mindspore.ops.gather

mindspore.ops.gather(input_params, input_indices, axis, batch_dims=0)

返回输入tensor在指定轴及指定索引上对应元素的切片。

下图展示了Gather常用的计算过程：

其中，params代表输入 input_params ，indices代表要切片的索引 input_indices 。

说明

• input_indices的值必须在 [0, input_params.shape[axis]) 范围内。CPU与GPU平台越界访问将会抛出异常，Ascend平台越界访问的返回结果是未定义的。

• Ascend平台上，input_params的数据类型不能是 mindspore.bool_ 。

• 返回tensor的shape为 $input\mathrm{\_}params.shape[:axis]+input\mathrm{\_}indices.shape[batch\mathrm{\_}dims:]+input\mathrm{\_}params.shape[axis+1:]$ 。

参数：

• input_params (Tensor) - 输入tensor。

• input_indices (Tensor) - 指定索引。

• axis (Union(int, Tensor[int])) - 指定轴。

• batch_dims (int) - batch维的数量。默认 0 。

返回：

Tensor

支持平台：

Ascend GPU CPU

样例：

>>> import mindspore
>>> # case1: input_indices is a Tensor with shape (5, ).
>>> input_params = mindspore.tensor([1, 2, 3, 4, 5, 6, 7], mindspore.float32)
>>> input_indices = mindspore.tensor([0, 2, 4, 2, 6], mindspore.int32)
>>> axis = 0
>>> output = mindspore.ops.gather(input_params, input_indices, axis)
>>> print(output)
[1. 3. 5. 3. 7.]
>>> # case2: input_indices is a Tensor with shape (2, 2). When the input_params has one dimension,
>>> # the output shape is equal to the input_indices shape.
>>> input_indices = mindspore.tensor([[0, 2], [2, 6]], mindspore.int32)
>>> axis = 0
>>> output = mindspore.ops.gather(input_params, input_indices, axis)
>>> print(output)
[[1. 3.]
 [3. 7.]]
>>> # case3: input_indices is a Tensor with shape (2, ) and
>>> # input_params is a Tensor with shape (3, 4) and axis is 0.
>>> input_params = mindspore.tensor([[1, 2, 3, 4], [5, 6, 7, 8], [9, 10, 11, 12]], mindspore.float32)
>>> input_indices = mindspore.tensor([0, 2], mindspore.int32)
>>> axis = 0
>>> output = mindspore.ops.gather(input_params, input_indices, axis)
>>> print(output)
[[ 1.  2.  3.  4.]
 [ 9. 10. 11. 12.]]
>>> # case4: input_indices is a Tensor with shape (2, ) and
>>> # input_params is a Tensor with shape (3, 4) and axis is 1, batch_dims is 1.
>>> input_params = mindspore.tensor([[1, 2, 3, 4], [5, 6, 7, 8], [9, 10, 11, 12]], mindspore.float32)
>>> input_indices = mindspore.tensor([0, 2, 1], mindspore.int32)
>>> axis = 1
>>> batch_dims = 1
>>> output = mindspore.ops.gather(input_params, input_indices, axis, batch_dims)
>>> print(output)
[ 1.  7. 10.]