!31150 Add document for clip by norm about the parallel mode

Merge pull request !31150 from huangxinjing/code_docs_add_clip_by_norm

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

@@ -6,7 +6,10 @@
     通过权重梯度总和的比率来裁剪多个Tensor的值。
 
     .. note::
-        输入'x'应为Tensor的tuple或list。否则，将引发错误。
+        输入 `x` 应为Tensor的tuple或list。否则，将引发错误。
+
+    .. note::
+        在半自动并行模式或自动并行模式下，如果输入是梯度，那么将会自动汇聚所有设备上的梯度的平方和。
 
     **参数：**
 

mindspore/python/mindspore/ops/composite/clip_ops.py

@@ -197,7 +197,12 @@ def clip_by_global_norm(x, clip_norm=1.0, use_norm=None):
     Clips tensor values by the ratio of the sum of their norms.
 
     Note:
-        Input 'x' should be a tuple or list of tensors. Otherwise, it will raise an error.
+        Input `x` should be a tuple or list of tensors. Otherwise, it will raise an error.
+
+    Note:
+        On the SEMI_AUTO_PARALLEL mode or AUTO_PARALLEL mode, if the input `x` is the gradient,
+        the gradient norm values on all devices will be automatically aggregated by allreduce inserted after the local
+        square sum of the gradients.
 
     Args:
         x (Union(tuple[Tensor], list[Tensor])): Input data to clip.