[Clean Fluid API]Remove API: label_smooth (#47952)

* Remove API: label_smooth (replace with paddle.nn.functional.label_smooth)
	Replace the paddle.fluid.layers.label_smooth with the
paddle.nn.functional.label_smooth

* modify the call of label_smooth from old style to new style

python/paddle/fluid/layers/nn.py

@@ -103,7 +103,6 @@ __all__ = [
     'lod_reset',
     'lod_append',
     'pad',
-    'label_smooth',
     'roi_pool',
     'roi_align',
     'image_resize',
@@ -5843,95 +5842,6 @@ def pad(x, paddings, pad_value=0.0, name=None):
     return out
 
 
-def label_smooth(
-    label, prior_dist=None, epsilon=0.1, dtype="float32", name=None
-):
-    r"""
-    :alias_main: paddle.nn.functional.label_smooth
-        :alias: paddle.nn.functional.label_smooth,paddle.nn.functional.common.label_smooth
-        :old_api: paddle.fluid.layers.label_smooth
-
-    Label smoothing is a mechanism to regularize the classifier layer and is called
-    label-smoothing regularization (LSR).
-
-    Label smoothing is proposed to encourage the model to be less confident,
-    since optimizing the log-likelihood of the correct label directly may
-    cause overfitting and reduce the ability of the model to adapt. Label
-    smoothing replaces the ground-truth label :math:`y` with the weighted sum
-    of itself and some fixed distribution :math:`\mu`. For class :math:`k`,
-    i.e.
-
-    .. math::
-
-        \\tilde{y_k} = (1 - \epsilon) * y_k + \epsilon * \mu_k,
-
-    where :math:`1 - \epsilon` and :math:`\epsilon` are the weights
-    respectively, and :math:`\\tilde{y}_k` is the smoothed label. Usually
-    uniform distribution is used for :math:`\mu`.
-
-    See more details about label smoothing in https://arxiv.org/abs/1512.00567.
-
-    Parameters:
-        label(Variable): The input variable containing the label data. The
-                        label data should use one-hot representation. It's
-                        a multidimensional tensor with a shape of
-                        :math:`[N_1, ..., Depth]`, where Depth is class number. The dtype can be "float32" and "float64".
-        prior_dist(Variable, optional): The prior distribution to be used to smooth
-                        labels. If not provided, an uniform distribution
-                        is used. It's a multidimensional tensor with a shape of
-                        :math:`[1, class\_num]` . The default value is None.
-        epsilon(float, optional): The weight used to mix up the original ground-truth
-                        distribution and the fixed distribution. The default value is
-                        0.1.
-        dtype(np.dtype|core.VarDesc.VarType|str, optional): The data type can be set
-                        as 'float32', 'float64'. The default value is 'float32'.
-        name(str, optional): The default value is None. Normally there is no need for user
-                        to set this property. For more information, please refer to
-                        :ref:`api_guide_Name`.
-
-    Returns:
-        Variable: The tensor variable containing the smoothed labels.
-
-    Examples:
-        .. code-block:: python
-
-            import paddle.fluid as fluid
-            import paddle.fluid.layers as layers
-
-            label = layers.data(name="label", shape=[1], dtype="int32")
-            one_hot_label = layers.one_hot(input=label, depth=10)
-            smooth_label = layers.label_smooth(
-                label=one_hot_label, epsilon=0.1, dtype="float32")
-    """
-    if in_dygraph_mode():
-        return _C_ops.label_smooth(label, prior_dist, float(epsilon))
-
-    if epsilon > 1.0 or epsilon < 0.0:
-        raise ValueError("The value of epsilon must be between 0 and 1.")
-
-    if _non_static_mode():
-        return _legacy_C_ops.label_smooth(
-            label, prior_dist, 'epsilon', float(epsilon)
-        )
-
-    check_variable_and_dtype(
-        label, 'label', ['float32', 'float64'], 'label_smooth'
-    )
-
-    helper = LayerHelper("label_smooth", **locals())
-    label.stop_gradient = True
-    smooth_label = helper.create_variable_for_type_inference(dtype)
-    helper.append_op(
-        type="label_smooth",
-        inputs={"X": label, "PriorDist": prior_dist}
-        if prior_dist
-        else {"X": label},
-        outputs={"Out": smooth_label},
-        attrs={"epsilon": float(epsilon)},
-    )
-    return smooth_label
-
-
 @templatedoc()
 def roi_pool(
     input,

