fix tutorial test=develop (#2601)

* fix tutorial test=develop

* fix pic test=develop

* update tutorial test=develop

doc/paddle/guides/model_save_load_cn.rst

@@ -21,8 +21,8 @@
 
 本文主要介绍飞桨2.0动态图存储载入体系，各接口关系如下图所示：
 
-.. image:: images/save_2.0.png
-.. image:: images/load_2.0.png
+.. image:: https://github.com/PaddlePaddle/FluidDoc/blob/develop/doc/paddle/guides/images/save_2.0.png?raw=true
+.. image:: https://github.com/PaddlePaddle/FluidDoc/blob/develop/doc/paddle/guides/images/load_2.0.png?raw=true
 
 1.2 静态图存储载入体系（飞桨1.x）
 ----------------------------
@@ -739,4 +739,4 @@ Layer更准确的语义是描述一个具有预测功能的模型对象，接收
     fluid.io.save_params(exe, model_path)
 
     # load 
-    state_dict = paddle.io.load_program_state(model_path)
+    state_dict = paddle.io.load_program_state(model_path)
\ No newline at end of file

doc/paddle/tutorial/cv_case/convnet_image_classification/convnet_image_classification.rst

@@ -4,7 +4,7 @@
 本示例教程将会演示如何使用飞桨的卷积神经网络来完成图像分类任务。这是一个较为简单的示例，将会使用一个由三个卷积层组成的网络完成\ `cifar10 <https://www.cs.toronto.edu/~kriz/cifar.html>`__\ 数据集的图像分类任务。
 
 设置环境
-----------
+--------
 
 我们将使用飞桨2.0beta版本。
 
@@ -18,18 +18,15 @@
     
     paddle.disable_static()
     print(paddle.__version__)
-    print(paddle.__git_commit__)
-
 
 
 .. parsed-literal::
 
-    0.0.0
-    264e76cae6861ad9b1d4bcd8c3212f7a78c01e4d
+    2.0.0-beta0
 
 
 加载并浏览数据集
--------------------
+----------------
 
 我们将会使用飞桨提供的API完成数据集的下载并为后续的训练任务准备好数据迭代器。cifar10数据集由60000张大小为32
 \*
@@ -49,7 +46,7 @@
         train_labels[i, 0] = train_label
 
 浏览数据集
--------------
+----------
 
 接下来我们从数据集中随机挑选一些图片并显示，从而对数据集有一个直观的了解。
 
@@ -70,11 +67,11 @@
 
 
 
-.. image:: convnet_image_classification_files/convnet_image_classification_6_0.png
+.. image:: https://github.com/PaddlePaddle/FluidDoc/tree/develop/doc/paddle/tutorial/cv_case/convnet_image_classification/convnet_image_classification_files/convnet_image_classification_001.png?raw=true
 
 
 组建网络
-----------
+--------
 
 接下来我们使用飞桨定义一个使用了三个二维卷积（\ ``Conv2d``)且每次卷积之后使用\ ``relu``\ 激活函数，两个二维池化层（\ ``MaxPool2d``\ ），和两个线性变换层组成的分类网络，来把一个\ ``(32, 32, 3)``\ 形状的图片通过卷积神经网络映射为10个输出，这对应着10个分类的类别。
 
@@ -165,8 +162,8 @@
                 if batch_id % 1000 == 0:
                     print("epoch: {}, batch_id: {}, loss is: {}".format(epoch, batch_id, avg_loss.numpy()))
                 avg_loss.backward()
-                opt.minimize(avg_loss)
-                model.clear_gradients()
+                opt.step()
+                opt.clear_grad()
     
             # evaluate model after one epoch
             model.eval()
@@ -198,36 +195,36 @@
 .. parsed-literal::
 
     start training ... 
-    epoch: 0, batch_id: 0, loss is: [2.3024805]
-    epoch: 0, batch_id: 1000, loss is: [1.1422595]
-    [validation] accuracy/loss: 0.5575079917907715/1.2516425848007202
-    epoch: 1, batch_id: 0, loss is: [0.9350736]
-    epoch: 1, batch_id: 1000, loss is: [1.3825703]
-    [validation] accuracy/loss: 0.5959464907646179/1.1320706605911255
-    epoch: 2, batch_id: 0, loss is: [0.979844]
-    epoch: 2, batch_id: 1000, loss is: [0.87730503]
-    [validation] accuracy/loss: 0.6607428193092346/0.9754576086997986
-    epoch: 3, batch_id: 0, loss is: [0.7345351]
-    epoch: 3, batch_id: 1000, loss is: [1.0982555]
-    [validation] accuracy/loss: 0.6671326160430908/0.9667007327079773
-    epoch: 4, batch_id: 0, loss is: [0.9291839]
-    epoch: 4, batch_id: 1000, loss is: [1.1812104]
-    [validation] accuracy/loss: 0.6895966529846191/0.9075900316238403
-    epoch: 5, batch_id: 0, loss is: [0.5072213]
-    epoch: 5, batch_id: 1000, loss is: [0.60360587]
-    [validation] accuracy/loss: 0.6944888234138489/0.8740479350090027
-    epoch: 6, batch_id: 0, loss is: [0.5917944]
-    epoch: 6, batch_id: 1000, loss is: [0.7963876]
-    [validation] accuracy/loss: 0.7072683572769165/0.8597638607025146
-    epoch: 7, batch_id: 0, loss is: [0.50116754]
-    epoch: 7, batch_id: 1000, loss is: [0.95844793]
-    [validation] accuracy/loss: 0.700579047203064/0.876727819442749
-    epoch: 8, batch_id: 0, loss is: [0.87496114]
-    epoch: 8, batch_id: 1000, loss is: [0.68749857]
-    [validation] accuracy/loss: 0.7198482155799866/0.8403064608573914
-    epoch: 9, batch_id: 0, loss is: [0.8548105]
-    epoch: 9, batch_id: 1000, loss is: [0.6488569]
-    [validation] accuracy/loss: 0.7106629610061646/0.874437153339386
+    epoch: 0, batch_id: 0, loss is: [2.331658]
+    epoch: 0, batch_id: 1000, loss is: [1.6067888]
+    [validation] accuracy/loss: 0.5676916837692261/1.2106356620788574
+    epoch: 1, batch_id: 0, loss is: [1.1509854]
+    epoch: 1, batch_id: 1000, loss is: [1.3777964]
+    [validation] accuracy/loss: 0.5818690061569214/1.1748384237289429
+    epoch: 2, batch_id: 0, loss is: [1.051642]
+    epoch: 2, batch_id: 1000, loss is: [1.0261706]
+    [validation] accuracy/loss: 0.6607428193092346/0.9685573577880859
+    epoch: 3, batch_id: 0, loss is: [0.8457774]
+    epoch: 3, batch_id: 1000, loss is: [0.6820123]
+    [validation] accuracy/loss: 0.6822084784507751/0.9241172075271606
+    epoch: 4, batch_id: 0, loss is: [0.9059805]
+    epoch: 4, batch_id: 1000, loss is: [0.587117]
+    [validation] accuracy/loss: 0.7012779712677002/0.8670551180839539
+    epoch: 5, batch_id: 0, loss is: [1.0894825]
+    epoch: 5, batch_id: 1000, loss is: [0.9055369]
+    [validation] accuracy/loss: 0.6954872012138367/0.8820587992668152
+    epoch: 6, batch_id: 0, loss is: [0.4162583]
+    epoch: 6, batch_id: 1000, loss is: [0.5274862]
+    [validation] accuracy/loss: 0.7074680328369141/0.8538646697998047
+    epoch: 7, batch_id: 0, loss is: [0.52636147]
+    epoch: 7, batch_id: 1000, loss is: [0.70929015]
+    [validation] accuracy/loss: 0.7107627987861633/0.8633227944374084
+    epoch: 8, batch_id: 0, loss is: [0.57556355]
+    epoch: 8, batch_id: 1000, loss is: [0.83717]
+    [validation] accuracy/loss: 0.69319087266922/0.903077244758606
+    epoch: 9, batch_id: 0, loss is: [0.88774866]
+    epoch: 9, batch_id: 1000, loss is: [0.91165334]
+    [validation] accuracy/loss: 0.7194488644599915/0.8668457865715027
 
 
 .. code:: ipython3
@@ -244,15 +241,15 @@
 
 .. parsed-literal::
 
-    <matplotlib.legend.Legend at 0x163d6ec50>
+    <matplotlib.legend.Legend at 0x167d186d0>
 
 
 
 
-.. image:: convnet_image_classification_files/convnet_image_classification_12_1.png
+.. image:: https://github.com/PaddlePaddle/FluidDoc/tree/develop/doc/paddle/tutorial/cv_case/convnet_image_classification/convnet_image_classification_files/convnet_image_classification_002.png?raw=true
 
 
 The End
 -------
 
-从上面的示例可以看到，在cifar10数据集上，使用简单的卷积神经网络，用飞桨可以达到71%以上的准确率。
+从上面的示例可以看到，在cifar10数据集上，使用简单的卷积神经网络，用飞桨可以达到71%以上的准确率。你也可以通过调整网络结构和参数，达到更好的效果。

doc/paddle/tutorial/cv_case/image_search/image_search.rst

@@ -25,13 +25,11 @@
     
     paddle.disable_static()
     print(paddle.__version__)
-    print(paddle.__git_commit__)
 
 
 .. parsed-literal::
 
-    0.0.0
-    89af2088b6e74bdfeef2d4d78e08461ed2aafee5
+    2.0.0-beta0
 
 
 数据集
@@ -127,12 +125,12 @@
 
 
 
-.. image:: image_search_files/image_search_8_0.png
+.. image:: https://github.com/PaddlePaddle/FluidDoc/blob/develop/doc/paddle/tutorial/cv_case/image_search/image_search_files/image_search_001.png?raw=true?raw=true
 
 
 
 构建训练数据
---------------
+------------
 
 图片检索的模型的训练样本跟我们常见的分类任务的训练样本不太一样的地方在于，每个训练样本并不是一个\ ``(image, class)``\ 这样的形式。而是（image0,
 image1,
@@ -205,12 +203,12 @@ similary_or_not)的形式，即，每一个训练样本由两张图片组成，
 
 
 
-.. image:: image_search_files/image_search_15_1.png
+.. image:: https://github.com/PaddlePaddle/FluidDoc/blob/develop/doc/paddle/tutorial/cv_case/image_search/image_search_files/image_search_002.png?raw=true
 
 
 
 把图片转换为高维的向量表示的网络
------------------------------------
+--------------------------------
 
 我们的目标是首先把图片转换为高维空间的表示，然后计算图片在高维空间表示时的相似度。
 下面的网络结构用来把一个形状为\ ``(3, 32, 32)``\ 的图片转换成形状为\ ``(8,)``\ 的向量。在有些资料中也会把这个转换成的向量称为\ ``Embedding``\ ，请注意，这与自然语言处理领域的词向量的区别。
@@ -267,8 +265,6 @@ similary_or_not)的形式，即，每一个训练样本由两张图片组成，
 
 .. code:: ipython3
 
-    # 定义训练过程
-    
     def train(model):
         print('start training ... ')
         model.train()
@@ -302,8 +298,8 @@ similary_or_not)的形式，即，每一个训练样本由两张图片组成，
                 if batch_id % 500 == 0:
                     print("epoch: {}, batch_id: {}, loss is: {}".format(epoch, batch_id, avg_loss.numpy()))
                 avg_loss.backward()
-                opt.minimize(avg_loss)
-                model.clear_gradients()
+                opt.step()
+                opt.clear_grad()
     
     model = MyNet()
     train(model)
@@ -312,46 +308,46 @@ similary_or_not)的形式，即，每一个训练样本由两张图片组成，
 .. parsed-literal::
 
     start training ... 
-    epoch: 0, batch_id: 0, loss is: [2.3080945]
-    epoch: 0, batch_id: 500, loss is: [2.326215]
-    epoch: 1, batch_id: 0, loss is: [2.0898924]
-    epoch: 1, batch_id: 500, loss is: [1.8754089]
-    epoch: 2, batch_id: 0, loss is: [2.2416227]
-    epoch: 2, batch_id: 500, loss is: [1.9024051]
-    epoch: 3, batch_id: 0, loss is: [1.841417]
-    epoch: 3, batch_id: 500, loss is: [2.1239076]
-    epoch: 4, batch_id: 0, loss is: [1.9291763]
-    epoch: 4, batch_id: 500, loss is: [2.2363486]
-    epoch: 5, batch_id: 0, loss is: [2.0078473]
-    epoch: 5, batch_id: 500, loss is: [2.0765374]
-    epoch: 6, batch_id: 0, loss is: [2.080376]
-    epoch: 6, batch_id: 500, loss is: [2.1759136]
-    epoch: 7, batch_id: 0, loss is: [1.908263]
-    epoch: 7, batch_id: 500, loss is: [1.7774136]
-    epoch: 8, batch_id: 0, loss is: [1.6335764]
-    epoch: 8, batch_id: 500, loss is: [1.5713912]
-    epoch: 9, batch_id: 0, loss is: [2.287479]
-    epoch: 9, batch_id: 500, loss is: [1.7719988]
-    epoch: 10, batch_id: 0, loss is: [1.2894523]
-    epoch: 10, batch_id: 500, loss is: [1.599735]
-    epoch: 11, batch_id: 0, loss is: [1.78816]
-    epoch: 11, batch_id: 500, loss is: [1.4773489]
-    epoch: 12, batch_id: 0, loss is: [1.6737808]
-    epoch: 12, batch_id: 500, loss is: [1.8889393]
-    epoch: 13, batch_id: 0, loss is: [1.6156021]
-    epoch: 13, batch_id: 500, loss is: [1.3851049]
-    epoch: 14, batch_id: 0, loss is: [1.3854092]
-    epoch: 14, batch_id: 500, loss is: [2.0325592]
-    epoch: 15, batch_id: 0, loss is: [1.9734558]
-    epoch: 15, batch_id: 500, loss is: [1.8050598]
-    epoch: 16, batch_id: 0, loss is: [1.7084911]
-    epoch: 16, batch_id: 500, loss is: [1.8919995]
-    epoch: 17, batch_id: 0, loss is: [1.3137552]
-    epoch: 17, batch_id: 500, loss is: [1.8817297]
-    epoch: 18, batch_id: 0, loss is: [1.9453808]
-    epoch: 18, batch_id: 500, loss is: [2.1317677]
-    epoch: 19, batch_id: 0, loss is: [1.6051079]
-    epoch: 19, batch_id: 500, loss is: [1.779858]
+    epoch: 0, batch_id: 0, loss is: [2.3078856]
+    epoch: 0, batch_id: 500, loss is: [1.9325346]
+    epoch: 1, batch_id: 0, loss is: [1.9889]
+    epoch: 1, batch_id: 500, loss is: [2.0410695]
+    epoch: 2, batch_id: 0, loss is: [2.2465641]
+    epoch: 2, batch_id: 500, loss is: [1.8171736]
+    epoch: 3, batch_id: 0, loss is: [1.9939486]
+    epoch: 3, batch_id: 500, loss is: [2.1440036]
+    epoch: 4, batch_id: 0, loss is: [2.1497147]
+    epoch: 4, batch_id: 500, loss is: [2.3686018]
+    epoch: 5, batch_id: 0, loss is: [1.938681]
+    epoch: 5, batch_id: 500, loss is: [1.7729127]
+    epoch: 6, batch_id: 0, loss is: [2.0061004]
+    epoch: 6, batch_id: 500, loss is: [1.6132584]
+    epoch: 7, batch_id: 0, loss is: [1.8874661]
+    epoch: 7, batch_id: 500, loss is: [1.6153599]
+    epoch: 8, batch_id: 0, loss is: [1.9407685]
+    epoch: 8, batch_id: 500, loss is: [2.1532288]
+    epoch: 9, batch_id: 0, loss is: [1.4792883]
+    epoch: 9, batch_id: 500, loss is: [1.857158]
+    epoch: 10, batch_id: 0, loss is: [2.1518302]
+    epoch: 10, batch_id: 500, loss is: [1.790559]
+    epoch: 11, batch_id: 0, loss is: [1.7292264]
+    epoch: 11, batch_id: 500, loss is: [1.8555079]
+    epoch: 12, batch_id: 0, loss is: [1.6968924]
+    epoch: 12, batch_id: 500, loss is: [1.4554331]
+    epoch: 13, batch_id: 0, loss is: [1.3950458]
+    epoch: 13, batch_id: 500, loss is: [1.7197256]
+    epoch: 14, batch_id: 0, loss is: [1.7336586]
+    epoch: 14, batch_id: 500, loss is: [2.0465684]
+    epoch: 15, batch_id: 0, loss is: [1.7675827]
+    epoch: 15, batch_id: 500, loss is: [2.6443417]
+    epoch: 16, batch_id: 0, loss is: [1.7331158]
+    epoch: 16, batch_id: 500, loss is: [1.6207634]
+    epoch: 17, batch_id: 0, loss is: [2.0908554]
+    epoch: 17, batch_id: 500, loss is: [1.7711265]
+    epoch: 18, batch_id: 0, loss is: [1.8717268]
+    epoch: 18, batch_id: 500, loss is: [1.5269613]
+    epoch: 19, batch_id: 0, loss is: [1.5681677]
+    epoch: 19, batch_id: 500, loss is: [1.7821472]
 
 
 模型预测
@@ -397,7 +393,7 @@ similary_or_not)的形式，即，每一个训练样本由两张图片组成，
 
 
 
-.. image:: image_search_files/image_search_22_0.png
+.. image:: https://github.com/PaddlePaddle/FluidDoc/blob/develop/doc/paddle/tutorial/cv_case/image_search/image_search_files/image_search_003.png?raw=true
 
 
 

doc/paddle/tutorial/cv_case/image_segmentation/pets_image_segmentation_U_Net_like.ipynb

