diff --git a/mindspore/common/tensor.py b/mindspore/common/tensor.py
index 2a03ab80f7..756202d1ec 100644
--- a/mindspore/common/tensor.py
+++ b/mindspore/common/tensor.py
@@ -125,10 +125,6 @@ class Tensor(Tensor_):
             return Tensor_.__repr__(self)
         return ''
 
-    def __add__(self, other):
-        out = tensor_operator_registry.get('__add__')(self, other)
-        return out
-
     def __eq__(self, other):
         if not isinstance(other, (int, float, Tensor)):
             return False
@@ -150,10 +146,6 @@ class Tensor(Tensor_):
     def __hash__(self):
         return hash(id(self))
 
-    def __mul__(self, other):
-        out = tensor_operator_registry.get('__mul__')(self, other)
-        return out
-
     def __neg__(self):
         out = tensor_operator_registry.get('__neg__')(self)
         return out
@@ -187,38 +179,59 @@ class Tensor(Tensor_):
     def __pos__(self):
         return self
 
+    def __add__(self, other):
+        return tensor_operator_registry.get('__add__')(self, other)
+
+    def __radd__(self, other):
+        return self.__add__(other)
+
     def __iadd__(self, other):
         return self.__add__(other)
 
-    def __radd__(self, other):
-        out = tensor_operator_registry.get('__add__')(self, other)
-        return out
+    def __sub__(self, other):
+        return tensor_operator_registry.get('__sub__')(self, other)
 
-    def __imul__(self, other):
-        return self.__mul__(other)
+    def __rsub__(self, other):
+        return tensor_operator_registry.get('__sub__')(other, self)
+
+    def __isub__(self, other):
+        return self.__sub__(other)
+
+    def __mul__(self, other):
+        return tensor_operator_registry.get('__mul__')(self, other)
 
     def __rmul__(self, other):
-        out = tensor_operator_registry.get('__mul__')(self, other)
-        return out
+        return self.__mul__(other)
+
+    def __imul__(self, other):
+        return self.__mul__(other)
 
     def __truediv__(self, other):
-        out = tensor_operator_registry.get('__truediv__')(self, other)
-        return out
+        return tensor_operator_registry.get('__truediv__')(self, other)
 
     def __rtruediv__(self, other):
-        out = tensor_operator_registry.get('__truediv__')(other, self)
-        return out
+        return tensor_operator_registry.get('__truediv__')(other, self)
 
-    def __sub__(self, other):
-        out = tensor_operator_registry.get('__sub__')(self, other)
-        return out
+    def __mod__(self, other):
+        return tensor_operator_registry.get('__mod__')(self, other)
 
-    def __isub__(self, other):
-        return self.__sub__(other)
+    def __rmod__(self, other):
+        return tensor_operator_registry.get('__mod__')(other, self)
 
-    def __rsub__(self, other):
-        out = tensor_operator_registry.get('__sub__')(other, self)
-        return out
+    def __imod__(self, other):
+        return self.__mod__(other)
+
+    def __pow__(self, other):
+        return tensor_operator_registry.get('__pow__')(self, other)
+
+    def __floordiv__(self, other):
+        return tensor_operator_registry.get('__floordiv__')(self, other)
+
+    def __rfloordiv__(self, other):
+        return tensor_operator_registry.get('__floordiv__')(other, self)
+
+    def __ifloordiv__(self, other):
+        return self.__floordiv__(other)
 
     def __lt__(self, other):
         out = tensor_operator_registry.get('__lt__')(self, other)
@@ -229,8 +242,6 @@ class Tensor(Tensor_):
         return out
 
     def __getitem__(self, index):
-        if isinstance(index, int) and not isinstance(index, bool) and self.shape and index >= self.shape[0]:
-            raise IndexError("index {} is out of bounds for axis 0 with size {}".format(index, self.shape[0]))
         out = tensor_operator_registry.get('__getitem__')(self, index)
         return out
 
@@ -253,27 +264,6 @@ class Tensor(Tensor_):
             return out[0]
         raise TypeError("Not support len of a 0-D tensor")
 
-    def __mod__(self, other):
-        return tensor_operator_registry.get('__mod__')(self, other)
-
-    def __imod__(self, other):
-        return self.__mod__(other)
-
-    def __rmod__(self, other):
-        return tensor_operator_registry.get('__mod__')(other, self)
-
-    def __pow__(self, other):
-        return tensor_operator_registry.get('__pow__')(self, other)
-
-    def __floordiv__(self, other):
-        return tensor_operator_registry.get('__floordiv__')(self, other)
-
-    def __ifloordiv__(self, other):
-        return self.__floordiv__(other)
-
-    def __rfloordiv__(self, other):
-        return tensor_operator_registry.get('__floordiv__')(other, self)
-
     def __str__(self):
         if self.dtype == mstype.type_none:
             return "Unknown Tensor type!"
diff --git a/mindspore/ops/composite/multitype_ops/_compile_utils.py b/mindspore/ops/composite/multitype_ops/_compile_utils.py
index e7a8d0d428..6c56358979 100644
--- a/mindspore/ops/composite/multitype_ops/_compile_utils.py
+++ b/mindspore/ops/composite/multitype_ops/_compile_utils.py
@@ -28,22 +28,31 @@ stack = P.Stack(axis=-1)
 
 def _tensor_getitem(self, index):
     """Handle tensor getitem"""
+    if isinstance(index, Tensor):
+        return tensor_index_by_tensor(self, index)
     if isinstance(index, list):
         return tensor_index_by_list(self, index)
     if isinstance(index, tuple):
         return tensor_index_by_tuple(self, index)
-    if isinstance(index, (Tensor, int, slice)) or index in (None, ...):
-        return tensor_index_by_tuple(self, (index,))
+    if isinstance(index, bool):
+        return _tensor_index_by_bool(self, index)
+    if isinstance(index, int):
+        return _tensor_index_by_integer(self, index)
+    if isinstance(index, slice):
+        return tensor_index_by_slice(self, index)
+    if index is None:
+        return F.expand_dims(self, 0)
+    if index is ...:
+        return self
     raise IndexError(f"Only support integers, slices(`:`), ellipsis(`...`), None, bool, tensor with int, "
                      f"list and tuple ,but got {index} with type {type(index)}.")
-tensor_operator_registry.register("__getitem__", _tensor_getitem)
+
 
 def _tensor_setitem(self, index, value):
     """Handle tensor setitem"""
     if not isinstance(value, (int, float, bool, list, tuple, Tensor)):
         raise ValueError(f"only support numbers, Tensor, tuple, list as value,"
                          f"but got {value} with type {type(value)}.")
-
     if isinstance(index, list):
         index = format_list_indices(index, self.shape[0])
     if isinstance(index, Tensor):
@@ -62,10 +71,75 @@ def _tensor_setitem(self, index, value):
     if index is ...:
         return tensor_setitem_by_ellipsis(self, index, value)
 
-    raise IndexError("Tensor setitem index only support integers, slices(`:`), ellipsis(`...`), None, bool\
-                         and tensor with int32, got {} with type{}".format(index, type(index)))
+    raise IndexError("Tensor setitem index only support integers, slices(`:`), ellipsis(`...`), bool, tensor, \
+        list and tuple, but got {index} with type{type(index)}")
+
+
+tensor_operator_registry.register("__getitem__", _tensor_getitem)
 tensor_operator_registry.register("__setitem__", _tensor_setitem)
 
