!48955 [bugfix][api] Fix I6FCIG

Merge pull request !48955 from shaojunsong/fix/I6FCIG

docs/api/api_python/nn/mindspore.nn.HingeEmbeddingLoss.rst

@@ -24,7 +24,7 @@ mindspore.nn.HingeEmbeddingLoss
     其中 :math:`L = \{l_1,\dots,l_N\}^\top`。
 
     参数：
-        - **margin** (float) - Hinge Embedding Loss公式定义的阈值 :math:`margin`。公式中表示为 :math:`\Delta`。默认值：1.0。
+        - **margin** (float, int) - Hinge Embedding Loss公式定义的阈值 :math:`margin`。公式中表示为 :math:`\Delta`。默认值：1.0。
         - **reduction** (str) - 指定应用于输出结果的计算方式，'none'、'mean'、'sum'，默认值：'mean'。
 
     输入：
@@ -35,8 +35,8 @@ mindspore.nn.HingeEmbeddingLoss
         Tensor或Tensor scalar，根据 :math:`reduction` 计算的loss。
 
     异常：
-        - **TypeError** - `logits` 不是Tensor。
-        - **TypeError** - `labels` 不是Tensor。
-        - **TypeError** - `margin` 不是float。
+        - **TypeError** - `logits` 不是数据类型为float的Tensor。
+        - **TypeError** - `labels` 不是数据类型为float的Tensor。
+        - **TypeError** - `margin` 不是float或int。
         - **ValueError** - `labels` 和 `logits` shape不一致。
         - **ValueError** - `reduction` 不是"none"、"mean"或者"sum"。

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

@@ -26,15 +26,15 @@ mindspore.ops.hinge_embedding_loss
     参数：
         - **inputs** (Tensor) - 预测值，公式中表示为 :math:`x`，shape为 :math:`(*)`。`*` 代表着任意数量的维度。
         - **targets** (Tensor) - 标签值，公式中表示为 :math:`y`，和 `logits` 具有相同shape，包含1或-1。
-        - **margin** (float) - Hinge Embedding Loss公式定义的阈值 :math:`margin`。公式中表示为 :math:`\Delta`。默认值：1.0。
+        - **margin** (float, int) - Hinge Embedding Loss公式定义的阈值 :math:`margin`。公式中表示为 :math:`\Delta`。默认值：1.0。
         - **reduction** (str) - 指定应用于输出结果的计算方式，'none'、'mean'、'sum'，默认值：'mean'。
 
     返回：
         Tensor或Tensor scalar，根据 :math:`reduction` 计算的loss。
 
     异常：
-        - **TypeError** - `inputs` 不是Tensor。
-        - **TypeError** - `targets` 不是Tensor。
-        - **TypeError** - `margin` 不是float。
+        - **TypeError** - `inputs` 不是数据类型为float的Tensor。
+        - **TypeError** - `targets` 不是数据类型为float的Tensor。
+        - **TypeError** - `margin` 不是float或者int。
         - **ValueError** - `inputs` 和 `targets` shape不一致。
         - **ValueError** - `reduction` 不是"none"、"mean"或者"sum"。

mindspore/python/mindspore/nn/loss/loss.py

@@ -2492,7 +2492,7 @@ class HingeEmbeddingLoss(LossBase):
     where :math:`L = \{l_1,\dots,l_N\}^\top`.
 
     Args:
-        margin (float): Threshold defined by Hinge Embedding Loss :math:`margin`.
+        margin (float, int): Threshold defined by Hinge Embedding Loss :math:`margin`.
             Represented as :math:`\Delta` in the formula. Default: 1.0.
         reduction (str): Specify the computing method to be applied to the outputs: 'none', 'mean', or 'sum'.
             Default: 'mean'.
@@ -2505,9 +2505,9 @@ class HingeEmbeddingLoss(LossBase):
         Tensor or Tensor scalar, the computed loss depending on `reduction`.
 
     Raises:
-        TypeError: If `logits` is not a Tensor.
-        TypeError: If `labels` is not a Tensor.
-        TypeError: If `margin` is not a float.
+        TypeError: If `logits` is not a Tensor of floats.
+        TypeError: If `labels` is not a Tensor of floats.
+        TypeError: If `margin` is not a float or int.
         ValueError: If `labels` does not have the same shape as `logits`.
         ValueError: If `reduction` is not one of 'none', 'mean', 'sum'.
 
