diff --git a/operators/cuda/cuda_ops.cc b/operators/cuda/cuda_ops.cc
index 8f684c29e..25913b8bf 100644
--- a/operators/cuda/cuda_ops.cc
+++ b/operators/cuda/cuda_ops.cc
@@ -7,11 +7,8 @@
 #include "cuda/add_mul.h"
 #include "cuda/fast_gelu.h"
 #include "cuda/negxplus1.h"
-<<<<<<< HEAD
 #include "cuda/scatter_nd_of_shape.h"
-=======
 #include "cuda/transpose_cast.h"
->>>>>>> 79f3b048d4d195b6684f2d1b6ca5bfe1ab9ea8d6
 #endif
 
 FxLoadCustomOpFactory LoadCustomOpClasses_Contrib = []() -> CustomOpArray& {
diff --git a/test/cuda/test_cudaops.py b/test/cuda/test_cudaops.py
index e55eb6e75..c5c8dc6e0 100644
--- a/test/cuda/test_cudaops.py
+++ b/test/cuda/test_cudaops.py
@@ -169,12 +169,6 @@ def _negxplus1_cuda(self, itype):
         got = sess.run(None, feeds1)[0]
         assert_almost_equal(expected, got, decimal=5)
 
-<<<<<<< HEAD
-    @unittest.skipIf(not has_cuda(), reason="cuda not available")
-    def test_cuda_negxplus1(self):
-        self._negxplus1_cuda(TensorProto.FLOAT)
-        self._negxplus1_cuda(TensorProto.FLOAT16)
-=======
     @unittest.skipIf(not has_cuda(), reason="CUDA is missing")
     def test_cuda_negxplus1(self):
         self._negxplus1_cuda(TensorProto.FLOAT)
@@ -289,7 +283,6 @@ def test_add_shared_input_cuda_broadcast2(self):
             shapeb=(3, 2, 3),
             shapec=(3, 2, 3),
         )
->>>>>>> 04029e13f7998574670e2bc7eb500825800654fa
 
     def _scatternd_of_shape_optimize_cuda(self, optimize, dim3, itype):
         indices_shape = ["i", "j", 1] if dim3 else ["j", 1]