer):
                 epsilon=self._label_smooth_eps,
             )
 
-        cost = fluid.layers.softmax_with_cross_entropy(
+        cost = paddle.nn.functional.softmax_with_cross_entropy(
             logits=predict,
             label=label_out,
             soft_label=True if self._label_smooth_eps else False,

python/paddle/fluid/tests/unittests/collective/fleet/test_communicator_half_async.py

@@ -33,7 +33,7 @@ class TestCommunicatorHalfAsyncEnd2End(unittest.TestCase):
         y_predict = fluid.layers.fc(input=x, size=1, act=None)
         y = fluid.layers.data(name='y', shape=[1], dtype='float32')
 
-        cost = fluid.layers.square_error_cost(input=y_predict, label=y)
+        cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
         avg_cost = paddle.mean(cost)
         return avg_cost, x, y
 

python/paddle/fluid/tests/unittests/collective/fleet/test_communicator_sync.py

@@ -29,7 +29,7 @@ class TestCommunicator(unittest.TestCase):
     def net(self):
         x = fluid.layers.data(name='x', shape=[1], dtype='float32')
         y = fluid.layers.data(name='y', shape=[1], dtype='float32')
-        cost = fluid.layers.square_error_cost(input=x, label=y)
+        cost = paddle.nn.functional.square_error_cost(input=x, label=y)
         avg_cost = paddle.mean(cost)
         return avg_cost
 

python/paddle/fluid/tests/unittests/collective/fleet/test_distributed_strategy.py

@@ -273,7 +273,7 @@ class TestDebugInfo(unittest.TestCase):
         x = fluid.layers.data(name='x', shape=[1], dtype='float32')
         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)
+        cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
         avg_cost = paddle.mean(cost)
 
         role = role_maker.UserDefinedRoleMaker(

python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_rolemaker_new.py

@@ -449,7 +449,7 @@ class TestGlooWithCloudRoleMaker(unittest.TestCase):
             x = paddle.fluid.layers.data(name='x', shape=[13], dtype='float32')
             y_predict = paddle.fluid.layers.fc(input=x, size=1, act=None)
             y = paddle.fluid.layers.data(name='y', shape=[1], dtype='float32')
-            cost = paddle.fluid.layers.square_error_cost(
+            cost = paddle.nn.functional.square_error_cost(
                 input=y_predict, label=y
             )
             avg_cost = paddle.mean(cost)

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

@@ -1585,7 +1585,7 @@ def transformer(
             epsilon=label_smooth_eps,
         )
 
-    cost = layers.softmax_with_cross_entropy(
+    cost = paddle.nn.functional.softmax_with_cross_entropy(
         logits=paddle.reshape(predict, shape=[-1, trg_vocab_size]),
         label=label,
         soft_label=True if label_smooth_eps else False,

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

@@ -410,7 +410,7 @@ class PretrainModelLayer(Layer):
         else:
             fc_out = self.out_fc(mask_trans_feat)
 
-        mask_lm_loss = fluid.layers.softmax_with_cross_entropy(
+        mask_lm_loss = paddle.nn.functional.softmax_with_cross_entropy(
             logits=fc_out, label=mask_label
         )
         mean_mask_lm_loss = paddle.mean(mask_lm_loss)
@@ -420,7 +420,7 @@ class PretrainModelLayer(Layer):
         (
             next_sent_loss,
             next_sent_softmax,
-        ) = fluid.layers.softmax_with_cross_entropy(
+        ) = paddle.nn.functional.softmax_with_cross_entropy(
             logits=next_sent_fc_out, label=labels, return_softmax=True
         )
 

python/paddle/fluid/tests/unittests/dygraph_to_static/seq2seq_dygraph_model.py

@@ -294,7 +294,7 @@ class BaseModel(fluid.dygraph.Layer):
 
         dec_output = paddle.stack(dec_output)
         dec_output = self.fc(self._transpose_batch_time(dec_output))
-        loss = fluid.layers.softmax_with_cross_entropy(
+        loss = paddle.nn.functional.softmax_with_cross_entropy(
             logits=dec_output, label=label, soft_label=False
         )
         loss = paddle.squeeze(loss, axis=[2])
@@ -828,7 +828,7 @@ class AttentionModel(fluid.dygraph.Layer):
 
         dec_output = paddle.stack(dec_output)
         dec_output = self.fc(self._transpose_batch_time(dec_output))
-        loss = fluid.layers.softmax_with_cross_entropy(
+        loss = paddle.nn.functional.softmax_with_cross_entropy(
             logits=dec_output, label=label, soft_label=False
         )
         loss = paddle.squeeze(loss, axis=[2])

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

@@ -379,7 +379,7 @@ def bmn_loss_func(
 
         weights = u_hmask + u_smmask + u_slmask
         weights.stop_gradient = True
-        loss = fluid.layers.square_error_cost(pred_score, gt_iou_map)
+        loss = paddle.nn.functional.square_error_cost(pred_score, gt_iou_map)
         loss = paddle.multiply(loss, weights)
         loss = 0.5 * paddle.sum(loss) / paddle.sum(weights)
 

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

@@ -216,7 +216,7 @@ class PtbModel(fluid.Layer):
         projection = fluid.layers.matmul(rnn_out, self.softmax_weight)
         projection = paddle.add(projection, self.softmax_bias)
 
-        loss = fluid.layers.softmax_with_cross_entropy(
+        loss = paddle.nn.functional.softmax_with_cross_entropy(
             logits=projection, label=label, soft_label=False
         )
         loss = paddle.reshape(loss, shape=[-1, self.num_steps])

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

@@ -576,7 +576,7 @@ class CrossEntropyCriterion:
                 epsilon=self.label_smooth_eps,
             )
 
-        cost = layers.softmax_with_cross_entropy(
+        cost = paddle.nn.functional.softmax_with_cross_entropy(
             logits=predict,
             label=label_out,
             soft_label=True if self.label_smooth_eps else False,

python/paddle/fluid/tests/unittests/ipu/test_dy2static_ipu.py

@@ -220,7 +220,7 @@ class TestWithoutIdentityLoss1(TestBase):
 
 class TestWithoutIdentityLoss2(TestBase):
     def set_op_attrs(self):
-        self.loss_op = paddle.fluid.layers.softmax_with_cross_entropy
+        self.loss_op = paddle.paddle.nn.functional.softmax_with_cross_entropy
 
     def set_data_feed(self):
         self.data = paddle.uniform((8, 3, 10, 10), dtype='float32')

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

@@ -103,4 +103,4 @@ def TestHuberLossOp3(TestHuberLossOp):
 
 
 if __name__ == '__main__':
-    unittest.main()
+    unittest.main()
\ No newline at end of file

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

@@ -143,7 +143,7 @@ class TestMomentumV2(unittest.TestCase):
             x = fluid.layers.data(name='x', shape=[13], dtype='float32')
             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)
+            cost =paddle.nn.functional.square_error_cost(input=y_predict, label=y)
             avg_cost = paddle.mean(cost)
 
             rms_optimizer = paddle.optimizer.Momentum(
@@ -268,7 +268,7 @@ class TestMomentumOpWithDecayAPI(unittest.TestCase):
             x = fluid.layers.data(name='x', shape=[13], dtype='float32')
             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)
+            cost =paddle.nn.functional.square_error_cost(input=y_predict, label=y)
             avg_cost = paddle.mean(cost)
 
             momentum_optimizer = paddle.fluid.contrib.optimizer.Momentum(

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

@@ -126,7 +126,7 @@ class TestPowNet(unittest.TestCase):
             fc_1 = fluid.layers.fc(input=z, size=128)
             prediction = fluid.layers.fc(input=fc_1, size=2)
 
-            cost = fluid.layers.softmax_with_cross_entropy(prediction, label)
+            cost = paddle.nn.functional.softmax_with_cross_entropy(prediction, label)
             loss = paddle.mean(cost)
             sgd = fluid.optimizer.SGD(learning_rate=0.01)
             sgd.minimize(loss)

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

@@ -111,7 +111,7 @@ class TestMomentumV2(unittest.TestCase):
             x = fluid.layers.data(name='x', shape=[13], dtype='float32')
             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)
+            cost =paddle.nn.functional.square_error_cost(input=y_predict, label=y)
             avg_cost = paddle.mean(cost)
 
             rms_optimizer = paddle.optimizer.Momentum(
@@ -239,7 +239,7 @@ class TestMomentumOpWithDecayAPI(unittest.TestCase):
             x = fluid.layers.data(name='x', shape=[13], dtype='float32')
             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)
+            cost =paddle.nn.functional.square_error_cost(input=y_predict, label=y)
             avg_cost = paddle.mean(cost)
 
             momentum_optimizer = paddle.fluid.contrib.optimizer.Momentum(

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

@@ -275,7 +275,7 @@ class TestSliceNet(unittest.TestCase):
 
             prediction = paddle.static.nn.fc(z, size=2, activation='softmax')
 
-            cost = paddle.fluid.layers.softmax_with_cross_entropy(
+            cost = paddle.paddle.nn.functional.softmax_with_cross_entropy(
                 logits=prediction, label=label
             )
             loss = paddle.mean(cost)

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

@@ -124,7 +124,7 @@ class TestPowNet(unittest.TestCase):
             fc_1 = fluid.layers.fc(input=z, size=128)
             prediction = fluid.layers.fc(input=fc_1, size=2)
 
-            cost = fluid.layers.softmax_with_cross_entropy(prediction, label)
+            cost = paddle.nn.functional.softmax_with_cross_entropy(prediction, label)
             loss = paddle.mean(cost)
             sgd = fluid.optimizer.SGD(learning_rate=0.01)
             sgd.minimize(loss)

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

@@ -68,7 +68,7 @@ class SimpleNet(fluid.Layer):
         fc = fluid.layers.matmul(x_emb, self.softmax_weight)
         fc = paddle.add(fc, self.softmax_bias)
         projection = paddle.reshape(fc, shape=[-1, self.vocab_size])
-        loss = fluid.layers.softmax_with_cross_entropy(
+        loss = paddle.nn.functional.softmax_with_cross_entropy(
             logits=projection, label=label, soft_label=False
         )
         loss = paddle.reshape(loss, shape=[-1, self.num_steps])

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

@@ -146,7 +146,9 @@ class TestAdadeltaV2(unittest.TestCase):
             x = fluid.layers.data(name='x', shape=[13], dtype='float32')
             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)
+            cost = paddle.nn.functional.square_error_cost(
+                input=y_predict, label=y
+            )
             avg_cost = paddle.mean(cost)
 
             rms_optimizer = paddle.optimizer.Adadelta(learning_rate=0.1)

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

@@ -941,7 +941,9 @@ class TestAdamOptimizer(unittest.TestCase):
             y_predict = fluid.layers.fc(
                 input=x, size=1, act=None, param_attr=weight_attr
             )
-            cost = fluid.layers.square_error_cost(input=y_predict, label=y)
+            cost = paddle.nn.functional.square_error_cost(
+                input=y_predict, label=y
+            )
             avg_cost = paddle.mean(cost)
 
             adam = fluid.optimizer.AdamOptimizer(

python/paddle/fluid/tests/unittests/test_adam_optimizer_fp32_fp64.py

@@ -33,7 +33,9 @@ def main_test_func(place, dtype):
             x = fluid.data(name='x', shape=[None, 13], dtype=dtype)
             y = fluid.data(name='y', shape=[None, 1], dtype=dtype)
             y_predict = fluid.layers.fc(input=x, size=1, act=None)
-            cost = fluid.layers.square_error_cost(input=y_predict, label=y)
+            cost = paddle.nn.functional.square_error_cost(
+                input=y_predict, label=y
+            )
             avg_cost = paddle.mean(cost)
 
             adam_optimizer = fluid.optimizer.AdamOptimizer(0.01)

