[Docathon][Add API Legend No.26-30]添加tensor_split系列API图例 #6809
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
感谢你贡献飞桨文档，文档预览构建中，Docs-New 跑完后即可预览，预览链接：http://preview-pr-6809.paddle-docs-preview.paddlepaddle.org.cn/documentation/docs/zh/api/index_cn.html |
|
@fufu0615 你是用啥画的图啊 |
fufu0615
force-pushed
the
develop
branch
3 times, most recently
from
eaef617 to
967a1e6
Compare
docs/api/paddle/dsplit_cn.rst
Outdated
| @@ -6,8 +6,12 @@ dsplit | |||
| .. py:function:: paddle.dsplit(x, num_or_indices, name=None) | |||
|
|
|||
|
|
|||
| ``dsplit`` 全称 Depth Split ，即深度分割，将输入 Tensor 沿着深度轴分割成多个子 Tensor，等价于将 :ref:`cn_api_paddle_tensor_split` API 的参数 axis 固定为 2。注意：使用 ``paddle.dsplit`` 进行变换的 Tensor 维度数量必须不少于 3。 | |||
There was a problem hiding this comment.
megemini
reviewed
Aug 7, 2024
docs/api/paddle/tensor_split_cn.rst
Outdated
| @@ -12,7 +12,7 @@ tensor_split | |||
| 参数 | |||
| ::::::::: | |||
| - **x** (Tensor) - 输入变量，数据类型为 bool、bfloat16、float16、float32、float64、uint8、int8、int32、int64 的多维 Tensor，其维度必须大于 0。 | |||
| - **num_or_indices** (int|list|tuple) - 如果 ``num_or_indices`` 是一个整数 ``n`` ，则 ``x`` 沿 ``axis`` 拆分为 ``n`` 部分。如果 ``x`` 可被 ``n`` 整除，则每个部分都是 ``x.shape[axis]/n`` 。如果 ``x`` 不能被 ``n`` 整除，则第一个 ``int（x.shape[axis]%n）`` 分割大小将为 ``int（x.shape[axis]/n）+1`` ，其余部分的大小将是 ``int（x.shape[axis]/n）`` 。如果 ``num_or_indices`` 是整数索引的列表或元组，则在每个索引处沿 ``axis`` 分割 ``x`` 。例如， ``num_or_indices=[2, 4]`` 在 ``axis=0`` 时将沿轴 0 将 ``x`` 拆分为 ``x[：2]`` 、 ``x[2:4]`` 和 ``x[4:]`` 。 | |||
| - **num_or_indices** (int|list|tuple) - 如果 ``num_or_indices`` 是一个整数 ``n`` ，则 ``x`` 沿 ``axis`` 拆分为 ``n`` 部分。如果 ``x`` 可被 ``n`` 整除，则每个部分都是 ``x.shape[axis]/n`` 。如果 ``x`` 不能被 ``n`` 整除，则第 ``int(x.shape[axis]/n)+1`` 个部分的大小将是 ``int(x.shape[axis]%n)`` ，其余部分的大小将是 ``int(x.shape[axis]/n)`` 。如果 ``num_or_indices`` 是整数索引的列表或元组，则在每个索引处沿 ``axis`` 分割 ``x`` 。例如， ``num_or_indices=[2, 4]`` 在 ``axis=0`` 时将沿轴 0 将 ``x`` 拆分为 ``x[：2]`` 、 ``x[2:4]`` 和 ``x[4:]`` 。 | |||
There was a problem hiding this comment.
