clear fluid api: kldiv_loss, kldiv_loss, mse_loss (#48147)

python/paddle/fluid/layers/loss.py

@@ -41,9 +41,6 @@ __all__ = [
     'nce',
     'softmax_with_cross_entropy',
     'sigmoid_cross_entropy_with_logits',
-    'kldiv_loss',
-    'npair_loss',
-    'mse_loss',
 ]
 
 kIgnoreIndex = -100
@@ -826,142 +823,3 @@ def sigmoid_cross_entropy_with_logits(
         outputs={"Out": out},
     )
     return out
-
-
-@deprecated(since="2.0.0", update_to="paddle.nn.functional.kl_div")
-@templatedoc()
-def kldiv_loss(x, target, reduction='mean', name=None):
-    """
-
-    ${comment}
-
-    Args:
-        x (Tensor): ${x_comment}
-        target (Tensor): ${target_comment}
-        reduction (Tensor): ${reduction_comment}
-        name(str, optional): For detailed information, please refer
-                             to :ref:`api_guide_Name`. Usually name is no need to set and
-                             None by default.
-
-    Returns:
-        Tensor: The KL divergence loss. The data type is same as input tensor
-
-    Examples:
-        .. code-block:: python
-
-            import paddle
-            import paddle.fluid as fluid
-
-            x = paddle.rand(shape=[3,4,2,2], dtype='float32')
-            target = paddle.rand(shape=[3,4,2,2], dtype='float32')
-
-            # 'batchmean' reduction, loss shape will be [1]
-            loss = fluid.layers.kldiv_loss(x=x, target=target, reduction='batchmean')
-            print(loss.shape) # shape=[1]
-
-            # 'mean' reduction, loss shape will be [1]
-            loss = fluid.layers.kldiv_loss(x=x, target=target, reduction='mean')
-            print(loss.shape) # shape=[1]
-
-            # 'sum' reduction, loss shape will be [1]
-            loss = fluid.layers.kldiv_loss(x=x, target=target, reduction='sum')
-            print(loss.shape) # shape=[1]
-
-            # 'none' reduction, loss shape is same with X shape
-            loss = fluid.layers.kldiv_loss(x=x, target=target, reduction='none')
-            print(loss.shape) # shape=[3, 4, 2, 2]
-
-    """
-    helper = LayerHelper('kldiv_loss', **locals())
-    check_variable_and_dtype(x, 'x', ['float32', 'float64'], 'kldiv_loss')
-    check_variable_and_dtype(
-        target, 'target', ['float32', 'float64'], 'kldiv_loss'
-    )
-    check_type(reduction, 'reduction', str, 'kldiv_loss')
-    loss = helper.create_variable_for_type_inference(dtype=x.dtype)
-    helper.append_op(
-        type='kldiv_loss',
-        inputs={'X': x, 'Target': target},
-        outputs={'Loss': loss},
-        attrs={'reduction': reduction},
-    )
-    return loss
-
-
-from .control_flow import equal
-
-
-def npair_loss(anchor, positive, labels, l2_reg=0.002):
-    """
-
-    Npair loss requires paired data. Npair loss has two parts: the first part is L2
-    regularizer on the embedding vector; the second part is cross entropy loss which
-    takes the similarity matrix of anchor and positive as logits.
-
-    For more information, please refer to:
-    `Improved Deep Metric Learning with Multi class N pair Loss Objective <http://www.nec-labs.com/uploads/images/Department-Images/MediaAnalytics/papers/nips16_npairmetriclearning.pdf>`_
-
-    Args:
-      anchor(Tensor): embedding vector for the anchor image. shape=[batch_size, embedding_dims],
-                        the data type is float32 or float64.
-      positive(Tensor): embedding vector for the positive image. shape=[batch_size, embedding_dims],
-                        the data type is float32 or float64.
-      labels(Tensor): 1-D tensor. shape=[batch_size], the data type is float32 or float64 or int64.
-      l2_reg(float32): L2 regularization term on embedding vector, default: 0.002.
-
-
-    Returns:
-      A Tensor representing the npair loss, the data type is the same as anchor, the shape is [1].
-
-    Examples:
-
-      .. code-block:: python
-
-          import paddle
-
-          DATATYPE = "float32"
-
-          anchor = paddle.rand(shape=(18, 6), dtype=DATATYPE)
-          positive = paddle.rand(shape=(18, 6), dtype=DATATYPE)
-          labels = paddle.rand(shape=(18,), dtype=DATATYPE)
-
-          npair_loss = paddle.nn.functional.npair_loss(anchor, positive, labels, l2_reg = 0.002)
-          print(npair_loss)
-
-    """
-    return paddle.nn.functional.npair_loss(anchor, positive, labels, l2_reg)
-
-
-def mse_loss(input, label):
-    """
-
-    This op accepts input predications and target label and returns the mean square error.
-
-    The loss can be described as:
-
-    .. math::
-
-        Out = MEAN((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 mean square error difference of input and label.
-
-    Return type: Tensor.
-
-    Examples:
-        .. code-block:: python
-
-            import paddle
-            input = paddle.to_tensor([1.1, 1.9])
-            label = paddle.to_tensor([1.0, 2.0])
-            output = paddle.fluid.layers.mse_loss(input, label)
-            print(output.numpy())
-            # [0.01]
-    """
-    check_variable_and_dtype(input, "input", ['float32', 'float64'], 'mse_loss')
-    check_variable_and_dtype(label, "label", ['float32', 'float64'], 'mse_loss')
-    return nn.reduce_mean(square_error_cost(input, label))

