remove mkl & fix commit

paddle/fluid/operators/data_norm_op.cc

@@ -15,9 +15,6 @@ limitations under the License. */
 #include "paddle/fluid/operators/data_norm_op.h"
 #include <string>
 #include "paddle/fluid/framework/data_layout.h"
-#ifdef PADDLE_WITH_MKLDNN
-#include "paddle/fluid/platform/mkldnn_helper.h"
-#endif
 
 namespace paddle {
 namespace operators {
@@ -97,13 +94,6 @@ class DataNormOp : public framework::OperatorWithKernel {
     // TODO(pzelazko-intel): enable MKLDNN layout when it's ready
     framework::LibraryType library = framework::LibraryType::kPlain;
     framework::DataLayout layout = framework::DataLayout::kAnyLayout;
-#ifdef PADDLE_WITH_MKLDNN
-    if (library == framework::LibraryType::kPlain &&
-        platform::CanMKLDNNBeUsed(ctx)) {
-      library = framework::LibraryType::kMKLDNN;
-      layout = framework::DataLayout::kMKLDNN;
-    }
-#endif
 
     return framework::OpKernelType(input_data_type, ctx.GetPlace(), layout,
                                    library);
@@ -140,9 +130,6 @@ class DataNormOpMaker : public framework::OpProtoAndCheckerMaker {
               "Scales of the history data batch, "
               "will apply to output when training")
         .AsIntermediate();
-    AddAttr<bool>("use_mkldnn",
-                  "(bool, default false) Only used in mkldnn kernel")
-        .SetDefault(false);
     AddComment(R"DOC(
 Data Normalization.
 
@@ -264,14 +251,6 @@ class DataNormGradOp : public framework::OperatorWithKernel {
     framework::LibraryType library = framework::LibraryType::kPlain;
     framework::DataLayout layout = framework::DataLayout::kAnyLayout;
 
-#ifdef PADDLE_WITH_MKLDNN
-    if (library == framework::LibraryType::kPlain &&
-        platform::CanMKLDNNBeUsed(ctx)) {
-      library = framework::LibraryType::kMKLDNN;
-      layout = framework::DataLayout::kMKLDNN;
-    }
-#endif
-
     return framework::OpKernelType(ctx.Input<Tensor>("X")->type(),
                                    ctx.GetPlace(), layout, library);
   }

paddle/fluid/operators/teacher_student_sigmoid_loss_op.cc

@@ -117,11 +117,11 @@ class TeacherStudentSigmoidLossOpMaker
               "[N x 1]. The teacher student sigmoid loss.");
     AddAttr<float>(
         "soft_max_up_bound",
-        "fp32, if input > soft_max_up_bound, will be bound, default 15.0")
+        "fp32, if input > soft_max_up_bound, input will be bound, default 15.0")
         .SetDefault(15.0);
-    AddAttr<float>(
+    AddAttr<float>("soft_max_lower_bound",
-        "soft_max_lower_bound",
+                   "fp32, if input < soft_max_lower_bound, input will be "
-        "fp32, if input < soft_max_lower_bound, will be bound, default -15.0")
+                   "bound, default -15.0")
         .SetDefault(-15.0);
     AddComment(R"DOC(
 TeacherStudentSigmoidLoss Operator.

python/paddle/fluid/layers/nn.py

@@ -2944,7 +2944,6 @@ def data_norm(input,
               param_attr=None,
               data_layout='NCHW',
               in_place=False,
-              use_mkldnn=False,
               name=None,
               moving_mean_name=None,
               moving_variance_name=None,
@@ -2978,7 +2977,6 @@ def data_norm(input,
         param_attr(ParamAttr): The parameter attribute for Parameter `scale`.
         data_layout(string, default NCHW): NCHW|NHWC
         in_place(bool, Default False): Make the input and output of batch norm reuse memory.
-        use_mkldnn(bool, Default false): ${use_mkldnn_comment}
         name(string, Default None): A name for this layer(optional). If set None, the layer
             will be named automatically.
         moving_mean_name(string, Default None): The name of moving_mean which store the global Mean.
@@ -3059,8 +3057,7 @@ def data_norm(input,
         outputs={"Y": data_norm_out,
                  "Means": means,
                  "Scales": scales},
-        attrs={"epsilon": epsilon,
+        attrs={"epsilon": epsilon})
-               "use_mkldnn": use_mkldnn})
 
     return helper.append_activation(data_norm_out)
 
@@ -9491,6 +9488,7 @@ def teacher_student_sigmoid_loss(input,
 
     Examples:
         .. code-block:: python
+
           cost = fluid.layers.teacher_student_sigmoid_loss(input=similarity, label=label)
     """
     helper = LayerHelper('teacher_student_sigmoid_loss', **locals())