diff --git a/backends/iluvatar_gpu/kernels/cuda_kernels/partial_concat_kernel_register.cu b/backends/iluvatar_gpu/kernels/cuda_kernels/partial_concat_kernel_register.cu
index f7198f7932..2c248da8a3 100644
--- a/backends/iluvatar_gpu/kernels/cuda_kernels/partial_concat_kernel_register.cu
+++ b/backends/iluvatar_gpu/kernels/cuda_kernels/partial_concat_kernel_register.cu
@@ -24,7 +24,7 @@
 
 #include "paddle/phi/backends/gpu/gpu_context.h"
 #include "paddle/phi/core/kernel_registry.h"
-#include "paddle/phi/kernels/gpu/partial_concat_kernel.cu"  //NOLINT
+#include "paddle/phi/kernels/partial_concat_kernel.h"
 PD_CUSTOM_KERNEL_REGISTER(partial_concat,
                           iluvatar_gpu,
                           ALL_LAYOUT,
diff --git a/backends/metax_gpu/CMakeLists.txt b/backends/metax_gpu/CMakeLists.txt
index 2e2f9aa0e1..836374f648 100755
--- a/backends/metax_gpu/CMakeLists.txt
+++ b/backends/metax_gpu/CMakeLists.txt
@@ -124,11 +124,7 @@ file(
   ${PADDLE_SOURCE_DIR}/paddle/phi/kernels/funcs/math/*.cu
   ${PADDLE_SOURCE_DIR}/paddle/phi/kernels/funcs/eigen/*.cu
   # kernels/gpu
-  ${PADDLE_SOURCE_DIR}/paddle/phi/kernels/gpu/partial_send_kernel.cu
-  ${PADDLE_SOURCE_DIR}/paddle/phi/kernels/gpu/partial_recv_kernel.cu
-  ${PADDLE_SOURCE_DIR}/paddle/phi/kernels/gpu/partial_concat_grad_kernel.cu
-  ${PADDLE_SOURCE_DIR}/paddle/phi/kernels/gpu/partial_allgather_kernel.cu
-  ${PADDLE_SOURCE_DIR}/paddle/phi/kernels/gpu/ctc_align_kernel.cu
+  ${PADDLE_SOURCE_DIR}/paddle/phi/kernels/gpu/partial_concat_kernel.cu
   ${PADDLE_SOURCE_DIR}/paddle/phi/kernels/gpu/yolo_box_head_kernel.cu
   ${PADDLE_SOURCE_DIR}/paddle/phi/kernels/gpu/stft_grad_kernel.cu
   ${PADDLE_SOURCE_DIR}/paddle/phi/kernels/gpu/soft_relu_grad_kernel.cu
diff --git a/backends/metax_gpu/kernels/cuda_kernels/partial_concat_kernel_register.cu b/backends/metax_gpu/kernels/cuda_kernels/partial_concat_kernel_register.cu
index 1d369d5d98..2a393b43c8 100644
--- a/backends/metax_gpu/kernels/cuda_kernels/partial_concat_kernel_register.cu
+++ b/backends/metax_gpu/kernels/cuda_kernels/partial_concat_kernel_register.cu
@@ -24,7 +24,7 @@
 
 #include "paddle/phi/backends/gpu/gpu_context.h"
 #include "paddle/phi/core/kernel_registry.h"
-#include "paddle/phi/kernels/gpu/partial_concat_kernel.cu"  //NOLINT
+#include "paddle/phi/kernels/partial_concat_kernel.h"
 PD_CUSTOM_KERNEL_REGISTER(partial_concat,
                           metax_gpu,
                           ALL_LAYOUT,