提交 04cb9dfb 编写于 作者: M mindspore-ci-bot 提交者: Gitee

!1793 fix doc for merge

Merge pull request !1793 from jiangjinsheng/issue_doc
......@@ -44,10 +44,11 @@ class Embedding(Cell):
dtype (:class:`mindspore.dtype`): Data type of input. Default: mindspore.float32.
Inputs:
- **input** (Tensor) - Tensor of shape :math:`(\text{vocab_size})`.
- **input** (Tensor) - Tensor of shape :math:`(\text{batch_size}, \text{input_length})`. The element of
the Tensor should be integer and not larger than vocab_size. else the corresponding embedding vector is zero
if larger than vocab_size.
Outputs:
Tensor of shape :math:`(\text{vocab_size}, \text{embedding_size})`.
Tensor of shape :math:`(\text{batch_size}, \text{input_length}, \text{embedding_size})`.
Examples:
>>> net = nn.Embedding(20000, 768, True)
......@@ -61,6 +62,7 @@ class Embedding(Cell):
def __init__(self, vocab_size, embedding_size, use_one_hot=False, embedding_table='normal', dtype=mstype.float32):
super(Embedding, self).__init__()
validator.check_subclass("dtype", dtype, mstype.number_type, self.cls_name)
validator.check_value_type('use_one_hot', use_one_hot, [bool], self.cls_name)
self.vocab_size = vocab_size
self.embedding_size = embedding_size
self.use_one_hot = use_one_hot
......
......@@ -144,7 +144,7 @@ class Merge(PrimitiveWithInfer):
One and only one of the inputs should be selected as the output
Inputs:
- **inputs** (Tuple) - The data to be merged. All tuple elements should have same data type.
- **inputs** (Union(Tuple, List)) - The data to be merged. All tuple elements should have same data type.
Outputs:
tuple. Output is tuple(`data`, `output_index`). The `data` has the same shape of `inputs` element.
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册