@@ -7,7 +7,7 @@
     "id": "ueGUN2EQeScw"
    },
    "source": [
-    "# 基于U型语义分割模型实现的宠物图像分割\n",
+    "# 基于U-Net卷积神经网络实现宠物图像分割\n",
     "\n",
     "本示例教程当前是基于2.0-beta版本Paddle做的案例实现，未来会随着2.0的系列版本发布进行升级。"
    ]
@@ -34,16 +34,18 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 1,
+   "execution_count": 21,
    "metadata": {},
    "outputs": [
     {
-     "output_type": "execute_result",
      "data": {
-      "text/plain": "'0.0.0'"
+      "text/plain": [
+       "'2.0.0-beta0'"
+      ]
      },
+     "execution_count": 21,
      "metadata": {},
-     "execution_count": 1
+     "output_type": "execute_result"
     }
    ],
    "source": [
@@ -92,7 +94,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 6,
    "metadata": {
     "colab": {
      "base_uri": "https://localhost:8080/",
@@ -103,7 +105,20 @@
     "outputId": "3985783f-7166-4afa-f511-16427b3e2a71",
     "tags": []
    },
-   "outputs": [],
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current\n",
+      "                                 Dload  Upload   Total   Spent    Left  Speed\n",
+      "100  755M  100  755M    0     0  1707k      0  0:07:32  0:07:32 --:--:-- 2865k0  0:12:48  524k   0  0:13:34  0:02:41  0:10:53  668k      0  0:12:45  0:03:06  0:09:39 1702k     0  1221k      0  0:10:33  0:03:25  0:07:08 3108k37  282M    0     0  1243k      0  0:10:21  0:03:52  0:06:29  719k0:05:53  566k0  1237k      0  0:10:25  0:04:43  0:05:42 1593k 0  0:09:46  0:05:28  0:04:18 2952k 1467k      0  0:08:47  0:06:43  0:02:04 1711k\n",
+      "  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current\n",
+      "                                 Dload  Upload   Total   Spent    Left  Speed\n",
+      "100 18.2M  100 18.2M    0     0  1602k      0  0:00:11  0:00:11 --:--:-- 3226k\n"
+     ]
+    }
+   ],
    "source": [
     "!curl -O http://www.robots.ox.ac.uk/~vgg/data/pets/data/images.tar.gz\n",
     "!curl -O http://www.robots.ox.ac.uk/~vgg/data/pets/data/annotations.tar.gz\n",
@@ -140,10 +155,10 @@
     "├── test.txt\n",
     "├── trainval.txt\n",
     "├── trimaps\n",
-    "│    ├── Abyssinian_1.png\n",
-    "│    ├── Abyssinian_10.png\n",
-    "│    ├── ......\n",
-    "│    └── yorkshire_terrier_99.png\n",
+    "│    ├── Abyssinian_1.png\n",
+    "│    ├── Abyssinian_10.png\n",
+    "│    ├── ......\n",
+    "│    └── yorkshire_terrier_99.png\n",
     "└── xmls\n",
     "      ├── Abyssinian_1.xml\n",
     "      ├── Abyssinian_10.xml\n",
@@ -158,7 +173,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 2,
+   "execution_count": 22,
    "metadata": {
     "colab": {
      "base_uri": "https://localhost:8080/",
@@ -171,9 +186,11 @@
    },
    "outputs": [
     {
-     "output_type": "stream",
      "name": "stdout",
-     "text": "用于训练的图片样本数量: 7390\n"
+     "output_type": "stream",
+     "text": [
+      "用于训练的图片样本数量: 7390\n"
+     ]
     }
    ],
    "source": [
@@ -218,7 +235,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 3,
+   "execution_count": 23,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -371,7 +388,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 4,
+   "execution_count": 24,
    "metadata": {
     "colab": {
      "base_uri": "https://localhost:8080/",
@@ -383,15 +400,16 @@
    },
    "outputs": [
     {
-     "output_type": "display_data",
      "data": {
-      "text/plain": "<Figure size 432x288 with 2 Axes>",
-      "image/svg+xml": "<?xml version=\"1.0\" encoding=\"utf-8\" standalone=\"no\"?>\n<!DOCTYPE svg PUBLIC \"-//W3C//DTD SVG 1.1//EN\"\n  \"http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd\">\n<!-- Created with matplotlib (https://matplotlib.org/) -->\n<svg height=\"181.699943pt\" version=\"1.1\" viewBox=\"0 0 349.2 181.699943\" width=\"349.2pt\" xmlns=\"http://www.w3.org/2000/svg\" xmlns:xlink=\"http://www.w3.org/1999/xlink\">\n <defs>\n  <style type=\"text/css\">\n*{stroke-linecap:butt;stroke-linejoin:round;}\n  </style>\n </defs>\n <g id=\"figure_1\">\n  <g id=\"patch_1\">\n   <path d=\"M 0 181.699943 \nL 349.2 181.699943 \nL 349.2 0 \nL 0 0 \nz\n\" style=\"fill:none;\"/>\n  </g>\n  <g id=\"axes_1\">\n   <g clip-path=\"url(#p58ad9a7e6d)\">\n    <image height=\"153\" id=\"image6a21407320\" transform=\"scale(1 -1)translate(0 -153)\" width=\"153\" x=\"7.2\" xlink:href=\"data:image/png;base64,\niVBORw0KGgoAAAANSUhEUgAAAJkAAACZCAYAAAA8XJi6AAAABHNCSVQICAgIfAhkiAAAIABJREFUeJx0vUuvZUl2HvatiNiP87qPzKysKnWzKTatNgk0YNMUQMCgKcMaGIIhzSRrbI0kAx7YHhmGQcCA/R/kIUH4B/SYgDywKUpNWObAzZdksLurWZVZlY9773nsvSNiebDWioh9MusUbuW955z9ilixHt/61gpiZgYAZi4/OUf8+Z//Jf7ZP/2n+MM//EPsNiOICJkAIgIAeO/RhYCcM3LOYGaEEMoPAKSUkHNGSgnee/RdD0dAYkbOGY4IOSbAETKzXCNn2Mt+DyHAM5Dsb+fhnEPXd2BkvQ4wDAN2ux3unz3HJ59+jtu7W3z66ae4u3+G3X6DcRxx2N8ALPcLMMgRuhDKcy1xwvl8xjAMiDGCiNB1PbwL6LoORA7eB3gv90BEYCwAE7puhPcdyANdGGQsfA/nHch5gKhcx16Pj0f827/8C/w//+b/xr/7t3+Kr776Am/fvcWyzAAnvU+UMQYTUuIyrgAQY9R5k/dSkuNs/OVYIKX84RgzA5C/iQjIDAJkrgH5XyIwMQgM1uOI5D5SBn77t/9j/PN//r/h+9//1dW42LM6Eyw78OnpCb//+/87fvM/+g38+U/+X9zd7OCDAzk5iFQgU4yY57k8kA1E1PftxcxwziHnjHmekGIqN7DEiIUzlhjL8e0gmYA555CTDJZzDsMwYLPZIKeMeZ7LIMuDJx3IhNPpiGk6w3sCMyPGBcfjA+b5gmW5IKUogq7XXeIkg6L3K5PEcORXwsEqpDaRJowpJeS4AFkWas4ZKS7IMYFzM5HlPPbMWe/bgZxbTZQjB0fynvcezhNCcOi6Dn3fw4eAvu/L3/bjvZfFqf+GIIvkY0Jg8woQiAKIHEAEDwJngB0BIDgGPKoMeAL6nvDHf/xj/PCHP8Tv/d7v4d27d6u5Z2aEdlWdz2f8/u//Pv6n//F/wH4cscwzQienjTkj51Q0mE0GESGqkBARnHMAgMvlUs7bdV1ZaZdlRp5ZB5DkYWwAASSIQCzLIoLjPMAZw2YsA+gcsMwzGAnzPIPIYbsVbeu8x3a/xf2zW2w2W4TQYZkjdts9iIG4JIAZzgM+EGKaRRupkNok5Jj0GQnkGAAhpQznMxw6HcCElCMADyIPgMDOI2VCII+cM8gTCAyS2ULKGYEIRA4RjK7v8eKTl/iV7/8qjqcHvH3/dbkHzgRU3QEQIRGDiZFZhCR4D2YPzhmcM5wjhODR9x1ijEi6OOuiyfCeihakxGB9ftPuDICYEXWhOyIwgASAVL0RsyxQBmKc4R3hv/mv/xmmecI/+a/+CYZhUA3pEGxwY4z42U9/iv/1f/mf4ZyYBNabYpaTm1bxvq5s01QAyuCU1aFaQh5MjiHnEADEZSnmI6r2ciBAtYOZX7vO6XTC5XLBMAzYbkfM8wznPYZhwDAMmKYFfd9jHEe8efMGm+0W2+1OBcfpoBKADO97EQgGYlxWK9s5BzAjxojQdyqAAJBlhbP8LVpMhC8EB6L6zERcNCqRg3MZzA7EGQQHBsuYOoe+77Hf77HZbMQyxSQTCJLJx/plmteRg3cO6uuAsxPt4v1qzHPm8neMcSVw9v3EVRCJHYCMnOnqyld3YuNS3iYsy4L//r/97/B3/7O/ix/84AcyXoAIGQA8PDzg7//9/wJxmkRlq8pc0iKmIHPRRiZQJmSmCWylmE9mwmYmsOs60V7MCD4Uk+hVMB1IhE/PK5rElWtsNhuklHC5XGSw9XqXywVdN5Tvv3jxAvf39+i6DtvtFsPQwXuC84RljgB1SEmeQ+7Xg1OG7zoZfPXVCE5NNcN5MYGAU3OSQSB4HwCwPh8h5QVBfVVZcBmZEyiLIIIYmRzI2/gwQh8wjiO22y122y3OxyfVQqLJ7PmLoDknyyWzahkCqztjc2Q+nAmZLFoTQC4Cx8yIWX23mPQYrBSEzaN6aVXsyMRLfulCh5gW/L2/95/jxz/+Y9ze3om5JhIJ/Mu//EtMpxPIEThlRLD4EU6lleRkZlLswvZgzrniy9QBbgbFAobGFEEdfyZZ1akZHDsmpVS0ZyvQ5sfFGGVy9js450WwNlt4/b3rOjjnxP/yHgzxDZkZwzDgfD5jP44g1QBiNlTzOg+AkTkiR11YGSCKADkQJGDpOg/vgJQjiAJyWsSM5SxCkhOYgpg0ynAuiMZwEB/Le4zjiKB+1TiOiMsMTgs4E6qbrpPNMt2qx+BIJpxhLotei8X0VYFzSJmRU4YPHpwZmTNC1ADNJzCLEFZf06lwMZir+bZzEzOIJSjMOQOc8frVK/zkJ3+K3/qt3xJNRkSY5xn/6e/8DvbbDbwnIGfAOThPSOW06kPpTZMJURM4LDGWVWYrKsZYNJsJR1HNRCLUzMiqvjMYXgW5FSoAxQRxFsG7XC4ggpjO8wW393cg7zHPETnLypABznCe9H25J2Jg5kn+zRmd90CKxcQwZYAWMFczKveXxdVQ8+gdIScCsxwbfAcwwKwLiXsw5FlSVpeCHMAZyKRmz4PIYRxHbDYbeO8QgkNcnPiBDmCWcfAgMEn0BzVHxFXHlKjZm5kHOHhkZuRMcBnqkmRkNZPe+TK+rSk1vzhnNNpRfooGs/lhRnYA4LEsC/7O3/kdvH37DofDASHnjP/jX/wLPH/+HGmZVr4FiICcdHVXrWQ3VLSUPpxpuHbVdV1XBNP8hzZAsIFpNSBR9ZFMGC2qzDkjJy6wSMoZ3gfsdzt0XYfQRFvDIDDCsiwABdFmTv2z3JxDhTcE09Aowm1a1e7RtHXRKgAyMTyrJues2k6c8BgXuMBwzACCmHgQXLPwAAYRcDmf8e7dO5zPZ8zzLNchCRzkeq4iATrBOqgrc3oNk5if6MiJwGZG8BrZ5gwODPCHQmYLvQYPjJyT/mR1OdpouSqFrgv4gz/4A/yDf/APEGKM+Mf/+B/hcNgXB7869TaY8litSSy4WuujacTh9OZsQOxBnRNn1R6i/bz1wTy5IgAm0JvNpgjmPEWklCQwcB7juMEwjtjtdvjkxSd4+fIl7u7vixNM1JWHl4FM8FSDFLu2QAq8Eu6cM7quK460DWQ7RkRUtAJhAReNr4EAEzITPNWwPnOGgy0mJzhWWWRU/BybQvvb8EQHgFiQqxa7sn9bP85MLAjIJKYNENelA8SEqi9tiz3nVHy4lLKa0YyUHFKq77c+n/3Y+PyX/+gf4vHpiOC9x+3tLXKOooaZQXrXibOulmom68qgIhT2kolxQHFqZXjaVdHekA1gbnA6MSupnLed7MvlLOau+GkBLnTwoUPoOuz3e3R9h3mZcb6cELoDco5YFqDrNoD6mSkmMHHx1RhcgOXiP6oJBwg5iVbKjhB80PtKGmXaOHj1HwmEoMIq36Pk4dUtIOeQOcFlAjsufu5mM+L+7h43N7d4enyH6VKDHvG1xDn25CUwaQSLyJnXpOPGEDFEq2jALILlWEBdhmmVqq2L9soJXmGRVoullJFyQk4ZLTxy/QMAfS+4XPjRj36kbzo4D6QY4VnsPkGdW/U0i3CoRruOegCRR7uIDRBUuEwzFcHU6Ig8IbbnaGAVM7kVx3JYFkaMGUQJ4zAgI8N5CRwYQOg6mWCOyAwsMeMyEZwjLEtETgm+70UASHTEMs8IfV8EO86Lmt+u4kcNCEtlPMQxds7re4Y0VWiHc0aODATxcC3Q4SyYF5FEZuO4reaQxISJD+p0xcpsEAgQpAX2rupi+RqrdBGqniPSRa/3XnQjChTSZgJ8CMUsgvFRQZIIOBXFYShCxeSAH/3oR/B/8id/8rvT5byyp845gBp/C9W0tdmBVtCuTaP5AeVRuPwPOgfV/DIjN6q2YECN/0Y6SNVPEiT7cHODrutwc3OD7XaLw80N7u/vMY4DgvcgJ4FNToK620C0EbLdhw8VVhF4wsN7EyKNspr7MOECKograyeJgDiolrFjJPqCI4lO9YfBWOYZX3/9NV599dc4HY+Ies8g0VrVVzaEXv+jOqbOXftl9fvreTGfG5DHsu+56tbAAh4Byr3zCAqaWwbhY7/Xfx1CCPiXf/RHCH/2Z3+Gl588XwlJttDYiT/BWQdOb5QB0QaNo28mpn3llEFcQVtZwlUgqVlB16bYVkcrzMwSScqEStjvFNDcbrc4HA7Y7XYFNGYIThSXGZfTCeM4luAhaiRs17AI2DSuKwCuwikQvy6nrOPDBZUnX4MXG7fEEYhitkSjMzwEU3Ps1EeqC7brO+x3O+x3B3Rdj34YkJZFoBodW1nMBhslEHn52wmMYD51WSYNrIHm3/Yl46vngwdYrU2WeyfnkMFgx3AKQmfVjy3w2+Kkoskkwv+Lv/gLhO1mXGNTEFBUJUPVv7wnOI9Ifpkgu1msH8iiSedd0QwOkOiLSCEG1sFzIKqwhqHufd+DmXE4HHA6HQv+JIEDoRsGZAY22x1ubm+x3e3gvVMhEj9KXA5GXM6YKcNrHs6ASu9tEWj2gpwAnLnNPABECSBCilF9H4KDA5NMtj1zNZEAfGuCogK81bSBAXay+Lq+x92z53j2/CXev3+LlBbM04yYosAhBQvlAp9Uy+BV49H6faom8dsEDFAo1DAwE3yX4Vj8NTIr5mQmHct5Ja3UEiuyQiUJOXfgzHjx/B5hu92swvL2xjjL0sgsoCKcK6ujrHgiRclRLiiTJv9aNESaSuHMYEfCTCgwQdVigrCL1jRfbJomxChAIeeq3fqhx2a7x83NLTabTTlHzhEp5+KvIEd0ISDFBfN0wTBsyrVFG1NZlXAKz9j954ycHZgngAT9lzFwyJTBiHAOxafs+x4xJnjXlXEomjlnsEf17XIGuwwiLzDM4QYvXrzEN998hePxAY/v38sTrOAJVCFSYS+rhCrozWi/96FglQgUAKfGrTHzS+KSlKwCNWcrXyV4VBfGewlKUhYgmhKje/4MwTQSUMHOVkhSSgKYqvAVTWWRYvOeo5qhN6e9TKblNIsZy2ut11yv86FENUSE8/ksT5VFCwTfFZhkt9thHEcsizjqKSXMiznAOkA5Iae5mNwQegRep6wK0AvCUpLmxshIClVknXSLqLkAm1CBTjnDqw+Zc0bo3ErQXLPyzdclQvFt+nFEyozzNCHlJHhrXs/Jhy+upkTvw3xpXd6Kq314iAlc8asL/FF91m8T1Bb6EetgEa4XReEkK+G+7QTmbHtfhbDFY8yPKZQUtz5VYTOoCYnGfVqWwtoAPoRFWozHhNWA35wzOo32NpsNQKR0o1gS5SEEFWhR2zFKNDlNc/W9iDAvC5gqqp1SAscE1myA+RVl2iiAyMPSLC3u5tWE2ABdL5prf8VMM+dWaBh9P2C/2+Pu/h7b7VYS7x+ZG72KarVcHXlUraJ3vfrXAi4TQiq/lcf8IIAr1KDr6zdphho8ETon3D/nHJx36LogCfJiKjRctQNFnRK882oy9T4VQ7ILF4Q4hHLLhqyXFeyc5ODU0bT35boK4jrhTtnkFCpRmisoCsb+sBcHfl5wPh9xPo94enoCABwOe+ScscwLHLEEcgAcROCXnBFyxtBpXtTrmueMOTMCBXhICN9GjeRlEDmz+mu5DCyrv+ScQ5wXEfLAyOxAQTSu8wCTmtBiJllBVVlEQxdwczig77oCQzjykspaCRo3v9vilgDAnP8qVQp8qMbEh3qlkTEVOQtnIXQmOaeQFldSD0kL5hIKiNsETwjkinIIQGVWBk1bmCksgsBcMu6F5qOmxlYoAFDOAhvoU7YIudyf0Hfg1uxMqOC2qapxHEvKJ/MCRxL99X2PZVmQAXz3l76L7W6Hvu8xDAOICPMywxFEMFNEcB7IQo0Jviv3l1JG72tEKYi3JNNd8ICTxWURtGggeY6krgCrd78sC7Jvvqv/835NEqhRcvVbW63NBPjgcDw+4nQ+Fq3XLlQ5TiWBDT9rhbAK2MqXYwKYyuet5ms1YZ2Q5jvUwr+tjBX7LK6JZurtOS2DEOzmvfdAqimidtLJzuca+EFvoE1FfCxVtMJkIBDAkuKVWq+TYEK8LAsulwu894hLwjhWKs/zl5/AeY95WbBpzBIAeOcxL2ekFOGpwfHIwTWQSx86dM7XqFInsMAfihsVn22pxMwKVxhSnhFjBZzbScscRT/wWFI3Zg1WPxKuiaY3AuSV6WrH00BgkcIP538tMFfOfiNo1+7JB4fqe+b8X39GJNCGeaXtMTauwSaOgMKsaDGfVbSYuEAYdgLBlLyatVQnu9VicgKFLdSsxCponDJy45PZZIcQCi15ThFdEOrO5XIRnljfI3QB+8MW+8NWtdxUgo+UU2GODEMPy0AwMxInUBTNaGNn92wOMKBUGBfgGIhpKdE3q6kQn65OmPHmZPwqNSolRhcki9F3vboD6qYQaRbFY+i3+OTF38AnL77E+ekBcV7AuaZ9WqGTC1+9BzTjTh985gMJHtbM7fV32pfJRtL7XH+/vuXIQRA0Kr5mmUsb2JhSMZfXN2wn5LoExBZnXkVL8hChJMHXjqJgLiJUqZyb9eYF1vAg1lXTRKiu7+GIMG5G+L4rE9n3veTQYsZ0mRDnWfJqMYIoin/ne4S+QwaQmODJlYgveQY7oB/GmnBXYclJ7supwCQkpLxAsj1ONWyGDwExLgq3CByRUoIQHvXZjaER5Rpi7yzpHJWWI77qsBkx7kYB2ljZE+5b5qUZ3g8EEOpjkZNxX8EZRhn/NhVo0aplMFrTSwUYbrWtMNS9HOcdjGhARAhW9OFcLRYxE1rVuWJOJGYkmY8TPDzUbyOC7zpAheP64Y2BUXjntgLUue+HQc1pTY4DRoRj3N3fo+/7Alm8fPlSuGqKpR2fnoT/7wFwFEH2EoHFSOi6QaqN4DBPEX3vMPgRIYgzbmbffAl7z7RusmRxYrjgyoKySTVz07KCa8bCF1Np40uEoqVdZngn2jE44O72Bnf3z/H+3TdwLuF8OpUoeK196u9mps0KWaKckWR9N6aMRTy+RcCg6imWIKNCNvXztem2c9b78t6e00moYQxSO6hNZBeaNSrY6i2nmWqmwCtKb3+bT2eRZ05S9CGDxLAwKPhqFpm58M9Mk0kelTBNUkn08PAA5xxevXqF0/mEx8fHlXAQALCwBeY5YZ4t1eM1DydRo+XpTCicc1iWZeW02mdCRgwAPJwPxQ8j/cysgUXUtshYiZNGB7Jn+iDoySylaHqOzWaH29t7nM4XnM7nWlTzLZrneqxDEX6JCtfycYXzlQxChSJEoXgQBQDuo9dtQV+iSmhtn8nOFSx1A5QAtCRHS/Rj8ENzcos+PVCot2Zy7Hvt6hLnm5FyLnWXRIQlKzPTAVFNSFl1juD6Drf7Pe7u7tB1HTabDeZ5lrKwPoCYkaJo46enR3BO8MTwzqPf7tD3Uj4XQleSt1ZK1vUjYmIMvVREjX2HnBNc18N4UoBhXYDvenmOrMn2HOUeEbRsLQjAmsXPhKfCtEhpAdCtimMMpjBBIQJc8KAwgILD3f0dkCccnx4wz40mK35hDZpaQahCrGjYletTrksZUoMkaSmwCVa9LxG2gpCWqLposua85uYUQdRfi+P/sVcRDiVAtr6BPRgTVpScktgGIWaL+GraqgUq5fxipkOnEe2Syrk61TZW3zmOo2g3CoXEuMwzaBFsSnJwWUq5OldMlEEifT8WzeOVMdv3fVkIswKxobdJybAxawtnbBEx1zxrzhnzMotP2kaYTRK85ci1/Drnqt/rnMfNzQHeEU7HI87nCSnV2tUygWhA1mau2ijxW7WfsWOITARUgABmdXUowwDAKjfropbqDnzoM7b3FVrJM4FwqNqnlHa5dRhcByug04ruxDVV5BgIWpSaIQ7+0pgKmQiBNbxCCUbBCV78I3iHzWaD25sbEYJ5xmazwTRN2PKAFAVIJgjAGe2hg4frOjgX4JxH3w/FPBYh05+a3nLwvtK+r/3TVTTMAkzaeOac1R/0xSFnZoS+k6iL6+JaO+7VkxG/xoFI8sHPnj3Hy08/gwPjzTevSvFLIydQMGuV2lvJ0rUGU0ExAiSp9gKtPxehEY9OJMytzvWt2NrVde334qWGIGG6DVqrcUgwyCJwzjmNJmqkkWMC+yZoAABHWJRX3/pv9WahbFyhInddJ8wGtedOTZzlJU3rbLaDUEkWE1gHT4S+60Deo+t6bDc77HZ7DINosK7r64qzcL5UU6vAqzklZJCv97zWXlZwTCWT0Gl9pnfCXwMBoRuQEzRT4Ao7xLRrO6lVeOXvYRiwGXeIMeN0mUTw3Lp6i5P4hEytdjWhNU22Npf1OwZPAUBeaUTzK733SkbwV5GrCZwdw8XnLE90pdWcre6WT3UtqSlpYahGEQ5rPyArBcjOZWAmoJpRI9frouCcpepJtKRQfIvbQEDX9wihx/5wKMlvcawXnDTiOp2OhbfPKih91yt3bMQwbtCZloFgga2wk2kCu08SwfDewanpuzY7YgYdyAfxV3MC4BBCB8Fora2AlPxLMENF4LTSpEyWaDHJtIAYfR/w/JOX+M53fgm3t7eFN2eTN46jAONurUm+HZJAyVMSrn20NT/fBE3ID1Ug68sYwkl/lJ61huVWr2BIe9JcZGvPq2kQjeq8IvJYh8oldE65lLgVHyJrGZlzyE3hLjOv/BxmQtAolYhAwWO320ui2HuwM56Y8KsMchjHDZAYFAJCN2Cz3WG736MfNwhNpVTbDOZjfoP9G2NC1wvijmbhtK0YdNaEceBr8tjK/yxCJkIDzKJQYYh84/TrtQEx0SyTXFIyyjK152iFrQpFOcOV1llrnIparDXNt2khO74k2qmei1VpEonQGbGq/Z7JQTAVDAa8rWp8+CqOJtXV0GoyU/vgmoohImROYjaoIuJ2cYMQRGgVPjFB8KIhlmWGpQCtkYhzXTE9RMJNk4KSgO1uj34YETptRtKYyZbxYf+KaZe8pLKrwCzXNoLhNT5VJsK3KLzCLx+ZMEC5cRDAWWCUdelfTU4LxOB9gA8BKWdE1fgmeO0YtsHAh84366xp7tGV3z7y+jBIIPOTqH0WKsJVjnMEcC2NtJdZC1e0EEsNHtxagMQ0VKYBmsmyi9hKKhcg6aoTU1LU+mMOKMp1RUOSIN8WxqsZSnHBNAmfzDSZXV9oO4DvgvaT2ElS22mbhaZvxfX1bSGIeyCNUzgnEGoLpsy1dM9+Wqo4WPAg86ey4WGo1zOsL2l3IO+CNEtxYSVkOn1aGyla/MWLT/Ds2TNs98I6kcBE4BDRkKlkFa61kf0HCJjtwPBAYbm2gmSQxMcEzcxYKSRXig+3X2BVTKqFGXklPyW6bM0I5YpjGRZiAF7OuQBv8n0Rde+9ZSHqZLCE5ERr6KJw1UJAUL+mpUWXlBQzkqtRYdBCD6kDTBgG8VWGccB2u8M4bopmtOtYCN5Gy/ZZMT+Z4b2luDyQZdQyV46+PXtLtgTLMS54jfIU8sgZcB4hdGXcylhq8YhFd/V982tEu0pxshAYY0xSZdWQEHa7DWKcYXanNelF2OTpi0QQVR8UH3zTAoFrc9t8S8fww+NNXyrwTWvcLngX4ByrGVInNNZEd1nFrR1fXZzKzXu9UVYBu27Kcm2mgg/oukGqelB9DhvMw+FQwNPW+XYuIGcRni4M2Ix7eN+h6waE4BFCBYFNoIC6cJjFtEmyWlJhTIxM1echEuGjRlPIfVfcqEAcaFNwddHmzOh6mRTvG8YDoLBB9aOMnm7aRJD/rVZuyTh5cjhfhDfXadkfc0sBaubIgmg25aOmXxdvK0TFmuVatng9b+34F7NejoVaLNFiaMZex6LTiTDbzUiUV5PUYj8EiDNPVdOxmVGuk1HyelcPX6JKIvjg4TvBhZBZtVotUdtst1iWpTjwlhi3ye66Af2wUc6+aRwTRK9atNJb2meyMF20lSTEfR9kcTgCHEBtHwwogBklhUIWiV4NaOYoGQySQAgsTFxZzEHXp2jIGuGiOtdmGULAzc0dXrz4BI+P75HTjOPDUibashDQhR0Lx60FTGVGMymLyIT5Soiuo+dWAA2Qrn+jWLdGGmE20zBWNGSLUCdfzENhqLZZgManMtttDrG8XFFuNqF2bQfpSrN6eJ0kS36DCD44OCdcq2Hoiwa7JvsZJXq332Icd9jvD8W8mKl1hRRZ77elGZnglESyCrw53BZVrgZcJ420eMYWg0Ef1kdDil20i5ALYDjNe1K9H98hCfSKYB4HZ2SWCihycj/7/QGb7RYxRVymCXO8IGtbp0U77nCWcrp5vqy0DsoVFYss9ZtVk1UZuQ4YrrG3MhKNoLVvWzCxVkY236HFRExw6Ip810Zn1xBHkcEGd5LKzSxsFRaud9ty03tfqrOdZgvIAT54zTG6lUDYKrXBkc89NttDiTjtJUJpKZvaLbEV1FajMqO4CYD0YRNKtrJIc/VlWgG1CanZA8lbOgc41wHk4PX5KuxhAYN2/Cn4ogOz5E9hpg5OcL5hg5vDDabTE5bpjLic4X3QCnqJVqWQZirz8q2BlmljkQuYuhAsULoIrJABZYu0FsLOZHhD0aao1s857emmrbfCitGqDYK9qsPr/hVr+g8XQcyo7At7AG9FqDkjJQVxLROg55aIKYgQ2Dw0CPg8z9IWSrP5xV9yPcgHzPOCoe8KKwKotG4iIRR6XxF2G7gqjMZ7A8gRUloQgpa0eY/OBy170yip5ZvlrAxW7cnRCN4wDHDBwbluBUBn7fGR84LgAsDSU4J8EAOahfuVIXhj3wfc3NwjhB7zkopgAdAeHnLOlGJxyE0Urs14ealGZ+s7BxTtX/AvfOj0ryNPM1VXste+mpxtyW9Y7jE4r30asjAJmpteRQwa6dnKtldMCR0BUfEUdOKPUM5SgQwUjr8Uh4ipDZ1OICplyJoOExEG7VMh6aUB3TCUwtwKCGdYHaIIk5igGOOqpN6unXNG14tJY4ZGR7rokhAVyResB6WdZ5b+r6QNWCywIB8kfeVxU0SAAAAgAElEQVQlgux7YXNI9yFGigk5RYCCCBJLWiqlKM+SJUKTajRpZfXs2TPs9zfYbnfSmfsxIqUFm+0epxPAfMH5fAFlBgoiYEU70ABgbQZl/iud2twlA1aJDH+rQRDQpo/kxDW/e+WXFj6bNKwJLbc9G15CFu7nYt5aU2OrpLR2aoLa4KVwwySfAaFXp4QMIfoFLdAwjRCCFmGoiTNYwXAxsPDRfAjohwH9OKLvOvRDb0OnAmU+mWonApgTKFTBK/WVgjWA2WnXHmFzJmOOeAdmi6DMB2FF7F0xwdYnAiCxAKTWAFDhFuCYIIXJGlrJ9SQUh0AZDvBN2kcX3jAMONzc4HA4YIkXUMyY5iPiMmFaJtzutzidTghkRSUS3cN/zL9q3gNDu+qs3q8KQ90Fg+Gppfh86DK1EW6rkJgBVzpPO+n4xwC60MPDSejb+CA1oUyrNM0QahUQEWEx05kyXJIrFT+oQasBRoyzVPvoyokxFTOZc8bxeGxQbmCzkV4X2bA0R5hjlIGAQ4wL5lnap7eAq73a3miWAkuJpd9HAVolUOBkxTR19TJntKwFOYZLKsgGdxxHWGJbzLIKOIQCTgzkmMr1gFQiOSvqdeSECTxI77UYI6b5IkDxEvHZp5/i5uZGLEWJglvYqUW9muDLBM6t368vIzbac1/7dx/ayetz2Lgxs6SVWnLhtttIZj9luK4H0rLSWtcnLf6Zc0iwgUQ5JqeMxvyvXjFGDQCEFx85FuGw1qCDMi+89xg2G/gulIkzJqskkC26q8W0FfOqlCUocdK7ToQIQN/JphNgCwgYRAnkCch1wMqCW8EPNtjVJ+tChxRlIiwh7r2Hg91XbZDHnCCQJMFDesp6DkhRMiFjP4IQ8Fc/+6lE/46kMgqM49MJn336Qu6Jcw2QGnzs+kWAMjfWc7l+FhYLQJLua5UCAF1UH5lQGwlXmdEA4Nq26d577Kwt5tCLyiVhJJgW+9iPCal3Hh15OH3AnDMihAlrNyrXE4fZePfWA2yeLiVtZGtoWRYcj0eBNLoRm3G3EvIYI+bLpBhaKHSgFmiUgTHmgGorVsA5M+Z0KQyPZZlF63DCnGZkKO0pA1k13vViu6Yot8GFvW9RfNWqBq1oloClOV7Ki1RIkQMyo+t6PHvxAt//lb+FcdyBXNAF73A8X4Rp3AQ7BNtgQjZ3cFwxOACAd9IpGwDFDMq8us/2uaqWrXiZEDVbAqfAKG3A0JZZqtWrxQ3b7RYANOrTwlo4OKoA5rWUrl5ZuioXyIIAj8pHYgDeB93EQTIMcm2Z4JSl0UlSMxVCKNz+0+mE7f5Og4UAIJe2BLZil2UB0do/SCkVmCNl2Sug5eQDWCH2IjAijJaqccGjruJcFheg/P9QeWnMkk3IzAiharFpPorpKL1ixR8sfi1HOHYgHa+olU2h67U96TO8efMKp4cHxOUM52UxPz4+SgRvwlCg/pXIfLAAPKgENSsw1pHUG5Cl44C271k7BhKkYaUyK55Zd54J4ziu0i19P8BrysW5gOlyBsd5BWFY+qGtzTRTstqEINdWBlA/bhgGpcJUG55zRlwWZNUmhnNdsyfGcdRJovIdZmn3VAtRPiz1KvsLdHUFZmS4rPUJ0e5DBDIpxsMkGiVFE6y1Zqx7SGV4X8fhcrmgHwcscZIoXIOYTgtm6mQI+8N7D2L9UbMcgvLfgsd2s8f5dMG42eDlZ5/hm9eEx8f3uL+/w8O7N9LLbJ4+6DQONFqsrR0xYcwojOfy/UZTOzIsTAS3eqb6/fJlWmnLa5wutCsSAJwP2Iw7pJSw3QJPj+8xnY/IKa6iMxtoO6nhUtXRls0VpCUoF6wpJYb3WUiBDTfKJivnjGVaSqsEKR7usNvtVqaHcwVp2THGvgcjS2Bb2h7VxSM+lzShs6gXLoCRtQCiLhwBRxN6Ly2QMiVpLqx+VNtMpqVyt4vNFodBKa2w25jJ942NIbhaPbZCBKHr8Gu//uv4V//6LV69+iukRZrHPLu7xXR6wGYz4lHfa5+ZFEbSwHgdBNgcsmQeiOrnZBCF2HFJu7E6+xZum3nEt2nPKnCBHKHzlr902G52GDd9YxaAJ0dYLhfVbgnmS7RMjVbzyPcAcoxpyUIq1BL8EskayswVqmAGliUiLVH78UvkuT/cYrs7YKfm3NpGtry2lDJCkPSUDiOct4S2OLIxRd0JRTWwk7YG0hrK10FLSQqNUwK7APisQroWHhMu08rmYnRdhxzFR0SWjcNaQRTQOKl/Kh28ycn2M0iW0ZAUX+c8druxKIG7u1s8Pci1Ht69A1JE5we0O8C1RFJmlqZ1Sgtv9nsQ08hcqowKlZvEl+OcC4hehI9VaInhraUVJJfdClqrzQJzQoYVnoofsSy6Y0c/Yhw28I4wjyMulxNisytb65fVlUnaCC5imk8Az0gEJEh/L+9qIartRtY68eIWifk7nU6lM7ZsXzMU8p5MqgYkRBrtEFJUv4ehCeqEvh+gjb4Q0wznJckeXMC8XEoaCyQ5VRiUAOshbYvomqgJTNMF3gektKDvO4TQS2TmCIyo2r1mF0wDViFQ2jlkYy2ZxARiWYSk4/Xd7/4y/ta/9zX+dI6YzicQebz+6itsNzKGornjChivWhwgXtMVi5/MBsvIrnCZ7D0ICYBblVUamYJASHpGzYl8K/gfzuezSr+asyhFFPO8YBxH7T8RELoOA8ZyIkPRW2qOMSRsNzlQ1vZMTtp1s3Sfdj7BkTAScqMZsvYIiznJNjgxoR8H6dOv3LFe97Q8n8+a+5Q7Ej79OrqRJJWa86baygTl6ekJ/WCZhmaQyAGaMqpmuwY8a9gnA5RFE+UIRFk8/RCwLAkhdCUrUFNCdavFlCKIpRCGc0Z2snMJUVQqum69OG6w291oasnhcNji+PgeS5wUZ6xtutrImvS80AVzXT8LmI8lm4sV38sE1OhORYnUZtUmfeZzm9C2CgcAwuVyKkJiu3ewgrCX8xNCCNiMW0321obBVtJmKQgzlzlHhKBRGTn0XYclLmB28oQ5IUVp71RbK4k5y2kRDMbMoPO4u3sm2tRL0CD05Q7jKBiTI/H5pDq8ppfKas2VnSr37AA4bLcb3QjMwZO1UReN4IJHipX9Kg17SYthnDQCdur4luu5Mm6iyefC9QdM8xmo3RWBEDRC+4q43Ai39PgyQen7EZ++/BT7ww0up/f43vd+CT/76b9DzFGCJkB3PSl5CVCuPXizhoptUx0TNmHJAJ4ZFBnW1I2Za+ag+M4G8daggCFjQFeCVjTZdJFwGMyIYEzTRVUlCY3ZB1ymC4Z+wNAPyHGpoB8q+9Q5QuYoiPsyqXZKuqlolDIyJ2Bp3f4vrHwHyblFabgXZQIPhxvBxqhqJfH9rKSNwDlKV0KysjHbsCo3k1lbLxAcTqcTnJMMQxdqMDEMI5Y5fuDflIwFZzj4VRTNbLWXA0CKfWXVjU3Ng/lstiBli8QFxNrWqsyfOOIlgmXSApUBn3/+Ob7+6heYpsmc6hWyWmNKFRZeN0zJ1xFA0UcqHA2A3YpU8cXKe/VKlqUs52o0KTMjoKHvTMsMkDT2JSJk55F8B7csSDEK1147DdqE+SCDQ0RIOaLuvcMAazoIDKG3aC9VzrB9jlrzlWwFUhvFeTBVUzyOoxIOnSbXqURRBhhaAGKC1S6IlBKGYcASVSPr1i5934MBTJe5wgBY06OJSDtBWsoFxZckqpXodr5rLV+yI7421pPKJQdnW/MAyEmi8gKLaIbh/v4eLz/7HON2gy+++EK1kBTSiIBQ8ZFYgdiVIJFkMFrnqY3ufaPlWp+toP0F/jBdpoAyCIVCbMJKNZ0XrF3jOk2UAfWXwOYMJiSnBa+oobrPhj1B+z20N56QGIBGZlT4VA6cE1KMq8Z0tiJtqzzvBa/b7gTlt0Ajxhned5jnBd4DwQuTwfj45iuaYLW+lHMO0zTBaMuuk023GNA2CKl0+yayaLVJKZVd3arPY5kMu7++HyVIcITgrCyP1Xdbm3Q5vrYKsOS2nbdqbad7mTu8/PRz/PUXfwXZfxPwoattvWxhwKJBU0HVi4JmAdqmdq3T3qbMiOp8Fm1HpJ0b6/ull4UcuDLHjo1guCryFCKF7Hu4CEYWI+I8IcZFdvTVIlvZz3vGskwlNWOrt7RmN1dRL+ydVM8wJ3BOdWAUf6FmNFLOZTNV4Yd5jdgUpNXJysmKV7JCEjXSbYMT+1sAVOHhCyvX43w+q28hzq7swlE3tW81r3O+0H8cuaK9xcxFgTz02UsnIWe9Z2XhWbAl8IJsvmqdh6wARVI2spUiEWPoevzNv/mrWhUvfLXtdlu0kKhC8UWrECheRkItN+JDGWUd9zL/aMxkI5jt70LlVpmSyAFtzq21UAFYh7Orahz7jIHEtbe9jk5ZIUSGYuPDF6lWKCZGhJhJd0lrrm9gn++8Mm1lRY7jiGmaCsV6nmdsxhFSl7ku/Rfz3WJ4rgCzovqrX1QAS5CaWRMmE3ihhJv/VD8XDpz3TX8ymIYj5MRYENEFhxCqqS2usy1kzpLFSYuUAhY2Ra3ot/MzCz54e3uHn/3sZ5paE790v9+VDS5Mi/EHk9GUB5LBDlV+rlt12icWzJDha/SxL3Ld1BVrrQjUli4rAWnD0HIbTd9Rwa4YBvTajZOepzqOTgXw+rKyuZP0YZBaz5oPVCDUubKxvWlH2TQiYplnxLgICZBqBsLMkF2t9uyqC+dymYqGk4BAMhzLEgtVxzAxhiHvGUQBts9QSwywJV7frxkA0t6vy7IUYoBouooxRo6A43KsjX8brLS+3H6/xzgOZdyd1gM4cmVj+laL2PEFoVcznHKSvRmIyp7nzQElfSTAq/2vasdW2DJaIVVvrRGoUH+vN9hqtnYFrvCR9jjdHuWDtUOMUoxJ9QZUpwMcdfUFeKdVQMSIKaJ30s8CJI3vjC3y9PSE0HW4TBfxg6KE/hbh2f6b4FSyCpamGYYetTW6V0G3AhJLC2n1FEMCHA8Q+cbngQYlQSEd4bI5J71zQ9chOGHIsmZGxJci2bRCAUzOQrHOLEl7uyfRklyAVUb6cI2C4UOPYdzifDri7ds3YqLVFNpIswUA1jpC0cCiwTVsNJuRqSl709RAmbYipfK/GmHSSsEQc/G9s45VCQmupb88ThMMyCQ2UYaqfettbyAemn9FXqtTaslW+ztn8e88EYLiUM5LEYZttpVSwu3tLYgI0zRjf9hhibOW1AU4bz33A4LvwMoBM9qQ0Wn6fijPYzBMzhkGSK+eialoHKsSFxehrcZicNO23LQYKYU9eK/PImVxhh1KZlqO8Yr7eW9FJ5bH1HSV+soxRszzXAKOm9sbJSEwvvzyy6JhWE2hmEDSzEMbEaI2DgbUdzNZUStFVMilZn6LBmsUmZlZhiEObKZMtKFG/C7rAF4zKtqX/E2NILY6i8pF7fiPvVq/ixqhK5MFaejiQOAkNYt1o3WUpnc5Zwz9AO877LbSkGXcjGVLaNso3mg6tRVmjU4tV2rpsWVZZAvp5j7b7+bGoWXO2r7AgE47RnKAJedLrAUedcH1/QgBbT1CGOBDj9AN8G6Ac31x+E2zWhBQIIXMmKdJun6HDjkzXr58qQFQNYVyjkYiIIwK23ALzT21eU6DP3LOqz1N2zm0+4AplSZQK+aVa9gDwPZWqvSTjwmY3YDZ2mttJSGr4GD1b15/B6IdvJMw3ZxZkdkM5z26LiAuCeBYose4LOBhwNdff43vfOc7uFwumKYFn3/2HczzInABSbWTRZWAcOL6YUTX98gsTNh1Ul8E6ebmBk9PT8XnucbWzNSymmIp/tViYSfEQu8ZzndgOPTdAIKki5ZlAYPRdwO6vlPKD+D8IEIprSEBCCXboZJHq+kkkAvI2fzRRavfF8xzxHYrMAsvlRlry5hLYEZF6yBX/+t6rhlqUhlIxCpA60o0ZKUo6fMXlWOajVHwOtJgxNnFWom29z5mPlWK1p+35uNbvm8ApjmrAAozALBGI4QQdCsZPd/lcsHxeCytN6Py+bdb2YnksD9ontR46TZgKLv5Bj/A6h8rE0K+dT6fcblcStPkNulv38uNybS0VHUDCEwOcE7SRfBq2gSLm6YTYlxgEAFRRgiA8wxh587IeQawwBHKtj81RcaImRGz5FoFevG4e3aPzWaLV69fiSA5J1CGZkE8ETo46XZpvhih1DWguEGNL55Z9kVqp0w/K64S6nsy31xKIk2bAqrdliTM2/WKzR8VLBKdX4TRHNqVaeS2dOwqslGPwGr7Sq2nqtxVcUqnvokeG0LAsJFGKl999RUOBynoPZ3OIBIhZGYE1YTXjX+NuWpcMbvneV5wPl904lLx4Voz+m2mvy1IFsDVwyGDc0TOETFOmJcLpumsEfGCZZlwmZ4AygJELwvm5VLglJxZkt3xDCCVe0pJUnPESavUGX3fYb/b4z/8D34DwQekmFamrQRrGoFmQV81KOK6+0nzKjw31VjJsE6qi83wCfPzbH4dCJ6k3b4pjQRGdkB2qF19PjagK5PXaqnV/Zkl9gDSt+AoVfJlmzqpj7SIiiBgZj8MipZ7Kdl3hBhnZO1Wk3PG4XDA09MJIUx4/vwZHh4eMI4jzqcn2CaqhqdlizhLrCH3Zo30cmYpJ9Oqq8vlolVGlQhgfmAbKFgNp3dOMU81USSCEuMCzhGcGJwSljihyx7e9Zh0DwE4Qt+NcBCWcEoLUpoETOYMRwuksYzc/+V0whJnnM5HpBRxPs74V3/0f4GgjNuYChaac4YDVZIkObBmKqwdVpnXq1cp/iASv4tR5tRaK3xwVNacrnNgMmGtghjE91nn1q6FTrAnjVhY9/5hi2YFjuAPL90EDEDtNQrZPY09NpsD7m5vMQwjLsuM9w/vkXLGJ59+hvfvHwAi9H0nHDTncHNzg2laFONKmCahN1cHPup4qFb1knjuhx5x0d4RKjjTtKy0twi8kAitiupau9v3AGUSN+bB0lYxJ8QpYro8wrsgKaJiKgnDsFUWBiDaXarGyRPAVKPjnHFJp6Lp3j+8xTevv8ai5IObm3t0nUecGJ4cnr98hq9ev8ayLNhtd3h+/wyAaPplMQ2tOx53EkzkmAqV6VoTdpYlIZS9Tm3Sc/21QGit9QDpM0MgoVISF2Ozx3gjYEbfgTU2YzXWBrKyheQf98eMi28T8+zZc3Rdj/P5iOPxiL/+6hcAOfgu4Ob2Bp9++jmCk16vRB1ev36N7ei1P6yQ9V6+fIl377/GwwPwySef4NWrV3Ba/SwTJyYzJcG1s5rCzThgmuQ+eq1If/fuXUHXY4x4enrCdjuu2nDKudYN8YBaLOxcQAiidUGyQx05h8vlgv3trQQ8FErjY/NJC5uV9Z4hzV66Qbt/L4SHhwe8ffsW59OxuBaAw+3tvfzOwOV8wU+Pv8Av//IvIeeML//6S3zxxRfFr5NOiFLyRyQt63W160a4lYdmwhKjtEEw2k8bQUIF6xoybTFUKYlRzX93s/1dGUiDN9aiYpGlhdak1dKwCEYPJt8KFBXMZ1kW3N7e4nA4YJ5nvH//Fo+P7zBNZ91BLcMHiWKmacKbr1/jm69f49nz5/js8+/i/vlL3N/f4XyZsN/LHuPLMgutRid6t9thmSccj6eGBSLVT1JjKdHaze0t+i6g70eM44C+78uOJrvdBtvtviD3IdQ+ILZji5lJg0uqj2d+3oQUF5xPZ5yfztjfHjBspPs2uaBwR4UOCkN4BQPVegPnPDabLW72NzI233xTtE6MC96+eY3Hp0f8+g9/CALw87/6KeKylM5MNn/y7xp4MqAUgO6TtabSk0aJDhD6NlWqVdF4RQquXKtWLojg7253vyurqva7al/i9KOGpGyRZ/slwDY8ELUrJ3/+/Dm89zgej0KlVoapIerEHrIbvPllGSDpqvz+3Xv84ou/xs3NLW5v7/D97/8Knp6OePPmG+z3e9ze3uLNmzdwzuFwOODx8QHHp0cAFfdaloiYhNj44sULbDeb0kzFKqeISO/N0HFB3NuyN6kNrUGFJdmtuFiizSymaZ5wOT6hGwcMmw3GYaM9MkYAwDBUvr6ZqJp0D/ChQz+MpY0Xs/bQyBnbzRan0xmXyxkpLXj7zddw5PGnf/ZnWOYJgQhxiSpQNTXYQhBsbo9GhYCg/+VedJ6dr5uieucKRFGAW2ZYjUURgkbI2pe/u93/rqyedfXRStAg2hbFwa0SLF8QA+207nAcBTh9eHhYtZgUkLLRs5pechCcDETIWbRmhvDT3r17gzdvXmO33+P166/xt//2b+L9+/f4+c9/gc8//xt48+YbTNOE73znu+iHAfM8oQsBUSMuRx7b7R53d8/lvDmV5sAh9LIBa4wYhk1heVg0ahvQy9bT1iutmlvr1gh4xOWCGCdMpzOICIfbW4AY47BD1w0AdI8AqjQh6WYk/pnTnYFJiZQWceeU8PjwWFrc397c4Hh8wu6ww7/+l/8nhk2P4/ERvMwy/w30wSzQhoHEAMpWkqapDI5iZu2zC+3ZYQTPCnkY+i/B6VUPOxNd4prHVIXkb9VctgUO1xgZgVbN8T6QVhU676XlwOVyFgpNA2xW5xk1RVFiYvnXey9bBqaInBk5OaVuM77++mv8+7/2a/jxj3+MEHr89m//J9hstrhcztjvDzgcDpqXlCXX9z36YcQ4bjEMG3RhEH9H6zKdCwidFaQEbLdbPD4+qhkL+uNLFkHoQVLXYH31rQRwnhekZcJymZAzsNlssd3foBtGOHLowohhFKyu7wfllAlOxiQNZUiZIilleCddlRat1np4/yBFzpTx9u07AIzj6YSf//T/w9t33wC5MmRSkw4yANY0EKEgTTVKbP1vM5VqvURIFRVQbr+Bx60gty8u5xKBJUfwdzdiLk3STSjag0v+Ch/aXtJeWnKO1ERj1Vdp4QOwJVBbTZjVxNae+CBfsZqc4X2P169f4wc/+AG+971fxqtXXyHGiB/+8IcAgMfHB9zf3yFn0g2+ZIOGvh9U4HoMg2xlaL0tQif5Sds7oBS0KHXIgoO+78E5SstR9c1MCFOKOD09ImuE67seruuw3W4Rl4i+7zAOW5CTfmyAdOsmLUiOyyxlgNY3TAFfy6kejwJZKNcBr776Ep0nMDu8e/s1Hsy/bQK3MuEqYLaHaPs+NRJR57LpIWfaFiTt7plLXrOVj7U2QwkUHItFYlKczAa43MS1c9b8+WEmwAnD0TVOY6MNeaW1Kqq2GhDVbjlG9OOAlBLmWfaChKZ7xrHHi08+wePDA4L3+M53v4dXr1/hJz/5CQ6HPe7v7/H49ASpiL/Fu3dvEEJfiIkAME1zacFgq3EYBgTl3u92O3z11VcAS/9aoxrlnLDMM4a+R6c9yELwWOYF87xgmU+4nM7YHu4Q1JSmVN2GJUWMWjXf9z2YHChYxEagUCnUrG3hjcNPRHh8fIR3DsfTAzwxTqcnzAvhZz/9aWlQcz13IiBcdpHhRqCcqrZrZdJOuL0t0Xku1gyt49/ITdFsmcEkgBexlOiHFgv62AXl/aS+mLYSV0VFJmSwOvGmSqi95aK91vvyWOK6njBimYGuG9EFj2WRSGoct9jtdgCkUPabb77B669f4/nze9DtAT/7+c9we7vH+/fvsN/f4pe/9yu4vb3Fw8MDcmZsNmPpvTHPF+x2O0yXC+YpYxxHbMaxMDFevHiBd9+8LR3AnXNwTEDM6Dah0L5TSjg+PuFyesTx/XskIgw5wvktuqEvnLigHblzivBezKdXDI5DJSfID5cKrr7vcTmfkeMCoozz6YTpeML5fBZe2c0O4zjgzdv3FUpgmYGiW4zsQVQ3sWV12htFUQpcCpuiTJGaylqVtI5R168SPGRB+rNQcOEPu/F3vy1H2fpRBAh8QfWzahahuNm1/2VmtNZB2rnqNdrftUdaFt56TAtSygofiAN7OZ/w+PgOD49vcZkuWJYF3/veL4m/0w04Xy44nU64ubnD3d0dmKW/xzhucDgcAAjKv91uxeHX9u2mgb33iIsQI512CUIWIuRm3OrnImTLdMbTu7c4PR2xORzQDSO22x2cC9jttzAAehgGhK5H1w8qfIqGs6XYjDJuboRM6nQ+43R8BIgQ5wVvv3mL4+lJ22wlvP36FY7HR0B3lGHmZlOPOk9mKhmAbwDkNqMBQBWFzXULrTSTVGxpmTU50jUWSv1izqIBPzCX13nHD/OQKMLVKiytM/rosbXesUGOG4HMmsMEaW8G4kJKDJ1TQU04Hh8RZ+my44LHOIxgZvziF7/A559/B8OwVSIi4auvvsTd3R0Oh5uSjzSn3ftQNNf5fC69a5kZXeiwHUe8fnzEsESkGMHBYdwMSDlhmhRwvUw4PT3izTffaKRlNO1Om/QBISgDl4FRgwdyTgiQKrgSUFiqS1nnIByfnvDmm6+Rk9RVztOEuMyYLhfkvofXtltOAc6cU3VFqve9mjNSv0o6kiu151vy1UV+PoKdrgROXR1OXHw9+7ZX31JKGr4FumgF68P3U0F4bYOAlS9/da5vy41a5AJyYAj6zSQ0mGVZMAySSzwenzBNE6Z5klrJrgcUUT6fLvji51/g8fEB2+0e4yjFwO/fP0gvjf0ejhzO5wne99hsdri5vUPX94gpYZ6WgoY7J/uhbzcb6f8xTcrrkwwCmBHnC85Pj3j36jWOTycE3b9J+GKyglOccJkuSJnRDaNwx7yHJ9UmQInWBRLZlGBlWRY8PDzgdHzCNF3w+PCA8/mIfuhwe3OAd8oMQU2d2XxbbziTASMxgrnuyKxJchOKVrNRnRxYE+MSbV5/JrgXYJT3bP4fo5X40Dri9nJF33C1Z8oIbWLOqpWMZt3CHrRuwtICj62WI9LmvoqT2Xcla5UxzxP6occyZ7x//4RxHOBzxvl4KoPQdUm1gPhgz5+9EBD4dMTbt2+x2Wyw2W0xXRjYRAMAACAASURBVC4oG7eGgMPhgMtlQlwV8wZsthvE+YA337zG5fSEzThimaUvRF5mnC8nPLx9g3dv3oDh0fUDuq6XHeugfiY5jJstNts9hn5Qs6T4l42Rc3CsewfA/LIFaVmwGUfE7Rbv378vhAKQUpFYC7tZuV0aHBUyQLP8GY2P1rzaiHL9gQpR6Vq8Ljppj9XZhyhmLgwduzJJT1LdLEK/fG0eqWBYH8aGBOl10VKJr7VVG0xYaC6pCl/OVo5TX8HaJdm1pumC8+UJtzfPMQxeOyGmwngNIWAcc9lmhpnx+PiIYeiVuDjgdD5jS2K6LpezbqMTVNBuC8Mh5yTtEzrBzY4PHZ7ev8fQD+jCVFb9Mk94enjA+XTC7fPP4EIPBimo24sPuNlisztgs9kJ6OoJHLk8t9RpxtU4pZRwuVyQkyT/j8cT5nlG13d4fDzhyy+/xKBa03mUBoHWcrNObitEXAMDVN+zgGfG5bcx/wiyYNmD1haywhnm4ki9wjo1VdpfiEpdS3WVRhR6bRHs2nAfNTAkLW9fMxY+uFtYiVUd2MwMjgui9aJXsxUzMC8iTM4TLpczDocbEHV4eHgsLAnZFifCeunnzFiWGXFZcLhx8JsNuhBwOYtwySCydgZCCQxiipLO0Q6PNGTc3Nzh8eEBp8cnBN/p5BIulxNOxyeAgXG7xW5/UC0mK3oYR+wPtwjdRuGRiJRzSUY776/GqI22xUIsy4J5nqSvyDLjcr6As+R3N5sNOGfM86y+5GW1yLMKV3HdCYahoiqJxkyCm62eGmFqcTRu5KMIIn8ok00QZf8Gs9lESpVlgB2tJF5sNRVzVhx/mOBLSPOtDmR5iWSnXBO4bR7UFpNUMdamcd57pLjg4eE9AJaO0OOI4/EowGgnLZus96xp5fPppJx/MWeyJU3Gu7cPSCljt9tjWWYMvbA+mDM4ZilS6Trs9gfc3z/H5XxGill3skt4eP8el9MJN7fPcHt3i8PNDTYHiSqHccTh9g59LzlKzhGcZ+TM6Met7lQs+F8hHTT7VHnvcdKO313f43x6QpoXbDcbTJctlkUYvMu8aL3DiCd6Wk2+gbBOhaf4XwUjs+9qNOmoNrMoNcIqld+OWKBVba1glU/1mWRrH3PAnROMY/VFO9laWrFahR8PGGr0KL5OXR65nIP4Y6V0DsxRHFmy0jVhORARNhs5/2effYZpWtCFXqnbWvy7LBIxdh3gCDFlDIMrxSVm2t69eyvbTjOw324hIYwFNBk3fY+UIv76Zz8FUsKyJCyXC9K8YLfd4/75C2wON3ChwzBssNvtBMLwQTRJToXLXFoneF/6skI3tyAXVj1WrUp+vlwwT7IRxPH4iGm66BaFsrNeSgmn8wmWQpLuPS2o3sAKjda69pGLWInGWWk8RnVlrrVWUQp2jvK9KnhiLhthMA21NnltUFpvuNwaMYQRK0vm2qmvEi59tYr6BqTB2gcQCZV/QeLLSKX4IhgTEZ6ennBW87fbHXC4uVGz1+Ozzz/HZZImfpvNBptxRPBek9wD5mWBCx6sfo8g6+LU932HtETs9htpn5AZ/dhjv93h8viAaZkxn89gIuwOdxj3B/TjCOc9NpsdtrsbDP0oUSgyEjJ8FzTzoJVHQuQv40zEpfLbWnfJziUJ8zSLWZym4oeacMQ54u7uFm/evpLNOJpqMzQTfQ0rtfNSP7gSngKGXn2X1n53y/m341pGsb3CCphrfK+aZFVHEusGHVbQK8G9L8LxMZ+sOvK1Jx998J32Exsg2SHEOhXKKg+lt6xRffphwG/8xm9KS08VqNNJnGbvHPrdTgHRDqQFuDGKtksp4fh0xOl0xN3NrWi2mOGD7Iw7jjvc3j3H49u3eP/mLZgZh2fPsdlusRm3GIcNbm5vG0eeYWVyFsVaQxZQU7CDjMiyQSxUe/kQMC8LLtMFzjt0Qfb1bNtNzfMs/Lkl4hdffLESMEAS0wlqlRiyHQ7MfFbXZJ2K0oQ2Sb2SkT9bX5FJG01cuUQlnWRRKK+VSxGyazt6XYMpDrIMhpU7WSQi3fnwwWpoz/eBr9b4eoABkKbUuFRXmyEVOo3DMI5gRmFFbDYbjOOIt2/e4E/+5N/g/v4Znj1/gc1mi77rpJo7yF7eMWc8vnsHZsb9/T3GzYicMtIiG4TFeSnpk5xZ0ki+wxICdre3uHv+AvPxCegCdocbbA57RN2CL3SdBkTaToAIXd9LwxbloVnKhyATb8lmWKUVhHXBWgaYYsS8SG40KQO3FY6n0xHsMvreYZ4SUmwCMX1lw69MWaj/nJsArQhDgSFaa6IChuqft/ewNo3Q7bWpLKamdZRNtNOaOl4JRjV1GkGi+lAfN68fvlYarvy/lUqbIIs8paOOlK9JC9AUxVxst9uyEG5vb/Hll69wOOzw5u03SAkYxy1ubm4xaDt2R4RpmmRPJ23vdLlccHNzh6H3mBSpzxvRymJWR0H6M2PoByycsb3Z4/nLl5gzY9jvEboRz+6fY9xu4IMkxUMIGDf9mvVK1YTIaqojaC3MM0To5un/r+zNenVL0juvX8Qa32lPZ8p0le02nS64cKsNtGU3FgIkREstcYXgAyDRXHKHEHf+BH0D4o4b+gsg07gFTXULWYBayCqpJzurypWZdTJPnmkP77imiODiiYgVa+13n2pWaufZ+33XECviiWd+/k/H9u6B5nDkl998BRjKqhSC82k/19fX9H1P2xzJnKXte0LTL2lvLRwrEPE5IJXUzTDqalMCmmoyaqLfPVpbPBaaziYMKxz5/OT5YYxJ0kQSjpf+HWX2YwPhLBdjSpDOBYMgiMxgfYXdZpJOHJIbb4zlzZu3rJbrWBjS911Miy7LmsvLC19tnVFWC5xzPqt0wPQtzcGQ5YrSux9QAsacKYk3aiRp8LjTLFcbTvUOrGNzsaG+WLParFksF6xWS+lc54H6qqJEZWNhTnizUuuY9pw2KnNGwGSkQUbGYAeWqwXGdBwOe3b7PcuyIluvORwOrNYLTqdj7EkV1k6YgOTyj41l07lnoqifk2LnfpeVdskyj7lk0YCDWNmUujEAcu0jQs66R7V4gM/eVF6pZyTnWeOe+Y5IBzPfNfiJP8/5ZKdrrSl0Rms6yWD1le7WSieSH/7wN3j37h0oR98bfvSvfcHLl5/zwx/+kBcvXuJ8ZdJgpCQs89DnNzfPGAbD/e17Pt6+ZbO5gqTuUynBEBsRtMVvtlytOC2XdMagc2nVkxeFbxNURk5WFAUOqTay1pJriYTkfoPkWkdL0jqpa0QJV+v7jvuHW47HA0VVUZiCTBfs9we2D/e4TLFer+n6XohyGHwlVAx+ECu17BhLTu33YC3OIQjmR6o2zZd7GsOWI4SuQnFySmh59IeoKeeZcqFQnW2T8yKOMqHaPu2fGAb5iMBScpqJZsAnERKhDJTKefn8BR8/fvRVQTl/8Ad/wHffvuN3f/evc319xTBYvvnla06nryTPfjBsLi6FGyV1k8ZblHW9oFosePbspcQuffWOzjPysqQuSwrfzMKUBRlw2m3JMs2qrtlcXXNxc8NqvY5iOGTQWmulX4DOyHUGHh0xJELawUSrDW9sGWPpO9HD+rbjdNhz+/GjiErTczjsWGUllOIGMdbSNEfKouRopKKe1FtqR9AV0aG9mLaPiWO+HoH4JlzwzPrNC1VSuglHSE/P5yc/JgJkgMHiiGatjoO3aoQnEkIZ+w89RWDTz9Jyee3hRxUGiTN++PDOE3fJfn/kH/7DH7NcLrm9+0CWKYqi5NVnn/Pq1WcY4zjsd6wvLlmt1pRlQdM0UR9zzvkM2SVtfcJpJ8Ho45FSZ1S6BONwOijEmtK3AmyXK4p6wc3z5yw3F+RFQV0vYkldSMdGCzcUi3z0fYleliDm+Dl2dkTuaU4NXduSaU1zPCIOZMNAT60KMRqM4XRqafuOXOcYF/qCTudXQeyxeW5t5wSSisG56JzTxzxZ8ZGO598HQKdUe84SVKSWoIJAXMqnrITU6zOyPbVQ0pebvuT0ZUPSnrOW0ne8zXJHlkm/oT/8w7/JZrMi8+Gd+/v7WC52f3/H9c0NVzc3XF5eij/MjRCeFxcXaK05nU589/23HPsTTom1ul6vOTYN2+29RBl8TySVa7QCMxhy72Kxvp4zizUCIziK1prSuy10lpElKkLqQ0o3oRkMh/2et2/fMgw9i8WC5Uq6rzSHI9pHSA67Pc5ZTu0B5STc1tthann62G/0dSV4I/M1CFZlkIVxbUB6TgW/mH2sr8+Nw3RN5+udi06gAjlNBqCUio7CAHHpxJ8xmrUxAHEeyecp1jxR9JNBhoJXrXIUUOUF7WAZBsswtHz55ZcAscgjtk7Mc9qm4fbje+E6bcNiWUvPJSftdLqu4+JiTdu2PHv2bNQZ8pIsK1guxcemihLjIFeKHMXgwDUd9thQ+nTt3BNWURSx74D21qwjVJiPLpgsWZBAbMHBut/v+f77N+x3DxyOO5rTSWARnMNYwewoi4KLiwtUnnF3dyfwDVYiBsZ65Tuuj8ywVWA8xwx95Z1z+BqPGBbM3GOGYHGTZl4qaFVKPTp3Qi9naCCPH7jHOllKjeEn7lil4iDP0dGc9U5dGFMCnLhRvP6nfAMr0anwhF6w2+1YrVYR4Uc62hpevHhBVQoBBFA9MwzoRY32pW1lWVJVBXVd03uId0HgGWICZN/3FLnGdA6nMzCWw3bL9v1bOBzIVmsKB4VP2VFuzAsLPin8wtqAXasVQ99H5Tms3dAPHI9H3r9/x/v37zgeDz4w3rHf79FWesIHp7H058zZb3eiR6bZF951YZy0pkaBco5cJdIoegOShfDjDR878HBRTzOIc6JxLmZT+gl965hzshHekrgzzpbD4fOGmA7qnI434XDxtcZ7OxeUT+dLxRR5UVHa0vvFLri7u4sYrCsPvb5er2iO0mOgLEva04n3797yKvuc5aJmdbEi1BcEcZyXBUM/YJRglEkrRMtiUeOsoR96MqVo7h94+9Of0334SK2A0wnbdth+IC8LcTzm4nS1IAgODt8QQ/kExgFjXKSu0AvUDIa7+3t5p+bEYXfPw919zOI1w8But6UoCxSOsq5ojG++oRSD6Udry1p6z2msCuVuoD1wxcRfpqa/Kq1wxiVR6LG+EiX1Acq5uEmeUn/Cv2lIyTlHHqLykZjCQIPXn2AUT8lCB9tZ7nTWJfG0qPSXJUaRPM9jrwZvsXMonyUb8+U9t3LW0p4alstltOyur68pfBnbfrdjsVhwsbmiH6Cuc99lxaMPaU1V1oQmB8MgsO1d00hP877l+PCR73/xNd/85Cd88dmvoxcVtmno7u4oS+k5pWrjLUopGraDla54ynn4+VZqSK24Q3pjpAMwjqbpuL/9yMPdB+7ubrm7vaNrW9FFfU+jxXKBRkWYBG0MJx+bdV6UKKVQTlE4FcFdrPJoh747n/J6VQRLUS627fZTP6o8fqMov7ZPKjyJrhbpIqnfDTQQnbEST5vKWLGAxM80fuW8pfdJ+nnC4pCAcDooYeJjLFPr0C88yHkxhbvOt4EeBjItNY3VqqI9nsA4qucBhl2ILNPCCZrTnrzM0CzIsoLGDDhr0XlOrivQCmOFIxhrMV3L/v6B1z/7c777+c/45l/+OVdlQffsFawWHLb3/Pk/+wmf/9Zv8a///u+zXn/h1QUl/Q36AYWjGzoO2w90hwN9e2LoWtr+xLHt6TsweYF1mcRY25Zh6NBKewirls16zWl/jKB2ocNJ6Gh89HMUw2BBSQ/uKBi5GW4Ui8F9khhcIfXcr7y4Vmbr+ZRPMyxkKAJOXR+B2CKRSW3dWLFCGGSkKJU4a30fxJmiN1cK3YTTBVYZ7i6/T4PlAelwfGHnBJs/t+IiCJ2Gy7Kk6zrZ9Ywm86k5CkexFodB4WiPDR+GlmOz49Q2KKe5efaCerWiqteU1YJM5VjlaE4Nr7/+BW++/gqz2/JsWVHrgsV6Q1Ev6bqGm3XNzbKkcAZlvb7jdaauazgdHth/+I6vfvrPefv65wxdR6nFD9cNhu19z15VXL38jKpacmxa0bH8a5dFQds0NK0Uuzy7vqEoC6qy4nZ7R0jsVEpJFoyfVitMC+20XypR3qUWUs4LQMIZTPLawpqFloapz+wphT6sYkqIqRsr/J10JJheEW/qd8aoFap48lxEprI6wBr5p8ddNz7usSHg6dFbsp5YrSHLxXrsu4GiyH0mhnjwVZbhkGzYh4eeqpGyuPVyydCvcMby7Vdf8v3rn3F/+5E6L8mLmvXNcy6ubsgXG158/htcbS558fnnODLWV1fkyyXqeKS8vuFqfclivSErS26uLtlcrqmvLikuNlAW0tXXCYG9f/stb776kq/+2T+hKhSHu/cYNLbQoDKyIsPSsb3b0fSW569eMRjHbrtDKQQxsqromobr62vJji0KirxAQXQeC7wm3sEqs6gT5V5Eo5suGaPHQgjAT7pYVnJlUiAUCOdTDvWUbp4KUSXW5RPyz4k+NIrIVPzJS+mzrBRv4Xj0+MCqHw1i5HwqIcYoLhHlucgrwYhA0o5TjIosz2makw8LwWJZo5SjbY58ePNL7r7/Baf9vSzIckG3u+f97TuGAXRVU29e8vz5c/7w3/sPuLx6QbVYcvXiJae2ZXl5yavnr6g3S8o8p6wqLl7esLy8pt5comKPT8f+7o7/+x/9mPtv/4JV3tPj2D7cs3U5L68vWOYa6zS950RSFOIwfUcWikS04nQ80DYNRR/CVFLy1rYNQ98LIUS/WyAiyYgJmrPoack6IPqZTop3A9/IRCny9/CEOGvIla7RxGsAk9K7sLYp55u0vXnCAIkefO/pmBGhGndL8pDRZeEf6AloSswjoU10AiS8lOdhJhy51gxeHJSFpDZvNpuIsBMcocvFAuUcx/2e+7tb7t/+kuH0gOl7UBn7U0OeaapyiR0M7z9+oFcfePtmTVbk/Lv//t+iWix59vIV275noTOWVy+oNitKpSiLnGq1obq4olhtkmb3ho/v3vJPf/JnPKsadKE5tSfA8ZffvGO1XJJZi3KKbHGJbnbUVU17bNnuH6ICZYaeoe0EKVtrbGYpioyu7yT12hqs95HNJcPcMy+czbua/H94MZhiwdqg6CeGX7DKIgdL3R6MTEL4znlPQrh//ujbuOwpWw0P0zEhb/pC8t0505ZI0fHps6fM2bDzInPMRyJ0XysqyZQ1+OokAcALfipJ8jOcjh2H3Z43b14zHB/A9FjTg87YnlqeX12Q5QNt39O0LftW0K9/9uVf8G//3t9kWT9jc3mNdlBnOav1FYvlgiLPqIqccrmgvLgkq2uUIjYsNdZgzMB2t6dVlrY5Ua4qPm73FMsL8rqmMxl1dcnpzS2XVxX7/c6rBNLwwjqL04hhUo6Ae23TgBO9VKvpRo6LHtZJxw9jvv+o47sggoQBMNKPcqPREKIxcU1mrqrJ+s4419ytMRWXgYDTgcNYocSY7hEGoVRy/ZnDYWM9XuBqU3Y6Eu04ZpU8Q9oWt23L1WpDUeQcT0dBmvaWIkqyHJxzDH3H6XBgt9tyOh0ZmhbTd4BFF3Boe1a9Iy8Np94wDNB0bayLPOweuLm6pqqXVK8WVGXO5fqCuhKIzyLPycuCoq7QRRGbIjgcg3PcPH/Jm5+9IaOnOx3RTcvQD1xefY7OV2AVXSOo24fDnsP+QFmXaJ+eXtcLqrLEOseiqqXiXXt91DeISGHgx4UNKrOQTnAJCYcb1zVwtmjgxTK2QGzeGIg463IVmpgMPZV8KjUyx3VP1jQK3pRuxxMt3jyJBubIrbxnPgEcPhdITTx8BJ/LSMf2EdU/PuQNRqTGnLIsKDPpgdQ2jVRma03Xtmwfttze3nI8nei6jlPT0nQDx7anawesBWM11y//Kn/1R7/DyVf9hPEfDwf6XtrLlFVFsVigqoJ8WVOuFyI2lwtUnkVQXwm8Gy6ur/mtL36b3Wng9qFh1zje3+7ZLNcYu6AbNDrP+Bf/4p9LZfhux2B6SSfKcuqyRuc5WVmyXC0pFzXlopZMhqwg05mU3qm5g4EJR5p6B8YlkJ9QKebAWd9dzsYTx/WwnoEkBPMkIxmjOqlXIfw+0ckeXaxAjfTtjYCUy4SM/PPFI/N7OxUyD8K0pOVwakLA49SJ/y4tpLi4uIBemmEFDNmmOXI8njjs9zzcP3BqTnTdUeAtLTgn7atzrSmqFVcvfoOvv/qXDL4Bl1QBCdpi057IckXet1i3EqyyqqIMfjUkqE0AkvESaL1Z84Pf/E1WF9e8+e6AHQaavqOocv7ff/J/cXF1zXZ3z4d3b3nx4jnG9CyXK6q6pioFsM9lCucMmfK6sBKITmsMPY6ri0vpw0Q7yQmLxOXG+SRwncDFZht6HvKLn3mwlehUDT/ApHhEefGaOGXPMZpHOf7pMQYX3LjcbmZlBMvlE97Z1B0SPGaBiFLxKIPUj69VjqIqYk7YsqjReeEhB3Lvn+o4Hg/s9jt2+x1t12LtIP4/J4HiwTiyQnBZnev4xS9+gbMpFlnO8bTncNhirbSRdk6gp6y1WGPpkHQeB2CtVE8jISJjHZurK/7N3/t3ePcnf8zDfsdgLE174KuvfkqVS4/Oq6tLtEggiVhUNeViSb1Y4pwR67LtMNaPvR9oTidQisVqSV5WcNqPk2SdePbjShHHKPaEiwq9bxcRvaHzdUuV/LjGEIt/z+ndKTz/OTdHhI6aU3QgiuQPEY2JwyWYsFEvmw+WVEENiY+ph8w9Oi+8WIh/iYsiA1VEULk8LyiritVy5Yt2HW0res7Dwz1tJ5gXkufu6xKCra8y2q7lm69/QXs80htDXS/ZXF5y/eyGV69eje4TpSgrzWK5IityjDUS4/SIPDjJzbdGiLkfGozrqcslD9t7tvs9u8MRO0jYqm97bm5uKOsKnedcXV2xXK1QOqOsakDRtCdOxwPHw5Guk35Kdhjo2obe9KAVD4c9p/1WjATnXUhq9IuJ33yiXHsCJH6vZmsUFzVeor0hQTQS1BMO2XPrnh4T6zKlxDGsNPKywL4nro6E4pk9IFB3QDYcX2M62OnA5GmRVTsFThpjBeJru45FvYhJgsPQczgc6DoJzWjlkQt9y2iHKMYWx9D3NO39BIF6sV5Q+CxaEb0niiJnuZTuKCISBvrW0Fmpz6SqIv6G9UaItZY8K7h4fsVf/xt/gzfff49SOafjMaYjkUke3mK5JMsLcMSiE+csQ9czdD12MFFfE2IvaQ8HHu7u2axW3GstWLH4KQpiUYW1cnFzpzOrAWeJmCOTyAwuKEbEcFW6nv7fR7UDTK3dR66UlBjm0fN04RUjmJrczDfkenTu9MG5N83l0qAfTF9+dCw6nJM6S2nLF86T9jjGGJq24XQ88u7dO7KiIFMZRV7FFyyKQpL+PAZGABpGZbFeU6kxmbEoClbLjU8ZGhiGjiyTFoNaCSS8wvrGKwJUZ8xAN3T0wwDGd4/z6UhlKbj9Lz//NX733/o9lssl5CNi9sXlBZuLCxGTZYFzoogPduB4OtCcBBo0NCoz1sYClaIoMG3HerkYlQ7fv8jqqSRK8igeOUqzbIwznuM8aQzyXOb0nKDOZeekBDy6MM4Qixq19DhglcQvo+3qH2rPKPwBEnPU40ZuNuec4fcMcE7gBFACjtIPrZTzH0xMFvzyyy9R2lEvqli+b62hyHP6oZfmWn0vldiZJsvEaSsWoUzgarWiqguqsuDyUpTq1WrDZnPBcn3Bol5RVQt0kWOHHtf3GOvIbQ7OYLzrQQizBJWByuj6nl/74Q/44kc/4rPdZ+RagJaVUtS1QBo458jKjLzIGdqO7nhkaAXmvTc9RZFTlxVt30XO2w49l3lNWVUYO/hGrYmbIir5BkXA3ZiQ0CMjMV2zKWd7QmcjKPvT6+bFKZPY5VPBT6/d+oGHfth4fSwMZprZel4EPn2kFo7zD7BSCeHrFJVPw5FMU+erk6QgpCYvffWPESW6riq6tqMfOvI8k8yMvIgoOYOxZLmmLHNWq3XkMPv9js9/7Qes12vBtFivqJfLCE6sM4E2CKVhEu5CeiIhWR86K+J7V/UClRU8f/EqjknmRfSdYRioFzVVXcu4OmnqKmVyVuo5M82paWjbVoD8tJaEy75js7mgaY7e+iNo+H4+z0kWn+WctBdMCeT/13oF8lJT+o26qjfwHjtjOUMkDpQSgLbw/Xzgkwck7ou5eZyGmeS64JtKXlRrSfsNRO/1vNDjUfn43tBL5Y9kLuRgQZfyLBGVS5yzNL4V4PF45Oj1oiDe63rpRYeAm2w2G3GPXF2yubhitZKayrquKYoctKZXCrTg2o7IkFb6JumMoqjiGOt6xdXNMz6+f8sGadwqRcaZj1BI5XjXtrSNYJ9prSXbohbxfzqdUL4KPaAbZVnG6dSwWl3y7t2bZO2IndqCeiMAeTZahUoprEl0pYTzpFXlT3kcxs+DZTqSQPzOeR9bNsLW5yQcZE6xstBIOpmz0jwUJv4ylMch9bLbJlwxWJ9hYeUaF38CgcV7Od9VNtEDskwqyPO8RFmHGwRV0RjDYlFTVxXKCcla71Iwg2C9VtWCqlzgHBG//9QELC9fwFuMXeaur6+Fs1UVq81FrCYvcmlaX+TSogetsUrgMTWC25rlBQEfTSlBD8p0QVEtIlz6crn0fcSFk+VFgfE6nrNORH5dg5Kw2cXVJaYf2O+k3eJ2uyXPc/YPD1z7lkIxhjmz8FPfpdc6HonJqHdlCu3xXefEFfSzVDcLKk9IOp2cHwjOSphNa00eAGoVko+Uzah3EmaC2HRTCGn0M9sZ55rL5jjAcJ+gt87keJZpD1ziLS7rdQqd49Tgi2F9gwrlGAbn/VkCP3A6nVgul9w8f+Yry1uKU0me5RwOB8q8IisylptNxD8zFnRWsNlcsllfslpvyFVGmRViyuuMDCjz53l7wQAAIABJREFUQtB+nOB8WW1RWUbhe4fj58ci3PfofVl5VlAuar59/YbNZoN1jmW1oD2esE5qLkPfdGsceVWiy1Kq4ruO9cUldjDU1YLtfkdeFrHxxHyO07l3sR3J4yNVzK0JEuX8eROnb8IQzqpZSvhoyhnzwGVCC+GQPzYONHU1jNQclfiZ8zQd/NTJmrhE1LgF0oE653zrGnF5hE5zee4tmKwkd6GeUDGiCSnv1ReCK8uS0+kUA+fPX7ygPTWoTMeMja43bC4v0VrzzetfioGSZdS19EJaLBYUZSkdSLSEsIwZpH+UcmIVKk2eFRD7hWsvtyzN8ch+v4/Pe//+jViaKIq6Fl2ulWKSFLu/qhcs12ssju2wZX15QaYc94c70GJhCiRpEUW9c85XlT1WfZ7St6OijuMRBJRSE+711D2fcsIaZyff57gRFVkx6kfzgSnGbIPg01HRIJi+4LmXii9HqslNj+CykEEHApRcKqUyikIQpofBsqxFDFk7oJRD5QoGwZYIxbyC2y/+MV3kXK+W0i3DKUrvcmjbFuVguV6R5wncQFVRFYXgS1iLynMPGep7USpJQsxVJn2RUJKRiuF0OPDhwzse7m65vrmk8QmItx/vpa6Smq4bMB53Lcu075xiY/isWtQ8f/4CZy0fPn7k1ec/oDmeyPMtt7cf6bqOYRir+IMMC3pRIPhzHC79W5DLg/ya6tOTyM4jo+4x25tbmBHVZ/JARawhHIlLRasuqO7yIDxBTCuJ5w8NfPhcOUI6qJGww3fhRTxKo8DGAErMf2NZrzeAELu1BquUoP8MI1ic1pq8KLh5tqFtpJtJWQrM+nq95uc//5k4MwdDXVcitvJM+gkpyDOpotJZRuaEyzonuK9RT8GJnuYMtu85Hk98eP+eTDnubu99H6aSly9fev3Pcv9wJx3lnCMri8Rnt6Jc1BR1BUh2yW9dXrHb7rh8vuJweJCZtM6rrh4O1LfGdgGWKi6Fm+hT6dwrJRvDBoh8v7zKK1xzhhOiC7Em/ZxK5C3MlBLy1H8fW/oyFvY6Z6V6RQVROorO0dIYfS9nOVjQ1ZRGYRgLgRUuKn1T5+94r8cW7DAMLBYeCHggpr9Y6wTa0/uP2rZjtVqz2VyKTja03Nw8RyklZXDDwGq1xNiOspQGErkvb9NKyuek9jPzSX4+8yRTqDyj8NCczovIIO6lS55Yg0PbMpiBqq6iCNoftr5MTlKVClt4rIyCopTwWblYUJU1CrEyF4sVg7Uc90cPU3DyhC4IR87ZEAAIU+43+MhRUm6kUWCsz4iN7k7PE9RkvsNmVSpZ/4Ro09y/tBW2iuLSk905mY0SBD6HpPpor4elOltKHOdcIZPzwps8ITSjv8yOkYTUAnVO+TaF4gWv6xqXZThjqarST/RA73H5Q3PV3XbrcSuqGOsUF8GBsqxEr7GK6+sb4Xw+rSZ0oRE4euXLy4iuDxWrxIXItV8YB2JFOst2u+X6+tqLMem00pxO4MNFfouitNRjtn1PYSwlkqCIEzVltVrG/qHKeagp5yMReFcFespd3WPHa0qFYRXiPo9LMg0pzUXndF29RGJa7ebCekPM4HlSFwsma/g5M9on3mJOuMFtkTGl56nzMFig6fVKST78crmMVlXwfwWOkfrbAjaFlNf1HhLUxy8TAlVKMVjLan3J5vLKg6eU/p00QTyPKeSjdR2yhjMtYTPj+5y3fUffdRwPR+7u7iOcZ5aJx785nQTJZzBSmheIwoY+7tJnqe86mqalHwYRn0VO25xEF9aawVfAj2G68f3nBDb3f0UHbqYEdUiNgDCfwkVJ12jyuwdPDJAWkeNpD5sVrUcXFjwRVXGUMtUxOS15gFJTCv7VhDYPrsakk1FP0KNV69wIZeScpFsHQsmyjDIv6LqWLCsk7WfQOCuhlqIssdYx9ANDPrBab1j4Rl6996NdXIjP7dWrz6gqARkehkHgCooc64SDZglHj1htns05JyLc9J1HCDrgzEBZL7i5usRaK4DDnnv1fU/fCWhf7v10fS/PVM0p4vxLM9icIstoTtLyxxnDqW1Y5LlYurR+A+iR65OypunaZJmOXCp4MZUWnXYknKlUCrcI6zy3Ji3SMzTUe4y+Mq/4Kzw8AL4IdEYSQQ1TSMcvZTX4pqg6dV+4x6QWZPnsw/MkmRC1vIRwEeFGMtjj8YDWGZeXl2PQOMvFI48QoxRbSDfcoiglqK+170ISxOSJi6tL8iIXzAmdcXlxjXHSEXcYBvqupapr6sWSIpdAtgvjRGFdQEk3dE3LyfdB+v77N+y29yyWNb0v22vblvv7e5/BmzFYS9O19G1LboI3Pydv2iiOcaGOoaRrWprmRNe2klqEo+9ajBmt+nNKfdCV030exLNW05rKiGjrRq4NoiYEl0RM0grcypfjOZy05skyjDUxTTs8dlZIMqXUeDOXdtcJIjWR9yqw4PN+mblpO6U5L4aYsXNSbjbqZql7QmuNygVSyjkwRvxkg8nJs4xlvaCuKgY7UFYFh+MJa6SnZd91DMNAURR89tln4qsyA8fdFlNVNFpTVRVFWXrdR4g+05nMhzc2zGDY77Yc9zuatuX+/pb2JIS6qGvatvGcVwyLvhVcscEa+mEQ562xDF2HKXJsprF5gaksxlppj22ddAg+Hr2eJvFbrRXDzMKb6ExKalgneo6b/APMQoLB6PMKqQ26KL4uIL2XaP++nXUoNgnPG5lJPlnURHk7G7sKBBb/8j6ZWCowdUmcUxTHjo4TSpvoFlFsh+u1Bie9ubVSNE1D5qHU21NDUYRuuZIZUngHap7nLFaSbaq0ou8Ni8UiwpQvlkuf0p2R5wX73Y7Oh4CsNTx7/oLr568kZ0yNprkDqUxqLUPXcn93R9OcOB6lOMQO0kJHedSgtm3JskyI8XDAOsNgZMdL+EnT9y1dN7aQxjnv7pAGEf3QowgNvBTGGrKsGOdMpYW8yttZdmJvRcnkVaOg+Y8MIRGRCk8oclJqTYb11cEsJYhH76hKfWskAXJhmdLRQk9JYLyxynz5uwKnUWrw33oUwVmAbO5jic/RiiwdnH/haeUynigF91WmMyPPcuEuHhag8EB3YiGK78gOYBgoi4Le9AI1VVRYqwhhkrZtWdYluZPelA7D7e1HmVCluLhY45xYiWVf43VpjJXuwYICJG0Jj8eDIFS3HW3Txo1iuo7jYYsxA6dTw8PDPYfDDq0zBp+5myl88yuH8cAtxhhMP9B1LfUik9BT29A3TSzoLYqCQXmDRo2iLDzb+6N8LBUBx2PUvUJE0PrMjMC9wvIF6ZP5TSXK/ZQeLGMFWvJF4jtQI5HNyAnOEJko/BJOEU4TKlymTevPhZPmv0flcHJ+2CVyRtgF2jsZwzlSOV5EKy/PC5QX17I4HZ1pWS6X5IVMUVUt6LoG5wx9P7BaLSiKjLY5YbtBHKk2KNagdc7zFzdYaySdRiuqaikcxPWYQZqc9l1H355om5amldbNpu942G/puxbtHB9v31EUJU3TYEwffWh5XtD1Q/Aa4nyYqG0bdJ6j0BRFjnOGrrPs9wdpcpbnDMZQLxaYZojoR37WHhFL1EXma+OIoMjJxMd5njvk01qOqE6hJtelngEcETXokU6WFiKknweymHwe6TE8+HHxZ6rfhahC9P6nFuzsBcc0msSBqMUyK+vKZ2Es/LkiFoZh4Hg4kOca56SUTKHo+5ZTs+X27o7nz19we/eRqytpqtq1Ldv9lt3unsViyXK5pioXHI8n6vpItX2g71oWyw5rxL1gnaVrWo6HPcfT0ddP7mjbA19//XP2+61EHbDsDlvKYkGRFxwPB8DRtkcGU5CpgsFI92DnJH2pbzuUOuGsIysynJbwldKK9WbNYb/HOUdRlZitB/KzAiwzEpishYs72YtOiGtkgkhUkrsf0rGt8TpyoCq/9lrpGEmIDe9tYATTDI40RAlJtVIUZ1ZjNeiUSPzgXOKCcASFPdQhz8zelESTHRSsjriJVBoHnVwUrd5QpZTirIYWMFYPOGdojgfJO9M6wkc5Z6nrBUPf8O03X6OznNevv6KqJPFP5xrbdrx9+8ZzuBXmqsOuLtjvdxSlpPjUdc397R3ff/+Gqq64uBC3hOl77u8+cDwcuL+/ZX/c8v79G7ruJK1rEIeqXThaEJdFnmPsIBhmmXDiYRCv/TBY+sFQOCdgxdbiBoPKJH37/tTSto3g2TqJ6RoTeaFHDNfR7xj2sMCfK088YT29cTi3Sj1CQFDbwrqFXqLES0b6GIZh6kWwI0dzzo0dScZD/EEq4TbS4Iq4S7R3so1iTuR+GrkPx5yInwRnSYjTWhtjg4ELBujN1WoTuVjww/SdQGWuqhrtu4OE3P48z/jlN1/z7u171qs1h9MRlUnLHJRFO9jfP2CBrjtKxGAYMEo4Tp6XPtPB8u23v6QsS549e0ldS3Ow7f0tD/d3nE4H7h7uOOz29EPDMPTRUYxzHE8Hqqqi78Xo2O124KBciPO378VAKMvSB80zhr7HlUJ4+8Oe4/EoE5Up8lISMNEZDJKxq9Bjy6C5GhIozo1xSqle8MZYYmGm6zH6yZJqdESPdHZMuz7XYjz8/mSOvw3mrIeknLs1VLRm0nJ4dXagE0JidFeMg5kSZXp2+Lvve1bLDafTifXFhqZpKPOCssjoMYClaVv0oDHWsLnYcH1zydt33/H9999y/eyG3XbHZrPGOoNSjv12i8XQ9SIKu1646ul0ZLd/4OOH91hrOJ4O4CRjt+97vnv9NZvNpaA62oGH+zu6rqNrjlJNPvR0doiTH5TtsixjaKiqKu+E7ajyGrxDV5zAhedUjlNzoh0G30JRCppRmmpRe67uCcovuohJO5lPHVwSPgN1DAPJEbwcc33aU2dUoazPssVzRq3URLJEqzRIIGfI0NMucXLypM5lItfD5wK2m4MyOBfUu9F8TY9JBCHcM7g+xOyJ7FdcB36CnFQVye/yAm3fcXWxIDQc1Sj22x3WDgymR1pWi2iRziM9b99+z3q9Ybu7Y7/fcTzsyPKMuipomoN/d7EujYNTc2AYOpr26EFQPMYrTmCwnEVhGdqj73oyeMeoQdkEuz4Uz/gkg2EYIoGlDRm6vkGhqCspLGm7jrquxc0ClFWBVRJGW9RLmuaE1orVciUWvg5QEsR1JOFgo6Iv5rFkbPh5tWJtat+2O9KVcyLyAtPzPkrtjYigOkke2tPxHgdYNROXMYnN+/e1ClEAX4AQsIW8LFcJ1aYWzORBiesiOurGLO5EJ9CR0MYvfXhNKd9yz7E/HLnYaGkkUZaUdUHXWDBSwaS05uLyksViwffvBLZ86ASKabO5wjnY7ba0zX4U/fFQo2thGHBNurPDFnMe+txKMa0b431hUsOGxHlHpd/AwzAImMrge2nmgtajnODXSi6cJCQWuRS3OAVlLbUGImIztMpB+bRxI6qNsUOsaEqXYBRjEue0niOLH1JgHlLEzNATJDDD0F4w81VPwpntyCH9mwf1yvqXd85RaFEVzhb3aseMOkcLLgg85wzYDKVDhcr0OBtc9btMnMl+wR6dH1VGFEEHlDGYvqd3jtPBSeFtUdB4fH4BJcl58fwZL1684OPte95+/x1de8IaEU99d+J4OrBeX7Dfd964GBXY8GhjTLTGIGXOKu4x00+La5RSArfOqDiP3oFRd1Fa0/dNvCZwrcNxx3KxIq8qDs0J5xQXaoNTiv5hB8ButxMu/UyyOpTrcVZqHrRKE+cf85a5L8skhAKp159oMYbrQhc7zSgenQvtFGVe7JxiPOFansAni/mW6UyRQBQQiGIUhWPFt4urEl4s6CbRYAhca7zbuWHgCLDuEmBeLBR5pkT/6Toe7rYoJQXEVVVxc3PDer3hcNzz+vU3viOckUZXStH51J++v0MpF+GmwljHkIqbTJoeMZPiqK1yk/kI40Ux1kF47hWgnowxwrm887PrOyqko/Cp63BtxmUlqd+mN5Jl4hzWuyMWiwVa51xdXvCwvZsSzkyvTY2t+VoIs5FVTgt8nRsXPFwf3slZN0mrJuF+lsdAxlGHd4+/e7zQqTHAzM8VFYHRCkxR18JOUXh2m9zp3G5Lj7SeQHBiM/bHhs6nuGgVsCpyikzz/PlzFqslRVXy+rvv0JnUVpqhQynHMHTgDOvNkizkh50R7fEzR0QfjM5NraZczf8TutYG8Yl1qLAgRS4N5xHYzOA5D07Uru+wTvo9te2Rh4d7uqFHlZIa1HatwBEYy6pecHmxZrPZsNvvGUKjW2/yWWP9z8iV03dyzid3hjI5CxmZdPe2/kVgkvOvGN1OwdqPnNk7ZPXsObJ+IwPS52KUjw8Xf6Ll4n0x1uNNSNB41FvCAOOOcONncRCPnWPTBbcW54w008pyqYHMq4ly23U9KhNguqurK/b7LXkmBRf397e+PmBsNaMcUXeZlIAlOqXym2QcO37nhk3nCO6coEcqpSS5EfFFheul4FXjjOBcODumC4naIWAxINZn15+4f7gFrCRlOlHOV+sF6/Way6srrLW0XTPOl0rXZZQyzuIJTsUf5dHsJOFZxXLAOOfnLMzkkHUTdd8GI+5XHDpeOLtRuNn4XQC8m+JlRBC85B7nOMTZ41cQeHh2UZSJjqQEn8v7wayz6Dxjvd7Qtid2u3sOh3vevPkGZoq5UpLmEyp8Ju/uHj/3iVE9+i3kujkHBjA6EZmAweEyja4KBqSrnvHNHAJnCFZjnmu6/sT7D2/QueLZs2cUi4pyuWBxsWaxWHB/f8/pdPB6pPMWb+A00znXITnPje8l8+E3gE7dUlMQlrnP89w6nuNe8/N0+uHcnTElUplIpZikdkCqqofPpkrxtDD0Uzslfdnxvjqk3RSF37E+CdBjiD17dsN6veT1t7/ku+9ec3d7K7j6g+8WMgwM1sS6gLOTlIzJJsUx+J/5xClvWYnPbXJHjP87Q4FOuKhX9FOuGQqScU70NZUJfGfvePP2Daf2xOXFJRebDZnS7Pd7bu9u6fvOtzSUZRQ9yUsclY4mmGp4w8SXF44yBZ1nolfpqd4dUsnjXJxZw/lannPm5pOTgt7uhzfZs1HGzxfBL5ArgMFjLWTxBYPITAeUfiafh6kIdsN4HghUkzHS/AEUTTOQ59Lq78WLV7x48ZJvv/2Gd+/eYMxocaXvVpdVTNd+lEgZxuTHpXRi7s84dHBNjAGRaeImDpSR6wYc+CYXDgnuD2YAX9GjtEKFOlLjQFm6U0NZV6jMcTzs+OoXP2Vzfc1nr37AcrEmz3LJ7GjbOI5QPR/aM4eEgaByhPx750SnDrHFsKGDby8YAUqpMe/fGwMpQ5nYg7O1RIWsmnHNn0S/9pd783i6NSTa7+LCqDDvEwTa4KSdWjvnFjgxSJMjBGwH2vbIarVhGHy1UpaxXqz4/Nc+58Xz5xyOW15/+zXGdhLATvCzwq4L3Xmf2nmPiOmTIjMd5uw8NQLxBnz8kHbdNI3sbC0GwEioI5fQmcRks6JAKuRb7u/e83B3z3p9wc3Nc5qTOKDHAXt9y6+t85p66qOM6tvoZo8iNFqCSmGHscAl3n68eHKPcQrG9XWJGyS8UwwrqUCBo96OKLVjUnZw4IVqbe0VQOVz3zOVCbdxEhZB20RZHh86Z7XnxehoLGgUz66usUp7/LCSly9ecX1zye7hjp/+9M9p2iN2MNgZF5tYOVo/Qo5Ox4Wa+oHi50yJMA2djdkKYS5kGQYcuVMUWcaAnbRmthA97LJBBTlJa/E3SYmbIysrHBpjeqzpMNtOIKb6o+dSwfzVPp1aQnzBIInz7adfnO0eXCWxPqPbxhMYQGhqrj0IzrljHkIEJj7WQEp5aAjlPPmlojLlMKluL/9qr6N5i8ULfUn/CBM6srewiIqpbyYw4GARzRmDtVLZ07cDl8+fsVzWlHlNvVjw7etvef36LzkethJMDpZgckRfzoy7hX8nnPUJO2QSmmFqEEm4iWT3y+dGQeYcWZ4lld7MGumGe4zzgJKGrcoKBKizjsVqzWAFdfFweEBlo9XrjWyi+1dJXcTEpaASdICEe8ZNrBRE+M5kUyk+SWBnK5tcQmhKKCk/nk5c+HL+cL9MjUSR3FW+tkwJIjQYcBZrBGbKKcETk401uuqiqqnSyVWM8UsIYhLCLrNgBw6nPepO050EN/bu7iMPu1uG/pSgND6ekLkYTI/UIAmOZOHACRmMdEXgEtGHl/aV8Up1ODNTYkG2fSfEPL6dF2nEs128N15fk3sEmKfmGDI4BixGqr5DKMuOye9WhVqI8d2VUpErhdlWWlwqcV7sjPMpFR3o8aJw7kRXOz+v8XPg1HbkbTdMb5hO+hPHxHUBkaOl3FBeViZOB9EQCPWMCEr/nj/fDgNNe6BpD2gPu2TsAMpghl7EN0nG5r+qPjUbw1wXUeF/bnp+fI5XiAOXSLmzvHe6UCp+HgnKz4UKYmN86iP+cTqdgFm1djSQxhT2MH+Td1PEtJw0EP4py/Dc4VIjINHnHh3RioLddo/6a3/td9z7d9+B7/wBoAKYmkoGHDhMsksmd0smSOkkVMGUnY4GqVT/BLl8ziR2Tgp1M6XJihxZJBtZdRC9+B7cqSiYT/ZcwZ1P9OS5KQOPYxxFUzw/6KR+XlJis4zvpJWPa0Yuo2JBTchscCGd3Y5pU+cWMxBZ+lkMwk+IIaxZoh5YEZc6jHc215NVDGOdcK4wx4nide5QgFEobVldvUT/0R/9kedAkm8UQiBeY/cPCH+HO0giY6pgBt+W0inRnXkRJ5wver7PHClxWGsZrPFB8BB77JOe3MGnF1b/X21Hzr3c8Wd+TpTxyeufG6/yxBmI2evkyvnFjVPqU5Ydo6XnnOf63veGwC/JfRN/5ExEhR8TfG3p3MkficEhYHeBRub3COeE93j0roE7RDr49KGUIssz/uv/5r9FdV3j/spv/joO6V4bM3oYHxQGdt7jL19GjqBtmG0/y1NO4bUNJIKQiV0wY+/nxJ3oCY9FqR5pC+vxW89NQuQqcz3s3POS1nqPRDqjjylwssn3UYebX5feXlYy3DvF309vpxMuGfTBxyhIT6sGKUGNitvj8+JcBO75ieNXbeD0yIqcP/1//inaGMvDw97HHT91ozF294j6k3PkJ2P0k83wZtXkH3+PqXh79ORgYifoNOO5oZJdPVrY9Pq5u+RTz5tfMx/nJz+b6XXhcEh1kFNMuM5kDPML45g//T5PjS2It5GLMoYxSRQdP+ZfRWDhub9K5xWJYPjw8Tb0Zs/5n/7e35PGBcE3FexsNR381Nj0u9CzdNBeVCrEBB1BctMGEzKA0aM052JPWYlaa5RT6CRxRJ6qQz+J8cMnJki+dhOdazIxyXNVYvafu96F7zzUeNRZA+efZZomdx45RphjRtGqJs9/rEvG+ROFLt7/3GHdY+swUlvgBwoPjXp+3ubqxKeel85dnuf8d//D/yjrNgyD2x/2vHr1kpurDcqNuPtxcsOYfF5TKmak6AQgA9Wf3f1Tl0XYP9mox50Z5Pw+WmuUR42xyscCtaZwAhI87x4bxzcnEjW+07miliA2Ytbn7JTUskw3iAo6mX9Fh4tl/UqldYvyb7rxglsDiI5apZQEvpWUqU3m3DlC1nuKpv6rvAL+JL87x039FP5++H4yN7PP5vOrMylmef3dW/7iL79juVyglVIs6gX/6Mf/mDzPI4TQ9GHTB04WUzLw4jnzh0s671ykjO87f4Hz8zLdRcGPEx2QZyYjXZQ50Qbl+9wx4WBndm/qFpiOL5mCZLHTBXTpZel4kudrYERzmeJUPFIZmKsuqdR5LFLDs0KbnLnxM5mfT6gKT/0uFxvINP/z3//fqBcLlPKyR2vNF198Qdv00/OTyfG3TNj2mHgorzvG0gIRyAAE/nJCaH5FRjcIj154TjRxssPETgc6cTjOOVpYnOm4plGAcJz77Nx9g0UZudjkmeO7PtWHaLqQ09oGZZ2EOhJ15ewmPKOrpeMLP3P3jMuUB/R7vPmAR5LsKc527rldZ7jfNvz2j/6NKCm0UgKAdn19zT/+P/+UrKwIzQykBxGoYAkqifAbM8SNqHyWpVYFipwAPRR3nxKnqbAHhUOjEWytwDVmI47IfCNxmshhFeJzyoA8gl2Jr11bb20mOz/VZwDxdH9CpXCM6M2oMen83CKLWZOEspwDZ3HKRE6DIjZOGPUxvKhTk7CWiGoHGnSehLvU+H3kkOPHca4DZPx0OmdqByp4qyLHns9Reo+nJIwi2QxOnPNWOerlgv/l7/8D1uvNONZQffLHf/zH/PjH/we///t/CDpjcAMOSV1On+McSWzMxgcmj37yJUczOdzrU9YdXlwBZL47riccK2nGzkbjPyrdYRhzEZmKr5C6HbqzzYcfcdfcWJ3zqQlP58Z7wcaXiCJrfDERm6HHUwhKu3HsWsX3BaK3fjo/0zjvfGxPGVBBDeCJ78M5n7JeRY+TphlKCTh/lgvMwt/5L/8rfv3X/0rcPEoplPWwe6EIdbfb87/+yT/g7/wX/zmff3ZD15xQaDJdxBePCruC0ecV9pYTX9n49sF9Sywp9Tj4RMV/KpY9aLsPgQQEv0EakeqxDF9uH6p0xiiD1Uwcneeq2s8eilGEzDrUTquyZ0cCY+nSsT1aQ+UzZp1wMT+HgUPLTDzmRIGxphJC+05vKrG2J/rfI72aR5+nx68yGs4TuWyGvKx4/+EDf/fv/vf8h3/rb1N5qNSU6wIC85jnBVdXF/xn/+l/wp/92U/44rd/h/cftuz30s9bdJuJ+hrmLnn440HFFw07yCu1cunTfhe5JnDHQCwu4S7nOao8ZpzMlP3PRcHcQIg3SM+ZjelJc16F/6mz8Fti1QqBqSc5yePFn4835RIp4YXKqLR24VPK/fz+c3fFOZ0unGeMYX888v5uyw9+4wv+5H//U/6jv/0fe2UfGL3MW1veAAAAB0lEQVRw/H+b2Lc3Rg46LgAAAABJRU5ErkJggg==\" y=\"-21.499943\"/>\n   </g>\n   <g id=\"text_1\">\n    <!-- Train Image -->\n    <defs>\n     <path d=\"M -0.296875 72.90625 \nL 61.375 72.90625 \nL 61.375 64.59375 \nL 35.5 64.59375 \nL 35.5 0 \nL 25.59375 0 \nL 25.59375 64.59375 \nL -0.296875 64.59375 \nz\n\" id=\"DejaVuSans-84\"/>\n     <path d=\"M 41.109375 46.296875 \nQ 39.59375 47.171875 37.8125 47.578125 \nQ 36.03125 48 33.890625 48 \nQ 26.265625 48 22.1875 43.046875 \nQ 18.109375 38.09375 18.109375 28.8125 \nL 18.109375 0 \nL 9.078125 0 \nL 9.078125 54.6875 \nL 18.109375 54.6875 \nL 18.109375 46.1875 \nQ 20.953125 51.171875 25.484375 53.578125 \nQ 30.03125 56 36.53125 56 \nQ 37.453125 56 38.578125 55.875 \nQ 39.703125 55.765625 41.0625 55.515625 \nz\n\" id=\"DejaVuSans-114\"/>\n     <path d=\"M 34.28125 27.484375 \nQ 23.390625 27.484375 19.1875 25 \nQ 14.984375 22.515625 14.984375 16.5 \nQ 14.984375 11.71875 18.140625 8.90625 \nQ 21.296875 6.109375 26.703125 6.109375 \nQ 34.1875 6.109375 38.703125 11.40625 \nQ 43.21875 16.703125 43.21875 25.484375 \nL 43.21875 27.484375 \nz\nM 52.203125 31.203125 \nL 52.203125 0 \nL 43.21875 0 \nL 43.21875 8.296875 \nQ 40.140625 3.328125 35.546875 0.953125 \nQ 30.953125 -1.421875 24.3125 -1.421875 \nQ 15.921875 -1.421875 10.953125 3.296875 \nQ 6 8.015625 6 15.921875 \nQ 6 25.140625 12.171875 29.828125 \nQ 18.359375 34.515625 30.609375 34.515625 \nL 43.21875 34.515625 \nL 43.21875 35.40625 \nQ 43.21875 41.609375 39.140625 45 \nQ 35.0625 48.390625 27.6875 48.390625 \nQ 23 48.390625 18.546875 47.265625 \nQ 14.109375 46.140625 10.015625 43.890625 \nL 10.015625 52.203125 \nQ 14.9375 54.109375 19.578125 55.046875 \nQ 24.21875 56 28.609375 56 \nQ 40.484375 56 46.34375 49.84375 \nQ 52.203125 43.703125 52.203125 31.203125 \nz\n\" id=\"DejaVuSans-97\"/>\n     <path d=\"M 9.421875 54.6875 \nL 18.40625 54.6875 \nL 18.40625 0 \nL 9.421875 0 \nz\nM 9.421875 75.984375 \nL 18.40625 75.984375 \nL 18.40625 64.59375 \nL 9.421875 64.59375 \nz\n\" id=\"DejaVuSans-105\"/>\n     <path d=\"M 54.890625 33.015625 \nL 54.890625 0 \nL 45.90625 0 \nL 45.90625 32.71875 \nQ 45.90625 40.484375 42.875 44.328125 \nQ 39.84375 48.1875 33.796875 48.1875 \nQ 26.515625 48.1875 22.3125 43.546875 \nQ 18.109375 38.921875 18.109375 30.90625 \nL 18.109375 0 \nL 9.078125 0 \nL 9.078125 54.6875 \nL 18.109375 54.6875 \nL 18.109375 46.1875 \nQ 21.34375 51.125 25.703125 53.5625 \nQ 30.078125 56 35.796875 56 \nQ 45.21875 56 50.046875 50.171875 \nQ 54.890625 44.34375 54.890625 33.015625 \nz\n\" id=\"DejaVuSans-110\"/>\n     <path id=\"DejaVuSans-32\"/>\n     <path d=\"M 9.8125 72.90625 \nL 19.671875 72.90625 \nL 19.671875 0 \nL 9.8125 0 \nz\n\" id=\"DejaVuSans-73\"/>\n     <path d=\"M 52 44.1875 \nQ 55.375 50.25 60.0625 53.125 \nQ 64.75 56 71.09375 56 \nQ 79.640625 56 84.28125 50.015625 \nQ 88.921875 44.046875 88.921875 33.015625 \nL 88.921875 0 \nL 79.890625 0 \nL 79.890625 32.71875 \nQ 79.890625 40.578125 77.09375 44.375 \nQ 74.3125 48.1875 68.609375 48.1875 \nQ 61.625 48.1875 57.5625 43.546875 \nQ 53.515625 38.921875 53.515625 30.90625 \nL 53.515625 0 \nL 44.484375 0 \nL 44.484375 32.71875 \nQ 44.484375 40.625 41.703125 44.40625 \nQ 38.921875 48.1875 33.109375 48.1875 \nQ 26.21875 48.1875 22.15625 43.53125 \nQ 18.109375 38.875 18.109375 30.90625 \nL 18.109375 0 \nL 9.078125 0 \nL 9.078125 54.6875 \nL 18.109375 54.6875 \nL 18.109375 46.1875 \nQ 21.1875 51.21875 25.484375 53.609375 \nQ 29.78125 56 35.6875 56 \nQ 41.65625 56 45.828125 52.96875 \nQ 50 49.953125 52 44.1875 \nz\n\" id=\"DejaVuSans-109\"/>\n     <path d=\"M 45.40625 27.984375 \nQ 45.40625 37.75 41.375 43.109375 \nQ 37.359375 48.484375 30.078125 48.484375 \nQ 22.859375 48.484375 18.828125 43.109375 \nQ 14.796875 37.75 14.796875 27.984375 \nQ 14.796875 18.265625 18.828125 12.890625 \nQ 22.859375 7.515625 30.078125 7.515625 \nQ 37.359375 7.515625 41.375 12.890625 \nQ 45.40625 18.265625 45.40625 27.984375 \nz\nM 54.390625 6.78125 \nQ 54.390625 -7.171875 48.1875 -13.984375 \nQ 42 -20.796875 29.203125 -20.796875 \nQ 24.46875 -20.796875 20.265625 -20.09375 \nQ 16.0625 -19.390625 12.109375 -17.921875 \nL 12.109375 -9.1875 \nQ 16.0625 -11.328125 19.921875 -12.34375 \nQ 23.78125 -13.375 27.78125 -13.375 \nQ 36.625 -13.375 41.015625 -8.765625 \nQ 45.40625 -4.15625 45.40625 5.171875 \nL 45.40625 9.625 \nQ 42.625 4.78125 38.28125 2.390625 \nQ 33.9375 0 27.875 0 \nQ 17.828125 0 11.671875 7.65625 \nQ 5.515625 15.328125 5.515625 27.984375 \nQ 5.515625 40.671875 11.671875 48.328125 \nQ 17.828125 56 27.875 56 \nQ 33.9375 56 38.28125 53.609375 \nQ 42.625 51.21875 45.40625 46.390625 \nL 45.40625 54.6875 \nL 54.390625 54.6875 \nz\n\" id=\"DejaVuSans-103\"/>\n     <path d=\"M 56.203125 29.59375 \nL 56.203125 25.203125 \nL 14.890625 25.203125 \nQ 15.484375 15.921875 20.484375 11.0625 \nQ 25.484375 6.203125 34.421875 6.203125 \nQ 39.59375 6.203125 44.453125 7.46875 \nQ 49.3125 8.734375 54.109375 11.28125 \nL 54.109375 2.78125 \nQ 49.265625 0.734375 44.1875 -0.34375 \nQ 39.109375 -1.421875 33.890625 -1.421875 \nQ 20.796875 -1.421875 13.15625 6.1875 \nQ 5.515625 13.8125 5.515625 26.8125 \nQ 5.515625 40.234375 12.765625 48.109375 \nQ 20.015625 56 32.328125 56 \nQ 43.359375 56 49.78125 48.890625 \nQ 56.203125 41.796875 56.203125 29.59375 \nz\nM 47.21875 32.234375 \nQ 47.125 39.59375 43.09375 43.984375 \nQ 39.0625 48.390625 32.421875 48.390625 \nQ 24.90625 48.390625 20.390625 44.140625 \nQ 15.875 39.890625 15.1875 32.171875 \nz\n\" id=\"DejaVuSans-101\"/>\n    </defs>\n    <g transform=\"translate(48.199347 16.318125)scale(0.12 -0.12)\">\n     <use xlink:href=\"#DejaVuSans-84\"/>\n     <use x=\"46.333984\" xlink:href=\"#DejaVuSans-114\"/>\n     <use x=\"87.447266\" xlink:href=\"#DejaVuSans-97\"/>\n     <use x=\"148.726562\" xlink:href=\"#DejaVuSans-105\"/>\n     <use x=\"176.509766\" xlink:href=\"#DejaVuSans-110\"/>\n     <use x=\"239.888672\" xlink:href=\"#DejaVuSans-32\"/>\n     <use x=\"271.675781\" xlink:href=\"#DejaVuSans-73\"/>\n     <use x=\"301.167969\" xlink:href=\"#DejaVuSans-109\"/>\n     <use x=\"398.580078\" xlink:href=\"#DejaVuSans-97\"/>\n     <use x=\"459.859375\" xlink:href=\"#DejaVuSans-103\"/>\n     <use x=\"523.335938\" xlink:href=\"#DejaVuSans-101\"/>\n    </g>\n   </g>\n  </g>\n  <g id=\"axes_2\">\n   <g clip-path=\"url(#pf02e2d733d)\">\n    <image height=\"153\" id=\"imageb081ed1ee7\" transform=\"scale(1 -1)translate(0 -153)\" width=\"153\" x=\"189.818182\" xlink:href=\"data:image/png;base64,\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\" y=\"-21.499943\"/>\n   </g>\n   <g id=\"text_2\">\n    <!