new representation docs (#225)

Author: yanghaihua

docs/locale/zh_CN/LC_MESSAGES/source/modules/models/representation.po

@@ -8,7 +8,7 @@ msgid ""
 msgstr ""
 "Project-Id-Version:  \n"
 "Report-Msgid-Bugs-To: \n"
-"POT-Creation-Date: 2022-11-01 15:19+0800\n"
+"POT-Creation-Date: 2022-11-03 19:00+0800\n"
 "PO-Revision-Date: YEAR-MO-DA HO:MI+ZONE\n"
 "Last-Translator: FULL NAME <EMAIL@ADDRESS>\n"
 "Language-Team: LANGUAGE <LL@li.org>\n"
@@ -18,12 +18,12 @@ msgstr ""
 "Generated-By: Babel 2.10.3\n"
 
 #: ../../source/modules/models/representation.rst:3
-#: 9b064944e0964bdbbb2915390845514f
+#: 2129e18757cd45339d97c4e5a75d789d
 msgid "Representation Model Tutorial"
 msgstr "表征模型使用教程"
 
 #: ../../source/modules/models/representation.rst:5
-#: abd881ced8334a62ad29b17617645b05
+#: 5896660fa6c1416ba7b4f5776d48051f
 msgid ""
 "The representation model (TS2Vec) is one of the self-supervised models, "
 "mainly hoping to learn a general feature expression for downstream tasks."
@@ -33,155 +33,155 @@ msgid ""
 msgstr "表征模型(TS2Vec)属于自监督模型里的一种，主要是希望能够学习到一种通用的特征表达用于下游任务；当前主流的自监督学习主要有基于生成式和基于对比学习的方法，当前案例使用的TS2Vec模型是一种基于对比学习的自监督模型"
 
 #: ../../source/modules/models/representation.rst:9
-#: 3025ac9aaf0446fc95a211d81437c9ce
+#: 09155f9a2ee447c7bb16e3599abba6b9
 msgid "The use of self-supervised models is divided into two phases:"
 msgstr "自监督模型的使用一般分为两个阶段"
 
 #: ../../source/modules/models/representation.rst:8
-#: 771b57b7651b4109bb60e44ec04edb0b
+#: 79f6b680fde64c7a81536cfdf190cb88
 msgid "Pre-training with unlabeled data, independent of downstream tasks"
 msgstr "不涉及任何下游任务，使用无标签的数据进行预训练"
 
 #: ../../source/modules/models/representation.rst:9
-#: 259b6005c1124e83a51050a77e625e80
+#: 49fb481f64bb4045809861500e6f6265
 msgid "Fine-tune on downstream tasks using labeled data"
 msgstr "使用带标签的数据在下游任务上 Fine-tune"
 
 #: ../../source/modules/models/representation.rst:13
-#: c847d4c3b8aa47fb914da3ecec97be17
+#: 95ca2edf84d247038fa0547759e6c9cb
 msgid "TS2Vec follows the usage paradigm of self-supervised models:"
 msgstr "TS2Vec结合下游任务的使用同样遵循自监督模型的使用范式，分为2个阶段："
 
 #: ../../source/modules/models/representation.rst:12
-#: 36866eb1d84442109043c8f4f1b5e07f
+#: 765c6f6c9a05406eac590aa81eab9de3
 msgid "Representational model training"
 msgstr "表征模型训练"
 
 #: ../../source/modules/models/representation.rst:13
-#: 71ec4f51da22462e8981f16fcca546a0
+#: fec2e8aa03ec49c4b19edb67a882af84
 msgid ""
 "Use the output of the representation model for the downstream task (the "
 "downstream task of the current case is the prediction task)"
 msgstr "将表征模型的输出用于下游任务(当前案例的下游任务为预测任务)"
 