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

@@ -621,7 +621,9 @@ class TestAdamWOpLayerwiseLR(TestAdamWOp):
                 fc2_b_mon1 = np.zeros((linear2.bias.shape)).astype("float32")
                 fc2_b_mon2 = np.zeros((linear2.bias.shape)).astype("float32")
 
-                cost = fluid.layers.square_error_cost(input=out, label=y)
+                cost = paddle.nn.functional.square_error_cost(
+                    input=out, label=y
+                )
                 avg_cost = paddle.mean(cost)
 
                 simple_lr_fun = partial(

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

@@ -262,7 +262,9 @@ class SimpleNet(BackwardNet):
             name='fc_no_use',
         )
         # loss
-        cost = fluid.layers.square_error_cost(input=predict, label=label)
+        cost = paddle.nn.functional.square_error_cost(
+            input=predict, label=label
+        )
         loss = paddle.mean(cost, name='mean_loss')
 
         return loss
@@ -330,7 +332,7 @@ class TestAppendBackwardWithError(unittest.TestCase):
         y = fluid.data(name='y', shape=[None, 1], dtype='float32')
         x_emb = fluid.embedding(x, size=[100, 256])
         y_predict = fluid.layers.fc(input=x_emb, size=1, name='my_fc')
-        loss = fluid.layers.square_error_cost(input=y_predict, label=y)
+        loss = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
         avg_loss = paddle.mean(loss)
         param_names = [
             param.name

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

@@ -17,6 +17,8 @@ import unittest
 import numpy as np
 from op_test import OpTest, randomize_probability
 
+import paddle
+
 
 class TestBprLossOp1(OpTest):
     """Test BprLoss with discrete one-hot labels."""
@@ -47,4 +49,5 @@ class TestBprLossOp1(OpTest):
 
 
 if __name__ == "__main__":
+    paddle.enable_static()
     unittest.main()

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

@@ -30,7 +30,7 @@ class TestCommunicator(unittest.TestCase):
         x = fluid.layers.data(name='x', shape=[1], dtype='float32')
         y = fluid.layers.data(name='y', shape=[1], dtype='float32')
 
-        cost = fluid.layers.square_error_cost(input=x, label=y)
+        cost = paddle.nn.functional.square_error_cost(input=x, label=y)
         avg_cost = paddle.mean(cost)
         return avg_cost
 

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

@@ -49,7 +49,7 @@ class TestCommunicatorGeoEnd2End(unittest.TestCase):
         y_predict = fluid.layers.fc(input=z, size=1, act=None)
         y = fluid.layers.data(name='y', shape=[1], dtype='float32')
 
-        cost = fluid.layers.square_error_cost(input=y_predict, label=y)
+        cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
         avg_cost = paddle.mean(cost)
         return avg_cost, x, x1, y
 

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

@@ -55,7 +55,7 @@ class TestCommunicator(unittest.TestCase):
         y = fluid.layers.data(name='y', shape=[1], dtype='float32')
         slots_vars = [x, y]
 
-        cost = fluid.layers.square_error_cost(input=x, label=y)
+        cost = paddle.nn.functional.square_error_cost(input=x, label=y)
         avg_cost = paddle.mean(cost)
 
         optimizer = fluid.optimizer.Adam(0.01)

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

@@ -47,7 +47,7 @@ class TestFleetGradientMergeMetaOptimizer(unittest.TestCase):
 
         x = paddle.fluid.layers.data(name='x', shape=[1], dtype='float32')
         y = paddle.fluid.layers.data(name='y', shape=[1], dtype='float32')
-        cost = paddle.fluid.layers.square_error_cost(input=x, label=y)
+        cost = paddle.nn.functional.square_error_cost(input=x, label=y)
         avg_cost = paddle.mean(cost)
 
         strategy = paddle.distributed.fleet.DistributedStrategy()
@@ -85,7 +85,7 @@ class TestFleetGradientMergeMetaOptimizer(unittest.TestCase):
 
         x = paddle.fluid.layers.data(name='x', shape=[1], dtype='float32')
         y = paddle.fluid.layers.data(name='y', shape=[1], dtype='float32')
-        cost = paddle.fluid.layers.square_error_cost(input=x, label=y)
+        cost = paddle.nn.functional.square_error_cost(input=x, label=y)
         avg_cost = paddle.mean(cost)
 
         strategy = paddle.distributed.fleet.DistributedStrategy()

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

@@ -40,7 +40,7 @@ class TestFleetGradientMergeMetaOptimizer(unittest.TestCase):
 
         x = paddle.fluid.layers.data(name='x', shape=[1], dtype='float32')
         y = paddle.fluid.layers.data(name='y', shape=[1], dtype='float32')
-        cost = paddle.fluid.layers.square_error_cost(input=x, label=y)
+        cost = paddle.nn.functional.square_error_cost(input=x, label=y)
         avg_cost = paddle.mean(cost)
 
         strategy = paddle.distributed.fleet.DistributedStrategy()

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

@@ -41,7 +41,7 @@ class TestDistStrategyTrainerDescConfig(unittest.TestCase):
 
         x = paddle.fluid.layers.data(name='x', shape=[1], dtype='float32')
         y = paddle.fluid.layers.data(name='y', shape=[1], dtype='float32')
-        cost = paddle.fluid.layers.square_error_cost(input=x, label=y)
+        cost = paddle.nn.functional.square_error_cost(input=x, label=y)
         avg_cost = paddle.mean(cost)
 
         strategy = paddle.distributed.fleet.DistributedStrategy()

