clang format

src/targets/gpu/device/layernorm.cpp

@@ -215,8 +215,11 @@ __device__ __half2 block_reduce_half2(
 
 // m = x - mean(x)
 // m / sqrt(mean(m ^ 2) + 1e-12)
-__device__ void layernorm_kernel_half2(__half2* in_data, __half2* in_data_reduce, 
-                                       __half2* out, index_int batch_item_num, index_int block_size,
+__device__ void layernorm_kernel_half2(__half2* in_data,
+                                       __half2* in_data_reduce,
+                                       __half2* out,
+                                       index_int batch_item_num,
+                                       index_int block_size,
                                        float rbatch_num)
 {
     auto rnum = __float2half2_rn(rbatch_num);
@@ -289,8 +292,12 @@ block_reduce_half(T* buffer, index_int batch_item_num, index_int tid, index_int
 
 // m = x - mean(x)
 // m / sqrt(mean(m ^ 2) + 1e-12)
-__device__ void layernorm_kernel_half(__half* in_data, __half* in_data_reduce, __half* out, 
-                                      index_int batch_item_num, index_int block_size, float rnum)
+__device__ void layernorm_kernel_half(__half* in_data,
+                                      __half* in_data_reduce,
+                                      __half* out,
+                                      index_int batch_item_num,
+                                      index_int block_size,
+                                      float rnum)
 {
     auto m = block_reduce_half(in_data_reduce, batch_item_num, threadIdx.x, block_size);
     m *= rnum;