diff --git a/backends/iluvatar_gpu/kernels/cuda_kernels/shuffle_channel_kernel_register.cu b/backends/iluvatar_gpu/kernels/cuda_kernels/shuffle_channel_kernel_register.cu
index 969745d02a..4ce43c1f62 100644
--- a/backends/iluvatar_gpu/kernels/cuda_kernels/shuffle_channel_kernel_register.cu
+++ b/backends/iluvatar_gpu/kernels/cuda_kernels/shuffle_channel_kernel_register.cu
@@ -14,7 +14,7 @@
 
 #include "paddle/phi/core/kernel_registry.h"
 #include "paddle/phi/kernels/gpu/shuffle_channel.h"
-#include "paddle/phi/kernels/gpu/shuffle_channel_kernel.cu"  //NOLINT
+#include "paddle/phi/kernels/gpu/shuffle_channel_kernel.h"
 PD_CUSTOM_KERNEL_REGISTER(shuffle_channel,
                           iluvatar_gpu,
                           ALL_LAYOUT,
diff --git a/backends/iluvatar_gpu/kernels/cuda_kernels/soft_relu_grad_kernel_register.cu b/backends/iluvatar_gpu/kernels/cuda_kernels/soft_relu_grad_kernel_register.cu
index a8c52dc662..068990ce3f 100644
--- a/backends/iluvatar_gpu/kernels/cuda_kernels/soft_relu_grad_kernel_register.cu
+++ b/backends/iluvatar_gpu/kernels/cuda_kernels/soft_relu_grad_kernel_register.cu
@@ -13,12 +13,12 @@
 // limitations under the License.
 
 #include "paddle/phi/core/kernel_registry.h"
-#include "paddle/phi/kernels/gpu/soft_relu_grad_kernel.cu"  //NOLINT
+#include "paddle/phi/kernels/soft_relu_grad_kernel.h"
 
 PD_CUSTOM_KERNEL_REGISTER(soft_relu_grad,
                           iluvatar_gpu,
                           ALL_LAYOUT,
-                          phi::SoftReluGradCudaKernel,
+                          phi::SoftReluGradKernel,
                           float,
                           double,
                           phi::dtype::float16,
diff --git a/backends/iluvatar_gpu/kernels/cuda_kernels/stft_grad_kernel_register.cu b/backends/iluvatar_gpu/kernels/cuda_kernels/stft_grad_kernel_register.cu
index 5ef89980d8..5e6b5755a6 100644
--- a/backends/iluvatar_gpu/kernels/cuda_kernels/stft_grad_kernel_register.cu
+++ b/backends/iluvatar_gpu/kernels/cuda_kernels/stft_grad_kernel_register.cu
@@ -13,8 +13,8 @@
 // limitations under the License.
 