+
+def _tensor_add(self, other):
+    if isinstance(other, (tuple, list)):
+        other = sequence_to_tensor(other, F.dtype(self))
+    return F.add(self, other)
+
+
+def _tensor_sub(self, other):
+    if isinstance(self, (tuple, list)):
+        self = sequence_to_tensor(self, F.dtype(other))
+    if isinstance(other, (tuple, list)):
+        other = sequence_to_tensor(other, F.dtype(self))
+    return F.sub(self, other)
+
+
+def _tensor_mul(self, other):
+    if isinstance(other, (tuple, list)):
+        other = sequence_to_tensor(other, F.dtype(self))
+    return F.mul(self, other)
+
+
+def _tensor_div(self, other):
+    if isinstance(self, (tuple, list)):
+        self = sequence_to_tensor(self, F.dtype(other))
+    if isinstance(other, (tuple, list)):
+        other = sequence_to_tensor(other, F.dtype(self))
+    return F.div(self, other)
+
+
+def _tensor_mod(self, other):
+    if isinstance(self, (tuple, list)):
+        self = sequence_to_tensor(self, F.dtype(other))
+    if isinstance(other, (tuple, list)):
+        other = sequence_to_tensor(other, F.dtype(self))
+    return F.floormod(self, other)
+
+
+def _tensor_pow(self, other):
+    if isinstance(self, (tuple, list)):
+        self = sequence_to_tensor(self, F.dtype(other))
+    if isinstance(other, (tuple, list)):
+        other = sequence_to_tensor(other, F.dtype(self))
+    return F.tensor_pow(self, other)
+
+
+def _tensor_floordiv(self, other):
+    if isinstance(self, (tuple, list)):
+        self = sequence_to_tensor(self, F.dtype(other))
+    if isinstance(other, (tuple, list)):
+        other = sequence_to_tensor(other, F.dtype(self))
+    return F.floordiv(self, other)
+
+
+tensor_operator_registry.register('__add__', _tensor_add)
+tensor_operator_registry.register('__sub__', _tensor_sub)
+tensor_operator_registry.register('__mul__', _tensor_mul)
+tensor_operator_registry.register('__truediv__', _tensor_div)
+tensor_operator_registry.register('__mod__', _tensor_mod)
+tensor_operator_registry.register('__pow__', _tensor_pow)
+tensor_operator_registry.register('__floordiv__', _tensor_floordiv)
+
+
 def _broadcast(broadcast_shape, x):
     """Broadcast tensor to the required shape."""
     if F.shape(x) == broadcast_shape:
@@ -120,7 +194,9 @@ def _expand_data_dims(data, tuple_index):
             tuple_index_new += (const_utils.make_empty_slice(),)
             expand_positions += (i,)
         elif const_utils.judge_index_type(index_type, mstype.bool_):
-            tuple_index_new += (const_utils.make_tensor([0] if index else[], mstype.int64),)
+            if not index:
+                const_utils.raise_index_error("Dose not support 'False'.")
+            tuple_index_new += (const_utils.make_tensor([0], mstype.int64),)
             expand_positions += (i,)
         else:
             tuple_index_new += (index,)
@@ -131,47 +207,66 @@ def _expand_data_dims(data, tuple_index):
     return data, tuple_index_new
 
 
+def tensor_index_by_slice(data, slice_index):
+    """Tensor getitem by a slice."""
+    min_data_dim, max_data_dim = 1, 8
+    const_utils.judge_data_dim(data.ndim, min_data_dim, max_data_dim)
+    data_shape = F.shape(data)
+    begin_strides, end_strides, step_strides = const_utils.get_stride_info_from_slice(data_shape, slice_index)
+    return F.strided_slice(data, begin_strides, end_strides, step_strides)
+
+
 def tensor_index_by_number(data, number_index):
     """Tensor getitem by a Number which may be integer/float/bool value"""
-    data_type = F.typeof(data)
-    if const_utils.judge_index_type(data_type, mstype.tensor_type):
-        data_shape = F.shape(data)
-        data_rank = len(data_shape)
-        min_data_rank, max_data_rank = 0, 8
-        const_utils.judge_data_rank(data_rank, min_data_rank, max_data_rank)
-
     number_type = const_utils.check_number_index_type(number_index)
     if number_type == const_utils.BOOL_:
-        return tensor_index_by_tuple(data, (number_index,))
+        return _tensor_index_by_bool(data, number_index)
     if number_type == const_utils.INT_:
         return _tensor_index_by_integer(data, number_index)
     return const_utils.raise_index_error("Only support integers, slices(`:`), ellipsis(`...`), None and bool.")
 
 
+def _tensor_index_by_bool(data, bool_value):
+    """Tensor getitem by a single bool value"""
+    min_data_dim, max_data_dim = 0, 7
+    const_utils.judge_data_dim(data.ndim, min_data_dim, max_data_dim)
+    if bool_value:
+        return F.expand_dims(data, 0)
+    return const_utils.raise_index_error("When tensor is indexed by a bool object, the value only support 'True'.")
+
+
 def _tensor_index_by_integer(data, int_index):
     """Tensor getitem by a single integer number"""
+    if const_utils.judge_index_type(F.typeof(data), mstype.tensor_type):
+        min_data_dim, max_data_dim = 1, 8
+        const_utils.judge_data_dim(data.ndim, min_data_dim, max_data_dim)
+
     data_shape = F.shape(data)
-    data_rank = len(data_shape)
-    if data_rank == 0:
-        return const_utils.raise_type_error("When tensor is indexed by an integer, the dimension of the tensor "
-                                            "cannot be 0.")
     transformed_number = const_utils.check_and_transform_int_index(int_index, data_shape[0], const_utils.TENSOR_GETITEM)
     begin_strides, end_strides, step_strides = const_utils.get_stride_info_from_integer(data_shape, transformed_number)
     shrink_axis_mask = 1
     return P.StridedSlice(0, 0, 0, 0, shrink_axis_mask)(data, begin_strides, end_strides, step_strides)
 
 
+def tensor_index_by_tensor(data, tensor_index):
+    """Tensor getitem by a single tensor"""
+    min_data_dim, max_data_dim = 0, 7
+    const_utils.judge_data_dim(data.ndim, min_data_dim, max_data_dim)
+    const_utils.check_type_valid(F.dtype(tensor_index), mstype.int_type, const_utils.TENSOR_GETITEM)
+    return F.gather(data, tensor_index, 0)
+
+
 def tensor_index_by_list(data, list_index):
     """Tensor getitem by list of int and bool"""
+    min_data_dim, max_data_dim = 1, 8
+    const_utils.judge_data_dim(data.ndim, min_data_dim, max_data_dim)
+
     data_shape = F.shape(data)
     indexes_types = hyper_map(F.typeof, list_index)
     if const_utils.judge_indexes_types(indexes_types, mstype.int_type + (mstype.bool_,)):
         sub_tuple_index = const_utils.transform_sequence_index(list_index, data_shape[0], const_utils.TENSOR_GETITEM)
         if not sub_tuple_index:
-            data_rank = len(data_shape)
-            if data_rank == 1:
-                return const_utils.make_tensor([], data.dtype, ())
-            return const_utils.make_tensor([], data.dtype, data_shape[1:])
+            const_utils.raise_index_error("Getitem does not support empty list, this will reference shape '0'.")
         tensor_index = const_utils.make_tensor(sub_tuple_index, mstype.int64)
         return F.gather(data, tensor_index, 0)
 
@@ -183,18 +278,15 @@ def tensor_index_by_list(data, list_index):
 
 def tensor_index_by_tuple(data, tuple_index):
     """Tensor getitem by tuple of various types with None"""
-    tuple_index_len = len(tuple_index)
-    if tuple_index_len == 0:
+    if not tuple_index:
         return data
 
     op_name = const_utils.TENSOR_GETITEM
     tuple_index = _transform_ellipsis_to_slice(data, tuple_index, op_name)
     data, tuple_index = _expand_data_dims(data, tuple_index)
 
-    data_shape = F.shape(data)
-    data_rank = len(data_shape)
-    min_data_rank, max_data_rank = 0, 8
-    const_utils.judge_data_rank(data_rank, min_data_rank, max_data_rank)
+    min_data_dim, max_data_dim = 1, 8
+    const_utils.judge_data_dim(data.ndim, min_data_dim, max_data_dim)
 
     indexes_types = hyper_map(F.typeof, tuple_index)
     contain_type = const_utils.tuple_index_type_cnt(indexes_types, op_name)
@@ -382,21 +474,27 @@ def _generate_updates_from_scalar(data, indices, value, op_type):
     return const_utils.convert_scalar_to_tensor(data_shape, data_dtype, indices_shape, value, op_type)
 