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

@@ -47,7 +47,7 @@ class TranspilerTest(unittest.TestCase):
             bias_attr=fluid.ParamAttr(name='fc_b'),
         )
         y = fluid.layers.data(name='y', shape=[1], dtype='float32')
-        cost = fluid.layers.square_error_cost(input=y_predict, label=y)
+        cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
         avg_cost = paddle.mean(cost)
         sgd_optimizer = fluid.optimizer.SGD(learning_rate=0.1)
         sgd_optimizer.minimize(avg_cost)
@@ -302,7 +302,7 @@ class TestLRDecay(TranspilerTest):
             bias_attr=fluid.ParamAttr(name='fc_b'),
         )
         y = fluid.layers.data(name='y', shape=[1], dtype='float32')
-        cost = fluid.layers.square_error_cost(input=y_predict, label=y)
+        cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
         avg_cost = paddle.mean(cost)
         sgd_optimizer = fluid.optimizer.SGD(
             learning_rate=fluid.layers.exponential_decay(
@@ -471,7 +471,7 @@ class TestDecayedAdagrad(TranspilerTest):
             bias_attr=fluid.ParamAttr(name='fc_b'),
         )
         y = fluid.layers.data(name='y', shape=[1], dtype='float32')
-        cost = fluid.layers.square_error_cost(input=y_predict, label=y)
+        cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
         avg_cost = paddle.mean(cost)
         opt = fluid.optimizer.DecayedAdagrad(learning_rate=0.1)
         opt.minimize(avg_cost)
@@ -492,7 +492,7 @@ class TestFtrl(TranspilerTest):
             bias_attr=fluid.ParamAttr(name='fc_b'),
         )
         y = fluid.layers.data(name='y', shape=[1], dtype='float32')
-        cost = fluid.layers.square_error_cost(input=y_predict, label=y)
+        cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
         avg_cost = paddle.mean(cost)
         opt = fluid.optimizer.Ftrl(learning_rate=0.1)
         opt.minimize(avg_cost)
@@ -513,7 +513,7 @@ class TestLRDecayConditional(TranspilerTest):
             bias_attr=fluid.ParamAttr(name='fc_b'),
         )
         y = fluid.layers.data(name='y', shape=[1], dtype='float32')
-        cost = fluid.layers.square_error_cost(input=y_predict, label=y)
+        cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
         avg_cost = paddle.mean(cost)
         sgd_optimizer = fluid.optimizer.SGD(
             learning_rate=fluid.layers.piecewise_decay(
@@ -579,7 +579,7 @@ class TestL2Decay(TranspilerTest):
             bias_attr=fluid.ParamAttr(name='fc_b'),
         )
         y = fluid.layers.data(name='y', shape=[1], dtype='float32')
-        cost = fluid.layers.square_error_cost(input=y_predict, label=y)
+        cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
         avg_cost = paddle.mean(cost)
         sgd_optimizer = fluid.optimizer.SGD(learning_rate=0.1)
 
@@ -616,7 +616,7 @@ class TestL2DecayWithPiecewise(TranspilerTest):
             bias_attr=fluid.ParamAttr(name='fc_b'),
         )
         y = fluid.layers.data(name='y', shape=[1], dtype='float32')
-        cost = fluid.layers.square_error_cost(input=y_predict, label=y)
+        cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
         avg_cost = paddle.mean(cost)
         base_lr = 1.0
         bd = [1, 10, 20, 30]
@@ -692,7 +692,7 @@ class TestEmptyPserverOptimizeBlocks(TranspilerTest):
             bias_attr=False,
         )
         y = fluid.layers.data(name='y', shape=[1], dtype='float32')
-        cost = fluid.layers.square_error_cost(input=y_predict, label=y)
+        cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
         avg_cost = paddle.mean(cost)
         sgd_optimizer = fluid.optimizer.SGD(learning_rate=1.0)
         sgd_optimizer.minimize(avg_cost)
@@ -1134,7 +1134,7 @@ class TestRMSPropOptimizer(TranspilerTest):
             bias_attr=fluid.ParamAttr(name='fc_b'),
         )
         y = fluid.layers.data(name='y', shape=[1], dtype='float32')
-        cost = fluid.layers.square_error_cost(input=y_predict, label=y)
+        cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
         avg_cost = paddle.mean(cost)
         optimizer = fluid.optimizer.RMSProp(learning_rate=0.1)
         optimizer.minimize(avg_cost)
@@ -1167,7 +1167,7 @@ class TestLoadSliceVar(TranspilerTest):
             bias_attr=fluid.ParamAttr(name='fc_b'),
         )
         y = fluid.layers.data(name='y', shape=[1], dtype='float32')
-        cost = fluid.layers.square_error_cost(input=y_predict, label=y)
+        cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
         avg_cost = paddle.mean(cost)
         optimizer = fluid.optimizer.RMSProp(learning_rate=0.1)
         optimizer.minimize(avg_cost)
@@ -1452,6 +1452,7 @@ class TestRemoteHsigmoid(TestDistLookupTableBase):
             path_table=path_table,
             path_code=path_code,
         )
+
         avg_cost = paddle.mean(cost)
         # optimizer
         optimizer = fluid.optimizer.SGD(learning_rate=0.003)

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

@@ -58,7 +58,9 @@ class TestListenAndServOp(unittest.TestCase):
             )
             y_predict = fluid.layers.fc(input=x_emb, size=1, act=None)
             y = fluid.layers.data(name='y', shape=[1], dtype='float32')
-            cost = fluid.layers.square_error_cost(input=y_predict, label=y)
+            cost = paddle.nn.functional.square_error_cost(
+                input=y_predict, label=y
+            )
             avg_cost = paddle.mean(cost)
 
             ps_param = pslib.PSParameter()
@@ -120,7 +122,9 @@ class TestListenAndServOp(unittest.TestCase):
             )
             y_predict = fluid.layers.fc(input=x_emb, size=1, act=None)
             y = fluid.layers.data(name='y', shape=[1], dtype='float32')
-            cost = fluid.layers.square_error_cost(input=y_predict, label=y)
+            cost = paddle.nn.functional.square_error_cost(
+                input=y_predict, label=y
+            )
             avg_cost = paddle.mean(cost)
 
             ps_param = pslib.PSParameter()
@@ -180,7 +184,9 @@ class TestListenAndServOp(unittest.TestCase):
             )
             y_predict = fluid.layers.fc(input=x_emb, size=1, act=None)
             y = fluid.layers.data(name='y', shape=[1], dtype='float32')
-            cost = fluid.layers.square_error_cost(input=y_predict, label=y)
+            cost = paddle.nn.functional.square_error_cost(
+                input=y_predict, label=y
+            )
             avg_cost = paddle.mean(cost)
 
             ps_param = pslib.PSParameter()