 #: ../../source/modules/models/representation.rst:17
-#: 550841919e924e2fb9e52fbd95245cd0
+#: 6d77625749df473f8ead84b68e0b31de
 msgid ""
 "For the sake of accommodating both beginners and experienced developers, "
 "there are two ways to use representation tasks:"
 msgstr "为兼顾初学者和有一定的经验的开发者，本文给出两种表征任务的使用方法:"
 
 #: ../../source/modules/models/representation.rst:16
-#: dbd70c46672c464abdd8e66e49feaf92
+#: 42e63162d16042c29c6341bfbfa4b097
 msgid ""
 "A pipeline that combines the representation model and downstream tasks, "
 "which is very friendly to beginners"
 msgstr "表征模型和下游任务相结合的pipeline，初学者容易上手使用"
 
 #: ../../source/modules/models/representation.rst:17
-#: 108a7dc6375249d48c832583ae037860
+#: c47fdd248d0d40a5b53b53401284fe94
 msgid ""
 "Decoupling the representational model and downstream tasks, showing in "
 "detail how the representational model and downstream tasks are used in "
 "combination"
 msgstr "表征模型和下游任务解耦，详细展示表征模型和下游任务如何相结合使用"
 
 #: ../../source/modules/models/representation.rst:20
-#: 2d31625f0fdd48609b97bd524cb2e7a5
-msgid "Method one"
-msgstr "使用方法一"
+#: 7795706790b54c6cb6254116d2a288c0
+msgid "1. Method one"
+msgstr "1. 方法一"
 
 #: ../../source/modules/models/representation.rst:22
-#: 18b9f26fe49b4f30b773d8fb3f34f639
+#: 8a8be0919e1f4431af5cb789ac8da2b3
 msgid "A pipeline that combines the representation model and downstream tasks"
 msgstr "表征模型和下游任务相结合的pipeline"
 
 #: ../../source/modules/models/representation.rst:25
-#: 26e917470da34204b8a544ab27cd7341
-msgid "Prepare the data"
+#: b52cd7e6c4b740929c5cdffa4f1302c5
+msgid "1.1 Prepare the data"
 msgstr "准备数据集"
 
-#: ../../source/modules/models/representation.rst:45
-#: b13f51b26893463ebc91a547a228ff75
-msgid "Training"
+#: ../../source/modules/models/representation.rst:44
+#: a91871d5d4a8494492c5dac6bab66299
+msgid "1.2 Training"
 msgstr "训练"
 
 #: ../../source/modules/models/representation.rst:61
-#: bd7972f40530496b9e2ec804ff408312
-msgid "Prediction"
+#: 71db106f238e4055871114df13d66948
+msgid "1.3 Prediction"
 msgstr "预测"
 
-#: ../../source/modules/models/representation.rst:67
-#: c343a6b1e1d640a08f30c310f93a9cff
-msgid "Backtest"
+#: ../../source/modules/models/representation.rst:68
+#: 520a95c9594449938c591ae52057a5bc
+msgid "1.4 Backtest"
 msgstr "回测"
 
-#: ../../source/modules/models/representation.rst:81
-#: bd7175abbd8c45ea8088352499788ea4
-msgid "Method two"
-msgstr "使用方法二"
-
 #: ../../source/modules/models/representation.rst:83
-#: e07ddc48cb9d4d28a6670b3b417f1084
+#: bc4059a9057b487e9617723ddd54b037
+msgid "2. Method two"
+msgstr "2. 方法二"
+
+#: ../../source/modules/models/representation.rst:85
+#: 9f53fba5287d473abe61e9945e8f88e6
 msgid ""
 "Decoupling the representational model and downstream tasks. It's divided "
 "into two stages, the first stage is representation model training and "
 "prediction, and the second stage is the training and prediction of "
 "downstream tasks"
-msgstr "表征模型和下游回归任务解藕. 分为两个阶段，第一阶段是表征模型的训练和预测，第二阶段是下游任务模型的训练和预测"
+msgstr "表征模型和下游回归任务解耦. 分为两个阶段，第一阶段是表征模型的训练和预测，第二阶段是下游任务模型的训练和预测"
 