 
-def _generate_updates_from_sequence(data, index, value, op_type):
+def sequence_to_tensor(value, dtype):
     """Generate an updates tensor from a tuple, can only handle 1-D tensor/non-tensor mixtures."""
     value_types = hyper_map(F.typeof, value)
     value_elements_type = const_utils.check_value_elements(value_types)
 
     if value_elements_type == const_utils.ALL_TENSOR:
-        value = F.stack(value).astype(data.dtype)
+        value = F.stack(value).astype(dtype)
     elif value_elements_type == const_utils.NO_TENSOR:
-        value = const_utils.make_tensor(value, data.dtype)
+        value = const_utils.make_tensor(value, dtype)
     else:
         new_value = ()
         for ele in value:
             ele = ele if isinstance(ele, Tensor) else const_utils.make_tensor(ele)
             new_value += (ele,)
-        value = F.stack(new_value).astype(data.dtype)
+        value = F.stack(new_value).astype(dtype)
+    return value
+
+
+def _generate_updates_from_sequence(data, index, value, op_type):
+    """Generate an updates tensor from a tuple, can only handle 1-D tensor/non-tensor mixtures."""
+    value = sequence_to_tensor(value, F.dtype(data))
     if op_type == const_utils.SET_ITEM_BY_NON_TENSOR:
         return value
     return _generate_updates_from_tensor(data, index, value, op_type)
diff --git a/mindspore/ops/composite/multitype_ops/_constexpr_utils.py b/mindspore/ops/composite/multitype_ops/_constexpr_utils.py
index 4ad5d92cee..7c43115ea4 100644
--- a/mindspore/ops/composite/multitype_ops/_constexpr_utils.py
+++ b/mindspore/ops/composite/multitype_ops/_constexpr_utils.py
@@ -157,10 +157,10 @@ def make_tensor(a, dtype=mstype.int32, data_shape=None):
 
 
 @constexpr
-def judge_data_rank(data_rank, min_data_rank=0, max_data_rank=8):
-    if data_rank < min_data_rank or data_rank > max_data_rank:
-        raise ValueError(f"The input data's rank should in the range of[{min_data_rank}, "
-                         f"{max_data_rank}], bug actually is '{data_rank}'")
+def judge_data_dim(data_dim, min_data_dim=0, max_data_dim=8):
+    if data_dim < min_data_dim or data_dim > max_data_dim:
+        raise ValueError(f"The input data's dim should in the range of[{min_data_dim}, "
+                         f"{max_data_dim}], bug actually is '{data_dim}'")
 
 
 @constexpr
@@ -636,15 +636,6 @@ def transform_slice_to_ele_list(slice_index, dim_len):
     return slice_ele_list
 
 
-@constexpr
-def check_tuple_index_len(data_rank, tuple_index_len, op_name):
-    """Check if the number of index tensor exceeds the dimension of the operated tensor."""
-    if tuple_index_len <= data_rank:
-        return True
-    raise IndexError(f"For '{op_name}', the number {tuple_index_len} of tuple_index size"
-                     f"is greater than the dimension  {data_rank} of the operated tensor.")
-
-
 @constexpr
 def generate_index_info_from_tuple_of_mixed_tensors(tensor_positions, tensor_indexes_shapes,
                                                     slice_shapes, op_name):
@@ -669,6 +660,7 @@ def generate_index_info_from_tuple_of_mixed_tensors(tensor_positions, tensor_ind
 
     return broadcast_shape, index_tensor_new_shape, final_shape, fancy_position
 
+
 def _judge_order_continuous(order_sequence):
     if not order_sequence:
         return False
@@ -710,6 +702,20 @@ def check_number_index_type(number):
                      .format(number, type(number)))
 
 
+@constexpr
+def get_stride_info_from_slice(data_shape, slice_index):
+    """Get stride info from a python slice"""
+    begin, end, step = get_slice_stride(data_shape[0], slice_index)
+    begin_strides = [begin]
+    end_strides = [end]
+    step_strides = [step]
+    for end in data_shape[1:]:
+        begin_strides.append(0)
+        end_strides.append(end)
+        step_strides.append(1)
+    return tuple(begin_strides), tuple(end_strides), tuple(step_strides)
+
+
 @constexpr
 def get_stride_info_from_integer(data_shape, number):
     """Get stride info from a integer"""
@@ -741,7 +747,7 @@ def get_stride_info_from_tuple(data_shape, tuple_index):
     """Get stride info from a tuple"""
     begin_strides, end_strides, step_strides = [], [], []
     tuple_index_len = len(tuple_index)
-    data_rank = len(data_shape)
+    data_dim = len(data_shape)
     shrink_axis, index_count, ellipsis_count = 0, 0, 0
     for idx, item in enumerate(tuple_index):
         if isinstance(item, slice):
@@ -760,7 +766,7 @@ def get_stride_info_from_tuple(data_shape, tuple_index):
             ellipsis_count = ellipsis_count + 1
             if ellipsis_count > 1:
                 raise IndexError("An index can have only one ellipsis (...)")
-            ellipsis_range_size = data_rank - tuple_index_len + 1
+            ellipsis_range_size = data_dim - tuple_index_len + 1
             begin_strides.extend([0] * (ellipsis_range_size))
             end_strides.extend(
                 [shape for shape in data_shape[index_count: index_count + ellipsis_range_size]])
@@ -769,7 +775,7 @@ def get_stride_info_from_tuple(data_shape, tuple_index):
         else:
             raise IndexError("Not supported index data type, got ",
                              item, " type is ", type(item))
-    for item in range(index_count, data_rank):
+    for item in range(index_count, data_dim):
         begin_strides.append(0)
         end_strides.append(data_shape[item])
         step_strides.append(1)
diff --git a/mindspore/ops/composite/multitype_ops/add_impl.py b/mindspore/ops/composite/multitype_ops/add_impl.py
index e52bb02635..174a3e6783 100644
--- a/mindspore/ops/composite/multitype_ops/add_impl.py
+++ b/mindspore/ops/composite/multitype_ops/add_impl.py
@@ -15,6 +15,7 @@
 
 """Implementation for internal polymorphism `add` operations."""
 
+from . import _compile_utils as utils
 from ...composite import base
 from ... import functional as F
 
@@ -114,6 +115,70 @@ def _tensor_add_scalar(x, y):
     return F.add(x, y)
 
 
+@add.register("Tuple", "Tensor")
+def _tuple_add_tensor(x, y):
+    """
+    Tuple is added to tensor.
+
+    Args:
+        x (Tuple): x
+        y (Tensor): The dtype is same as x.
+
+    Returns:
+        Tensor, has the same dtype as x.
+    """
+    x = utils.sequence_to_tensor(x, y.dtype)
+    return F.tensor_add(x, y)
+
+
+@add.register("Tensor", "Tuple")
+def _tensor_add_tuple(x, y):
+    """
+    Tensor is added to number.
+
+    Args:
+        x (Tensor): x
+        y (Tuple): The dtype is same as x.
+
+    Returns:
+        Tensor, has the same dtype as x.
+    """
+    y = utils.sequence_to_tensor(y, x.dtype)
+    return F.tensor_add(x, y)
+
+
+@add.register("List", "Tensor")
+def _list_add_tensor(x, y):
+    """
+    Tuple is added to tensor.
+
+    Args:
+        x (List): x
+        y (Tensor): The dtype is same as x.
+
+    Returns:
+        Tensor, has the same dtype as x.
+    """
+    x = utils.sequence_to_tensor(x, y.dtype)
+    return F.tensor_add(x, y)
+
+
+@add.register("Tensor", "List")
+def _tensor_add_list(x, y):
+    """
+    Tensor is added to number.
+
+    Args:
+        x (Tensor): x
+        y (List): The dtype is same as x.
+
+    Returns:
+        Tensor, has the same dtype as x.
+    """
+    y = utils.sequence_to_tensor(y, x.dtype)
+    return F.tensor_add(x, y)
+
+
 @add.register("Tensor", "Tensor")
 def _tensor_add_tensor(x, y):
     """
diff --git a/mindspore/ops/composite/multitype_ops/div_impl.py b/mindspore/ops/composite/multitype_ops/div_impl.py
index d5d98d62af..478476513d 100644
--- a/mindspore/ops/composite/multitype_ops/div_impl.py
+++ b/mindspore/ops/composite/multitype_ops/div_impl.py
@@ -15,6 +15,7 @@
 
 """Implementation for internal polymorphism `div` operations."""
 