 #include "paddle/phi/core/kernel_registry.h"
-#include "paddle/phi/kernels/gpu/stft_grad_kernel.cu"  //NOLINT
 #include "paddle/phi/kernels/impl/stft_grad_kernel_impl.h"
+#include "paddle/phi/kernels/stft_grad_kernel.h"
 #include "paddle/phi/kernels/stft_kernel.h"
 PD_CUSTOM_KERNEL_REGISTER(
     stft_grad, iluvatar_gpu, ALL_LAYOUT, phi::StftGradKernel, float, double) {
diff --git a/backends/iluvatar_gpu/kernels/cuda_kernels/uniform_random_batch_size_like_kernel_register.cu b/backends/iluvatar_gpu/kernels/cuda_kernels/uniform_random_batch_size_like_kernel_register.cu
index c30b701f38..0dee643f1c 100644
--- a/backends/iluvatar_gpu/kernels/cuda_kernels/uniform_random_batch_size_like_kernel_register.cu
+++ b/backends/iluvatar_gpu/kernels/cuda_kernels/uniform_random_batch_size_like_kernel_register.cu
@@ -14,7 +14,7 @@
 // clang-format off
 #include "paddle/phi/core/kernel_registry.h"
 #include "paddle/phi/kernels/funcs/uniform_random_functor.h"
-#include "paddle/phi/kernels/gpu/uniform_random_batch_size_like_kernel.cu" // NOLINT
+#include "paddle/phi/kernels/uniform_random_batch_size_like_kernel.h"
 // clang-format on
 PD_CUSTOM_KERNEL_REGISTER(uniform_random_batch_size_like,
                           iluvatar_gpu,
diff --git a/backends/iluvatar_gpu/kernels/cuda_kernels/yolo_box_head_kernel_register.cu b/backends/iluvatar_gpu/kernels/cuda_kernels/yolo_box_head_kernel_register.cu
index e15feeb4c5..9eaa6a3180 100644
--- a/backends/iluvatar_gpu/kernels/cuda_kernels/yolo_box_head_kernel_register.cu
+++ b/backends/iluvatar_gpu/kernels/cuda_kernels/yolo_box_head_kernel_register.cu
@@ -15,7 +15,7 @@
 #include "paddle/phi/core/kernel_registry.h"
 #include "paddle/phi/kernels/funcs/math_function.h"
 #include "paddle/phi/kernels/funcs/yolo_box_util.h"
-#include "paddle/phi/kernels/gpu/yolo_box_head_kernel.cu"  //NOLINT
+#include "paddle/phi/kernels/gpu/yolo_box_head_kernel.h"
 
 PD_CUSTOM_KERNEL_REGISTER(
     yolo_box_head, iluvatar_gpu, ALL_LAYOUT, phi::YoloBoxHeadKernel, float) {}
diff --git a/backends/metax_gpu/CMakeLists.txt b/backends/metax_gpu/CMakeLists.txt
index 1397340c2a..711aeaac4c 100755
--- a/backends/metax_gpu/CMakeLists.txt
+++ b/backends/metax_gpu/CMakeLists.txt
@@ -124,7 +124,10 @@ 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/ctc_align_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
+  ${PADDLE_SOURCE_DIR}/paddle/phi/kernels/gpu/shuffle_channel_kernel.cu
   ${PADDLE_SOURCE_DIR}/paddle/phi/kernels/gpu/yolo_box_post_kernel.cu
   ${PADDLE_SOURCE_DIR}/paddle/phi/kernels/gpu/spectral_norm_grad_kernel.cu
   ${PADDLE_SOURCE_DIR}/paddle/phi/kernels/gpu/spectral_norm_kernel.cu
diff --git a/backends/metax_gpu/kernels/cuda_kernels/shuffle_channel_kernel_register.cu b/backends/metax_gpu/kernels/cuda_kernels/shuffle_channel_kernel_register.cu
index 584a089f79..8f265db520 100644
--- a/backends/metax_gpu/kernels/cuda_kernels/shuffle_channel_kernel_register.cu
+++ b/backends/metax_gpu/kernels/cuda_kernels/shuffle_channel_kernel_register.cu
@@ -14,7 +14,7 @@
 
 #include "paddle/phi/core/kernel_registry.h"
 #include "paddle/phi/kernels/gpu/shuffle_channel.h"
-#include "paddle/phi/kernels/gpu/shuffle_channel_kernel.cu"  //NOLINT
+#include "paddle/phi/kernels/gpu/shuffle_channel_kernel.h"
 PD_CUSTOM_KERNEL_REGISTER(shuffle_channel,
                           metax_gpu,
                           ALL_LAYOUT,
diff --git a/backends/metax_gpu/kernels/cuda_kernels/soft_relu_grad_kernel_register.cu b/backends/metax_gpu/kernels/cuda_kernels/soft_relu_grad_kernel_register.cu
index 86a535d389..0b456b9efc 100644
--- a/backends/metax_gpu/kernels/cuda_kernels/soft_relu_grad_kernel_register.cu
+++ b/backends/metax_gpu/kernels/cuda_kernels/soft_relu_grad_kernel_register.cu
@@ -13,12 +13,12 @@
 // limitations under the License.
 