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

@@ -463,7 +463,7 @@ def lm_model(
     projection = paddle.add(projection, softmax_bias)
     projection = paddle.reshape(projection, shape=[-1, vocab_size])
 
-    loss = layers.softmax_with_cross_entropy(
+    loss = paddle.nn.functional.softmax_with_cross_entropy(
         logits=projection, label=y, soft_label=False
     )
 

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

@@ -43,7 +43,7 @@ class TestExceptionNoCStack(unittest.TestCase):
         x = fluid.layers.data(name='X', shape=[-1, 13], dtype='float32')
         y = fluid.layers.data(name='Y', shape=[-1, 1], dtype='float32')
         predict = fluid.layers.fc(input=x, size=1, act=None)
-        loss = fluid.layers.square_error_cost(input=predict, label=y)
+        loss = paddle.nn.functional.square_error_cost(input=predict, label=y)
         avg_loss = paddle.mean(loss)
 
         fluid.optimizer.SGD(learning_rate=0.01).minimize(avg_loss)

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

@@ -25,7 +25,7 @@ class TestExecutor(unittest.TestCase):
         y = fluid.data(name="y", shape=[None, 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)
+        cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
         avg_cost = paddle.mean(cost)
 
         opt = fluid.optimizer.Adam(learning_rate=lr)

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

@@ -27,7 +27,7 @@ class TestExecutor(unittest.TestCase):
         y = fluid.data(name="y", shape=[None, 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)
+        cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
         avg_cost = paddle.mean(cost)
 
         opt = fluid.optimizer.Adam(learning_rate=lr)

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

@@ -96,7 +96,9 @@ class TestDygraphGNN(unittest.TestCase):
             logits = paddle.reshape(logits, logits.shape[1:])
             # In other example, it's nll with log_softmax. However, paddle's
             # log_loss only supports binary classification now.
-            loss = fluid.layers.softmax_with_cross_entropy(logits, labels)
+            loss = paddle.nn.functional.softmax_with_cross_entropy(
+                logits, labels
+            )
             loss = paddle.sum(loss)
 
             adam = AdamOptimizer(learning_rate=1e-3)
@@ -134,7 +136,7 @@ class TestDygraphGNN(unittest.TestCase):
             logits = paddle.reshape(logits, logits.shape[1:])
             # In other example, it's nll with log_softmax. However, paddle's
             # log_loss only supports binary classification now.
-            loss = fluid.layers.softmax_with_cross_entropy(
+            loss = paddle.nn.functional.softmax_with_cross_entropy(
                 logits, to_variable(labels)
             )
             loss = paddle.sum(loss)
@@ -162,7 +164,7 @@ class TestDygraphGNN(unittest.TestCase):
             logits2 = paddle.reshape(logits2, logits2.shape[1:])
             # In other example, it's nll with log_softmax. However, paddle's
             # log_loss only supports binary classification now.
-            loss2 = fluid.layers.softmax_with_cross_entropy(
+            loss2 = paddle.nn.functional.softmax_with_cross_entropy(
                 logits2, to_variable(labels2)
             )
             loss2 = paddle.sum(loss2)

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

@@ -69,7 +69,7 @@ class SimpleNet(fluid.Layer):
         )
         projection = paddle.add(projection, self.softmax_bias)
         projection = paddle.reshape(projection, shape=[-1, self.vocab_size])
-        loss = fluid.layers.softmax_with_cross_entropy(
+        loss = paddle.nn.functional.softmax_with_cross_entropy(
             logits=projection, label=label, soft_label=False
         )
         loss = paddle.reshape(loss, shape=[-1, self.num_steps])

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

@@ -228,7 +228,7 @@ class PtbModel(fluid.Layer):
         projection = fluid.layers.matmul(rnn_out, self.softmax_weight)
         projection = paddle.add(projection, self.softmax_bias)
         projection = paddle.reshape(projection, shape=[-1, self.vocab_size])
-        loss = fluid.layers.softmax_with_cross_entropy(
+        loss = paddle.nn.functional.softmax_with_cross_entropy(
             logits=projection, label=label, soft_label=False
         )
         loss = paddle.reshape(loss, shape=[-1, self.num_steps])

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

@@ -224,7 +224,7 @@ class PtbModel(fluid.Layer):
         projection = fluid.layers.matmul(rnn_out, self.softmax_weight)
         projection = paddle.add(projection, self.softmax_bias)
         projection = paddle.reshape(projection, shape=[-1, self.vocab_size])
-        loss = fluid.layers.softmax_with_cross_entropy(
+        loss = paddle.nn.functional.softmax_with_cross_entropy(
             logits=projection, label=label, soft_label=False
         )
         loss = paddle.reshape(loss, shape=[-1, self.num_steps])

python/paddle/fluid/tests/unittests/test_imperative_save_load_v2.py

@@ -225,7 +225,7 @@ class PtbModel(fluid.Layer):
         projection = fluid.layers.matmul(rnn_out, self.softmax_weight)
         projection = paddle.add(projection, self.softmax_bias)
         projection = paddle.reshape(projection, shape=[-1, self.vocab_size])
-        loss = fluid.layers.softmax_with_cross_entropy(
+        loss = paddle.nn.functional.softmax_with_cross_entropy(
             logits=projection, label=label, soft_label=False
         )
         loss = paddle.reshape(loss, shape=[-1, self.num_steps])

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

@@ -78,7 +78,7 @@ class SimpleNet(fluid.Layer):
             fc, paddle.transpose(self.embedding.weight, perm=[1, 0])
         )
         projection = paddle.reshape(projection, shape=[-1, self.vocab_size])
-        loss = fluid.layers.softmax_with_cross_entropy(
+        loss = paddle.nn.functional.softmax_with_cross_entropy(
             logits=projection, label=label, soft_label=False
         )
         loss = paddle.reshape(loss, shape=[-1, self.num_steps])