+from . import _compile_utils as utils
 from ...composite import base
 from ... import functional as F
 
@@ -84,3 +85,67 @@ def _tensor_div_scalar(x, y):
         Tensor, has the same dtype as x.
     """
     return F.tensor_div(x, y)
+
+
+@div.register("Tuple", "Tensor")
+def _tuple_div_tensor(x, y):
+    """
+    Tuple divided by tensor.
+
+    Args:
+        x (Tuple): x
+        y (Tensor): The dtype is same as x.
+
+    Returns:
+        Tensor, has the same dtype as x.
+    """
+    x = utils.sequence_to_tensor(x, y.dtype)
+    return F.tensor_div(x, y)
+
+
+@div.register("Tensor", "Tuple")
+def _tensor_div_tuple(x, y):
+    """
+    Tensor divided by tuple.
+
+    Args:
+        x (Tensor): x
+        y (Tuple): The dtype is same as x.
+
+    Returns:
+        Tensor, has the same dtype as x.
+    """
+    y = utils.sequence_to_tensor(y, x.dtype)
+    return F.tensor_div(x, y)
+
+
+@div.register("List", "Tensor")
+def _list_div_tensor(x, y):
+    """
+    List divided by tensor.
+
+    Args:
+        x (List): x
+        y (Tensor): The dtype is same as x.
+
+    Returns:
+        Tensor, has the same dtype as x.
+    """
+    x = utils.sequence_to_tensor(x, y.dtype)
+    return F.tensor_div(x, y)
+
+
+@div.register("Tensor", "List")
+def _tensor_div_list(x, y):
+    """
+    Tensor divided by list
+
+    Args:
+        x (Tensor): x
+        y (List): The dtype is same as x.
+
+    Returns:
+        Tensor, has the same dtype as x.
+    """
+    y = utils.sequence_to_tensor(y, x.dtype)
+    return F.tensor_div(x, y)
diff --git a/mindspore/ops/composite/multitype_ops/floordiv_impl.py b/mindspore/ops/composite/multitype_ops/floordiv_impl.py
index 238867f457..cd1680b9ff 100644
--- a/mindspore/ops/composite/multitype_ops/floordiv_impl.py
+++ b/mindspore/ops/composite/multitype_ops/floordiv_impl.py
@@ -15,6 +15,7 @@
 
 """Implementation for internal polymorphism `floordiv` operations."""
 
+from . import _compile_utils as utils
 from ...composite import base
 from ... import functional as F
 
@@ -48,3 +49,31 @@ def _tensor_floordiv_scalar(x, y):
 def _scalar_floordiv_tensor(x, y):
     """Returns x // y where x is a scalar and y is a tensor. x and y should have same dtype."""
     return F.tensor_floordiv(x, y)
+
+
+@floordiv.register("Tuple", "Tensor")
+def _tuple_floordiv_tensor(x, y):
+    """Returns x // y where x is a tuple and y is a tensor. """
+    x = utils.sequence_to_tensor(x, y.dtype)
+    return F.tensor_floordiv(x, y)
+
+
+@floordiv.register("Tensor", "Tuple")
+def _tensor_floordiv_tuple(x, y):
+    """Returns x // y where x is a tensor and y is a tuple. """
+    y = utils.sequence_to_tensor(y, x.dtype)
+    return F.tensor_floordiv(x, y)
+
+
+@floordiv.register("List", "Tensor")
+def _list_floordiv_tensor(x, y):
+    """Returns x // y where x is a list and y is a tensor. """
+    x = utils.sequence_to_tensor(x, y.dtype)
+    return F.tensor_floordiv(x, y)
+
+
+@floordiv.register("Tensor", "List")
+def _tensor_floordiv_list(x, y):
+    """Returns x // y where x is a tensor and y is a list. """
+    y = utils.sequence_to_tensor(y, x.dtype)
+    return F.tensor_floordiv(x, y)
diff --git a/mindspore/ops/composite/multitype_ops/getitem_impl.py b/mindspore/ops/composite/multitype_ops/getitem_impl.py
index 189b991f41..e34e86bf48 100644
--- a/mindspore/ops/composite/multitype_ops/getitem_impl.py
+++ b/mindspore/ops/composite/multitype_ops/getitem_impl.py
@@ -163,7 +163,7 @@ def _tensor_getitem_by_number(data, number_index):
 
 
 @getitem.register("Tensor", "None")
-def _tensor_getitem_by_none(data, none_index):
+def _tensor_getitem_by_none(data, _):
     """
     For none indexing , expand data with one dim.
 
@@ -174,7 +174,7 @@ def _tensor_getitem_by_none(data, none_index):
     Outputs:
         Tensor, element type is as same as the element type of data.
     """
-    return compile_utils.tensor_index_by_tuple(data, (none_index,))
+    return F.expand_dims(data, 0)
 
 
 @getitem.register("Tensor", "Slice")
@@ -189,7 +189,7 @@ def _tensor_getitem_by_slice(data, slice_index):
     Outputs:
         Tensor, element type is the same as the element type of data.
     """
-    return compile_utils.tensor_index_by_tuple(data, (slice_index,))
+    return compile_utils.tensor_index_by_slice(data, slice_index)
 
 
 @getitem.register("Tensor", "Tensor")
@@ -204,11 +204,11 @@ def _tensor_getitem_by_tensor(data, tensor_index):
     Outputs:
         Tensor, element type is the same as the element type of data.
     """
-    return compile_utils.tensor_index_by_tuple(data, (tensor_index,))
+    return compile_utils.tensor_index_by_tensor(data, tensor_index)
 
 
 @getitem.register("Tensor", "Ellipsis")
-def _tensor_getitem_by_ellipsis(data, ellipsis_index):
+def _tensor_getitem_by_ellipsis(data, _):
     """
     Getting item of tensor by Ellipsis.
 
@@ -219,7 +219,7 @@ def _tensor_getitem_by_ellipsis(data, ellipsis_index):
     Outputs:
         Tensor, same as data.
     """
-    return compile_utils.tensor_index_by_tuple(data, (ellipsis_index,))
+    return data
 
 
 @getitem.register("Tensor", "List")
diff --git a/mindspore/ops/composite/multitype_ops/mod_impl.py b/mindspore/ops/composite/multitype_ops/mod_impl.py
index 3becc239e0..31ce81b189 100644
--- a/mindspore/ops/composite/multitype_ops/mod_impl.py
+++ b/mindspore/ops/composite/multitype_ops/mod_impl.py
@@ -15,6 +15,7 @@
 
 """Implementation for internal polymorphism `mod` operations."""
 
+from . import _compile_utils as utils
 from ...composite import base
 from ... import functional as F
 
@@ -48,3 +49,31 @@ def _tensor_mod_scalar(x, y):
 def _scalar_mod_tensor(x, y):
     """Returns x % y where x is a scalar and y is a tensor. x and y should have same dtype."""
     return F.tensor_mod(x, y)
+
+
+@mod.register("Tuple", "Tensor")
+def _tuple_mod_tensor(x, y):
+    """Returns x % y where x is a tuple and y is a tensor. """
+    x = utils.sequence_to_tensor(x, y.dtype)
+    return F.tensor_mod(x, y)
+
+
+@mod.register("Tensor", "Tuple")
+def _tensor_mod_tuple(x, y):
+    """Returns x % y where x is a tensor and y is a tuple. """
+    y = utils.sequence_to_tensor(y, x.dtype)
+    return F.tensor_mod(x, y)
+
+
+@mod.register("List", "Tensor")
+def _list_mod_tensor(x, y):
+    """Returns x % y where x is a list and y is a tensor. """
+    x = utils.sequence_to_tensor(x, y.dtype)
+    return F.tensor_mod(x, y)
+
+
+@mod.register("Tensor", "List")
+def _tensor_mod_list(x, y):
+    """Returns x % y where x is a tensor and y is a list. """
+    y = utils.sequence_to_tensor(y, x.dtype)
+    return F.tensor_mod(x, y)
diff --git a/mindspore/ops/composite/multitype_ops/mul_impl.py b/mindspore/ops/composite/multitype_ops/mul_impl.py
index 143975baad..b281daa6e5 100644
--- a/mindspore/ops/composite/multitype_ops/mul_impl.py
+++ b/mindspore/ops/composite/multitype_ops/mul_impl.py
@@ -16,6 +16,7 @@
 """Implementation for internal polymorphism `mul` operations."""
 
 from . import _constexpr_utils as const_utils
+from . import _compile_utils as utils
 from ...composite import base
 from ... import functional as F
 
@@ -82,26 +83,26 @@ def _list_mul_scalar(x, y):
     return const_utils.sequence_mul_int(x, y)
 
 
-@mul.register("Tuple", "Number")
-def _tuple_mul_scalar(x, y):
+@mul.register("Number", "List")
+def _scalar_mul_list(x, y):
     """
-    Returns x * y where x is a tuple and y is a number. y must be integer.
+    Returns x * y where x is a number and y is a list. x must be integer.
 
     Outputs:
-        Tuple.
+        List.
     """
-    return const_utils.sequence_mul_int(x, y)
+    return const_utils.sequence_mul_int(y, x)
 
 
-@mul.register("Number", "List")
-def _scalar_mul_list(x, y):
+@mul.register("Tuple", "Number")
+def _tuple_mul_scalar(x, y):
     """
-    Returns x * y where x is a number and y is a list. x must be integer.
+    Returns x * y where x is a tuple and y is a number. y must be integer.
 