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

@@ -238,7 +238,7 @@ class TestWithoutIdentityLoss2(TestBase):
 
 class TestWithoutIdentityLoss3(TestBase):
     def set_op_attrs(self):
-        self.loss_op = partial(paddle.fluid.layers.kldiv_loss, reduction="none")
+        self.loss_op = partial(paddle.nn.functional.kl_div, reduction="none")
 
     def set_data_feed(self):
         self.data = paddle.uniform((8, 3, 10, 10), dtype='float32')

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

@@ -57,7 +57,7 @@ class TestBase(IPUOpTest):
         target = paddle.static.data(
             name=self.feed_list[1], shape=self.feed_shape[1], dtype='float32'
         )
-        out = paddle.fluid.layers.kldiv_loss(x, target, **self.attrs)
+        out = paddle.nn.functional.kl_div(x, target, **self.attrs)
         self.fetch_list = [out.name]
 
     def run_model(self, exec_mode):

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

@@ -3731,7 +3731,9 @@ class TestBook(LayerTest):
                 dtype="float32",
                 append_batch_size=False,
             )
-            loss = layers.kldiv_loss(x=x, target=target, reduction='batchmean')
+            loss = paddle.nn.functional.kl_div(
+                input=x, label=target, reduction='batchmean'
+            )
             return loss
 
     def make_temporal_shift(self):
@@ -3773,7 +3775,7 @@ 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.mse_loss(input=x, label=y)
+            out = paddle.nn.functional.mse_loss(input=x, label=y)
             return out
 
     def make_square_error_cost(self):

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

@@ -17,7 +17,6 @@ import numpy as np
 import paddle
 import paddle.fluid.core as core
 import paddle.fluid as fluid
-import paddle.fluid.layers as layers
 from paddle.fluid.executor import Executor
 
 
@@ -32,7 +31,7 @@ class TestMseLoss(unittest.TestCase):
         input_var = fluid.data(name="input", shape=[-1, 3], dtype="float32")
         label_var = fluid.data(name="label", shape=[-1, 3], dtype="float32")
 
-        output = layers.mse_loss(input=input_var, label=label_var)
+        output = paddle.nn.functional.mse_loss(input=input_var, label=label_var)
         for use_cuda in (
             [False, True] if core.is_compiled_with_cuda() else [False]
         ):
@@ -52,14 +51,14 @@ class TestMseInvalidInput(unittest.TestCase):
         def test_invalid_input():
             input = [256, 3]
             label = fluid.data(name='label1', shape=[None, 3], dtype='float32')
-            loss = fluid.layers.mse_loss(input, label)
+            loss = paddle.nn.functional.mse_loss(input, label)
 
         self.assertRaises(TypeError, test_invalid_input)
 
         def test_invalid_label():
             input = fluid.data(name='input1', shape=[None, 3], dtype='float32')
             label = [256, 3]
-            loss = fluid.layers.mse_loss(input, label)
+            loss = paddle.nn.functional.mse_loss(input, label)
 
         self.assertRaises(TypeError, test_invalid_label)
 

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

@@ -14,6 +14,7 @@
 
 import unittest
 import paddle.fluid as fluid
+import paddle
 import paddle.fluid.core as core
 import numpy as np
 from paddle.fluid import Program, program_guard
@@ -99,7 +100,7 @@ class TestNpairLossOp(unittest.TestCase):
             append_batch_size=False,
         )
 
-        npair_loss_op = fluid.layers.npair_loss(
+        npair_loss_op = paddle.nn.functional.npair_loss(
             anchor=anc, positive=pos, labels=lab, l2_reg=reg_lambda
         )
         out_tensor = exe.run(
@@ -140,19 +141,19 @@ class TestNpairLossOpError(unittest.TestCase):
 
             def test_anchor_Variable():
                 # the anchor type must be Variable
-                fluid.layers.npair_loss(
+                paddle.nn.functional.npair_loss(
                     anchor=anchor_np, positive=positive_data, labels=labels_data
                 )
 
             def test_positive_Variable():
                 # the positive type must be Variable
-                fluid.layers.npair_loss(
+                paddle.nn.functional.npair_loss(
                     anchor=anchor_data, positive=positive_np, labels=labels_data
                 )
 
             def test_labels_Variable():
                 # the labels type must be Variable
-                fluid.layers.npair_loss(
+                paddle.nn.functional.npair_loss(
                     anchor=anchor_data, positive=positive_data, labels=labels_np
                 )
 
@@ -165,7 +166,7 @@ class TestNpairLossOpError(unittest.TestCase):
                 anchor_data1 = fluid.data(
                     name='anchor1', shape=[2, 4], dtype='int32'
                 )
-                fluid.layers.npair_loss(
+                paddle.nn.functional.npair_loss(
                     anchor=anchor_data, positive=positive_data, labels=labels_np
                 )
 
@@ -174,7 +175,7 @@ class TestNpairLossOpError(unittest.TestCase):
                 positive_data1 = fluid.data(
                     name='positive1', shape=[2, 4], dtype='int32'
                 )
-                fluid.layers.npair_loss(
+                paddle.nn.functional.npair_loss(
                     anchor=anchor_data,
                     positive=positive_data1,
                     labels=labels_np,
@@ -185,7 +186,7 @@ class TestNpairLossOpError(unittest.TestCase):
                 labels_data1 = fluid.data(
                     name='labels1', shape=[2], dtype='int32'
                 )
-                fluid.layers.npair_loss(
+                paddle.nn.functional.npair_loss(
                     anchor=anchor_data,
                     positive=positive_data,
                     labels=labels_data1,