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

@@ -1099,7 +1099,7 @@ class TransFormer(Layer):
                 epsilon=self._label_smooth_eps,
             )
 
-        cost = fluid.layers.softmax_with_cross_entropy(
+        cost = paddle.nn.functional.softmax_with_cross_entropy(
             logits=predict,
             label=label_out,
             soft_label=True if self._label_smooth_eps else False,

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

@@ -59,7 +59,9 @@ class TestBook(unittest.TestCase):
 
             y_predict = layers.fc(input=x, size=1, act=None)
 
-            cost = layers.square_error_cost(input=y_predict, label=y)
+            cost = paddle.nn.functional.square_error_cost(
+                input=y_predict, label=y
+            )
             avg_cost = paddle.mean(cost)
 
             sgd_optimizer = optimizer.SGDOptimizer(learning_rate=0.001)
@@ -153,7 +155,9 @@ class TestSaveInferenceModel(unittest.TestCase):
 
             y_predict = layers.fc(input=x, size=1, act=None)
 
-            cost = layers.square_error_cost(input=y_predict, label=y)
+            cost = paddle.nn.functional.square_error_cost(
+                input=y_predict, label=y
+            )
             avg_cost = paddle.mean(cost)
 
         place = core.CPUPlace()
@@ -209,7 +213,9 @@ class TestInstance(unittest.TestCase):
 
             y_predict = layers.fc(input=x, size=1, act=None)
 
-            cost = layers.square_error_cost(input=y_predict, label=y)
+            cost = paddle.nn.functional.square_error_cost(
+                input=y_predict, label=y
+            )
             avg_cost = paddle.mean(cost)
 
         place = core.CPUPlace()
@@ -245,7 +251,9 @@ class TestSaveInferenceModelNew(unittest.TestCase):
 
             y_predict = layers.fc(input=x, size=1, act=None)
 
-            cost = layers.square_error_cost(input=y_predict, label=y)
+            cost = paddle.nn.functional.square_error_cost(
+                input=y_predict, label=y
+            )
             avg_cost = paddle.mean(cost)
 
             sgd_optimizer = optimizer.SGDOptimizer(learning_rate=0.001)
@@ -422,7 +430,9 @@ class TestSaveInferenceModelNew(unittest.TestCase):
 
             y_predict = layers.fc(input=x, size=1, act=None)
 
-            cost = layers.square_error_cost(input=y_predict, label=y)
+            cost = paddle.nn.functional.square_error_cost(
+                input=y_predict, label=y
+            )
             avg_cost = paddle.mean(cost)
 
             sgd_optimizer = optimizer.SGDOptimizer(learning_rate=0.001)
@@ -469,7 +479,9 @@ class TestSaveInferenceModelNew(unittest.TestCase):
 
             y_predict = layers.fc(input=x, size=1, act=None)
 
-            cost = layers.square_error_cost(input=y_predict, label=y)
+            cost = paddle.nn.functional.square_error_cost(
+                input=y_predict, label=y
+            )
             avg_cost = paddle.mean(cost)
 
             sgd_optimizer = optimizer.SGDOptimizer(learning_rate=0.001)

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

@@ -16,6 +16,7 @@ import unittest
 
 import numpy as np
 
+import paddle
 import paddle.fluid as fluid
 
 
@@ -48,7 +49,7 @@ class TestSoftmaxWithXe(unittest.TestCase):
                     dtype='int64' if not self.soft_label else self.dtype,
                     append_batch_size=False,
                 )
-                z_d, s_d = fluid.layers.softmax_with_cross_entropy(
+                z_d, s_d = paddle.nn.functional.softmax_with_cross_entropy(
                     x_d,
                     y_d,
                     soft_label=self.soft_label,

python/paddle/fluid/tests/unittests/test_lambv2_op.py

@@ -126,7 +126,9 @@ class TestLambOpWithCombinedOp(unittest.TestCase):
                 x = fluid.layers.data(name='X', shape=[13], dtype='float32')
                 y = fluid.layers.data(name='Y', shape=[1], dtype='float32')
                 prediction = fluid.layers.fc(input=x, size=1, act=None)
-                loss = fluid.layers.square_error_cost(input=prediction, label=y)
+                loss = paddle.nn.functional.square_error_cost(
+                    input=prediction, label=y
+                )
                 avg_loss = paddle.mean(loss)
             return avg_loss
 

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

@@ -3037,7 +3037,9 @@ class TestBook(LayerTest):
             x = self._get_data(name='x', shape=[13], dtype='float32')
             y_predict = layers.fc(input=x, size=1, act=None)
             y = self._get_data(name='y', shape=[1], dtype='float32')
-            cost = layers.square_error_cost(input=y_predict, label=y)
+            cost = paddle.nn.functional.square_error_cost(
+                input=y_predict, label=y
+            )
             avg_cost = paddle.mean(cost)
             return avg_cost
 
@@ -3256,23 +3258,31 @@ class TestBook(LayerTest):
         ):
             x = self._get_data(name='x', shape=[16], dtype='float32')
             y = self._get_data(name='label', shape=[1], dtype='int64')
-            loss, softmax = layers.softmax_with_cross_entropy(
+            loss, softmax = paddle.nn.functional.softmax_with_cross_entropy(
                 x, y, return_softmax=True
             )
             self.assertIsNotNone(loss)
             self.assertIsNotNone(softmax)
 