-- Label -->\n    <defs>\n     <path d=\"M 9.8125 72.90625 \nL 19.671875 72.90625 \nL 19.671875 8.296875 \nL 55.171875 8.296875 \nL 55.171875 0 \nL 9.8125 0 \nz\n\" id=\"DejaVuSans-76\"/>\n     <path d=\"M 48.6875 27.296875 \nQ 48.6875 37.203125 44.609375 42.84375 \nQ 40.53125 48.484375 33.40625 48.484375 \nQ 26.265625 48.484375 22.1875 42.84375 \nQ 18.109375 37.203125 18.109375 27.296875 \nQ 18.109375 17.390625 22.1875 11.75 \nQ 26.265625 6.109375 33.40625 6.109375 \nQ 40.53125 6.109375 44.609375 11.75 \nQ 48.6875 17.390625 48.6875 27.296875 \nz\nM 18.109375 46.390625 \nQ 20.953125 51.265625 25.265625 53.625 \nQ 29.59375 56 35.59375 56 \nQ 45.5625 56 51.78125 48.09375 \nQ 58.015625 40.1875 58.015625 27.296875 \nQ 58.015625 14.40625 51.78125 6.484375 \nQ 45.5625 -1.421875 35.59375 -1.421875 \nQ 29.59375 -1.421875 25.265625 0.953125 \nQ 20.953125 3.328125 18.109375 8.203125 \nL 18.109375 0 \nL 9.078125 0 \nL 9.078125 75.984375 \nL 18.109375 75.984375 \nz\n\" id=\"DejaVuSans-98\"/>\n     <path d=\"M 9.421875 75.984375 \nL 18.40625 75.984375 \nL 18.40625 0 \nL 9.421875 0 \nz\n\" id=\"DejaVuSans-108\"/>\n    </defs>\n    <g transform=\"translate(249.721278 16.318125)scale(0.12 -0.12)\">\n     <use xlink:href=\"#DejaVuSans-76\"/>\n     <use x=\"55.712891\" xlink:href=\"#DejaVuSans-97\"/>\n     <use x=\"116.992188\" xlink:href=\"#DejaVuSans-98\"/>\n     <use x=\"180.46875\" xlink:href=\"#DejaVuSans-101\"/>\n     <use x=\"241.992188\" xlink:href=\"#DejaVuSans-108\"/>\n    </g>\n   </g>\n  </g>\n </g>\n <defs>\n  <clipPath id=\"p58ad9a7e6d\">\n   <rect height=\"152.181818\" width=\"152.181818\" x=\"7.2\" y=\"22.318125\"/>\n  </clipPath>\n  <clipPath id=\"pf02e2d733d\">\n   <rect height=\"152.181818\" width=\"152.181818\" x=\"189.818182\" y=\"22.318125\"/>\n  </clipPath>\n </defs>\n</svg>\n",
-      "image/png": "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\n"
+      "image/png": "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\n",
+      "text/plain": [
+       "<Figure size 432x288 with 2 Axes>"
+      ]
      },
      "metadata": {
       "needs_background": "light"
-     }
+     },
+     "output_type": "display_data"
     }
    ],
    "source": [
@@ -446,7 +464,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 6,
+   "execution_count": 25,
    "metadata": {
     "colab": {},
     "colab_type": "code",
@@ -509,7 +527,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 7,
+   "execution_count": 26,
    "metadata": {
     "colab": {},
     "colab_type": "code",
@@ -569,7 +587,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 8,
+   "execution_count": 27,
    "metadata": {
     "colab": {},
     "colab_type": "code",
@@ -591,7 +609,7 @@
     "                                                           kernel_size=3, \n",
     "                                                           padding='same')\n",
     "        self.bn = paddle.nn.BatchNorm2d(out_channels)\n",
-    "        self.upsample = paddle.nn.UpSample(scale_factor=2.0)\n",
+    "        self.upsample = paddle.nn.Upsample(scale_factor=2.0)\n",
     "        self.residual_conv = paddle.nn.Conv2d(in_channels, \n",
     "                                              out_channels, \n",
     "                                              kernel_size=1, \n",
@@ -630,7 +648,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 9,
+   "execution_count": 30,
    "metadata": {
     "colab": {},
     "colab_type": "code",
@@ -638,9 +656,9 @@
    },
    "outputs": [],
    "source": [
-    "class PetModel(paddle.nn.Layer):\n",
+    "class PetNet(paddle.nn.Layer):\n",
     "    def __init__(self, num_classes):\n",
-    "        super(PetModel, self).__init__()\n",
+    "        super(PetNet, self).__init__()\n",
     "\n",
     "        self.conv_1 = paddle.nn.Conv2d(3, 32, \n",
     "                                       kernel_size=3,\n",
@@ -706,7 +724,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 11,
+   "execution_count": 31,
    "metadata": {
     "colab": {
      "base_uri": "https://localhost:8080/",
@@ -719,17 +737,56 @@
    },
    "outputs": [
     {
-     "output_type": "stream",
      "name": "stdout",
-     "text": "--------------------------------------------------------------------------------\n   Layer (type)          Input Shape         Output Shape         Param #\n================================================================================\n      Conv2d-22    [-1, 3, 160, 160]     [-1, 32, 80, 80]             896\n  BatchNorm2d-9     [-1, 32, 80, 80]     [-1, 32, 80, 80]              64\n         ReLU-9     [-1, 32, 80, 80]     [-1, 32, 80, 80]               0\n        ReLU-12    [-1, 256, 20, 20]    [-1, 256, 20, 20]               0\n      Conv2d-33    [-1, 128, 20, 20]    [-1, 128, 20, 20]           1,152\n      Conv2d-34    [-1, 128, 20, 20]    [-1, 256, 20, 20]          33,024\nSeparableConv2d-11    [-1, 128, 20, 20]    [-1, 256, 20, 20]               0\n BatchNorm2d-12    [-1, 256, 20, 20]    [-1, 256, 20, 20]             512\n      Conv2d-35    [-1, 256, 20, 20]    [-1, 256, 20, 20]           2,304\n      Conv2d-36    [-1, 256, 20, 20]    [-1, 256, 20, 20]          65,792\nSeparableConv2d-12    [-1, 256, 20, 20]    [-1, 256, 20, 20]               0\n    MaxPool2d-6    [-1, 256, 20, 20]    [-1, 256, 10, 10]               0\n      Conv2d-37    [-1, 128, 20, 20]    [-1, 256, 10, 10]          33,024\n      Encoder-6    [-1, 128, 20, 20]    [-1, 256, 10, 10]               0\n        ReLU-16     [-1, 32, 80, 80]     [-1, 32, 80, 80]               0\nConvTranspose2d-15     [-1, 64, 80, 80]     [-1, 32, 80, 80]          18,464\n BatchNorm2d-16     [-1, 32, 80, 80]     [-1, 32, 80, 80]              64\nConvTranspose2d-16     [-1, 32, 80, 80]     [-1, 32, 80, 80]           9,248\n     UpSample-8     [-1, 64, 80, 80]   [-1, 64, 160, 160]               0\n      Conv2d-41   [-1, 64, 160, 160]   [-1, 32, 160, 160]           2,080\n      Decoder-8     [-1, 64, 80, 80]   [-1, 32, 160, 160]               0\n      Conv2d-42   [-1, 32, 160, 160]    [-1, 4, 160, 160]           1,156\n================================================================================\nTotal params: 167,780\nTrainable params: 167,780\nNon-trainable params: 0\n--------------------------------------------------------------------------------\nInput size (MB): 0.29\nForward/backward pass size (MB): 43.16\nParams size (MB): 0.64\nEstimated Total Size (MB): 44.10\n--------------------------------------------------------------------------------\n\n"
+     "output_type": "stream",
+     "text": [
+      "--------------------------------------------------------------------------------\n",
+      "   Layer (type)          Input Shape         Output Shape         Param #\n",
+      "================================================================================\n",
+      "      Conv2d-38    [-1, 3, 160, 160]     [-1, 32, 80, 80]             896\n",
+      " BatchNorm2d-14     [-1, 32, 80, 80]     [-1, 32, 80, 80]             128\n",
+      "        ReLU-14     [-1, 32, 80, 80]     [-1, 32, 80, 80]               0\n",
+      "        ReLU-17    [-1, 256, 20, 20]    [-1, 256, 20, 20]               0\n",
+      "      Conv2d-49    [-1, 128, 20, 20]    [-1, 128, 20, 20]           1,152\n",
+      "      Conv2d-50    [-1, 128, 20, 20]    [-1, 256, 20, 20]          33,024\n",
+      "SeparableConv2d-17    [-1, 128, 20, 20]    [-1, 256, 20, 20]               0\n",
+      " BatchNorm2d-17    [-1, 256, 20, 20]    [-1, 256, 20, 20]           1,024\n",
+      "      Conv2d-51    [-1, 256, 20, 20]    [-1, 256, 20, 20]           2,304\n",
+      "      Conv2d-52    [-1, 256, 20, 20]    [-1, 256, 20, 20]          65,792\n",
+      "SeparableConv2d-18    [-1, 256, 20, 20]    [-1, 256, 20, 20]               0\n",
+      "    MaxPool2d-9    [-1, 256, 20, 20]    [-1, 256, 10, 10]               0\n",
+      "      Conv2d-53    [-1, 128, 20, 20]    [-1, 256, 10, 10]          33,024\n",
+      "      Encoder-9    [-1, 128, 20, 20]    [-1, 256, 10, 10]               0\n",
+      "        ReLU-21     [-1, 32, 80, 80]     [-1, 32, 80, 80]               0\n",
+      "ConvTranspose2d-17     [-1, 64, 80, 80]     [-1, 32, 80, 80]          18,464\n",
+      " BatchNorm2d-21     [-1, 32, 80, 80]     [-1, 32, 80, 80]             128\n",
+      "ConvTranspose2d-18     [-1, 32, 80, 80]     [-1, 32, 80, 80]           9,248\n",
+      "     Upsample-8     [-1, 64, 80, 80]   [-1, 64, 160, 160]               0\n",
+      "      Conv2d-57   [-1, 64, 160, 160]   [-1, 32, 160, 160]           2,080\n",
+      "      Decoder-9     [-1, 64, 80, 80]   [-1, 32, 160, 160]               0\n",
+      "      Conv2d-58   [-1, 32, 160, 160]    [-1, 4, 160, 160]           1,156\n",
+      "================================================================================\n",
+      "Total params: 168,420\n",
+      "Trainable params: 167,140\n",
+      "Non-trainable params: 1,280\n",
+      "--------------------------------------------------------------------------------\n",
+      "Input size (MB): 0.29\n",
+      "Forward/backward pass size (MB): 43.16\n",
+      "Params size (MB): 0.64\n",
+      "Estimated Total Size (MB): 44.10\n",
+      "--------------------------------------------------------------------------------\n",
+      "\n"
+     ]
     },
     {
-     "output_type": "execute_result",
      "data": {
-      "text/plain": "{'total_params': 167780, 'trainable_params': 167780}"
+      "text/plain": [
+       "{'total_params': 168420, 'trainable_params': 167140}"
+      ]
      },
+     "execution_count": 31,
      "metadata": {},
-     "execution_count": 11
+     "output_type": "execute_result"
     }
    ],
    "source": [
@@ -737,7 +794,7 @@
     "\n",
     "paddle.disable_static()\n",
     "num_classes = 4\n",
-    "model = paddle.Model(PetModel(num_classes))\n",
+    "model = paddle.Model(PetNet(num_classes))\n",
     "model.summary((3, 160, 160))"
    ]
   },
@@ -765,7 +822,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 13,
+   "execution_count": 18,
    "metadata": {
     "colab": {},
     "colab_type": "code",
@@ -793,7 +850,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 13,
+   "execution_count": 19,
    "metadata": {
     "colab": {},
     "colab_type": "code",
@@ -846,15 +903,12 @@
     "                                 epsilon=1e-07, \n",
     "                                 centered=False,\n",
     "                                 parameters=model.parameters())\n",
-    "model = paddle.Model(PetModel(num_classes, model_tools))\n",
-    "model.prepare(optim, \n",
-    "              SoftmaxWithCrossEntropy())\n",
-    "\n",
+    "model = paddle.Model(PetModel(num_classes))\n",
+    "model.prepare(optim, SoftmaxWithCrossEntropy())\n",
     "model.fit(train_dataset, \n",
-    "    val_dataset, \n",
-    "    epochs=EPOCHS, \n",
-    "    batch_size=BATCH_SIZE\n",
-    ")"
+    "          val_dataset, \n",
+    "          epochs=EPOCHS, \n",
+    "          batch_size=BATCH_SIZE)"
    ]
   },
   {
@@ -887,7 +941,8 @@
    "metadata": {
     "colab": {},
     "colab_type": "code",
-    "id": "Ur088_vjeSdR"
+    "id": "Ur088_vjeSdR",
+    "tags": []
    },
    "outputs": [],
    "source": [
@@ -912,10 +967,12 @@
    "metadata": {
     "colab": {},
     "colab_type": "code",
-    "id": "1mfaFkO5S1PU"
+    "id": "1mfaFkO5S1PU",
+    "tags": []
    },
    "outputs": [],
    "source": [
+    "print(len(predict_results))\n",
     "plt.figure(figsize=(10, 10))\n",
     "\n",
     "i = 0\n",
@@ -934,8 +991,9 @@
     "    plt.title('Label')\n",
     "    plt.axis(\"off\")\n",
     "    \n",
-    "    \n",
-    "    data = val_preds[0][mask_idx][0].transpose((1, 2, 0))\n",
+    "    # 模型只有一个输出，所以我们通过predict_results[0]来取出1000个预测的结果\n",
+    "    # 映射原始图片的index来取出预测结果，提取mask进行展示\n",
+    "    data = predict_results[0][mask_idx][0].transpose((1, 2, 0))\n",
     "    mask = np.argmax(data, axis=-1)\n",
     "    mask = np.expand_dims(mask, axis=-1)\n",
     "\n",
@@ -961,7 +1019,7 @@
   "kernelspec": {
    "display_name": "Python 3.7.4 64-bit",
    "language": "python",
-   "name": "python_defaultSpec_1599452401282"
+   "name": "python37464bitc4da1ac836094043840bff631bedbf7f"
   },
   "language_info": {
    "codemirror_mode": {
@@ -973,9 +1031,9 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.7.4-final"
+   "version": "3.7.4"
   }
  },
  "nbformat": 4,
- "nbformat_minor": 1
-}
+ "nbformat_minor": 4
+}
\ No newline at end of file

