Hackathon No.30 #4644

yangguohao · 2022-04-17T12:43:47Z

TripletMarginLoss API 中文文档

PADDLEPADDLE_PR=40545

【Hackathon No.30】PaddlePaddle/Paddle#40545

paddle-bot-old · 2022-04-17T12:43:52Z

Thanks for your contribution!

Ligoml · 2022-04-19T05:16:50Z

docs/api/paddle/nn/Overview_cn.rst

@@ -256,6 +256,7 @@ Loss层
    " :ref:`paddle.nn.MSELoss <cn_api_paddle_nn_MSELoss>` ", "均方差误差损失层"
    " :ref:`paddle.nn.NLLLoss <cn_api_nn_loss_NLLLoss>` ", "NLLLoss层"
    " :ref:`paddle.nn.SmoothL1Loss <cn_api_paddle_nn_SmoothL1Loss>` ", "平滑L1损失层"
+    " :ref:`paddle.nn.TripletMarginWithDistanceLoss <cn_api_paddle_nn_TripletMarginWithDistanceLoss>` TripletMarginWithDistanceLoss层"


少了标点符号

Ligoml · 2022-04-19T05:17:40Z

docs/api/paddle/nn/Overview_cn.rst

@@ -473,6 +474,7 @@ Embedding相关函数
    " :ref:`paddle.nn.functional.smooth_l1_loss <cn_paddle_nn_functional_loss_smooth_l1>` ", "用于计算平滑L1损失"
    " :ref:`paddle.nn.functional.softmax_with_cross_entropy <cn_api_fluid_layers_softmax_with_cross_entropy>` ", "将softmax操作、交叉熵损失函数的计算过程进行合并"
    " :ref:`paddle.nn.functional.margin_cross_entropy <cn_api_paddle_nn_functional_margin_cross_entropy>` ", "支持 ``Arcface``，``Cosface``，``Sphereface`` 的结合 Margin 损失函数"
+    " :ref:`paddle.nn.functional.triplet_margin_with_distance_loss <cn_paddle_nn_functional_triplet_margin_with_distance_loss>` ", "用户自定义距离函数用于计算triplet margin loss 损失"


这里是不是缺少了nn.functional.triplet_margin_with_distance_loss的API文档？

Ligoml · 2022-04-19T05:17:57Z

docs/api/paddle/nn/TripletMarginWithDistanceLoss_cn.rst

+
+.. py:class:: paddle.nn.TripletMarginWithDistanceLoss(margin: float = 1.0, swap: bool = False, reduction: str = 'mean')
+
+该OP可创建一个TripletMarginWithDistanceLoss的可调用类，计算输入 `input` 和 `positive` 和 `negative` 间的 `triplet margin loss` 损失。


去掉「该OP」的说法

Ligoml · 2022-04-19T05:21:58Z

docs/api/paddle/nn/TripletMarginWithDistanceLoss_cn.rst

+    - **swap** (bool，可选) 
+    - **margin** (float，可选) - 手动指定间距，默认为1
+    - **reduction**(str,可选) -指定应用于输出结果的计算方式，可选值有: ``'none'``, ``'mean'``, ``'sum'`` 。默认为 ``'mean'``，计算 Loss 的均值；设置为 ``'sum'`` 时，计算 Loss 的总和；设置为 ``'none'`` 时，则返回原始Loss。
+    - **name** (str，可选) - 操作的名称（可选，默认值为None）。更多信息请参见 :ref:`api_guide_Name` 。


- 同TripletMarginLoss，需要注意中英文标点符号问题 - reduction显示不正常

Ligoml · 2022-04-19T05:23:19Z

docs/api/paddle/nn/TripletMarginWithDistanceLoss_cn.rst

+代码示例
+:::::::::
+
+.. code-block:: python


代码建议使用copy-from的方式，参考文档

Ligoml · 2022-05-09T08:51:07Z

docs/api/paddle/nn/TripletMarginWithDistanceLoss_cn.rst

+    d(x_i, y_i) = \left\lVert {\bf x}_i - {\bf y}_i \right\rVert_2
+
+
+其中 ``distance_function`` 为距离函数，默认为2范数。 ``margin`` 为（input,positive）与（input,negative）的距离间隔， ``swap`` 的内容可以看论文 `Learning shallow convolutional feature descriptors with triplet losses <http://www.bmva.org/bmvc/2016/papers/paper119/paper119.pdf>`_。