     Outputs:
-        List.
+        Tuple.
     """
-    return const_utils.sequence_mul_int(y, x)
+    return const_utils.sequence_mul_int(x, y)
 
 
 @mul.register("Number", "Tuple")
@@ -113,3 +114,67 @@ def _scalar_mul_tuple(x, y):
         Tuple.
     """
     return const_utils.sequence_mul_int(y, x)
+
+
+@mul.register("Tensor", "Tuple")
+def _tensor_mul_tuple(x, y):
+    """
+    Returns x * y where x is a tensor and y is a tuple.
+
+    Args:
+        x (Tensor): x
+        y (Tuple): The dtype is same as x.
+
+    Returns:
+        Tensor, has the same dtype as x.
+    """
+    y = utils.sequence_to_tensor(y, x.dtype)
+    return F.tensor_mul(x, y)
+
+
+@mul.register("Tuple", "Tensor")
+def _tuple_mul_tensor(x, y):
+    """
+    Returns x * y where x is a tuple and y is a tensor.
+
+    Args:
+        x (Tuple): x
+        y (Tensor): The dtype is same as x.
+
+    Returns:
+        Tensor, has the same dtype as x.
+    """
+    x = utils.sequence_to_tensor(x, y.dtype)
+    return F.tensor_mul(x, y)
+
+
+@mul.register("Tensor", "List")
+def _tensor_mul_list(x, y):
+    """
+    Returns x * y where x is a tensor and y is a list.
+
+    Args:
+        x (Tensor): x
+        y (List): The dtype is same as x.
+
+    Returns:
+        Tensor, has the same dtype as x.
+    """
+    y = utils.sequence_to_tensor(y, x.dtype)
+    return F.tensor_mul(x, y)
+
+
+@mul.register("List", "Tensor")
+def _list_mul_tensor(x, y):
+    """
+    Returns x * y where x is a list and y is a tensor.
+
+    Args:
+        x (List): x
+        y (Tensor): The dtype is same as x.
+
+    Returns:
+        Tensor, has the same dtype as x.
+    """
+    x = utils.sequence_to_tensor(x, y.dtype)
+    return F.tensor_mul(x, y)
diff --git a/mindspore/ops/composite/multitype_ops/pow_impl.py b/mindspore/ops/composite/multitype_ops/pow_impl.py
index 76b4c62d61..9361748f24 100644
--- a/mindspore/ops/composite/multitype_ops/pow_impl.py
+++ b/mindspore/ops/composite/multitype_ops/pow_impl.py
@@ -15,6 +15,7 @@
 
 """Implementation for internal polymorphism `pow` operations."""
 
+from . import _compile_utils as utils
 from ...composite import base
 from ... import functional as F
 
@@ -48,3 +49,31 @@ def _tensor_pow_scalar(x, y):
 def _scalar_pow_tensor(x, y):
     """Returns x ** y where x is a scalar and y is a tensor. x and y should have same dtype."""
     return F.tensor_pow(x, y)
+
+
+@pow_.register("Tuple", "Tensor")
+def _tuple_pow_tensor(x, y):
+    """Returns x ** y where x is a tuple and y is a tensor. """
+    x = utils.sequence_to_tensor(x, y.dtype)
+    return F.tensor_pow(x, y)
+
+
+@pow_.register("Tensor", "Tuple")
+def _tensor_pow_tuple(x, y):
+    """Returns x ** y where x is a tensor and y is a tuple. """
+    y = utils.sequence_to_tensor(y, x.dtype)
+    return F.tensor_pow(x, y)
+
+
+@pow_.register("List", "Tensor")
+def _list_pow_tensor(x, y):
+    """Returns x ** y where x is a list and y is a tensor. """
+    x = utils.sequence_to_tensor(x, y.dtype)
+    return F.tensor_pow(x, y)
+
+
+@pow_.register("Tensor", "List")
+def _tensor_pow_list(x, y):
+    """Returns x ** y where x is a tensor and y is a list. """
+    y = utils.sequence_to_tensor(y, x.dtype)
+    return F.tensor_pow(x, y)
diff --git a/mindspore/ops/composite/multitype_ops/sub_impl.py b/mindspore/ops/composite/multitype_ops/sub_impl.py
index 430a360955..90239dabb0 100644
--- a/mindspore/ops/composite/multitype_ops/sub_impl.py
+++ b/mindspore/ops/composite/multitype_ops/sub_impl.py
@@ -15,6 +15,7 @@
 
 """Implementation for internal polymorphism `sub` operations."""
 
+from . import _compile_utils as utils
 from ...composite import base
 from ... import functional as F
 
@@ -48,3 +49,31 @@ def _scalar_sub_tensor(x, y):
 def _tensor_sub_scalar(x, y):
     """Returns x - y where x is a tensor and y is a scalar. x and y should have same dtype."""
     return F.tensor_sub(x, y)
+
+
+@sub.register("Tuple", "Tensor")
+def _tuple_sub_tensor(x, y):
+    """Returns x - y where x is a tuple and y is a tensor. """
+    x = utils.sequence_to_tensor(x, y.dtype)
+    return F.tensor_sub(x, y)
+
+
+@sub.register("Tensor", "Tuple")
+def _tensor_sub_tuple(x, y):
+    """Returns x - y where x is a tensor and y is a tuple. """
+    y = utils.sequence_to_tensor(y, x.dtype)
+    return F.tensor_sub(x, y)
+
+
+@sub.register("List", "Tensor")
+def _list_sub_tensor(x, y):
+    """Returns x - y where x is a list and y is a tensor. """
+    x = utils.sequence_to_tensor(x, y.dtype)
+    return F.tensor_sub(x, y)
+
+
+@sub.register("Tensor", "List")
+def _tensor_sub_list(x, y):
+    """Returns x - y where x is a tensor and y is a list. """
+    y = utils.sequence_to_tensor(y, x.dtype)
+    return F.tensor_sub(x, y)
diff --git a/mindspore/ops/functional.py b/mindspore/ops/functional.py
index c42530111d..2abf1f3567 100644
--- a/mindspore/ops/functional.py
+++ b/mindspore/ops/functional.py
@@ -131,11 +131,13 @@ scatter_update = P.ScatterUpdate()
 scatter_nd_update = P.ScatterNdUpdate()
 stack = P.Stack()
 
+
 def pack(x):
     print("WARNING: 'pack' is deprecated from version 1.1 and will be removed in a future version, use 'stack' instead"
           ".")
     return stack(x)
 
+
 partial = P.Partial()
 # depend: mount a node to another node
 depend = P.Depend()
@@ -221,13 +223,6 @@ sparse_tensor_get_values = Primitive('SparseTensorGetValues')
 sparse_tensor_get_indices = Primitive('SparseTensorGetIndices')
 sparse_tensor_get_dense_shape = Primitive('SparseTensorGetDenseShape')
 