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

@@ -26,6 +26,7 @@ from paddle.fluid.dygraph import (
 from paddle.optimizer.lr import NoamDecay
 
 from test_dist_base import runtime_main, TestParallelDyGraphRunnerBase
+import paddle.nn.functional as F
 
 """
 Note(chenweihang): To compare loss of single-card and multi-card
@@ -934,7 +935,7 @@ class TransFormer(Layer):
         enc_output = self._wrap_encoder_layer(enc_inputs)
         predict = self._wrap_decoder_layer(dec_inputs, enc_output)
         if self._label_smooth_eps:
-            label_out = fluid.layers.label_smooth(
+            label_out = F.label_smooth(
                 label=fluid.layers.one_hot(
                     input=label, depth=self._trg_vocab_size
                 ),

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

@@ -27,6 +27,7 @@ import paddle
 import paddle.fluid as fluid
 import paddle.fluid.layers as layers
 from test_dist_base import TestDistRunnerBase, runtime_main, RUN_STEP
+import paddle.nn.functional as F
 import paddle
 
 const_para_attr = fluid.ParamAttr(initializer=fluid.initializer.Constant(0.001))
@@ -1580,7 +1581,7 @@ def transformer(
     # cancel padding index in calculating the loss.
     label, weights = make_all_inputs(label_data_input_fields)
     if label_smooth_eps:
-        label = layers.label_smooth(
+        label = F.label_smooth(
             label=layers.one_hot(input=label, depth=trg_vocab_size),
             epsilon=label_smooth_eps,
         )

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

@@ -26,6 +26,7 @@ from paddle.fluid.dygraph import (
 from paddle.fluid.dygraph.jit import dygraph_to_static_func
 from paddle.fluid.layers.utils import map_structure
 import paddle
+import paddle.nn.functional as F
 
 
 def position_encoding_init(n_position, d_pos_vec):
@@ -574,7 +575,7 @@ class CrossEntropyCriterion:
 
     def __call__(self, predict, label, weights):
         if self.label_smooth_eps:
-            label_out = layers.label_smooth(
+            label_out = F.label_smooth(
                 label=layers.one_hot(input=label, depth=predict.shape[-1]),
                 epsilon=self.label_smooth_eps,
             )

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

@@ -22,6 +22,7 @@ from test_imperative_base import new_program_scope
 from paddle.fluid.framework import _in_legacy_dygraph, _test_eager_guard
 from paddle.fluid import core
 import numpy as np
+import paddle.nn.functional as F
 
 np.set_printoptions(suppress=True)
 
@@ -1088,7 +1089,7 @@ class TransFormer(Layer):
         enc_output = self._wrap_encoder_layer(enc_inputs)
         predict = self._wrap_decoder_layer(dec_inputs, enc_output)
         if self._label_smooth_eps:
-            label_out = fluid.layers.label_smooth(
+            label_out = F.label_smooth(
                 label=fluid.layers.one_hot(
                     input=label, depth=self._trg_vocab_size
                 ),

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

@@ -48,11 +48,10 @@ class LabelSmoothTestCase(unittest.TestCase):
                 label_var = fluid.data(
                     "input", self.label_shape, dtype=self.dtype
                 )
-                y_var = fluid.layers.label_smooth(
+                y_var = F.label_smooth(
                     label_var,
                     prior_dist=self.prior_dist,
                     epsilon=self.epsilon,
-                    dtype=self.dtype,
                 )
         feed_dict = {"input": self.label}
         exe = fluid.Executor(place)

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

@@ -33,6 +33,7 @@ from paddle.fluid.dygraph import nn
 from paddle.fluid.dygraph import base
 from paddle.fluid.dygraph import to_variable
 from paddle.fluid.framework import _test_eager_guard
+import paddle.nn.functional as F
 
 
 class LayerTest(unittest.TestCase):
@@ -3359,9 +3360,7 @@ class TestBook(LayerTest):
         with fluid.framework._dygraph_place_guard(place=fluid.CPUPlace()):
             label = self._get_data(name="label", shape=[1], dtype="int32")
             one_hot_label = layers.one_hot(input=label, depth=10)
-            smooth_label = layers.label_smooth(
-                label=one_hot_label, epsilon=0.1, dtype="int32"
-            )
+            smooth_label = F.label_smooth(label=one_hot_label, epsilon=0.1)
             return smooth_label
 
     def make_topk(self):