doc/paddle/tutorial/cv_case/image_segmentation/pets_image_segmentation_U_Net_like.rst

-基于U型语义分割模型实现的宠物图像分割
+基于U-Net卷积神经网络实现宠物图像分割
 =====================================
 
 本示例教程当前是基于2.0-beta版本Paddle做的案例实现，未来会随着2.0的系列版本发布进行升级。
@@ -33,7 +33,7 @@
 
 .. parsed-literal::
 
-    '0.0.0'
+    '2.0.0-beta0'
 
 
 
@@ -65,6 +65,17 @@ Pet数据集，官网：https://www.robots.ox.ac.uk/~vgg/data/pets 。
     !tar -xf images.tar.gz
     !tar -xf annotations.tar.gz
 
+
+.. parsed-literal::
+
+      % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
+                                     Dload  Upload   Total   Spent    Left  Speed
+    100  755M  100  755M    0     0  1707k      0  0:07:32  0:07:32 --:--:-- 2865k0  0:12:48  524k   0  0:13:34  0:02:41  0:10:53  668k      0  0:12:45  0:03:06  0:09:39 1702k     0  1221k      0  0:10:33  0:03:25  0:07:08 3108k37  282M    0     0  1243k      0  0:10:21  0:03:52  0:06:29  719k0:05:53  566k0  1237k      0  0:10:25  0:04:43  0:05:42 1593k 0  0:09:46  0:05:28  0:04:18 2952k 1467k      0  0:08:47  0:06:43  0:02:04 1711k
+      % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
+                                     Dload  Upload   Total   Spent    Left  Speed
+    100 18.2M  100 18.2M    0     0  1602k      0  0:00:11  0:00:11 --:--:-- 3226k
+
+
 3.2 数据集概览
 ~~~~~~~~~~~~~~
 
@@ -89,10 +100,10 @@ Pet数据集，官网：https://www.robots.ox.ac.uk/~vgg/data/pets 。
    ├── test.txt
    ├── trainval.txt
    ├── trimaps
-   │    ├── Abyssinian_1.png
-   │    ├── Abyssinian_10.png
-   │    ├── ......
-   │    └── yorkshire_terrier_99.png
+   │    ├── Abyssinian_1.png
+   │    ├── Abyssinian_10.png
+   │    ├── ......
+   │    └── yorkshire_terrier_99.png
    └── xmls
          ├── Abyssinian_1.xml
          ├── Abyssinian_10.xml
@@ -315,7 +326,7 @@ DataLoader（多进程数据集加载）。
 
 
 
-.. image:: pets_image_segmentation_U_Net_like_files/pets_image_segmentation_U_Net_like_12_0.svg
+.. image:: https://github.com/PaddlePaddle/FluidDoc/tree/develop/doc/paddle/tutorial/cv_case/image_segmentation/pets_image_segmentation_U_Net_like_files/pets_image_segmentation_U_Net_like_001.png?raw=true
 
 
 4.模型组网
@@ -436,7 +447,7 @@ Layer类，整个过程是把\ ``filter_size * filter_size * num_filters``\ 的C
                                                                kernel_size=3, 
                                                                padding='same')
             self.bn = paddle.nn.BatchNorm2d(out_channels)
-            self.upsample = paddle.nn.UpSample(scale_factor=2.0)
+            self.upsample = paddle.nn.Upsample(scale_factor=2.0)
             self.residual_conv = paddle.nn.Conv2d(in_channels, 
                                                   out_channels, 
                                                   kernel_size=1, 
@@ -467,9 +478,9 @@ Layer类，整个过程是把\ ``filter_size * filter_size * num_filters``\ 的C
 
 .. code:: ipython3
 
-    class PetModel(paddle.nn.Layer):
+    class PetNet(paddle.nn.Layer):
         def __init__(self, num_classes):
-            super(PetModel, self).__init__()
+            super(PetNet, self).__init__()
     
             self.conv_1 = paddle.nn.Conv2d(3, 32, 
                                            kernel_size=3,
@@ -531,7 +542,7 @@ Layer类，整个过程是把\ ``filter_size * filter_size * num_filters``\ 的C
     
     paddle.disable_static()
     num_classes = 4
-    model = paddle.Model(PetModel(num_classes))
+    model = paddle.Model(PetNet(num_classes))
     model.summary((3, 160, 160))
 
 
@@ -540,32 +551,32 @@ Layer类，整个过程是把\ ``filter_size * filter_size * num_filters``\ 的C
     --------------------------------------------------------------------------------
        Layer (type)          Input Shape         Output Shape         Param #
     ================================================================================
-          Conv2d-22    [-1, 3, 160, 160]     [-1, 32, 80, 80]             896
-      BatchNorm2d-9     [-1, 32, 80, 80]     [-1, 32, 80, 80]              64
-             ReLU-9     [-1, 32, 80, 80]     [-1, 32, 80, 80]               0
-            ReLU-12    [-1, 256, 20, 20]    [-1, 256, 20, 20]               0
-          Conv2d-33    [-1, 128, 20, 20]    [-1, 128, 20, 20]           1,152
-          Conv2d-34    [-1, 128, 20, 20]    [-1, 256, 20, 20]          33,024
-    SeparableConv2d-11    [-1, 128, 20, 20]    [-1, 256, 20, 20]               0
-     BatchNorm2d-12    [-1, 256, 20, 20]    [-1, 256, 20, 20]             512
-          Conv2d-35    [-1, 256, 20, 20]    [-1, 256, 20, 20]           2,304
-          Conv2d-36    [-1, 256, 20, 20]    [-1, 256, 20, 20]          65,792
-    SeparableConv2d-12    [-1, 256, 20, 20]    [-1, 256, 20, 20]               0
-        MaxPool2d-6    [-1, 256, 20, 20]    [-1, 256, 10, 10]               0
-          Conv2d-37    [-1, 128, 20, 20]    [-1, 256, 10, 10]          33,024
-          Encoder-6    [-1, 128, 20, 20]    [-1, 256, 10, 10]               0
-            ReLU-16     [-1, 32, 80, 80]     [-1, 32, 80, 80]               0
-    ConvTranspose2d-15     [-1, 64, 80, 80]     [-1, 32, 80, 80]          18,464
-     BatchNorm2d-16     [-1, 32, 80, 80]     [-1, 32, 80, 80]              64
-    ConvTranspose2d-16     [-1, 32, 80, 80]     [-1, 32, 80, 80]           9,248
-         UpSample-8     [-1, 64, 80, 80]   [-1, 64, 160, 160]               0
-          Conv2d-41   [-1, 64, 160, 160]   [-1, 32, 160, 160]           2,080
-          Decoder-8     [-1, 64, 80, 80]   [-1, 32, 160, 160]               0
-          Conv2d-42   [-1, 32, 160, 160]    [-1, 4, 160, 160]           1,156
+          Conv2d-38    [-1, 3, 160, 160]     [-1, 32, 80, 80]             896
+     BatchNorm2d-14     [-1, 32, 80, 80]     [-1, 32, 80, 80]             128
+            ReLU-14     [-1, 32, 80, 80]     [-1, 32, 80, 80]               0
+            ReLU-17    [-1, 256, 20, 20]    [-1, 256, 20, 20]               0
+          Conv2d-49    [-1, 128, 20, 20]    [-1, 128, 20, 20]           1,152
+          Conv2d-50    [-1, 128, 20, 20]    [-1, 256, 20, 20]          33,024
+    SeparableConv2d-17    [-1, 128, 20, 20]    [-1, 256, 20, 20]               0
+     BatchNorm2d-17    [-1, 256, 20, 20]    [-1, 256, 20, 20]           1,024
+          Conv2d-51    [-1, 256, 20, 20]    [-1, 256, 20, 20]           2,304
+          Conv2d-52    [-1, 256, 20, 20]    [-1, 256, 20, 20]          65,792
+    SeparableConv2d-18    [-1, 256, 20, 20]    [-1, 256, 20, 20]               0
+        MaxPool2d-9    [-1, 256, 20, 20]    [-1, 256, 10, 10]               0
+          Conv2d-53    [-1, 128, 20, 20]    [-1, 256, 10, 10]          33,024
+          Encoder-9    [-1, 128, 20, 20]    [-1, 256, 10, 10]               0
+            ReLU-21     [-1, 32, 80, 80]     [-1, 32, 80, 80]               0
+    ConvTranspose2d-17     [-1, 64, 80, 80]     [-1, 32, 80, 80]          18,464
+     BatchNorm2d-21     [-1, 32, 80, 80]     [-1, 32, 80, 80]             128
+    ConvTranspose2d-18     [-1, 32, 80, 80]     [-1, 32, 80, 80]           9,248
+         Upsample-8     [-1, 64, 80, 80]   [-1, 64, 160, 160]               0
+          Conv2d-57   [-1, 64, 160, 160]   [-1, 32, 160, 160]           2,080
+          Decoder-9     [-1, 64, 80, 80]   [-1, 32, 160, 160]               0
+          Conv2d-58   [-1, 32, 160, 160]    [-1, 4, 160, 160]           1,156
     ================================================================================
-    Total params: 167,780
-    Trainable params: 167,780
-    Non-trainable params: 0
+    Total params: 168,420
+    Trainable params: 167,140
+    Non-trainable params: 1,280
     --------------------------------------------------------------------------------
     Input size (MB): 0.29
     Forward/backward pass size (MB): 43.16
@@ -579,7 +590,7 @@ Layer类，整个过程是把\ ``filter_size * filter_size * num_filters``\ 的C
 
 .. parsed-literal::
 
-    {'total_params': 167780, 'trainable_params': 167780}
+    {'total_params': 168420, 'trainable_params': 167140}
 
 
 
@@ -629,15 +640,12 @@ Layer类，整个过程是把\ ``filter_size * filter_size * num_filters``\ 的C
                                      epsilon=1e-07, 
                                      centered=False,
                                      parameters=model.parameters())
-    model = paddle.Model(PetModel(num_classes, model_tools))
-    model.prepare(optim, 
-                  SoftmaxWithCrossEntropy())
-    
+    model = paddle.Model(PetModel(num_classes))
+    model.prepare(optim, SoftmaxWithCrossEntropy())
     model.fit(train_dataset, 
-        val_dataset, 
-        epochs=EPOCHS, 
-        batch_size=BATCH_SIZE
-    )
+              val_dataset, 
+              epochs=EPOCHS, 
+              batch_size=BATCH_SIZE)
 
 6.模型预测
 ----------
@@ -660,6 +668,7 @@ Layer类，整个过程是把\ ``filter_size * filter_size * num_filters``\ 的C
 
 .. code:: ipython3
 
+    print(len(predict_results))
     plt.figure(figsize=(10, 10))
     
     i = 0
@@ -678,8 +687,9 @@ Layer类，整个过程是把\ ``filter_size * filter_size * num_filters``\ 的C
         plt.title('Label')
         plt.axis("off")
         
-        
-        data = val_preds[0][mask_idx][0].transpose((1, 2, 0))
+        # 模型只有一个输出，所以我们通过predict_results[0]来取出1000个预测的结果
+        # 映射原始图片的index来取出预测结果，提取mask进行展示
+        data = predict_results[0][mask_idx][0].transpose((1, 2, 0))
         mask = np.argmax(data, axis=-1)
         mask = np.expand_dims(mask, axis=-1)
     

doc/paddle/tutorial/cv_case/image_segmentation/pets_image_segmentation_U_Net_like_files/pets_image_segmentation_U_Net_like_12_0.svg

-<?xml version="1.0" encoding="utf-8" standalone="no"?>
-<!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN"
-  "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd">
-<!-- Created with matplotlib (https://matplotlib.org/) -->
-<svg height="181.699943pt" version="1.1" viewBox="0 0 349.2 181.699943" width="349.2pt" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink">
- <defs>
-  <style type="text/css">
-*{stroke-linecap:butt;stroke-linejoin:round;}
-  </style>
- </defs>
- <g id="figure_1">
-  <g id="patch_1">
-   <path d="M 0 181.699943 
-L 349.2 181.699943 
-L 349.2 0 
-L 0 0 
-z
-" style="fill:none;"/>
-  </g>
-  <g id="axes_1">
-   <g clip-path="url(#p58ad9a7e6d)">
-    <image height="153" id="image6a21407320" transform="scale(1 -1)translate(0 -153)" width="153" x="7.2" xlink:href="data:image/png;base64,
-iVBORw0KGgoAAAANSUhEUgAAAJkAAACZCAYAAAA8XJi6AAAABHNCSVQICAgIfAhkiAAAIABJREFUeJx0vUuvZUl2HvatiNiP87qPzKysKnWzKTatNgk0YNMUQMCgKcMaGIIhzSRrbI0kAx7YHhmGQcCA/R/kIUH4B/SYgDywKUpNWObAzZdksLurWZVZlY9773nsvSNiebDWioh9MusUbuW955z9ilixHt/61gpiZgYAZi4/OUf8+Z//Jf7ZP/2n+MM//EPsNiOICJkAIgIAeO/RhYCcM3LOYGaEEMoPAKSUkHNGSgnee/RdD0dAYkbOGY4IOSbAETKzXCNn2Mt+DyHAM5Dsb+fhnEPXd2BkvQ4wDAN2ux3unz3HJ59+jtu7W3z66ae4u3+G3X6DcRxx2N8ALPcLMMgRuhDKcy1xwvl8xjAMiDGCiNB1PbwL6LoORA7eB3gv90BEYCwAE7puhPcdyANdGGQsfA/nHch5gKhcx16Pj0f827/8C/w//+b/xr/7t3+Kr776Am/fvcWyzAAnvU+UMQYTUuIyrgAQY9R5k/dSkuNs/OVYIKX84RgzA5C/iQjIDAJkrgH5XyIwMQgM1uOI5D5SBn77t/9j/PN//r/h+9//1dW42LM6Eyw78OnpCb//+/87fvM/+g38+U/+X9zd7OCDAzk5iFQgU4yY57k8kA1E1PftxcxwziHnjHmekGIqN7DEiIUzlhjL8e0gmYA555CTDJZzDsMwYLPZIKeMeZ7LIMuDJx3IhNPpiGk6w3sCMyPGBcfjA+b5gmW5IKUogq7XXeIkg6L3K5PEcORXwsEqpDaRJowpJeS4AFkWas4ZKS7IMYFzM5HlPPbMWe/bgZxbTZQjB0fynvcezhNCcOi6Dn3fw4eAvu/L3/bjvZfFqf+GIIvkY0Jg8woQiAKIHEAEDwJngB0BIDgGPKoMeAL6nvDHf/xj/PCHP8Tv/d7v4d27d6u5Z2aEdlWdz2f8/u//Pv6n//F/wH4cscwzQienjTkj51Q0mE0GESGqkBARnHMAgMvlUs7bdV1ZaZdlRp5ZB5DkYWwAASSIQCzLIoLjPMAZw2YsA+gcsMwzGAnzPIPIYbsVbeu8x3a/xf2zW2w2W4TQYZkjdts9iIG4JIAZzgM+EGKaRRupkNok5Jj0GQnkGAAhpQznMxw6HcCElCMADyIPgMDOI2VCII+cM8gTCAyS2ULKGYEIRA4RjK7v8eKTl/iV7/8qjqcHvH3/dbkHzgRU3QEQIRGDiZFZhCR4D2YPzhmcM5wjhODR9x1ijEi6OOuiyfCeihakxGB9ftPuDICYEXWhOyIwgASAVL0RsyxQBmKc4R3hv/mv/xmmecI/+a/+CYZhUA3pEGxwY4z42U9/iv/1f/mf4ZyYBNabYpaTm1bxvq5s01QAyuCU1aFaQh5MjiHnEADEZSnmI6r2ciBAtYOZX7vO6XTC5XLBMAzYbkfM8wznPYZhwDAMmKYFfd9jHEe8efMGm+0W2+1OBcfpoBKADO97EQgGYlxWK9s5BzAjxojQdyqAAJBlhbP8LVpMhC8EB6L6zERcNCqRg3MZzA7EGQQHBsuYOoe+77Hf77HZbMQyxSQTCJLJx/plmteRg3cO6uuAsxPt4v1qzHPm8neMcSVw9v3EVRCJHYCMnOnqyld3YuNS3iYsy4L//r/97/B3/7O/ix/84AcyXoAIGQA8PDzg7//9/wJxmkRlq8pc0iKmIHPRRiZQJmSmCWylmE9mwmYmsOs60V7MCD4Uk+hVMB1IhE/PK5rElWtsNhuklHC5XGSw9XqXywVdN5Tvv3jxAvf39+i6DtvtFsPQwXuC84RljgB1SEmeQ+7Xg1OG7zoZfPXVCE5NNcN5MYGAU3OSQSB4HwCwPh8h5QVBfVVZcBmZEyiLIIIYmRzI2/gwQh8wjiO22y122y3OxyfVQqLJ7PmLoDknyyWzahkCqztjc2Q+nAmZLFoTQC4Cx8yIWX23mPQYrBSEzaN6aVXsyMRLfulCh5gW/L2/95/jxz/+Y9ze3om5JhIJ/Mu//EtMpxPIEThlRLD4EU6lleRkZlLswvZgzrniy9QBbgbFAobGFEEdfyZZ1akZHDsmpVS0ZyvQ5sfFGGVy9js450WwNlt4/b3rOjjnxP/yHgzxDZkZwzDgfD5jP44g1QBiNlTzOg+AkTkiR11YGSCKADkQJGDpOg/vgJQjiAJyWsSM5SxCkhOYgpg0ynAuiMZwEB/Le4zjiKB+1TiOiMsMTgs4E6qbrpPNMt2qx+BIJpxhLotei8X0VYFzSJmRU4YPHpwZmTNC1ADNJzCLEFZf06lwMZir+bZzEzOIJSjMOQOc8frVK/zkJ3+K3/qt3xJNRkSY5xn/6e/8DvbbDbwnIGfAOThPSOW06kPpTZMJURM4LDGWVWYrKsZYNJsJR1HNRCLUzMiqvjMYXgW5FSoAxQRxFsG7XC4ggpjO8wW393cg7zHPETnLypABznCe9H25J2Jg5kn+zRmd90CKxcQwZYAWMFczKveXxdVQ8+gdIScCsxwbfAcwwKwLiXsw5FlSVpeCHMAZyKRmz4PIYRxHbDYbeO8QgkNcnPiBDmCWcfAgMEn0BzVHxFXHlKjZm5kHOHhkZuRMcBnqkmRkNZPe+TK+rSk1vzhnNNpRfooGs/lhRnYA4LEsC/7O3/kdvH37DofDASHnjP/jX/wLPH/+HGmZVr4FiICcdHVXrWQ3VLSUPpxpuHbVdV1XBNP8hzZAsIFpNSBR9ZFMGC2qzDkjJy6wSMoZ3gfsdzt0XYfQRFvDIDDCsiwABdFmTv2z3JxDhTcE09Aowm1a1e7RtHXRKgAyMTyrJues2k6c8BgXuMBwzACCmHgQXLPwAAYRcDmf8e7dO5zPZ8zzLNchCRzkeq4iATrBOqgrc3oNk5if6MiJwGZG8BrZ5gwODPCHQmYLvQYPjJyT/mR1OdpouSqFrgv4gz/4A/yDf/APEGKM+Mf/+B/hcNgXB7869TaY8litSSy4WuujacTh9OZsQOxBnRNn1R6i/bz1wTy5IgAm0JvNpgjmPEWklCQwcB7juMEwjtjtdvjkxSd4+fIl7u7vixNM1JWHl4FM8FSDFLu2QAq8Eu6cM7quK460DWQ7RkRUtAJhAReNr4EAEzITPNWwPnOGgy0mJzhWWWRU/BybQvvb8EQHgFiQqxa7sn9bP85MLAjIJKYNENelA8SEqi9tiz3nVHy4lLKa0YyUHFKq77c+n/3Y+PyX/+gf4vHpiOC9x+3tLXKOooaZQXrXibOulmom68qgIhT2kolxQHFqZXjaVdHekA1gbnA6MSupnLed7MvlLOau+GkBLnTwoUPoOuz3e3R9h3mZcb6cELoDco5YFqDrNoD6mSkmMHHx1RhcgOXiP6oJBwg5iVbKjhB80PtKGmXaOHj1HwmEoMIq36Pk4dUtIOeQOcFlAjsufu5mM+L+7h43N7d4enyH6VKDHvG1xDn25CUwaQSLyJnXpOPGEDFEq2jALILlWEBdhmmVqq2L9soJXmGRVoullJFyQk4ZLTxy/QMAfS+4XPjRj36kbzo4D6QY4VnsPkGdW/U0i3CoRruOegCRR7uIDRBUuEwzFcHU6Ig8IbbnaGAVM7kVx3JYFkaMGUQJ4zAgI8N5CRwYQOg6mWCOyAwsMeMyEZwjLEtETgm+70UASHTEMs8IfV8EO86Lmt+u4kcNCEtlPMQxds7re4Y0VWiHc0aODATxcC3Q4SyYF5FEZuO4reaQxISJD+p0xcpsEAgQpAX2rupi+RqrdBGqniPSRa/3XnQjChTSZgJ8CMUsgvFRQZIIOBXFYShCxeSAH/3oR/B/8id/8rvT5byyp845gBp/C9W0tdmBVtCuTaP5AeVRuPwPOgfV/DIjN6q2YECN/0Y6SNVPEiT7cHODrutwc3OD7XaLw80N7u/vMY4DgvcgJ4FNToK620C0EbLdhw8VVhF4wsN7EyKNspr7MOECKograyeJgDiolrFjJPqCI4lO9YfBWOYZX3/9NV599dc4HY+Ies8g0VrVVzaEXv+jOqbOXftl9fvreTGfG5DHsu+56tbAAh4Byr3zCAqaWwbhY7/Xfx1CCPiXf/RHCH/2Z3+Gl588XwlJttDYiT/BWQdOb5QB0QaNo28mpn3llEFcQVtZwlUgqVlB16bYVkcrzMwSScqEStjvFNDcbrc4HA7Y7XYFNGYIThSXGZfTCeM4luAhaiRs17AI2DSuKwCuwikQvy6nrOPDBZUnX4MXG7fEEYhitkSjMzwEU3Ps1EeqC7brO+x3O+x3B3Rdj34YkJZFoBodW1nMBhslEHn52wmMYD51WSYNrIHm3/Yl46vngwdYrU2WeyfnkMFgx3AKQmfVjy3w2+Kkoskkwv+Lv/gLhO1mXGNTEFBUJUPVv7wnOI9Ifpkgu1msH8iiSedd0QwOkOiLSCEG1sFzIKqwhqHufd+DmXE4HHA6HQv+JIEDoRsGZAY22x1ubm+x3e3gvVMhEj9KXA5GXM6YKcNrHs6ASu9