-tensor_operator_registry.register('__add__', tensor_add)
-tensor_operator_registry.register('__sub__', tensor_sub)
-tensor_operator_registry.register('__mul__', tensor_mul)
-tensor_operator_registry.register('__truediv__', tensor_div)
-tensor_operator_registry.register('__mod__', tensor_mod)
-tensor_operator_registry.register('__pow__', tensor_pow)
-tensor_operator_registry.register('__floordiv__', tensor_floordiv)
 tensor_operator_registry.register('all', P.ReduceAll)
 tensor_operator_registry.register('any', P.ReduceAny)
 tensor_operator_registry.register('abs', P.Abs)
diff --git a/tests/st/pynative/test_tensor_augassign.py b/tests/st/pynative/test_tensor_augassign.py
new file mode 100644
index 0000000000..a9c9a4dd44
--- /dev/null
+++ b/tests/st/pynative/test_tensor_augassign.py
@@ -0,0 +1,651 @@
+# Copyright 2021 Huawei Technologies Co., Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ============================================================================
+""" test_tensor_setitem """
+import numpy as np
+import pytest
+
+from mindspore import Tensor, context
+from mindspore import dtype as mstype
+
+
+def setup_module():
+    context.set_context(mode=context.PYNATIVE_MODE)
+
+
+# GPU: does not supported op "FloorMod"
+@pytest.mark.level0
+@pytest.mark.platform_arm_ascend_training
+@pytest.mark.platform_x86_ascend_training
+@pytest.mark.env_onecard
+def test_tesnsor_augassign_by_slice():
+    input_np_3d = np.arange(120).reshape(4, 5, 6).astype(np.float32)
+    input_tensor_3d = Tensor(input_np_3d, mstype.float32)
+    index_slice_1 = slice(1, None, None)
+    index_slice_2 = slice(None, 4, None)
+    index_slice_3 = slice(-3, 4, None)
+    index_slice_4 = slice(2, -1, None)
+    index_slice_7 = slice(1, 5, None)
+    index_slice_8 = slice(-5, 3, None)
+
+    value_number = 3
+    value_list_1_ele = [2]
+    value_list_mul_ele = [10, 20, 30, 40, 50, 60]
+    value_list_much_ele = [10, 20, 30, 40, 50, 60, 70]
+
+    input_tensor_3d[index_slice_1] += value_number
+    input_np_3d[index_slice_1] += value_number
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[index_slice_2] -= value_list_1_ele
+    input_np_3d[index_slice_2] -= value_list_1_ele
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[index_slice_3] *= value_list_mul_ele
+    input_np_3d[index_slice_3] *= value_list_mul_ele
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[index_slice_4] /= value_number
+    input_np_3d[index_slice_4] /= value_number
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[index_slice_7] /= value_number
+    input_np_3d[index_slice_7] /= value_number
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[index_slice_8] += value_number
+    input_np_3d[index_slice_8] += value_number
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    with pytest.raises(ValueError):
+        input_tensor_3d[index_slice_8] /= value_list_much_ele
+
+
+# GPU: does not supported op "FloorMod"
+@pytest.mark.level0
+@pytest.mark.platform_arm_ascend_training
+@pytest.mark.platform_x86_ascend_training
+@pytest.mark.env_onecard
+def test_tesnsor_augassign_by_ellipsis():
+    input_np_3d = np.arange(24).reshape(2, 3, 4).astype(np.float32)
+    input_tensor_3d = Tensor(input_np_3d, mstype.float32)
+
+    value_number_1, value_number_2 = 1, 2.0
+
+    value_np_1 = np.array([1])
+    value_np_2 = np.array([1, 2, 3, 4])
+    value_np_3 = np.arange(12).reshape(3, 4)
+    value_tensor_1 = Tensor(value_np_1)
+    value_tensor_2 = Tensor(value_np_2)
+    value_tensor_3 = Tensor(value_np_3)
+
+    value_tuple_1_ele = (0.5,)
+    value_tuple_4_ele = (0.1, 0.2, 0.3, 0.4)
+
+    value_list_1_ele = [1.5]
+    value_list_4_ele = [1.1, 1.2, 1.3, 1.4]
+
+    input_tensor_3d[...] += value_number_1
+    input_np_3d[...] += value_number_1
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[...] -= value_number_2
+    input_np_3d[...] -= value_number_2
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[...] *= value_tensor_1
+    input_np_3d[...] *= value_np_1
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[...] /= value_tensor_2
+    input_np_3d[...] /= value_np_2
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[...] /= value_tensor_3
+    input_np_3d[...] /= value_np_3
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[...] -= value_tuple_1_ele
+    input_np_3d[...] -= value_tuple_1_ele
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[...] *= value_tuple_4_ele
+    input_np_3d[...] *= value_tuple_4_ele
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[...] -= value_list_1_ele
+    input_np_3d[...] -= value_list_1_ele
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[...] *= value_list_4_ele
+    input_np_3d[...] *= value_list_4_ele
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+
+# GPU: does not supported op "FloorMod"
+@pytest.mark.level0
+@pytest.mark.platform_arm_ascend_training
+@pytest.mark.platform_x86_ascend_training
+@pytest.mark.env_onecard
+def test_tesnsor_augassign_by_bool():
+    input_np_3d = np.arange(120).reshape(4, 5, 6).astype(np.float32)
+    input_tensor_3d = Tensor(input_np_3d, mstype.float32)
+
+    index_bool_1 = True
+    index_bool_2 = False
+
+    value_number = 1
+
+    value_np_1 = np.array([1], np.float32)
+    value_np_2 = np.array([1, 2, 3, 4, 5, 6], np.float32)
+    value_np_3 = np.arange(1, 31).astype(np.float32).reshape(5, 6)
+    value_np_4 = np.arange(1, 121).astype(np.float32).reshape(4, 5, 6)
+    value_tensor_1 = Tensor(value_np_1, mstype.float32)
+    value_tensor_2 = Tensor(value_np_2, mstype.float32)
+    value_tensor_3 = Tensor(value_np_3, mstype.float32)
+    value_tensor_4 = Tensor(value_np_4, mstype.float32)
+
+    value_tuple_1_ele = (0.5,)
+    value_tuple_6_ele = (0.1, 0.2, 0.3, 0.4, 0.5, 0.6)
+
+    value_list_1_ele = [1.5]
+    value_list_6_ele = [1.1, 1.2, 1.3, 1.4, 1.5, 1.6]
+
+    input_tensor_3d[index_bool_1] += value_number
+    input_np_3d[index_bool_1] += value_number
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[index_bool_1] -= value_tensor_1
+    input_np_3d[index_bool_1] -= value_np_1
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[index_bool_1] *= value_tensor_2
+    input_np_3d[index_bool_1] *= value_np_2
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[index_bool_1] -= value_tensor_3
+    input_np_3d[index_bool_1] -= value_np_3
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[index_bool_1] //= value_tensor_4
+    input_np_3d[index_bool_1] //= value_np_4
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[index_bool_1] %= value_tuple_1_ele
+    input_np_3d[index_bool_1] %= value_tuple_1_ele
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[index_bool_1] %= value_tuple_6_ele
+    input_np_3d[index_bool_1] %= value_tuple_6_ele
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[index_bool_1] %= value_list_1_ele
+    input_np_3d[index_bool_1] %= value_list_1_ele
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[index_bool_1] -= value_list_6_ele
+    input_np_3d[index_bool_1] -= value_list_6_ele
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    with pytest.raises(IndexError):
+        input_tensor_3d[index_bool_2] *= value_tensor_2
+
+
+# GPU: does not supported op "FloorMod"
+@pytest.mark.level0
+@pytest.mark.platform_arm_ascend_training
+@pytest.mark.platform_x86_ascend_training
+@pytest.mark.env_onecard
+def test_tesnsor_augassign_by_number():
+    input_np_1d = np.arange(4).astype(np.float32)
+    input_tensor_1d = Tensor(input_np_1d, mstype.float32)
+    input_np_3d = np.arange(80).reshape(4, 5, 4).astype(np.float32)
+    input_tensor_3d = Tensor(input_np_3d, mstype.float32)
+
+    number_index_1, number_index_2, number_index_3, number_index_4 = 0, 3, 4, 3.4
+
+    value_number = 2
+
+    value_np_scalar = np.array(5)
+    value_np_1_ele = np.array([1])
+    value_np_1d = np.array([1, 2, 3, 4])
+    value_np_2d = np.arange(20).reshape(5, 4)
+    value_tensor_scalar = Tensor(value_np_scalar, mstype.float32)
+    value_tensor_1_ele = Tensor(value_np_1_ele, mstype.float32)
+    value_tensor_1d = Tensor(value_np_1d, mstype.float32)
+    value_tensor_2d = Tensor(value_np_2d, mstype.float32)
+
+    value_tuple_1_ele = (100,)
+    value_tuple_mul_ele = (10, 20, 30, 40)
+    value_tuple_much_ele = (10, 20, 30, 40, 10)
+    value_tuple_empty = ()
+
+    value_list_1_ele = [101]
+    value_list_mul_ele = [11, 21, 31, 41]
+    value_list_much_ele = [12, 22, 33, 43, 18]
+    value_list_empty = []
+
+    input_tensor_1d[number_index_1] += value_number
+    input_np_1d[number_index_1] += value_number
