!43675 shield SparseApplyAdagrad Primitive interface

Merge pull request !43675 from 李林杰/1011_fix_some_issues_master

docs/api/api_python/ops/mindspore.ops.SparseApplyAdagrad.rst

@@ -3,36 +3,5 @@ mindspore.ops.SparseApplyAdagrad
 
 .. py:class:: mindspore.ops.SparseApplyAdagrad(lr, update_slots=True, use_locking=False)
 
-    根据Adagrad算法更新相关参数。
+    已弃用
 
-    .. math::
-        \begin{array}{ll} \\
-            accum += grad * grad \\
-            var -= lr * grad * \frac{1}{\sqrt{accum}}
-        \end{array}
-
-    `var` 、 `accum` 和 `grad` 的输入遵循隐式类型转换规则，使数据类型一致。如果它们具有不同的数据类型，则低精度数据类型将转换为相对最高精度的数据类型。
-
-    参数：
-        - **lr** (float) - 学习率。
-        - **update_slots** (bool) - 如果为True，则将更新 `accum` 。默认值：True。
-        - **use_locking** (bool) - 是否对参数更新加锁保护。默认值：False。
-
-    输入：
-        - **var** (Parameter) - 要更新的变量。为任意维度，其数据类型为float16或float32。
-        - **accum** (Parameter) - 要更新的累积。shape和数据类型必须与 `var` 相同。
-        - **grad** (Tensor) - 梯度，为一个Tensor。shape和数据类型必须与 `var` 相同，且需要满足 :math:`grad.shape[1:] = var.shape[1:] if var.shape > 1`。
-        - **indices** (Tensor) - `var` 和 `accum` 第一维度的索引向量，数据类型为int32，且需要保证 :math:`indices.shape[0] = grad.shape[0]`。
-
-    输出：
-        2个Tensor组成的tuple，更新后的参数。
-
-        - **var** (Tensor) - shape和数据类型与 `var` 相同。
-        - **accum** (Tensor) - shape和数据类型与 `accum` 相同。
-
-    异常：
-        - **TypeError** - 如果 `lr` 不是float类型。
-        - **TypeError** - 如果 `update_slots` 或者 `use_locking` 不是布尔值。
-        - **TypeError** - 如果 `var` 、 `accum` 、 `lr` 或 `grad` 的数据类型既不是float16也不是float32。
-        - **TypeError** - 如果 `indices` 的数据类型不是int32。
-        - **RuntimeError** - 如果 `var` 、 `accum` 和 `grad` 不支持数据类型转换。

mindspore/python/mindspore/ops/_utils/utils.py

@@ -76,8 +76,8 @@ def get_broadcast_shape(x_shape, y_shape, prim_name, shape_type="", arg_name1="x
                 raise ValueError(f"For '{prim_name}', {arg_name1}.shape and {arg_name2}.shape need to "
                                  f"broadcast. The value of {arg_name1}.shape[{i}] or {arg_name2}.shape[{i}]"
                                  f" must be 1 or -1 when they are not the same, "
-                                 f"but got {arg_name1}.shape[{i}] = {x_shape} "
+                                 f"but got {arg_name1}.shape = {x_shape} "
-                                 f"and {arg_name2}.shape[{i}] = {y_shape}.")
+                                 f"and {arg_name2}.shape = {y_shape}.")
 
     broadcast_shape_front = y_shape[0: y_len - length] if length == x_len else x_shape[0: x_len - length]
     broadcast_shape = list(broadcast_shape_front) + broadcast_shape_back

mindspore/python/mindspore/ops/operations/nn_ops.py

@@ -5982,71 +5982,8 @@ class ApplyAdagradV2(Primitive):
 
 
 class SparseApplyAdagrad(Primitive):
-    r"""
+    """
-    Updates relevant entries according to the adagrad scheme.
+    Deprecated
-
-    .. math::
-        \begin{array}{ll} \\
-            accum += grad * grad \\
-            var -= lr * grad * (1 / sqrt(accum))
-        \end{array}
-
-    Inputs of `var`, `accum` and `grad` comply with the implicit type conversion rules
-    to make the data types consistent.
-    If they have different data types, the lower priority data type will be converted to
-    the relatively highest priority data type.
-
-    Args:
-        lr (float): Learning rate.
-        update_slots (bool): If `True`, `accum` will be updated. Default: True.
-        use_locking (bool): If true, the `var` and `accum` tensors will be protected from being updated.
-            Default: False.
-
-    Inputs:
-        - **var** (Parameter) - Variable to be updated. The data type must be float16 or float32.
-          The shape is :math:`(N, *)` where :math:`*` means, any number of additional dimensions.
-        - **accum** (Parameter) - Accumulation to be updated. The shape and data type must be the same as `var`.
-        - **grad** (Tensor) - Gradients has the same data type as `var` and
-          grad.shape[1:] = var.shape[1:] if var.shape > 1.
-        - **indices** (Tensor) - A vector of indices into the first dimension of `var` and `accum`.
-          The type must be int32 and indices.shape[0] = grad.shape[0].
-
-    Outputs:
-        Tuple of 2 tensors, the updated parameters.
-
-        - **var** (Tensor) - The same shape and data type as `var`.
-        - **accum** (Tensor) - The same shape and data type as `accum`.
-
-    Raises:
-        TypeError: If `lr` is not a float.
-        TypeError: If neither `update_slots` nor `use_locking` is a bool.
-        TypeError: If dtype of `var`, `accum` or `grad` is neither float16 nor float32.
-        TypeError: If dtype of `indices` is not int32.
-        RuntimeError: If the data type of `var`, `accum` and `grad` conversion of Parameter is not supported.
-
-
-    Supported Platforms:
-        ``Ascend`` ``CPU`` ``GPU``
-
-    Examples:
-        >>> class Net(nn.Cell):
-        ...     def __init__(self):
-        ...         super(Net, self).__init__()
-        ...         self.sparse_apply_adagrad = ops.SparseApplyAdagrad(lr=1e-8)
-        ...         self.var = Parameter(Tensor(np.array([[[0.2]]]).astype(np.float32)), name="var")
-        ...         self.accum = Parameter(Tensor(np.array([[[0.1]]]).astype(np.float32)), name="accum")
-        ...     def construct(self, grad, indices):
-        ...         out = self.sparse_apply_adagrad(self.var, self.accum, grad, indices)
-        ...         return out
-        ...
-        >>> net = Net()
-        >>> grad = Tensor(np.array([[[0.7]]]).astype(np.float32))
-        >>> indices = Tensor([1], mindspore.int32)
-        >>> output = net(grad, indices)
-        >>> print(output)
-        (Tensor(shape=[1, 1, 1], dtype=Float32, value=
-        [[[ 2.00000003e-01]]]), Tensor(shape=[1, 1, 1], dtype=Float32, value=
-        [[[ 1.00000001e-01]]]))
     """
 
     __mindspore_signature__ = (
@@ -6056,6 +5993,7 @@ class SparseApplyAdagrad(Primitive):
         sig.make_sig('indices', dtype=sig.sig_dtype.T1)
     )
 
+    @deprecated("1.9", "SparseApplyAdagrad", False)
     @prim_attr_register
     def __init__(self, lr, update_slots=True, use_locking=False):
         """Initialize SparseApplyAdagrad."""