-            loss = layers.softmax_with_cross_entropy(x, y)
+            loss = paddle.nn.functional.softmax_with_cross_entropy(x, y)
             self.assertIsNotNone(loss)
 
             x1 = self._get_data(name='x1', shape=[16, 32, 64], dtype='float32')
             y1 = self._get_data(name='label1', shape=[1, 32, 64], dtype='int64')
             y2 = self._get_data(name='label2', shape=[16, 1, 64], dtype='int64')
             y3 = self._get_data(name='label3', shape=[16, 32, 1], dtype='int64')
-            loss1 = layers.softmax_with_cross_entropy(x1, y1, axis=1)
-            loss2 = layers.softmax_with_cross_entropy(x1, y2, axis=2)
-            loss3 = layers.softmax_with_cross_entropy(x1, y3, axis=3)
-            loss4 = layers.softmax_with_cross_entropy(x1, y3, axis=-1)
+            loss1 = paddle.nn.functional.softmax_with_cross_entropy(
+                x1, y1, axis=1
+            )
+            loss2 = paddle.nn.functional.softmax_with_cross_entropy(
+                x1, y2, axis=2
+            )
+            loss3 = paddle.nn.functional.softmax_with_cross_entropy(
+                x1, y3, axis=3
+            )
+            loss4 = paddle.nn.functional.softmax_with_cross_entropy(
+                x1, y3, axis=-1
+            )
             self.assertIsNotNone(loss1)
             self.assertIsNotNone(loss2)
             self.assertIsNotNone(loss3)
@@ -3694,7 +3704,7 @@ class TestBook(LayerTest):
         ):
             x = self._get_data(name="X", shape=[1], dtype="float32")
             y = self._get_data(name="Y", shape=[1], dtype="float32")
-            out = layers.square_error_cost(input=x, label=y)
+            out = paddle.nn.functional.square_error_cost(input=x, label=y)
             return out
 
     def test_dynamic_lstmp(self):

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

@@ -36,7 +36,7 @@ def run_pserver(use_cuda, sync_mode, ip, port, trainers, trainer_id):
     y = fluid.layers.data(name='y', shape=[1], dtype='float32')
 
     # loss function
-    cost = fluid.layers.square_error_cost(input=y_predict, label=y)
+    cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
     avg_cost = paddle.mean(cost)
 
     # optimizer
@@ -73,7 +73,7 @@ def run_pserver_with_empty_block(
     y = fluid.layers.data(name='y', shape=[1], dtype='float32')
 
     # loss function
-    cost = fluid.layers.square_error_cost(input=y_predict, label=y)
+    cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
     avg_cost = paddle.mean(cost)
 
     # optimizer

python/paddle/fluid/tests/unittests/test_lookup_table_v2_op.py

@@ -216,7 +216,7 @@ class TestLookupTableIsSparse(unittest.TestCase):
             )
             y = paddle.sum(emb, axis=-1)
 
-            loss = fluid.layers.square_error_cost(input=y, label=y_)
+            loss = paddle.nn.functional.square_error_cost(input=y, label=y_)
             loss = paddle.mean(loss)
 
             sgd_optimizer = fluid.optimizer.SGD(learning_rate=1e-4)

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

@@ -30,7 +30,7 @@ def train_simulator(test_batch_size=10):
     y_predict = fluid.layers.fc(input=x, size=1, act=None)
     y = fluid.layers.data(name='y', shape=[1], dtype='float32')
 
-    cost = fluid.layers.square_error_cost(input=y_predict, label=y)
+    cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
     avg_cost = paddle.mean(cost)
 
     sgd_optimizer = fluid.optimizer.SGD(learning_rate=0.001)

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

@@ -532,7 +532,9 @@ class TestMomentumV2(unittest.TestCase):
             x = fluid.layers.data(name='x', shape=[13], dtype='float32')
             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)
+            cost = paddle.nn.functional.square_error_cost(
+                input=y_predict, label=y
+            )
             avg_cost = paddle.mean(cost)
 
             rms_optimizer = paddle.optimizer.Momentum(
@@ -673,7 +675,9 @@ class TestMomentumOpWithDecayAPI(unittest.TestCase):
             x = fluid.layers.data(name='x', shape=[13], dtype='float32')
             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)
+            cost = paddle.nn.functional.square_error_cost(
+                input=y_predict, label=y
+            )
             avg_cost = paddle.mean(cost)
 
             momentum_optimizer = paddle.fluid.contrib.optimizer.Momentum(

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

@@ -33,7 +33,9 @@ class TestNetWithDtype(unittest.TestCase):
             x = fluid.layers.data(name='x', shape=[13], dtype=self.dtype)
             y = fluid.layers.data(name='y', shape=[1], dtype=self.dtype)
             y_predict = fluid.layers.fc(input=x, size=1, act=None)
-            cost = fluid.layers.square_error_cost(input=y_predict, label=y)
+            cost = paddle.nn.functional.square_error_cost(
+                input=y_predict, label=y
+            )
             avg_cost = paddle.mean(cost)
             sgd_optimizer = fluid.optimizer.SGD(learning_rate=0.001)
             sgd_optimizer.minimize(avg_cost)

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

@@ -82,7 +82,7 @@ def static(
 
         def fn_2(opt, avg_loss=None, pred=None, label=None):
             if avg_loss is None:
-                loss = layers.softmax_with_cross_entropy(
+                loss = paddle.nn.functional.softmax_with_cross_entropy(
                     logits=pred, label=label
                 )
                 avg_loss = paddle.mean(loss, name='mean_softmax_loss')
@@ -108,7 +108,7 @@ def static(
         else:
             loss_1 = layers.cross_entropy(input=prediction, label=label)
             avg_loss_1 = paddle.mean(loss_1)
-            loss_2 = layers.softmax_with_cross_entropy(
+            loss_2 = paddle.nn.functional.softmax_with_cross_entropy(
                 logits=prediction, label=label
             )
             avg_loss_2 = paddle.mean(loss_2)
@@ -193,7 +193,7 @@ def dynamic(train_data, use_cuda=False, use_parallel_exe=False):
                 loss.backward()
                 adam.minimize(loss)
             else:
-                softmax_loss = layers.softmax_with_cross_entropy(
+                softmax_loss = paddle.nn.functional.softmax_with_cross_entropy(
                     prediction, var_label
                 )
                 loss = paddle.mean(softmax_loss)

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

@@ -92,7 +92,9 @@ def cond_net(use_feed=None):
         return avg_loss
 
     def loss2(pred, label):
-        loss = fluid.layers.softmax_with_cross_entropy(logits=pred, label=label)
+        loss = paddle.nn.functional.softmax_with_cross_entropy(
+            logits=pred, label=label
+        )
         avg_loss = paddle.mean(loss, name='mean_softmax_loss')
         return avg_loss
 
