msadapter/pytorch/tensor.py

@@ -867,8 +867,8 @@ class Tensor(ms.Tensor):
     def sum(self, dim=None, keepdim=False, dtype=None):
         input = cast_to_ms_tensor(self)
         if dtype is not None:
-            input = input.astype(dtype) if dtype != mstype.bool_ else input.astype(mstype.int32)
-        elif self.dtype in msdapter_dtype.all_int_type:
+            input = input.astype(dtype) if dtype != mstype.bool_ else input.astype(mstype.bool_).astype(mstype.int64)
+        elif input.dtype in (msdapter_dtype.all_int_type, mstype.bool_):
             dtype = mstype.int64
             input = input.astype(dtype)
 
@@ -880,14 +880,11 @@ class Tensor(ms.Tensor):
         else:
             dim = validator.check_and_canonicalize_axes(dim, input.ndim)
 
-        if not validator.check_type_support(input.dtype, 'GPU', (mstype.float64, mstype.float32, mstype.float16)):
-            input = input.astype(mstype.float32)
         if 0 in self.shape:
             input = tensor_operator_registry.get('make_tensor')([0], input.dtype)
         res = tensor_operator_registry.get('sum')(bool(keepdim))(input, dim)
         if dtype == mstype.bool_:
             res = res.astype(mstype.bool_)
-
         return cast_to_adapter_tensor(res)
 
     def sum_to_size(self, *size):

testing/ut/pytorch/tensor/test_tensor.py

@@ -414,6 +414,29 @@ def test_sum2():
     assert np.allclose(ms_output.asnumpy(), torch_output.numpy())
     assert ms_output.asnumpy().dtype == torch_output.numpy().dtype
 
+def test_sum3():
+
+    torch_tensor = torch.tensor([True, True, False])
+    ms_tensor = pytorch.tensor([True, True, False])
+    torch_output = torch_tensor.sum()
+    ms_output = ms_tensor.sum()
+    assert np.allclose(ms_output.asnumpy(), torch_output.numpy())
+    assert ms_output.asnumpy().dtype == torch_output.numpy().dtype
+
+    torch_tensor = torch.tensor([-1, 1], dtype=torch.bool)
+    ms_tensor = pytorch.tensor([-1, 1], dtype=pytorch.bool)
+    torch_output = torch_tensor.sum()
+    ms_output = ms_tensor.sum()
+    assert np.allclose(ms_output.asnumpy(), torch_output.numpy())
+    assert ms_output.asnumpy().dtype == torch_output.numpy().dtype
+
+    torch_tensor = torch.tensor([-1, 1])
+    ms_tensor = pytorch.tensor([-1, 1])
+    torch_output = torch_tensor.sum(dtype=torch.bool)
+    ms_output = ms_tensor.sum(dtype=pytorch.bool)
+    assert np.allclose(ms_output.asnumpy(), torch_output.numpy())
+    assert ms_output.asnumpy().dtype == torch_output.numpy().dtype
+
 def test_split():
 
     tensor = np.random.random((3, 3)).astype(np.float32)
@@ -4907,6 +4930,7 @@ if __name__ == '__main__':
     test_numel()
     test_sum()
     test_sum2()
+    test_sum3()
     test_split()
     test_numpy()
     test_ndimension()