这里有点问题，我这里翻译的可能不太好，原文是
the first
int(x.shape[axis] % n)sections will have sizeint(x.shape[axis] / n) + 1
翻译为 则第一个 .... ，应该翻译为 则前 ... 个 。
参考代码：
In [5]: x = paddle.rand([13])
In [6]: out = paddle.tensor_split(x, 3)
In [7]: out
Out[7]:
[Tensor(shape=[5], dtype=float32, place=Place(gpu:0), stop_gradient=True,
[0.40005636, 0.65518010, 0.31436527, 0.23026623, 0.36780968]),
Tensor(shape=[4], dtype=float32, place=Place(gpu:0), stop_gradient=True,
[0.55613798, 0.12409472, 0.20364396, 0.83700526]),
Tensor(shape=[4], dtype=float32, place=Place(gpu:0), stop_gradient=True,
[0.03752962, 0.71997887, 0.79592997, 0.49080023])]
In [8]: x = paddle.rand([8])
In [9]: out = paddle.tensor_split(x, 3)
In [10]: out
Out[10]:
[Tensor(shape=[3], dtype=float32, place=Place(gpu:0), stop_gradient=True,
[0.75365561, 0.43469149, 0.56143743]),
Tensor(shape=[3], dtype=float32, place=Place(gpu:0), stop_gradient=True,
[0.89418393, 0.64861047, 0.43100974]),
Tensor(shape=[2], dtype=float32, place=Place(gpu:0), stop_gradient=True,
[0.99590099, 0.27822134])]
In [11]: 13%3
Out[11]: 1
In [12]: 8%3
Out[12]: 2
In [13]: 13//3
Out[13]: 4
In [14]: 8//3
Out[14]: 2- 当长度为
13，分割3份，则前13%3=1个的大小为13//3+1=5，余下2份，大小为13//3=4，即：[5, 4, 4] - 当长度为
8，分割3份，则前8%3=2个的大小为8//3+1=3，余下1份，大小为8//3=2，即：[3, 3, 2]
因此，如需修改，可以修改一下原文中的翻译，但是，这里修改后的描述有问题，因为
则第
int(x.shape[axis]/n)+1个部分的大小将是int(x.shape[axis]%n)
表示只有一个部分，即 int(x.shape[axis]/n)+1 是 int(x.shape[axis]%n) 这样大小，与实际不符。
另外，图例也需要修改。
There was a problem hiding this comment.
了解了，感谢师傅讲解，改完的图例和文档已推送🫡
| @@ -22,7 +22,42 @@ tensor_split | |||
| list[Tensor]，分割后的 Tensor 列表。 | |||
|
|
|||
|
|
|||
| 代码示例 | |||
| 代码示例 1 | |||
There was a problem hiding this comment.
这里分割开了示例代码，paddle 那边也需要同步修改 ～
docs/api/paddle/vsplit_cn.rst
Outdated
Comment on lines
10
to
23
| ``vsplit`` 全称 Vertical Split 即垂直分割，将输入 Tensor 沿着垂直轴分割成多个子 Tensor ，等价于将 :ref:`cn_api_paddle_tensor_split` API 的参数 axis 固定为 0。 | ||
|
|
||
| 当 Tensor 维度数量等于 1 时： | ||
|
|
||
| .. image:: ../../images/api_legend/vsplit/vsplit-1.png | ||
| :alt: vsplit 图例-1 | ||
|
|
||
| 当 Tensor 维度数量大于 1 时： | ||
|
|
||
| .. image:: ../../images/api_legend/vsplit/vsplit-2.png | ||
| :alt: vsplit 图例-2 |
megemini
reviewed
Aug 8, 2024
抱歉没注意到，已经重新推送了orz |
38 tasks
sunzhongkai588
changed the title
Aoraki-Dream
pushed a commit
to Aoraki-Dream/docs
that referenced
this pull request
Sep 4, 2024
…e#6809) * 修改tensor_split文档描述并增加图例 * 修改hsplit文档描述并增加图例 * 修改dsplit文档描述并增加图例 * 修改vsplit文档描述并增加图例 * 修改代码块引用和图例参数注释
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
PADDLEPADDLE_PR=66991
PR types
Add api legends
PR changes
Docs
Description
为tensor_split，hsplit，dsplit，vsplit API文档增加图例
（tensor_split文档的示例代码部分有修改，需要等待英文文档pr合并后更新）
对应的英文文档修复为：PaddlePaddle/Paddle#66991