diff --git a/paddle/fluid/operators/lstm_cudnn_op.cc b/paddle/fluid/operators/cudnn_lstm_op.cc
similarity index 98%
rename from paddle/fluid/operators/lstm_cudnn_op.cc
rename to paddle/fluid/operators/cudnn_lstm_op.cc
index c9a4a31738190e02139e9a9200961c199a19942e..86632fc9fb4f1488271c9823b01613ce865fa696 100644
--- a/paddle/fluid/operators/lstm_cudnn_op.cc
+++ b/paddle/fluid/operators/cudnn_lstm_op.cc
@@ -12,7 +12,7 @@ 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. */
 
-#include "paddle/fluid/operators/lstm_cudnn_op.h"
+#include "paddle/fluid/operators/cudnn_lstm_op.h"
 #include <string>
 
 #ifdef PADDLE_WITH_CUDA
@@ -205,12 +205,12 @@ class CudnnLSTMGradOp : public framework::OperatorWithKernel {
 }  // namespace paddle
 
 namespace ops = paddle::operators;
-REGISTER_OPERATOR(lstm_cudnn, ops::CudnnLSTMOp, ops::CudnnLSTMOpMaker,
+REGISTER_OPERATOR(cudnn_lstm, ops::CudnnLSTMOp, ops::CudnnLSTMOpMaker,
                   paddle::framework::DefaultGradOpDescMaker<true>);
 REGISTER_OPERATOR(lstm_cudnn_grad, ops::CudnnLSTMGradOp);
 
 REGISTER_OP_CPU_KERNEL(
-    lstm_cudnn,
+    cudnn_lstm,
     ops::CudnnLSTMKernel<paddle::platform::CPUDeviceContext, float>);
 
 REGISTER_OP_CPU_KERNEL(
diff --git a/paddle/fluid/operators/lstm_cudnn_op.cu.cc b/paddle/fluid/operators/cudnn_lstm_op.cu.cc
similarity index 99%
rename from paddle/fluid/operators/lstm_cudnn_op.cu.cc
rename to paddle/fluid/operators/cudnn_lstm_op.cu.cc
index 353ab1759732969625501bcf333573aebc51bf61..811975a9f3d0321aa01941e3c862ae1743269940 100644
--- a/paddle/fluid/operators/lstm_cudnn_op.cu.cc
+++ b/paddle/fluid/operators/cudnn_lstm_op.cu.cc
@@ -12,7 +12,7 @@ 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. */
 
-#include "paddle/fluid/operators/lstm_cudnn_op.h"
+#include "paddle/fluid/operators/cudnn_lstm_op.h"
 #include "paddle/fluid/platform/cudnn_helper.h"
 
 namespace paddle {
@@ -487,8 +487,8 @@ class CudnnLSTMGPUGradKernel : public framework::OpKernel<T> {
 
 namespace ops = paddle::operators;
 REGISTER_OP_CUDA_KERNEL(
-    lstm_cudnn,
+    cudnn_lstm,
     ops::CudnnLSTMGPUKernel<paddle::platform::CUDADeviceContext, float>);
 REGISTER_OP_CUDA_KERNEL(
-    lstm_cudnn_grad,
+    cudnn_lstm_grad,
     ops::CudnnLSTMGPUGradKernel<paddle::platform::CUDADeviceContext, float>);
diff --git a/paddle/fluid/operators/lstm_cudnn_op.h b/paddle/fluid/operators/cudnn_lstm_op.h
similarity index 100%
rename from paddle/fluid/operators/lstm_cudnn_op.h
rename to paddle/fluid/operators/cudnn_lstm_op.h