 #include "paddle/phi/core/kernel_registry.h"
-#include "paddle/phi/kernels/gpu/soft_relu_grad_kernel.cu"  //NOLINT
+#include "paddle/phi/kernels/soft_relu_grad_kernel.h"
 
 PD_CUSTOM_KERNEL_REGISTER(soft_relu_grad,
                           metax_gpu,
                           ALL_LAYOUT,
-                          phi::SoftReluGradCudaKernel,
+                          phi::SoftReluGradKernel,
                           float,
                           double,
                           phi::dtype::float16,
diff --git a/backends/metax_gpu/kernels/cuda_kernels/stft_grad_kernel_register.cu b/backends/metax_gpu/kernels/cuda_kernels/stft_grad_kernel_register.cu
index 15f6c19e72..7ac1b67543 100644
--- a/backends/metax_gpu/kernels/cuda_kernels/stft_grad_kernel_register.cu
+++ b/backends/metax_gpu/kernels/cuda_kernels/stft_grad_kernel_register.cu
@@ -13,8 +13,8 @@
 // limitations under the License.
 
 #include "paddle/phi/core/kernel_registry.h"
-#include "paddle/phi/kernels/gpu/stft_grad_kernel.cu"  //NOLINT
 #include "paddle/phi/kernels/impl/stft_grad_kernel_impl.h"
+#include "paddle/phi/kernels/stft_grad_kernel.h"
 #include "paddle/phi/kernels/stft_kernel.h"
 PD_CUSTOM_KERNEL_REGISTER(
     stft_grad, metax_gpu, ALL_LAYOUT, phi::StftGradKernel, float, double) {
diff --git a/backends/metax_gpu/kernels/cuda_kernels/uniform_random_batch_size_like_kernel_register.cu b/backends/metax_gpu/kernels/cuda_kernels/uniform_random_batch_size_like_kernel_register.cu
index aa25417c2a..ade154f353 100755
--- a/backends/metax_gpu/kernels/cuda_kernels/uniform_random_batch_size_like_kernel_register.cu
+++ b/backends/metax_gpu/kernels/cuda_kernels/uniform_random_batch_size_like_kernel_register.cu
@@ -15,7 +15,7 @@
 // clang-format off
 #include "paddle/phi/core/kernel_registry.h"
 #include "paddle/phi/kernels/funcs/uniform_random_functor.h"
-#include "paddle/phi/kernels/gpu/uniform_random_batch_size_like_kernel.cu" // NOLINT
+#include "paddle/phi/kernels/uniform_random_batch_size_like_kernel.h"
 // clang-format on
 PD_CUSTOM_KERNEL_REGISTER(uniform_random_batch_size_like,
                           metax_gpu,
diff --git a/backends/metax_gpu/kernels/cuda_kernels/yolo_box_head_kernel_register.cu b/backends/metax_gpu/kernels/cuda_kernels/yolo_box_head_kernel_register.cu
index fb8a82cd3a..09d3f3273e 100644
--- a/backends/metax_gpu/kernels/cuda_kernels/yolo_box_head_kernel_register.cu
+++ b/backends/metax_gpu/kernels/cuda_kernels/yolo_box_head_kernel_register.cu
@@ -15,7 +15,7 @@
 #include "paddle/phi/core/kernel_registry.h"
 #include "paddle/phi/kernels/funcs/math_function.h"
 #include "paddle/phi/kernels/funcs/yolo_box_util.h"
-#include "paddle/phi/kernels/gpu/yolo_box_head_kernel.cu"  //NOLINT
+#include "paddle/phi/kernels/gpu/yolo_box_head_kernel.h"
 
 PD_CUSTOM_KERNEL_REGISTER(
     yolo_box_head, metax_gpu, ALL_LAYOUT, phi::YoloBoxHeadKernel, float) {}