swap 的内容可以看论文 --> swap xxxx，详见论文（xxxx里简单描述一下这个参数吧）

Ligoml · 2022-05-09T08:52:41Z

docs/api/paddle/nn/TripletMarginWithDistanceLoss_cn.rst

+    - **swap** (bool，可选) - 默认为False
+    - **margin** (float，可选) - 手动指定间距，默认为1。
+    - **reduction** (str,可选) - 指定应用于输出结果的计算方式，可选值有: ``'none'``, ``'mean'``, ``'sum'`` 。默认为 ``'mean'``，计算 Loss 的均值；设置为 ``'sum'`` 时，计算 Loss 的总和；设置为 ``'none'`` 时，则返回原始Loss。
+    - **name** (str，可选) - 操作的名称（可选，默认值为None）。更多信息请参见 :ref:`api_guide_Name` 。


参数需在括号里写清数据类型和是否可选，可选参数需要注明默认值

, --> ，，句尾加。

参数要和源码顺序保持一致

没有name参数就不要写

中英文字符之间最好使用一个空格隔开

Ligoml · 2022-05-09T08:53:54Z

docs/api/paddle/nn/functional/triplet_margin_with_distance_loss_cn.rst

+
+.. py:class:: paddle.nn.functional.triplet_margin_with_distance_loss(input, positive, negative, distance_function=None, margin: float = 1.0, swap: bool = False, reduction: str = 'mean')
+
+该api计算输入 `input` 和 `positive` 和 `negative` 间的 `triplet margin loss` 损失。


去掉「该api」

Ligoml · 2022-05-09T08:57:50Z

docs/api/paddle/nn/functional/triplet_margin_with_distance_loss_cn.rst

+triplet_margin_with_distance_loss
+-------------------------------
+
+.. py:class:: paddle.nn.functional.triplet_margin_with_distance_loss(input, positive, negative, distance_function=None, margin: float = 1.0, swap: bool = False, reduction: str = 'mean')


这里不是class哦

Ligoml · 2022-05-09T08:59:33Z

docs/api/paddle/nn/functional/triplet_margin_with_distance_loss_cn.rst

+
+
+参数
+:::::::::


参考TripletMarginWithDistanceLoss进行修改

Ligoml · 2022-05-13T06:14:53Z

docs/api/paddle/nn/TripletMarginWithDistanceLoss_cn.rst

+    - **distance_function** (可选) - 手动指定范数，默认为None, 使用欧式距离。
+    - **margin** (float，可选) - 手动指定间距，默认为1。
+    - **swap** (bool，可选) - 默认为False。
+    - **reduction** (str,可选) - 指定应用于输出结果的计算方式，可选值有: ``'none'``， ``'mean'``， ``'sum'`` 。默认为 ``'mean'``，计算 Loss 的均值；设置为 ``'sum'`` 时，计算 Loss 的总和；设置为 ``'none'`` 时，则返回原始 Loss。


