diff --git a/colossalai/kernel/cuda_native/csrc/kernels/include/cross_entropy_layer.h b/colossalai/kernel/cuda_native/csrc/kernels/include/cross_entropy_layer.h
index af7c9c04d..f4e9befc6 100644
--- a/colossalai/kernel/cuda_native/csrc/kernels/include/cross_entropy_layer.h
+++ b/colossalai/kernel/cuda_native/csrc/kernels/include/cross_entropy_layer.h
@@ -8,8 +8,9 @@
 
 #include "cuda_util.h"
 
-template <typename T> class CrossEntropyLayer {
-public:
+template <typename T>
+class CrossEntropyLayer {
+ public:
   CrossEntropyLayer(float epsilon, int padding_idx, int max_batch_tokens);
 
   virtual ~CrossEntropyLayer();
@@ -22,7 +23,7 @@ public:
 
   void set_cur_batch_shape(int batch_size, int seq_len, int vocab_size);
 
-private:
+ private:
   void allocate_mem_buffer() {
     // allocate local gpu memory
     _loss_buffer = cuda_malloc<float>(_max_batch_tokens * 2);