+    assert np.allclose(input_tensor_1d.asnumpy(), input_np_1d, 0.0001, 0.0001)
+
+    input_tensor_1d[number_index_2] -= value_number
+    input_np_1d[number_index_2] -= value_number
+    assert np.allclose(input_tensor_1d.asnumpy(), input_np_1d, 0.0001, 0.0001)
+
+    input_tensor_3d[number_index_1] *= value_number
+    input_np_3d[number_index_1] *= value_number
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[number_index_2] /= value_number
+    input_np_3d[number_index_2] /= value_number
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_1d[number_index_1] //= value_tensor_scalar
+    input_np_1d[number_index_1] //= value_np_scalar
+    assert np.allclose(input_tensor_1d.asnumpy(), input_np_1d, 0.0001, 0.0001)
+
+    input_tensor_3d[number_index_1] *= value_tensor_scalar
+    input_np_3d[number_index_1] *= value_np_scalar
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[number_index_2] %= value_tensor_1_ele
+    input_np_3d[number_index_2] %= value_np_1_ele
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[number_index_1] += value_tensor_1d
+    input_np_3d[number_index_1] += value_np_1d
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[number_index_2] -= value_tensor_2d
+    input_np_3d[number_index_2] -= value_np_2d
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_1d[number_index_1] += value_tuple_1_ele
+    input_np_1d[number_index_1] += value_tuple_1_ele
+    assert np.allclose(input_tensor_1d.asnumpy(), input_np_1d, 0.0001, 0.0001)
+
+    input_tensor_3d[number_index_1] -= value_tuple_1_ele
+    input_np_3d[number_index_1] -= value_tuple_1_ele
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[number_index_1] *= value_tuple_mul_ele
+    input_np_3d[number_index_1] *= value_tuple_mul_ele
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_1d[number_index_2] += value_list_1_ele
+    input_np_1d[number_index_2] += value_list_1_ele
+    assert np.allclose(input_tensor_1d.asnumpy(), input_np_1d, 0.0001, 0.0001)
+
+    input_tensor_3d[number_index_1] -= value_list_1_ele
+    input_np_3d[number_index_1] -= value_list_1_ele
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[number_index_2] *= value_list_mul_ele
+    input_np_3d[number_index_2] *= value_list_mul_ele
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    with pytest.raises(IndexError):
+        input_tensor_1d[number_index_3] += value_number
+    with pytest.raises(IndexError):
+        input_tensor_3d[number_index_3] -= value_number
+    with pytest.raises(IndexError):
+        input_tensor_1d[number_index_4] *= value_number
+    with pytest.raises(IndexError):
+        input_tensor_3d[number_index_4] /= value_number
+
+    with pytest.raises(ValueError):
+        input_tensor_1d[number_index_1] *= value_tuple_mul_ele
+    with pytest.raises(ValueError):
+        input_tensor_3d[number_index_1] *= value_tuple_much_ele
+    with pytest.raises(RuntimeError):
+        input_tensor_1d[number_index_1] /= value_tuple_empty
+
+    with pytest.raises(ValueError):
+        input_tensor_3d[number_index_2] //= value_list_much_ele
+    with pytest.raises(ValueError):
+        input_tensor_3d[number_index_2] *= value_list_empty
+
+
+# GPU: does not supported op "FloorMod"
+@pytest.mark.level0
+@pytest.mark.platform_arm_ascend_training
+@pytest.mark.platform_x86_ascend_training
+@pytest.mark.env_onecard
+def test_tesnsor_augassign_by_tensor():
+    input_np_3d = np.arange(120).reshape(4, 5, 6).astype(np.float32)
+    input_tensor_3d = Tensor(input_np_3d, mstype.float32)
+
+    index_np_1d_1ele = np.random.randint(4, size=1)
+    index_np_1d = np.random.randint(4, size=6)
+    index_np_2d = np.random.randint(4, size=(5, 6))
+    index_np_3d = np.random.randint(4, size=(4, 5, 6))
+
+    index_tensor_1d_1ele = Tensor(index_np_1d_1ele, mstype.int32)
+    index_tensor_1d = Tensor(index_np_1d, mstype.int32)
+    index_tensor_2d = Tensor(index_np_2d, mstype.int32)
+    index_tensor_3d = Tensor(index_np_3d, mstype.int32)
+
+    value_number = 1
+
+    value_np_1 = np.array([1])
+    value_np_2 = np.array([1, 2, 3, 4, 5, 6])
+    value_np_3 = np.arange(1, 31).reshape(5, 6)
+    value_np_4 = np.arange(1, 181).reshape(6, 5, 6)
+    value_tensor_1 = Tensor(value_np_1)
+    value_tensor_2 = Tensor(value_np_2)
+    value_tensor_3 = Tensor(value_np_3)
+    value_tensor_4 = Tensor(value_np_4)
+
+    value_tuple_1_ele = (0.5,)
+    value_tuple_6_ele = (0.1, 0.2, 0.3, 0.4, 0.5, 0.6)
+
+    value_list_1_ele = [1.5]
+    value_list_6_ele = [1.1, 1.2, 1.3, 1.4, 1.5, 1.6]
+
+    input_tensor_3d[index_tensor_1d_1ele] += value_number
+    input_np_3d[index_np_1d_1ele] += value_number
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[index_tensor_1d_1ele] -= value_tensor_2
+    input_np_3d[index_np_1d_1ele] -= value_np_2
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[index_tensor_1d_1ele] /= value_tuple_6_ele
+    input_np_3d[index_np_1d_1ele] /= value_tuple_6_ele
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[index_tensor_1d_1ele] *= value_list_1_ele
+    input_np_3d[index_np_1d_1ele] *= value_list_1_ele
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[index_tensor_1d] += value_number
+    input_np_3d[index_np_1d] += value_number
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[index_tensor_1d] -= value_tensor_1
+    input_np_3d[index_np_1d] -= value_np_1
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[index_tensor_1d] /= value_tuple_1_ele
+    input_np_3d[index_np_1d] /= value_tuple_1_ele
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[index_tensor_1d] += value_list_6_ele
+    input_np_3d[index_np_1d] += value_list_6_ele
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[index_tensor_2d] -= value_number
+    input_np_3d[index_np_2d] -= value_number
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[index_tensor_2d] *= value_tensor_2
+    input_np_3d[index_np_2d] *= value_np_2
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[index_tensor_2d] /= value_tensor_4
+    input_np_3d[index_np_2d] /= value_np_4
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[index_tensor_2d] += value_tuple_6_ele
+    input_np_3d[index_np_2d] += value_tuple_6_ele
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[index_tensor_2d] -= value_list_1_ele
+    input_np_3d[index_np_2d] -= value_list_1_ele
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[index_tensor_3d] *= value_number
+    input_np_3d[index_np_3d] *= value_number
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[index_tensor_3d] /= value_tensor_1
+    input_np_3d[index_np_3d] /= value_np_1
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[index_tensor_3d] += value_tensor_3
+    input_np_3d[index_np_3d] += value_np_3
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[index_tensor_3d] /= value_tuple_1_ele
+    input_np_3d[index_np_3d] /= value_tuple_1_ele
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[index_tensor_3d] -= value_list_6_ele
+    input_np_3d[index_np_3d] -= value_list_6_ele
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+
+# GPU: does not supported op "FloorMod"
+@pytest.mark.level0
+@pytest.mark.platform_arm_ascend_training
+@pytest.mark.platform_x86_ascend_training
+@pytest.mark.env_onecard
+def test_tesnsor_augassign_by_list():
+    input_np_3d = np.arange(120).reshape(4, 5, 6).astype(np.float32)
+    input_tensor_3d = Tensor(input_np_3d, mstype.float32)
+
+    list_index_empty = []
+    list_index_int_1 = [2]
+    list_index_int_2 = [3, 1]
+    list_index_int_overflow = [4, 2]
+    list_index_bool_1 = [False, False, False, False]
+    list_index_bool_2 = [True, True, True, True]
+    list_index_bool_3 = [True, False, True, False]
+    list_index_mix_1 = [True, 0]
+    list_index_mix_2 = [3, False]
+
+    value_number = 2
+
+    value_np_scalar = np.array(100)
+    value_np_1_ele = np.array([1])
+    value_np_1d = np.array([1, 2, 3, 4, 5, 6])
+    value_np_2d = np.arange(1, 31).reshape(5, 6)
+    value_np_3d = np.arange(1, 61).reshape(2, 5, 6)
+    value_tensor_scalar = Tensor(value_np_scalar, mstype.float32)
+    value_tensor_1_ele = Tensor(value_np_1_ele, mstype.float32)
+    value_tensor_1d = Tensor(value_np_1d, mstype.float32)
+    value_tensor_2d = Tensor(value_np_2d, mstype.float32)
+    value_tensor_3d = Tensor(value_np_3d, mstype.float32)
+
+    input_tensor_3d[list_index_int_1] += value_number
+    input_np_3d[list_index_int_1] += value_number
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[list_index_int_1] += value_tensor_scalar
+    input_np_3d[list_index_int_1] += value_np_scalar