, --> ，

Ligoml · 2022-05-18T08:15:41Z

docs/api/paddle/nn/TripletMarginWithDistanceLoss_cn.rst

+
+参数
+:::::::::
+    - **distance_function** (可选) - 手动指定范数，默认为None, 使用欧式距离。


参数的数据类型需要增加

, --> ，

Ligoml

LGTM

yangguohao added 2 commits April 17, 2022 20:37

Create TripletMarginWithDistanceLoss_cn.rst

de3d418

Merge branch 'PaddlePaddle:develop' into triplet_margin_distance_loss

8e80927

paddle-bot-old bot added contributor status: proposed labels Apr 17, 2022

yangguohao mentioned this pull request Apr 17, 2022

【Hachathon No.30】 PaddlePaddle/Paddle#40545

Merged

yangguohao added 2 commits April 17, 2022 20:47

Update Overview_cn.rst

b1eab7c

Merge branch 'PaddlePaddle:develop' into triplet_margin_distance_loss

0b59483

Ligoml reviewed Apr 19, 2022

View reviewed changes

yangguohao added 15 commits April 27, 2022 21:36

Merge branch 'PaddlePaddle:develop' into triplet_margin_distance_loss

6d80ea4

Update Overview_cn.rst

817a924

Update TripletMarginWithDistanceLoss_cn.rst

f271539

Create triplet_margin_with_distance_loss_cn.rst

8ea441d

Update TripletMarginWithDistanceLoss_cn.rst

9654503

Update triplet_margin_with_distance_loss_cn.rst

7c7f772

Merge branch 'PaddlePaddle:develop' into triplet_margin_distance_loss

36cb64c

Merge branch 'PaddlePaddle:develop' into triplet_margin_distance_loss

21674d5

Update TripletMarginWithDistanceLoss_cn.rst

8fd9e9e

Update triplet_margin_with_distance_loss_cn.rst

0aedc77

Update TripletMarginWithDistanceLoss_cn.rst

ced2866

Update triplet_margin_with_distance_loss_cn.rst

58fe4d3

Update triplet_margin_with_distance_loss_cn.rst

824d904

Update TripletMarginWithDistanceLoss_cn.rst

35687d7

Update TripletMarginWithDistanceLoss_cn.rst

b2204f3

Ligoml reviewed May 9, 2022

View reviewed changes

yangguohao added 4 commits May 9, 2022 17:47

Update triplet_margin_with_distance_loss_cn.rst

2f7d3f8

Update TripletMarginWithDistanceLoss_cn.rst

6aa92a1

Update triplet_margin_with_distance_loss_cn.rst

ba6b33d

Update triplet_margin_with_distance_loss_cn.rst

a5e4d8d

Ligoml reviewed May 13, 2022

View reviewed changes

yangguohao added 3 commits May 17, 2022 11:52

Update triplet_margin_with_distance_loss_cn.rst

988c79f

Update TripletMarginWithDistanceLoss_cn.rst

899c9b5

Update Overview_cn.rst

532ad3e

Ligoml reviewed May 18, 2022

View reviewed changes

yangguohao added 2 commits May 18, 2022 18:32

Update TripletMarginWithDistanceLoss_cn.rst

580cc9d

Update triplet_margin_with_distance_loss_cn.rst

ed91ef5

Ligoml approved these changes May 20, 2022

View reviewed changes

Update triplet_margin_with_distance_loss_cn.rst

bf3ae2c

Ligoml closed this Jun 10, 2022

Ligoml reopened this Jun 10, 2022

Merge branch 'develop' into triplet_margin_distance_loss

f1d8878

Ligoml merged commit 7f5cbbd into PaddlePaddle:develop Jun 13, 2022

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Hackathon No.30 #4644

Hackathon No.30 #4644

yangguohao commented Apr 17, 2022 •

edited by Ligoml

Loading

paddle-bot-old bot commented Apr 17, 2022

Ligoml Apr 19, 2022

yangguohao May 5, 2022

yangguohao May 5, 2022

Ligoml Apr 19, 2022

yangguohao May 5, 2022

yangguohao May 5, 2022

Ligoml Apr 19, 2022

yangguohao May 5, 2022

yangguohao May 5, 2022

Ligoml Apr 19, 2022

yangguohao May 5, 2022

yangguohao May 5, 2022

Ligoml Apr 19, 2022

yangguohao May 5, 2022

yangguohao May 5, 2022

Ligoml May 9, 2022

Ligoml May 9, 2022

yangguohao May 11, 2022

Ligoml May 9, 2022

yangguohao May 11, 2022

Ligoml May 9, 2022

yangguohao May 11, 2022

Ligoml May 9, 2022

yangguohao May 11, 2022

Ligoml May 13, 2022

yangguohao May 17, 2022

Ligoml May 18, 2022

Ligoml left a comment


		.. py:class:: paddle.nn.TripletMarginWithDistanceLoss(margin: float = 1.0, swap: bool = False, reduction: str = 'mean')

		该OP可创建一个TripletMarginWithDistanceLoss的可调用类，计算输入 `input` 和 `positive` 和 `negative` 间的 `triplet margin loss` 损失。

		d(x_i, y_i) = \left\lVert {\bf x}_i - {\bf y}_i \right\rVert_2


		其中 ``distance_function`` 为距离函数，默认为2范数。 ``margin`` 为（input,positive）与（input,negative）的距离间隔， ``swap`` 的内容可以看论文 `Learning shallow convolutional feature descriptors with triplet losses <http://www.bmva.org/bmvc/2016/papers/paper119/paper119.pdf>`_。


		.. py:class:: paddle.nn.functional.triplet_margin_with_distance_loss(input, positive, negative, distance_function=None, margin: float = 1.0, swap: bool = False, reduction: str = 'mean')

		该api计算输入 `input` 和 `positive` 和 `negative` 间的 `triplet margin loss` 损失。



		参数
		:::::::::

Hackathon No.30 #4644

Hackathon No.30 #4644

Conversation

yangguohao commented Apr 17, 2022 • edited by Ligoml Loading

paddle-bot-old bot commented Apr 17, 2022

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Ligoml left a comment

Choose a reason for hiding this comment

yangguohao commented Apr 17, 2022 •

edited by Ligoml

Loading