@@ -2531,7 +2531,7 @@ class HingeEmbeddingLoss(LossBase):
 
     def __init__(self, margin=1.0, reduction='mean'):
         super(HingeEmbeddingLoss, self).__init__()
-        validator.check_value_type('margin', margin, [float], self.cls_name)
+        validator.check_value_type('margin', margin, [float, int], self.cls_name)
         validator.check_string(reduction, ['none', 'sum', 'mean'], 'reduction', self.cls_name)
         self.margin = margin
         self.reduction = reduction

mindspore/python/mindspore/ops/function/math_func.py

@@ -1731,7 +1731,7 @@ def signbit(x):
 
 def sgn(x):
     r"""
-    Extension of :func:`mindspore.ops.sign` in the plural.
+    Extension of :func:`mindspore.ops.sign` in complex domain.
     For real number input, this function is the same as :func:`mindspore.ops.sign`.
     For complex input, this function is calculated according to the following formula.
 
@@ -7185,7 +7185,7 @@ def norm(A, ord=None, dim=None, keepdim=False, *, dtype=None):
     ====================== ================================ ==========================================
 
     .. note::
-        Currently, plural is not supported.
+        Currently, complex numbers are not supported.
 
     Args:
         A (Tensor): Tensor of shape :math:`(*, n)` or :math:`(*, m, n)` where * is zero or more batch dimensions.

mindspore/python/mindspore/ops/function/nn_func.py

@@ -4090,7 +4090,7 @@ def hinge_embedding_loss(inputs, targets, margin=1.0, reduction='mean'):
     Args:
         inputs (Tensor): Tensor of shape :math:`(*)` where :math:`*` means any number of dimensions.
         targets (Tensor): Same shape as the logits, contains -1 or 1.
-        margin (float): Threshold defined by Hinge Embedding Loss :math:`margin`.
+        margin (float, int): Threshold defined by Hinge Embedding Loss :math:`margin`.
             Represented as :math:`\Delta` in the formula. Default: 1.0.
         reduction (str): Specify the computing method to be applied to the outputs: 'none', 'mean', or 'sum'.
             Default: 'mean'.
@@ -4099,9 +4099,9 @@ def hinge_embedding_loss(inputs, targets, margin=1.0, reduction='mean'):
         Tensor or Tensor scalar, the computed loss depending on `reduction`.
 
     Raises:
-        TypeError: If `inputs` is not a Tensor.
-        TypeError: If `targets` is not a Tensor.
-        TypeError: If `margin` is not a float.
+        TypeError: If `inputs` is not a Tensor of floats.
+        TypeError: If `targets` is not a Tensor of floats.
+        TypeError: If `margin` is not a float or int.
         ValueError: If `targets` does not have the same shape as `inputs`.
         ValueError: If `reduction` is not one of 'none', 'mean', 'sum'.
 
@@ -4121,13 +4121,25 @@ def hinge_embedding_loss(inputs, targets, margin=1.0, reduction='mean'):
         >>> print(loss)
         0.16666666
     """
-    _dtype = inputs.dtype
-    min_val = Tensor(0, _dtype)
+    if not isinstance(margin, (float, int)):
+        raise TypeError(f"For 'HingeEmbeddingLoss', 'margin' must be a float or int, but got {type(margin)}.")
+    if reduction not in ['none', 'mean', 'sum']:
+        raise ValueError(f"For 'HingeEmbeddingLoss', 'reduction' must be one of 'none', 'mean', 'sum',"
+                         f"but got {reduction}.")
+    inputs_dtype = inputs.dtype
+    targets_dtype = targets.dtype
+    if inputs_dtype not in mstype.float_type:
+        raise TypeError(f"For 'HingeEmbeddingLoss', the dtype of the first input must be float, but got "
+                        f"{inputs_dtype}.")
+    if targets_dtype not in mstype.float_type:
+        raise TypeError(f"For 'HingeEmbeddingLoss', the dtype of the second input must be float, but got "
+                        f"{targets_dtype}.")
+    min_val = Tensor(0, inputs_dtype)
     pos_index = targets > 0
     neg_index = targets < 0
     pos = pos_index * inputs
     neg = neg_index * inputs
-    m = ops.cast(margin, _dtype)
+    m = ops.cast(margin, inputs_dtype)
     margin_matrix = m * neg_index
     neg = margin_matrix - neg
     neg = ops.clip_by_value(neg, min_val)