-#: ../../source/modules/models/representation.rst:86
-#: bec3f1fdcff040eca38c6d9b92102ef0
-msgid "1. The first stage:"
+#: ../../source/modules/models/representation.rst:87
+#: da94ae067fa54515b66dfd73a0a426df
+msgid "The first stage:"
 msgstr "第一阶段:"
 
-#: ../../source/modules/models/representation.rst:87
-#: 56576ed054584365b4e5985058b0b7c8
+#: ../../source/modules/models/representation.rst:89
+#: a5b0f8777bf64ba6bc19bcc51fb61a89
 msgid "Training of the representation model"
 msgstr "表征模型训练"
 
-#: ../../source/modules/models/representation.rst:88
-#: 6fd9a537483d47a69fd323596fbbdf18
+#: ../../source/modules/models/representation.rst:90
+#: 47194a7a5d8d4b23bf90bbad849191be
 msgid "Output of training set and test set representation results"
 msgstr "输出训练集和测试集的表征结果"
 
 #: ../../source/modules/models/representation.rst:92
-#: 230b62e6ade9443493cafb3b84f7374d
-msgid "1.1 Prepare the data"
-msgstr "准备数据集"
-
-#: ../../source/modules/models/representation.rst:111
-#: f5c8ed59f85244b8934e9df6882b63a5
-msgid "1.2 Training of the representation model"
-msgstr "表征模型训练"
+#: 9b8d6802816b4626866c736827d7c88b
+msgid "The second stage:"
+msgstr "第二阶段:"
 
-#: ../../source/modules/models/representation.rst:126
-#: 1f2ca0e0cae941f9a95091f2dfb1d196
-msgid "1.3 Output of training set and test set representation results"
-msgstr "输出训练集和测试集的表征结果"
-
-#: ../../source/modules/models/representation.rst:136
-#: d194f7b4827741abb800c8f8fa463d1c
-msgid "2. The second stage:"
-msgstr "第二阶段"
-
-#: ../../source/modules/models/representation.rst:137
-#: 725503101ee545309d47377830437ec8
+#: ../../source/modules/models/representation.rst:94
+#: 78d575ebea984274a3148c6c2dba837f
 msgid "Build training and test samples for regression models"
 msgstr "构建回归模型的训练和测试样本"
 
-#: ../../source/modules/models/representation.rst:138
-#: 2d5a609b4f8545eb8f54c1c3e0bb5636
+#: ../../source/modules/models/representation.rst:95
+#: b973d5de54c64be8b5943c319aa34d75
 msgid "training and prediction"
 msgstr "训练和预测"
 
-#: ../../source/modules/models/representation.rst:141
-#: 73afc3be00df4cc7941c7acf12e358be
-msgid "2.1 Build training and test samples for regression models"
+#: ../../source/modules/models/representation.rst:99
+#: d9201c888f4a4a14b3a2fc008322b158
+msgid "2.1 Prepare the data"
+msgstr "准备数据集"
+
+#: ../../source/modules/models/representation.rst:117
+#: 262ab9ad28c24431956b37d12996ef0f
+msgid "2.2 Training of the representation model"
+msgstr "表征模型训练"
+
+#: ../../source/modules/models/representation.rst:133
+#: 16eb6ad99a3848fdbe1bc734eecc2e0b
+msgid "2.3 Output of training set and test set representation results"
+msgstr "输出训练集和测试集的表征结果"
+
+#: ../../source/modules/models/representation.rst:145
+#: 2a0b3a5f49c14282a3604b330715115b
+msgid "2.4 Build training and test samples for regression models"
 msgstr "构建回归模型的训练和测试样本"
 