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[list_index_int_1] -= value_tensor_1_ele
+    input_np_3d[list_index_int_1] -= value_np_1_ele
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[list_index_int_1] *= value_tensor_1d
+    input_np_3d[list_index_int_1] *= value_np_1d
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[list_index_int_1] /= value_tensor_2d
+    input_np_3d[list_index_int_1] /= value_np_2d
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[list_index_int_2] += value_number
+    input_np_3d[list_index_int_2] += value_number
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[list_index_int_2] //= value_tensor_scalar
+    input_np_3d[list_index_int_2] //= value_np_scalar
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[list_index_int_2] *= value_tensor_1_ele
+    input_np_3d[list_index_int_2] *= value_np_1_ele
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[list_index_int_2] %= value_tensor_1d
+    input_np_3d[list_index_int_2] %= value_np_1d
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[list_index_int_2] += value_tensor_2d
+    input_np_3d[list_index_int_2] += value_np_2d
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[list_index_int_2] -= value_tensor_3d
+    input_np_3d[list_index_int_2] -= value_np_3d
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[list_index_bool_2] += value_number
+    input_np_3d[list_index_bool_2] += value_number
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[list_index_bool_2] *= value_tensor_scalar
+    input_np_3d[list_index_bool_2] *= value_np_scalar
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[list_index_bool_2] /= value_tensor_1_ele
+    input_np_3d[list_index_bool_2] /= value_np_1_ele
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[list_index_bool_2] //= value_tensor_1d
+    input_np_3d[list_index_bool_2] //= value_np_1d
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[list_index_bool_2] %= value_tensor_2d
+    input_np_3d[list_index_bool_2] %= value_np_2d
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[list_index_bool_3] += value_number
+    input_np_3d[list_index_bool_3] += value_number
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[list_index_bool_3] *= value_tensor_scalar
+    input_np_3d[list_index_bool_3] *= value_np_scalar
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[list_index_bool_3] += value_tensor_1_ele
+    input_np_3d[list_index_bool_3] += value_np_1_ele
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[list_index_bool_3] -= value_tensor_1d
+    input_np_3d[list_index_bool_3] -= value_np_1d
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[list_index_bool_3] *= value_tensor_2d
+    input_np_3d[list_index_bool_3] *= value_np_2d
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[list_index_bool_3] /= value_tensor_3d
+    input_np_3d[list_index_bool_3] /= value_np_3d
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[list_index_mix_1] += value_number
+    input_np_3d[list_index_mix_1] += value_number
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[list_index_mix_1] *= value_tensor_scalar
+    input_np_3d[list_index_mix_1] *= value_np_scalar
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[list_index_mix_1] += value_tensor_1_ele
+    input_np_3d[list_index_mix_1] += value_np_1_ele
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[list_index_mix_1] -= value_tensor_1d
+    input_np_3d[list_index_mix_1] -= value_np_1d
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[list_index_mix_1] *= value_tensor_2d
+    input_np_3d[list_index_mix_1] *= value_np_2d
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[list_index_mix_1] /= value_tensor_3d
+    input_np_3d[list_index_mix_1] /= value_np_3d
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[list_index_mix_2] += value_number
+    input_np_3d[list_index_mix_2] += value_number
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[list_index_mix_2] *= value_tensor_scalar
+    input_np_3d[list_index_mix_2] *= value_np_scalar
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[list_index_mix_2] += value_tensor_1_ele
+    input_np_3d[list_index_mix_2] += value_np_1_ele
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[list_index_mix_2] -= value_tensor_1d
+    input_np_3d[list_index_mix_2] -= value_np_1d
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[list_index_mix_2] *= value_tensor_2d
+    input_np_3d[list_index_mix_2] *= value_np_2d
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[list_index_mix_2] /= value_tensor_3d
+    input_np_3d[list_index_mix_2] /= value_np_3d
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    with pytest.raises(IndexError):
+        input_tensor_3d[list_index_empty] += value_number
+    with pytest.raises(IndexError):
+        input_tensor_3d[list_index_int_overflow] += value_number
+    with pytest.raises(IndexError):
+        input_tensor_3d[list_index_bool_1] += value_number
+
+
+# GPU: does not supported op "FloorMod"
+@pytest.mark.level0
+@pytest.mark.platform_arm_ascend_training
+@pytest.mark.platform_x86_ascend_training
+@pytest.mark.env_onecard
+def test_tesnsor_augassign_by_tuple():
+    input_np_3d = np.arange(120).reshape(4, 5, 6).astype(np.float32)
+    input_tensor_3d = Tensor(input_np_3d, mstype.float32)
+
+    index_tuple_1 = (slice(1, 3, 1), ..., [1, 3, 2])
+    index_tuple_2 = (2, 3, 4)
+    index_tuple_4 = ([2, 3], True)
+    index_tuple_5 = (False, 3)
+    index_tuple_6 = (False, slice(3, 1, -1))
+    index_tuple_7 = (..., slice(None, 6, 2))
+
+    value_number = 2
+
+    value_np_scalar = np.array(100)
+    value_tensor_scalar = Tensor(value_np_scalar, mstype.float32)
+
+    input_tensor_3d[index_tuple_1] += value_number
+    input_np_3d[index_tuple_1] += value_number
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[index_tuple_1] -= Tensor(np.ones((2, 5, 3)), mstype.float32)
+    input_np_3d[index_tuple_1] -= np.ones((2, 5, 3))
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[index_tuple_2] *= value_tensor_scalar
+    input_np_3d[index_tuple_2] *= value_np_scalar
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[index_tuple_4] //= value_number
+    input_np_3d[index_tuple_4] //= value_number
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    input_tensor_3d[index_tuple_7] += value_number
+    input_np_3d[index_tuple_7] += value_number
+    assert np.allclose(input_tensor_3d.asnumpy(), input_np_3d, 0.0001, 0.0001)
+
+    with pytest.raises(IndexError):
+        input_tensor_3d[index_tuple_5] *= value_number
+
+    with pytest.raises(IndexError):
+        input_tensor_3d[index_tuple_6] %= value_number
diff --git a/tests/ut/python/ir/test_tensor.py b/tests/ut/python/ir/test_tensor.py
index 9ed92b418d..803e146dd6 100644
--- a/tests/ut/python/ir/test_tensor.py
+++ b/tests/ut/python/ir/test_tensor.py
@@ -470,8 +470,8 @@ def test_tensor_operation():
     assert np.all(res.asnumpy() == np.ones((3, 3)))
     x %= 3
     assert np.all(x.asnumpy() == np.ones((3, 3)))
-    with pytest.raises(ValueError):
-        res = x * (2, 3)
+    res = x * (2, 3, 4)
+    assert np.all(res.asnumpy() == np.ones((3, 3)) * (2, 3, 4))
     res = 5 % x
     assert np.all(x.asnumpy() == np.ones((3, 3)))
     res = 5 // x
diff --git a/tests/ut/python/pynative_mode/test_implicit_conversion.py b/tests/ut/python/pynative_mode/test_implicit_conversion.py
index 33a02d71fb..8ca75acf31 100644
--- a/tests/ut/python/pynative_mode/test_implicit_conversion.py
+++ b/tests/ut/python/pynative_mode/test_implicit_conversion.py
@@ -117,17 +117,17 @@ def test_float_tensor_and_str_add():
 def test_float_tensor_and_tuple_add():
     x = Tensor(np.array([[0.1, 0.2, 0.3], [0.4, 0.5, 0.6]], dtype=np.float32))
     y = (1, 2, 3)
-    with pytest.raises(TypeError) as er:
-        ret = x + y
-    assert "For 'Add', the 1th input is a not support implicit conversion type: tuple" in str(er.value)
+    ret_actual = x + y
+    ret_expect = Tensor(np.array([[1.1, 2.2, 3.3], [1.4, 2.5, 3.6]], dtype=np.float32))
+    assert (ret_actual.asnumpy() == ret_expect.asnumpy()).all()
 
 
 def test_float_tensor_and_list_add():
     x = Tensor(np.array([[0.1, 0.2, 0.3], [0.4, 0.5, 0.6]], dtype=np.float32))
     y = [1, 2, 3]
-    with pytest.raises(TypeError) as er:
-        ret = x + y
-    assert "For 'Add', the 1th input is a not support implicit conversion type: list" in str(er.value)
+    ret_actual = x + y
+    ret_expect = Tensor(np.array([[1.1, 2.2, 3.3], [1.4, 2.5, 3.6]], dtype=np.float32))
+    assert (ret_actual.asnumpy() == ret_expect.asnumpy()).all()
 
 
 def test_float_tensor_and_bool_tensors_add_grad():