diff --git a/python/paddle/fluid/tests/unittests/op_test.py b/python/paddle/fluid/tests/unittests/op_test.py
index cc2a54b95e224f31714c326eb7b5c434fbdab155..a54c8998ef06cedd3e76d247cc0e2e95eae2f46b 100644
--- a/python/paddle/fluid/tests/unittests/op_test.py
+++ b/python/paddle/fluid/tests/unittests/op_test.py
@@ -33,7 +33,7 @@ from paddle.fluid.executor import Executor
 from paddle.fluid.framework import Program, OpProtoHolder, Variable
 from testsuite import create_op, set_input, append_input_output, append_loss_ops
 from paddle.fluid import unique_name
-from white_list import op_accuracy_white_list, op_check_grad_white_list, check_shape_white_list
+from white_list import op_accuracy_white_list, op_check_grad_white_list, check_shape_white_list, compile_vs_runtime_white_list
 
 
 def _set_use_system_allocator(value=None):
@@ -1102,7 +1102,7 @@ class OpTest(unittest.TestCase):
                      equal_nan=False,
                      check_dygraph=True,
                      inplace_atol=None,
-                     check_compile_vs_runtime=False):
+                     check_compile_vs_runtime=True):
         self.__class__.op_type = self.op_type
         places = self._get_places()
         for place in places:
@@ -1112,7 +1112,9 @@ class OpTest(unittest.TestCase):
                 outs, dygraph_outs, fetch_list = res
             else:
                 outs, fetch_list = res
-            if check_compile_vs_runtime:
+            if check_compile_vs_runtime and (
+                    self.op_type not in
+                    compile_vs_runtime_white_list.COMPILE_RUN_OP_WHITE_LIST):
                 self.check_compile_vs_runtime(fetch_list, outs)
 
     def check_output_customized(self, checker):
diff --git a/python/paddle/fluid/tests/unittests/white_list/compile_vs_runtime_white_list.py b/python/paddle/fluid/tests/unittests/white_list/compile_vs_runtime_white_list.py
new file mode 100644
index 0000000000000000000000000000000000000000..c7184633f1dba80937e443d48745ddcdcf2a3e0b
--- /dev/null
+++ b/python/paddle/fluid/tests/unittests/white_list/compile_vs_runtime_white_list.py
@@ -0,0 +1,28 @@
+#   Copyright (c) 2018 PaddlePaddle Authors. All Rights Reserve
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+# If the output after infershape() is a lod_tensor, commenly its lod_level
+# should be equal during compile time and run time.
+# For ops in this whitelist, the equality check of lod_level between
+# compiletime&runtime will not be required.
+COMPILE_RUN_OP_WHITE_LIST = [
+    'lod_reset', 'sequence_pool', 'sequence_slice', 'generate_mask_labels',
+    'sequence_reshape', 'generate_proposals', 'mine_hard_examples',
+    'retinanet_detection_output', 'ctc_align', 'fusion_seqpool_cvm_concat',
+    'gru', 'sequence_erase', 'rpn_target_assign', 'retinanet_target_assign',
+    'filter_by_instag', 'fusion_seqpool_concat', 'multiclass_nms',
+    'multiclass_nms2', 'im2sequence', 'generate_proposal_labels',
+    'distribute_fpn_proposals', 'detection_map', 'locality_aware_nms',
+    'var_conv_2d'
+]
diff --git a/tools/check_api_approvals.sh b/tools/check_api_approvals.sh
index faedc75cc7608d91c16f841e59d13bc6ce3fe329..03d9335564520cb9217de2e5bc6d38852c2c4d69 100644
--- a/tools/check_api_approvals.sh
+++ b/tools/check_api_approvals.sh
@@ -168,6 +168,13 @@ if [ "${HAS_INPLACE_TESTS}" != "" ] && [ "${GIT_PR_ID}" != "" ]; then
     check_approval 1 46782768 47554610 43953930 6836917
 fi
 
+HAS_COMPILERUNTIME_NOT_TEST=`git diff -U0 --diff-filter=A upstream/$BRANCH |grep "+" |grep -E "check_output[(]*check_compile_vs_runtime=False" || true`
+if [ "${HAS_COMPILERUNTIME_NOT_TEST}" != "" ] && [ "${GIT_PR_ID}" != "" ]; then
+    echo_line="If the operator's output after infershape() is a LodTensor, the output's Lod-Level during compile-time and runtime must be equal. Please set check_compile_vs_runtime=True in op_test.check_output function to enable compile&runtime lod-level check.\n
+If you do not need check_compile_vs_runtime, you must have one RD (lanxianghit, phlrain, luotao1) approval.\nThe corresponding lines are as follows:\n${HAS_COMPILERUNTIME_NOT_TEST}\n"
+    check_approval 1 47554610 43953930 6836917
+fi
+
 OP_FILE_CHANGED=`git diff --name-only --diff-filter=AMR upstream/$BRANCH |grep -oE ".+_op..*" || true`
 if [ "${OP_FILE_CHANGED}" != "" ] && [ "${GIT_PR_ID}" != "" ]; then
     for OP_FILE in ${OP_FILE_CHANGED};