-#: ../../source/modules/models/representation.rst:167
-#: ed3d7f9d693d4fbba33ef4a5196ffe81
-msgid "2.2 Training and prediction"
+#: ../../source/modules/models/representation.rst:172
+#: ff0a0ac6b9c24b9caf5a1841744766eb
+msgid "2.5 Training and prediction"
 msgstr "训练和预测"
 

docs/source/modules/models/representation.rst

@@ -16,13 +16,14 @@ For the sake of accommodating both beginners and experienced developers, there a
     - A pipeline that combines the representation model and downstream tasks, which is very friendly to beginners
     - Decoupling the representational model and downstream tasks, showing in detail how the representational model and downstream tasks are used in combination
 
-Method one
+1. Method one
 =================
 
 A pipeline that combines the representation model and downstream tasks
 
-Prepare the data
-================
+1.1 Prepare the data
+--------------------
+
 .. code-block:: python
 
    import numpy as np
@@ -39,8 +40,9 @@ Prepare the data
    data, _ = data.split('2016-09-22 06:00:00')
    train_data, test_data = data.split('2016-09-21 05:00:00')
 
-Training 
-========
+1.2 Training 
+------------
+
 .. code-block:: python
 
    ts2vec_params = {"segment_size": 200, 
@@ -55,14 +57,16 @@ Training
                            repr_model_params=ts2vec_params)
    model.fit(train_data)
 
-Prediction
-==========
+1.3 Prediction
+--------------
+
 .. code-block:: python
 
    model.predict(train_data)
 
-Backtest
-========
+1.4 Backtest
+------------
+
 .. code-block:: python
 
    from paddlets.utils.backtest import backtest
@@ -75,19 +79,25 @@ Backtest
             return_predicts=True)
 
 
-Method two
+2. Method two
 =================
 
 Decoupling the representational model and downstream tasks. It's divided into two stages, the first stage is representation model training and prediction, and the second stage is the training and prediction of downstream tasks
 
-1. The first stage:
-===================
-    - Training of the representation model
-    - Output of training set and test set representation results
+The first stage:
 
+- Training of the representation model
+- Output of training set and test set representation results
+
+The second stage:
+
+- Build training and test samples for regression models
+- training and prediction
+
+
+2.1 Prepare the data
+--------------------
 
-1.1 Prepare the data
-====================
 .. code-block:: python
 
    import numpy as np
@@ -103,8 +113,9 @@ Decoupling the representational model and downstream tasks. It's divided into tw
    data, _ = data.split('2016-09-22 06:00:00')
    train_data, test_data = data.split('2016-09-21 05:00:00')
 
-1.2 Training of the representation model
-========================================
+2.2 Training of the representation model
+----------------------------------------
+
 .. code-block:: python
 
    # initialize the TS2Vect object
@@ -118,8 +129,9 @@ Decoupling the representational model and downstream tasks. It's divided into tw
    # training
    ts2vec.fit(train_data)
 
-1.3 Output of training set and test set representation results
-==============================================================
+2.3 Output of training set and test set representation results
+--------------------------------------------------------------
+
 .. code-block:: python
 
    sliding_len = 200 # Use past sliding_len length points to infer the representation of the current point in time
@@ -128,13 +140,10 @@ Decoupling the representational model and downstream tasks. It's divided into tw
    train_reprs = all_reprs[:, :split_tag]
    test_reprs = all_reprs[:, split_tag:]
 
-2. The second stage:
-=======================
-    - Build training and test samples for regression models
-    - training and prediction
 
-2.1 Build training and test samples for regression models
-=========================================================
+2.4 Build training and test samples for regression models
+---------------------------------------------------------
+
 .. code-block:: python
 
    # generate samples
@@ -159,8 +168,9 @@ Decoupling the representational model and downstream tasks. It's divided into tw
    test_to_numpy = np.expand_dims(test_to_numpy, 0) 
    test_features, test_labels = generate_pred_samples(test_reprs, test_to_numpy, pre_len) 
 
-2.2 Training and prediction
-===========================
+2.5 Training and prediction
+---------------------------
+
 .. code-block:: python
 
    # training