@@ -119,7 +121,9 @@ def optimization_in_cond_net(with_optimize=False):
         return avg_loss
 
     def loss2(opt, pred, label, with_optimize):
-        loss = fluid.layers.softmax_with_cross_entropy(logits=pred, label=label)
+        loss = paddle.nn.functional.softmax_with_cross_entropy(
+            logits=pred, label=label
+        )
         avg_loss = paddle.mean(loss, name='mean_softmax_loss')
         if with_optimize:
             opt.minimize(avg_loss)

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

@@ -280,7 +280,9 @@ class TestRMSPropV2(unittest.TestCase):
             x = fluid.layers.data(name='x', shape=[13], dtype='float32')
             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)
+            cost = paddle.nn.functional.square_error_cost(
+                input=y_predict, label=y
+            )
             avg_cost = paddle.mean(cost)
 
             rms_optimizer = paddle.optimizer.RMSProp(learning_rate=0.1)

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

@@ -631,7 +631,7 @@ def def_seq2seq_model(
     )
 
     # loss
-    loss = layers.softmax_with_cross_entropy(
+    loss = paddle.nn.functional.softmax_with_cross_entropy(
         logits=logits, label=label, soft_label=False
     )
     loss = layers.unsqueeze(loss, axes=[2])

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

@@ -204,7 +204,7 @@ class TestSGDOpWithLargeInput(unittest.TestCase):
         emb = fluid.embedding(input=data, size=(10000000, 150), dtype='float32')
         out = fluid.layers.l2_normalize(x=emb, axis=-1)
 
-        cost = fluid.layers.square_error_cost(input=out, label=label)
+        cost = paddle.nn.functional.square_error_cost(input=out, label=label)
         avg_cost = paddle.mean(cost)
         sgd_optimizer = fluid.optimizer.SGD(learning_rate=0.001)
         sgd_optimizer.minimize(avg_cost)

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

@@ -16,6 +16,7 @@ import unittest
 
 import numpy as np
 
+import paddle
 import paddle.fluid as fluid
 import paddle.fluid.core as core
 import paddle.fluid.layers as layers
@@ -32,7 +33,9 @@ class TestSquareErrorCost(unittest.TestCase):
 
         input_var = layers.create_tensor(dtype="float32", name="input")
         label_var = layers.create_tensor(dtype="float32", name="label")
-        output = layers.square_error_cost(input=input_var, label=label_var)
+        output = paddle.nn.functional.square_error_cost(
+            input=input_var, label=label_var
+        )
 
         for use_cuda in (
             [False, True] if core.is_compiled_with_cuda() else [False]
@@ -54,14 +57,14 @@ class TestSquareErrorInvalidInput(unittest.TestCase):
         def test_invalid_input():
             input = [256, 3]
             label = fluid.data(name='label1', shape=[None, 3], dtype='float32')
-            loss = fluid.layers.square_error_cost(input, label)
+            loss = paddle.nn.functional.square_error_cost(input, label)
 
         self.assertRaises(TypeError, test_invalid_input)
 
         def test_invalid_label():
             input = fluid.data(name='input2', shape=[None, 3], dtype='float32')
             label = [256, 3]
-            loss = fluid.layers.square_error_cost(input, label)
+            loss = paddle.nn.functional.square_error_cost(input, label)
 
         self.assertRaises(TypeError, test_invalid_label)
 

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

@@ -237,7 +237,7 @@ class PtbModel(fluid.Layer):
         projection = fluid.layers.matmul(rnn_out, self.softmax_weight)
         projection = paddle.add(projection, self.softmax_bias)
         projection = paddle.reshape(projection, shape=[-1, self.vocab_size])
-        loss = fluid.layers.softmax_with_cross_entropy(
+        loss = paddle.nn.functional.softmax_with_cross_entropy(
             logits=projection, label=label, soft_label=False
         )
         loss = paddle.reshape(loss, shape=[-1, self.num_steps])

python/paddle/fluid/tests/unittests/xpu/test_adadelta_op_xpu.py

@@ -171,7 +171,9 @@ class XPUTestAdadelta(XPUOpTestWrapper):
                 x = fluid.layers.data(name='x', shape=[13], dtype=self.dtype)
                 y = fluid.layers.data(name='y', shape=[1], dtype=self.dtype)
                 y_predict = fluid.layers.fc(input=x, size=1, act=None)
-                cost = fluid.layers.square_error_cost(input=y_predict, label=y)
+                cost = paddle.nn.functional.square_error_cost(
+                    input=y_predict, label=y
+                )
                 avg_cost = paddle.mean(cost)
 
                 rms_optimizer = paddle.optimizer.Adadelta(learning_rate=0.1)

python/paddle/fluid/tests/unittests/xpu/test_sgd_op_xpu.py

@@ -75,7 +75,7 @@ class TestSGDOpWithLargeInput(unittest.TestCase):
         emb = fluid.embedding(input=data, size=(10000, 150), dtype='float32')
         out = fluid.layers.l2_normalize(x=emb, axis=-1)
 
-        cost = fluid.layers.square_error_cost(input=out, label=label)
+        cost = paddle.nn.functional.square_error_cost(input=out, label=label)
         avg_cost = paddle.mean(cost)
         sgd_optimizer = fluid.optimizer.SGD(learning_rate=0.001)
         sgd_optimizer.minimize(avg_cost)

python/paddle/fluid/transpiler/distribute_transpiler.py

@@ -283,11 +283,16 @@ class DistributeTranspiler:
     Examples:
         .. code-block:: python
 
+            import paddle
+            import paddle.fluid as fluid
+
+            paddle.enable_static()
+
             x = fluid.data(name='x', shape=[13], dtype='float32')
             y = fluid.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)
+            cost =paddle.nn.functional.square_error_cost(input=y_predict, label=y)
             avg_loss = fluid.layers.mean(cost)
 
             sgd_optimizer = fluid.optimizer.SGD(learning_rate=0.001)