tEWj2gpwAnLnNPABECSBCilF9H4KDA5NMtj1zNZEAfGuCogK81bSBAXay+Lq+x92z53j2/CXev3+LlBbM04yYosAhBQvlAp9Uy+BV49H6faom8dsEDFAo1DAwE3yX4Vj8NTIr5mQmHct5Ja3UEiuyQiUJOXfgzHjx/B5hu92swvL2xjjL0sgsoCKcK6ujrHgiRclRLiiTJv9aNESaSuHMYEfCTCgwQdVigrCL1jRfbJomxChAIeeq3fqhx2a7x83NLTabTTlHzhEp5+KvIEd0ISDFBfN0wTBsyrVFG1NZlXAKz9j954ycHZgngAT9lzFwyJTBiHAOxafs+x4xJnjXlXEomjlnsEf17XIGuwwiLzDM4QYvXrzEN998hePxAY/v38sTrOAJVCFSYS+rhCrozWi/96FglQgUAKfGrTHzS+KSlKwCNWcrXyV4VBfGewlKUhYgmhKje/4MwTQSUMHOVkhSSgKYqvAVTWWRYvOeo5qhN6e9TKblNIsZy2ut11yv86FENUSE8/ksT5VFCwTfFZhkt9thHEcsizjqKSXMiznAOkA5Iae5mNwQegRep6wK0AvCUpLmxshIClVknXSLqLkAm1CBTjnDqw+Zc0bo3ErQXLPyzdclQvFt+nFEyozzNCHlJHhrXs/Jhy+upkTvw3xpXd6Kq314iAlc8asL/FF91m8T1Bb6EetgEa4XReEkK+G+7QTmbHtfhbDFY8yPKZQUtz5VYTOoCYnGfVqWwtoAPoRFWozHhNWA35wzOo32NpsNQKR0o1gS5SEEFWhR2zFKNDlNc/W9iDAvC5gqqp1SAscE1myA+RVl2iiAyMPSLC3u5tWE2ABdL5prf8VMM+dWaBh9P2C/2+Pu/h7b7VYS7x+ZG72KarVcHXlUraJ3vfrXAi4TQiq/lcf8IIAr1KDr6zdphho8ETon3D/nHJx36LogCfJiKjRctQNFnRK882oy9T4VQ7ILF4Q4hHLLhqyXFeyc5ODU0bT35boK4jrhTtnkFCpRmisoCsb+sBcHfl5wPh9xPo94enoCABwOe+ScscwLHLEEcgAcROCXnBFyxtBpXtTrmueMOTMCBXhICN9GjeRlEDmz+mu5DCyrv+ScQ5wXEfLAyOxAQTSu8wCTmtBiJllBVVlEQxdwczig77oCQzjykspaCRo3v9vilgDAnP8qVQp8qMbEh3qlkTEVOQtnIXQmOaeQFldSD0kL5hIKiNsETwjkinIIQGVWBk1bmCksgsBcMu6F5qOmxlYoAFDOAhvoU7YIudyf0Hfg1uxMqOC2qapxHEvKJ/MCRxL99X2PZVmQAXz3l76L7W6Hvu8xDAOICPMywxFEMFNEcB7IQo0Jviv3l1JG72tEKYi3JNNd8ICTxWURtGggeY6krgCrd78sC7Jvvqv/835NEqhRcvVbW63NBPjgcDw+4nQ+Fq3XLlQ5TiWBDT9rhbAK2MqXYwKYyuet5ms1YZ2Q5jvUwr+tjBX7LK6JZurtOS2DEOzmvfdAqimidtLJzuca+EFvoE1FfCxVtMJkIBDAkuKVWq+TYEK8LAsulwu894hLwjhWKs/zl5/AeY95WbBpzBIAeOcxL2ekFOGpwfHIwTWQSx86dM7XqFInsMAfihsVn22pxMwKVxhSnhFjBZzbScscRT/wWFI3Zg1WPxKuiaY3AuSV6WrH00BgkcIP538tMFfOfiNo1+7JB4fqe+b8X39GJNCGeaXtMTauwSaOgMKsaDGfVbSYuEAYdgLBlLyatVQnu9VicgKFLdSsxCponDJy45PZZIcQCi15ThFdEOrO5XIRnljfI3QB+8MW+8NWtdxUgo+UU2GODEMPy0AwMxInUBTNaGNn92wOMKBUGBfgGIhpKdE3q6kQn65OmPHmZPwqNSolRhcki9F3vboD6qYQaRbFY+i3+OTF38AnL77E+ekBcV7AuaZ9WqGTC1+9BzTjTh985gMJHtbM7fV32pfJRtL7XH+/vuXIQRA0Kr5mmUsb2JhSMZfXN2wn5LoExBZnXkVL8hChJMHXjqJgLiJUqZyb9eYF1vAg1lXTRKiu7+GIMG5G+L4rE9n3veTQYsZ0mRDnWfJqMYIoin/ne4S+QwaQmODJlYgveQY7oB/GmnBXYclJ7supwCQkpLxAsj1ONWyGDwExLgq3CByRUoIQHvXZjaER5Rpi7yzpHJWWI77qsBkx7kYB2ljZE+5b5qUZ3g8EEOpjkZNxX8EZRhn/NhVo0aplMFrTSwUYbrWtMNS9HOcdjGhARAhW9OFcLRYxE1rVuWJOJGYkmY8TPDzUbyOC7zpAheP64Y2BUXjntgLUue+HQc1pTY4DRoRj3N3fo+/7Alm8fPlSuGqKpR2fnoT/7wFwFEH2EoHFSOi6QaqN4DBPEX3vMPgRIYgzbmbffAl7z7RusmRxYrjgyoKySTVz07KCa8bCF1Np40uEoqVdZngn2jE44O72Bnf3z/H+3TdwLuF8OpUoeK196u9mps0KWaKckWR9N6aMRTy+RcCg6imWIKNCNvXztem2c9b78t6e00moYQxSO6hNZBeaNSrY6i2nmWqmwCtKb3+bT2eRZ05S9CGDxLAwKPhqFpm58M9Mk0kelTBNUkn08PAA5xxevXqF0/mEx8fHlXAQALCwBeY5YZ4t1eM1DydRo+XpTCicc1iWZeW02mdCRgwAPJwPxQ8j/cysgUXUtshYiZNGB7Jn+iDoySylaHqOzWaH29t7nM4XnM7nWlTzLZrneqxDEX6JCtfycYXzlQxChSJEoXgQBQDuo9dtQV+iSmhtn8nOFSx1A5QAtCRHS/Rj8ENzcos+PVCot2Zy7Hvt6hLnm5FyLnWXRIQlKzPTAVFNSFl1juD6Drf7Pe7u7tB1HTabDeZ5lrKwPoCYkaJo46enR3BO8MTwzqPf7tD3Uj4XQleSt1ZK1vUjYmIMvVREjX2HnBNc18N4UoBhXYDvenmOrMn2HOUeEbRsLQjAmsXPhKfCtEhpAdCtimMMpjBBIQJc8KAwgILD3f0dkCccnx4wz40mK35hDZpaQahCrGjYletTrksZUoMkaSmwCVa9LxG2gpCWqLposua85uYUQdRfi+P/sVcRDiVAtr6BPRgTVpScktgGIWaL+GraqgUq5fxipkOnEe2Syrk61TZW3zmOo2g3CoXEuMwzaBFsSnJwWUq5OldMlEEifT8WzeOVMdv3fVkIswKxobdJybAxawtnbBEx1zxrzhnzMotP2kaYTRK85ci1/Drnqt/rnMfNzQHeEU7HI87nCSnV2tUygWhA1mau2ijxW7WfsWOITARUgABmdXUowwDAKjfropbqDnzoM7b3FVrJM4FwqNqnlHa5dRhcByug04ruxDVV5BgIWpSaIQ7+0pgKmQiBNbxCCUbBCV78I3iHzWaD25sbEYJ5xmazwTRN2PKAFAVIJgjAGe2hg4frOjgX4JxH3w/FPBYh05+a3nLwvtK+r/3TVTTMAkzaeOac1R/0xSFnZoS+k6iL6+JaO+7VkxG/xoFI8sHPnj3Hy08/gwPjzTevSvFLIydQMGuV2lvJ0rUGU0ExAiSp9gKtPxehEY9OJMytzvWt2NrVde334qWGIGG6DVqrcUgwyCJwzjmNJmqkkWMC+yZoAABHWJRX3/pv9WahbFyhInddJ8wGtedOTZzlJU3rbLaDUEkWE1gHT4S+60Deo+t6bDc77HZ7DINosK7r64qzcL5UU6vAqzklZJCv97zWXlZwTCWT0Gl9pnfCXwMBoRuQEzRT4Ao7xLRrO6lVeOXvYRiwGXeIMeN0mUTw3Lp6i5P4hEytdjWhNU22Npf1OwZPAUBeaUTzK733SkbwV5GrCZwdw8XnLE90pdWcre6WT3UtqSlpYahGEQ5rPyArBcjOZWAmoJpRI9frouCcpepJtKRQfIvbQEDX9wihx/5wKMlvcawXnDTiOp2OhbfPKih91yt3bMQwbtCZloFgga2wk2kCu08SwfDewanpuzY7YgYdyAfxV3MC4BBCB8Fora2AlPxLMENF4LTSpEyWaDHJtIAYfR/w/JOX+M53fgm3t7eFN2eTN46jAONurUm+HZJAyVMSrn20NT/fBE3ID1Ug68sYwkl/lJ61huVWr2BIe9JcZGvPq2kQjeq8IvJYh8oldE65lLgVHyJrGZlzyE3hLjOv/BxmQtAolYhAwWO320ui2HuwM56Y8KsMchjHDZAYFAJCN2Cz3WG736MfNwhNpVTbDOZjfoP9G2NC1wvijmbhtK0YdNaEceBr8tjK/yxCJkIDzKJQYYh84/TrtQEx0SyTXFIyyjK152iFrQpFOcOV1llrnIparDXNt2khO74k2qmei1VpEonQGbGq/Z7JQTAVDAa8rWp8+CqOJtXV0GoyU/vgmoohImROYjaoIuJ2cYMQRGgVPjFB8KIhlmWGpQCtkYhzXTE9RMJNk4KSgO1uj34YETptRtKYyZbxYf+KaZe8pLKrwCzXNoLhNT5VJsK3KLzCLx+ZMEC5cRDAWWCUdelfTU4LxOB9gA8BKWdE1fgmeO0YtsHAh84366xp7tGV3z7y+jBIIPOTqH0WKsJVjnMEcC2NtJdZC1e0EEsNHtxagMQ0VKYBmsmyi9hKKhcg6aoTU1LU+mMOKMp1RUOSIN8WxqsZSnHBNAmfzDSZXV9oO4DvgvaT2ElS22mbhaZvxfX1bSGIeyCNUzgnEGoLpsy1dM9+Wqo4WPAg86ey4WGo1zOsL2l3IO+CNEtxYSVkOn1aGyla/MWLT/Ds2TNs98I6kcBE4BDRkKlkFa61kf0HCJjtwPBAYbm2gmSQxMcEzcxYKSRXig+3X2BVTKqFGXklPyW6bM0I5YpjGRZiAF7OuQBv8n0Rde+9ZSHqZLCE5ERr6KJw1UJAUL+mpUWXlBQzkqtRYdBCD6kDTBgG8VWGccB2u8M4bopmtOtYCN5Gy/ZZMT+Z4b2luDyQZdQyV46+PXtLtgTLMS54jfIU8sgZcB4hdGXcylhq8YhFd/V982tEu0pxshAYY0xSZdWQEHa7DWKcYXanNelF2OTpi0QQVR8UH3zTAoFrc9t8S8fww+NNXyrwTWvcLngX4ByrGVInNNZEd1nFrR1fXZzKzXu9UVYBu27Kcm2mgg/oukGqelB9DhvMw+FQwNPW+XYuIGcRni4M2Ix7eN+h6waE4BFCBYFNoIC6cJjFtEmyWlJhTIxM1echEuGjRlPIfVfcqEAcaFNwddHmzOh6mRTvG8YDoLBB9aOMnm7aRJD/rVZuyTh5cjhfhDfXadkfc0sBaubIgmg25aOmXxdvK0TFmuVatng9b+34F7NejoVaLNFiaMZex6LTiTDbzUiUV5PUYj8EiDNPVdOxmVGuk1HyelcPX6JKIvjg4TvBhZBZtVotUdtst1iWpTjwlhi3ye66Af2wUc6+aRwTRK9atNJb2meyMF20lSTEfR9kcTgCHEBtHwwogBklhUIWiV4NaOYoGQySQAgsTFxZzEHXp2jIGuGiOtdmGULAzc0dXrz4BI+P75HTjOPDUibashDQhR0Lx60FTGVGMymLyIT5Soiuo+dWAA2Qrn+jWLdGGmE20zBWNGSLUCdfzENhqLZZgManMtttDrG8XFFuNqF2bQfpSrN6eJ0kS36DCD44OCdcq2Hoiwa7JvsZJXq332Icd9jvD8W8mKl1hRRZ77elGZnglESyCrw53BZVrgZcJ420eMYWg0Ef1kdDil20i5ALYDjNe1K9H98hCfSKYB4HZ2SWCihycj/7/QGb7RYxRVymCXO8IGtbp0U77nCWcrp5vqy0DsoVFYss9ZtVk1UZuQ4YrrG3MhKNoLVvWzCxVkY236HFRExw6Ip810Zn1xBHkcEGd5LKzSxsFRaud9ty03tfqrOdZgvIAT54zTG6lUDYKrXBkc89NttDiTjtJUJpKZvaLbEV1FajMqO4CYD0YRNKtrJIc/VlWgG1CanZA8lbOgc41wHk4PX5KuxhAYN2/Cn4ogOz5E9hpg5OcL5hg5vDDabTE5bpjLic4X3QCnqJVqWQZirz8q2BlmljkQuYuhAsULoIrJABZYu0FsLOZHhD0aao1s857emmrbfCitGqDYK9qsPr/hVr+g8XQcyo7At7AG9FqDkjJQVxLROg55aIKYgQ2Dw0CPg8z9IWSrP5xV9yPcgHzPOCoe8KKwKotG4iIRR6XxF2G7gqjMZ7A8gRUloQgpa0eY/OBy170yip5ZvlrAxW7cnRCN4wDHDBwbluBUBn7fGR84LgAsDSU4J8EAOahfuVIXhj3wfc3NwjhB7zkopgAdAeHnLOlGJxyE0Urs14ealGZ+s7BxTtX/AvfOj0ryNPM1VXste+mpxtyW9Y7jE4r30asjAJmpteRQwa6dnKtldMCR0BUfEUdOKPUM5SgQwUjr8Uh4ipDZ1OICplyJoOExEG7VMh6aUB3TCUwtwKCGdYHaIIk5igGOOqpN6unXNG14tJY4ZGR7rokhAVyResB6WdZ5b+r6QNWCywIB8kfeVxU0SAAAAgAElEQVQlgux7YXNI9yFGigk5RYCCCBJLWiqlKM+SJUKTajRpZfXs2TPs9zfYbnfSmfsxIqUFm+0epxPAfMH5fAFlBgoiYEU70ABgbQZl/iud2twlA1aJDH+rQRDQpo/kxDW/e+WXFj6bNKwJLbc9G15CFu7nYt5aU2OrpLR2aoLa4KVwwySfAaFXp4QMIfoFLdAwjRCCFmGoiTNYwXAxsPDRfAjohwH9OKLvOvRDb0OnAmU+mWonApgTKFTBK/WVgjWA2WnXHmFzJmOOeAdmi6DMB2FF7F0xwdYnAiCxAKTWAFDhFuCYIIXJGlrJ9SQUh0AZDvBN2kcX3jAMONzc4HA4YIkXUMyY5iPiMmFaJtzutzidTghkRSUS3cN/zL9q3gNDu+qs3q8KQ90Fg+Gppfh86DK1EW6rkJgBVzpPO+n4xwC60MPDSejb+CA1oUyrNM0QahUQEWEx05kyXJIrFT+oQasBRoyzVPvoyokxFTOZc8bxeGxQbmCzkV4X2bA0R5hjlIGAQ4wL5lnap7eAq73a3miWAkuJpd9HAVolUOBkxTR19TJntKwFOYZLKsgGdxxHWGJbzLIKOIQCTgzkmMr1gFQiOSvqdeSECTxI77UYI6b5IkDxEvHZp5/i5uZGLEWJglvYqUW9muDLBM6t368vIzbac1/7dx/ayetz2Lgxs6SVWnLhtttIZj9luK4H0rLSWtcnLf6Zc0iwgUQ5JqeMxvyvXjFGDQCEFx85FuGw1qCDMi+89xg2G/gulIkzJqskkC26q8W0FfOqlCUocdK7ToQIQN/JphNgCwgYRAnkCch1wMqCW8EPNtjVJ+tChxRlIiwh7r2Hg91XbZDHnCCQJMFDesp6DkhRMiFjP4IQ8Fc/+6lE/46kMgqM49MJn336Qu6Jcw2QGnzs+kWAMjfWc7l+FhYLQJLua5UCAF1UH5lQGwlXmdEA4Nq26d577Kwt5tCLyiVhJJgW+9iPCal3Hh15OH3AnDMihAlrNyrXE4fZePfWA2yeLiVtZGtoWRYcj0eBNLoRm3G3EvIYI+bLpBhaKHSgFmiUgTHmgGorVsA5M+Z0KQyPZZlF63DCnGZkKO0pA1k13vViu6Yot8GFvW9RfNWqBq1oloClOV7Ki1RIkQMyo+t6PHvxAt//lb+FcdyBXNAF73A8X4Rp3AQ7BNtgQjZ3cFwxOACAd9IpGwDFDMq8us/2uaqWrXiZEDVbAqfAKG3A0JZZqtWrxQ3b7RYANOrTwlo4OKoA5rWUrl5ZuioXyIIAj8pHYgDeB93EQTIMcm2Z4JSl0UlSMxVCKNz+0+mE7f5Og4UAIJe2BLZil2UB0do/SCkVmCNl2Sug5eQDWCH2IjAijJaqccGjruJcFheg/P9QeWnMkk3IzAiharFpPorpKL1ixR8sfi1HOHYgHa+olU2h67U96TO8efMKp4cHxOUM52UxPz4+SgRvwlCg/pXIfLAAPKgENSsw1pHUG5Cl44C271k7BhKkYaUyK55Zd54J4ziu0i19P8BrysW5gOlyBsd5BWFY+qGtzTRTstqEINdWBlA/bhgGpcJUG55zRlwWZNUmhnNdsyfGcdRJovIdZmn3VAtRPiz1KvsLdHUFZmS4rPUJ0e5DBDIpxsMkGiVFE6y1Zqx7SGV4X8fhcrmgHwcscZIoXIOYTgtm6mQI+8N7D2L9UbMcgvLfgsd2s8f5dMG42eDlZ5/hm9eEx8f3uL+/w8O7N9LLbJ4+6DQONFqsrR0xYcwojOfy/UZTOzIsTAS3eqb6/fJlWmnLa5wutCsSAJwP2Iw7pJSw3QJPj+8xnY/IKa6iMxtoO6nhUtXRls0VpCUoF6wpJYb3WUiBDTfKJivnjGVaSqsEKR7usNvtVqaHcwVp2THGvgcjS2Bb2h7VxSM+lzShs6gXLoCRtQCiLhwBRxN6Ly2QMiVpLqx+VNtMpqVyt4vNFodBKa2w25jJ942NIbhaPbZCBKHr8Gu//uv4V//6LV69+iukRZrHPLu7xXR6wGYz4lHfa5+ZFEbSwHgdBNgcsmQeiOrnZBCF2HFJu7E6+xZum3nEt2nPKnCBHKHzlr902G52GDd9YxaAJ0dYLhfVbgnmS7RMjVbzyPcAcoxpyUIq1BL8EskayswVqmAGliUiLVH78UvkuT/cYrs7YKfm3NpGtry2lDJCkPSUDiOct4S2OLIxRd0JRTWwk7YG0hrK10FLSQqNUwK7APisQroWHhMu08rmYnRdhxzFR0SWjcNaQRTQOKl/Kh28ycn2M0iW0ZAUX+c8druxKIG7u1s8Pci1Ht69A1JE5we0O8C1RFJmlqZ1Sgtv9nsQ08hcqowKlZvEl+OcC4hehI9VaInhraUVJJfdClqrzQJzQoYVnoofsSy6Y0c/Yhw28I4wjyMulxNisytb65fVlUnaCC5imk8Az0gEJEh/L+9qIartRtY68eIWifk7nU6lM7ZsXzMU8p5MqgYkRBrtEFJUv4ehCeqEvh+gjb4Q0wznJckeXMC8XEoaCyQ5VRiUAOshbYvomqgJTNMF3gektKDvO4TQS2TmCIyo2r1mF0wDViFQ2jlkYy2ZxARiWYSk4/Xd7/4y/ta/9zX+dI6YzicQebz+6itsNzKGornjChivWhwgXtMVi5/MBsvIrnCZ7D0ICYBblVUamYJASHpGzYl8K/gfzuezSr+asyhFFPO8YBxH7T8RELoOA8ZyIkPRW2qOMSRsNzlQ1vZMTtp1s3Sfdj7BkTAScqMZsvYIiznJNjgxoR8H6dOv3LFe97Q8n8+a+5Q7Ej79OrqRJJWa86baygTl6ekJ/WCZhmaQyAGaMqpmuwY8a9gnA5RFE+UIRFk8/RCwLAkhdCUrUFNCdavFlCKIpRCGc0Z2snMJUVQqum69OG6w291oasnhcNji+PgeS5wUZ6xtutrImvS80AVzXT8LmI8lm4sV38sE1OhORYnUZtUmfeZzm9C2CgcAwuVyKkJiu3ewgrCX8xNCCNiMW0321obBVtJmKQgzlzlHhKBRGTn0XYclLmB28oQ5IUVp71RbK4k5y2kRDMbMoPO4u3sm2tRL0CD05Q7jKBiTI/H5pDq8ppfKas2VnSr37AA4bLcb3QjMwZO1UReN4IJHipX9Kg17SYthnDQCdur4luu5Mm6iyefC9QdM8xmo3RWBEDRC+4q43Ai39PgyQen7EZ++/BT7ww0up/f43vd+CT/76b9DzFGCJkB3PSl5CVCuPXizhoptUx0TNmHJAJ4ZFBnW1I2Za+ag+M4G8daggCFjQFeCVjTZdJFwGMyIYEzTRVUlCY3ZB1ymC4Z+wNAPyHGpoB8q+9Q5QuYoiPsyqXZKuqlolDIyJ2Bp3f4vrHwHyblFabgXZQIPhxvBxqhqJfH9rKSNwDlKV0KysjHbsCo3k1lbLxAcTqcTnJMMQxdqMDEMI5Y5fuDflIwFZzj4VRTNbLWXA0CKfWXVjU3Ng/lstiBli8QFxNrWqsyfOOIlgmXSApUBn3/+Ob7+6heYpsmc6hWyWmNKFRZeN0zJ1xFA0UcqHA2A3YpU8cXKe/VKlqUs52o0KTMjoKHvTMsMkDT2JSJk55F8B7csSDEK1147DdqE+SCDQ0RIOaLuvcMAazoIDKG3aC9VzrB9jlrzlWwFUhvFeTBVUzyOoxIOnSbXqURRBhhaAGKC1S6IlBKGYcASVSPr1i5934MBTJe5wgBY06OJSDtBWsoFxZckqpXodr5rLV+yI7421pPKJQdnW/MAyEmi8gKLaIbh/v4eLz/7HON2gy+++EK1kBTSiIBQ8ZFYgdiVIJFkMFrnqY3ufaPlWp+toP0F/jBdpoAyCIVCbMJKNZ0XrF3jOk2UAfWXwOYMJiSnBa+oobrPhj1B+z20N56QGIBGZlT4VA6cE1KMq8Z0tiJtqzzvBa/b7gTlt0Ajxhned5jnBd4DwQuTwfj45iuaYLW+lHMO0zTBaMuuk023GNA2CKl0+yayaLVJKZVd3arPY5kMu7++HyVIcITgrCyP1Xdbm3Q5vrYKsOS2nbdqbad7mTu8/PRz/PUXfwXZfxPwoattvWxhwKJBU0HVi4JmAdqmdq3T3qbMiOp8Fm1HpJ0b6/ull4UcuDLHjo1guCryFCKF7Hu4CEYWI+I8IcZFdvTVIlvZz3vGskwlNWOrt7RmN1dRL+ydVM8wJ3BOdWAUf6FmNFLOZTNV4Yd5jdgUpNXJysmKV7JCEjXSbYMT+1sAVOHhCyvX43w+q28hzq7swlE3tW81r3O+0H8cuaK9xcxFgTz02UsnIWe9Z2XhWbAl8IJsvmqdh6wARVI2spUiEWPoevzNv/mrWhUvfLXtdlu0kKhC8UWrECheRkItN+JDGWUd9zL/aMxkI5jt70LlVpmSyAFtzq21UAFYh7Orahz7jIHEtbe9jk5ZIUSGYuPDF6lWKCZGhJhJd0lrrm9gn++8Mm1lRY7jiGmaCsV6nmdsxhFSl7ku/Rfz3WJ4rgCzovqrX1QAS5CaWRMmE3ihhJv/VD8XDpz3TX8ymIYj5MRYENEFhxCqqS2usy1kzpLFSYuUAhY2Ra3ot/MzCz54e3uHn/3sZ5paE790v9+VDS5Mi/EHk9GUB5LBDlV+rlt12icWzJDha/SxL3Ld1BVrrQjUli4rAWnD0HIbTd9Rwa4YBvTajZOepzqOTgXw+rKyuZP0YZBaz5oPVCDUubKxvWlH2TQiYplnxLgICZBqBsLMkF2t9uyqC+dymYqGk4BAMhzLEgtVxzAxhiHvGUQBts9QSwywJV7frxkA0t6vy7IUYoBouooxRo6A43KsjX8brLS+3H6/xzgOZdyd1gM4cmVj+laL2PEFoVcznHKSvRmIyp7nzQElfSTAq/2vasdW2DJaIVVvrRGoUH+vN9hqtnYFrvCR9jjdHuWDtUOMUoxJ9QZUpwMcdfUFeKdVQMSIKaJ30s8CJI3vjC3y9PSE0HW4TBfxg6KE/hbh2f6b4FSyCpamGYYetTW6V0G3AhJLC2n1FEMCHA8Q+cbngQYlQSEd4bI5J71zQ9chOGHIsmZGxJci2bRCAUzOQrHOLEl7uyfRklyAVUb6cI2C4UOPYdzifDri7ds3YqLVFNpIswUA1jpC0cCiwTVsNJuRqSl709RAmbYipfK/GmHSSsEQc/G9s45VCQmupb88ThMMyCQ2UYaqfettbyAemn9FXqtTaslW+ztn8e88EYLiUM5LEYZttpVSwu3tLYgI0zRjf9hhibOW1AU4bz33A4LvwMoBM9qQ0Wn6fijPYzBMzhkGSK+eialoHKsSFxehrcZicNO23LQYKYU9eK/PImVxhh1KZlqO8Yr7eW9FJ5bH1HSV+soxRszzXAKOm9sbJSEwvvzyy6JhWE2hmEDSzEMbEaI2DgbUdzNZUStFVMilZn6LBmsUmZlZhiEObKZMtKFG/C7rAF4zKtqX/E2NILY6i8pF7fiPvVq/ixqhK5MFaejiQOAkNYt1o3WUpnc5Zwz9AO877LbSkGXcjGVLaNso3mg6tRVmjU4tV2rpsWVZZAvp5j7b7+bGoWXO2r7AgE47RnKAJedLrAUedcH1/QgBbT1CGOBDj9AN8G6Ac31x+E2zWhBQIIXMmKdJun6HDjkzXr58qQFQNYVyjkYiIIwK23ALzT21eU6DP3LOqz1N2zm0+4AplSZQK+aVa9gDwPZWqvSTjwmY3YDZ2mttJSGr4GD1b15/B6IdvJMw3ZxZkdkM5z26LiAuCeBYose4LOBhwNdff43vfOc7uFwumKYFn3/2HczzInABSbWTRZWAcOL6YUTX98gsTNh1Ul8E6ebmBk9PT8XnucbWzNSymmIp/tViYSfEQu8ZzndgOPTdAIKki5ZlAYPRdwO6vlPKD+D8IEIprSEBCCXboZJHq+kkkAvI2fzRRavfF8xzxHYrMAsvlRlry5hLYEZF6yBX/+t6rhlqUhlIxCpA60o0ZKUo6fMXlWOajVHwOtJgxNnFWom29z5mPlWK1p+35uNbvm8ApjmrAAozALBGI4QQdCsZPd/lcsHxeCytN6Py+bdb2YnksD9ontR46TZgKLv5Bj/A6h8rE0K+dT6fcblcStPkNulv38uNybS0VHUDCEwOcE7SRfBq2gSLm6YTYlxgEAFRRgiA8wxh587IeQawwBHKtj81RcaImRGz5FoFevG4e3aPzWaLV69fiSA5J1CGZkE8ETo46XZpvhih1DWguEGNL55Z9kVqp0w/K64S6nsy31xKIk2bAqrdliTM2/WKzR8VLBKdX4TRHNqVaeS2dOwqslGPwGr7Sq2nqtxVcUqnvokeG0LAsJFGKl999RUOBynoPZ3OIBIhZGYE1YTXjX+NuWpcMbvneV5wPl904lLx4Voz+m2mvy1IFsDVwyGDc0TOETFOmJcLpumsEfGCZZlwmZ4AygJELwvm5VLglJxZkt3xDCCVe0pJUnPESavUGX3fYb/b4z/8D34DwQekmFamrQRrGoFmQV81KOK6+0nzKjw31VjJsE6qi83wCfPzbH4dCJ6k3b4pjQRGdkB2qF19PjagK5PXaqnV/Zkl9gDSt+AoVfJlmzqpj7SIiiBgZj8MipZ7Kdl3hBhnZO1Wk3PG4XDA09MJIUx4/vwZHh4eMI4jzqcn2CaqhqdlizhLrCH3Zo30cmYpJ9Oqq8vlolVGlQhgfmAbKFgNp3dOMU81USSCEuMCzhGcGJwSljihyx7e9Zh0DwE4Qt+NcBCWcEoLUpoETOYMRwuksYzc/+V0whJnnM5HpBRxPs74V3/0f4GgjNuYChaac4YDVZIkObBmKqwdVpnXq1cp/iASv4tR5tRaK3xwVNacrnNgMmGtghjE91nn1q6FTrAnjVhY9/5hi2YFjuAPL90EDEDtNQrZPY09NpsD7m5vMQwjLsuM9w/vkXLGJ59+hvfvHwAi9H0nHDTncHNzg2laFONKmCahN1cHPup4qFb1knjuhx5x0d4RKjjTtKy0twi8kAitiupau9v3AGUSN+bB0lYxJ8QpYro8wrsgKaJiKgnDsFUWBiDaXarGyRPAVKPjnHFJp6Lp3j+8xTevv8ai5IObm3t0nUecGJ4cnr98hq9ev8ayLNhtd3h+/wyAaPplMQ2tOx53EkzkmAqV6VoTdpYlIZS9Tm3Sc/21QGit9QDpM0MgoVISF2Ozx3gjYEbfgTU2YzXWBrKyheQf98eMi28T8+zZc3Rdj/P5iOPxiL/+6hcAOfgu4Ob2Bp9++jmCk16vRB1ev36N7ei1P6yQ9V6+fIl377/GwwPwySef4NWrV3Ba/SwTJyYzJcG1s5rCzThgmuQ+eq1If/fuXUHXY4x4enrCdjuu2nDKudYN8YBaLOxcQAiidUGyQx05h8vlgv3trQQ8FErjY/NJC5uV9Z4hzV66Qbt/L4SHhwe8ffsW59OxuBaAw+3tvfzOwOV8wU+Pv8Av//IvIeeML//6S3zxxRfFr5NOiFLyRyQt63W160a4lYdmwhKjtEEw2k8bQUIF6xoybTFUKYlRzX93s/1dGUiDN9aiYpGlhdak1dKwCEYPJt8KFBXMZ1kW3N7e4nA4YJ5nvH//Fo+P7zBNZ91BLcMHiWKmacKbr1/jm69f49nz5/js8+/i/vlL3N/f4XyZsN/LHuPLMgutRid6t9thmSccj6eGBSLVT1JjKdHaze0t+i6g70eM44C+78uOJrvdBtvtviD3IdQ+ILZji5lJg0uqj2d+3oQUF5xPZ5yfztjfHjBspPs2uaBwR4UOCkN4BQPVegPnPDabLW72NzI233xTtE6MC96+eY3Hp0f8+g9/CALw87/6KeKylM5MNn/y7xp4MqAUgO6TtabSk0aJDhD6NlWqVdF4RQquXKtWLojg7253vyurqva7al/i9KOGpGyRZ/slwDY8ELUrJ3/+/Dm89zgej0KlVoapIerEHrIbvPllGSDpqvz+3Xv84ou/xs3NLW5v7/D97/8Knp6OePPmG+z3e9ze3uLNmzdwzuFwOODx8QHHp0cAFfdaloiYhNj44sULbDeb0kzFKqeISO/N0HFB3NuyN6kNrUGFJdmtuFiizSymaZ5wOT6hGwcMmw3GYaM9MkYAwDBUvr6ZqJp0D/ChQz+MpY0Xs/bQyBnbzRan0xmXyxkpLXj7zddw5PGnf/ZnWOYJgQhxiSpQNTXYQhBsbo9GhYCg/+VedJ6dr5uieucKRFGAW2ZYjUURgkbI2pe/u93/rqyedfXRStAg2hbFwa0SLF8QA+207nAcBTh9eHhYtZgUkLLRs5pechCcDETIWbRmhvDT3r17gzdvXmO33+P166/xt//2b+L9+/f4+c9/gc8//xt48+YbTNOE73znu+iHAfM8oQsBUSMuRx7b7R53d8/lvDmV5sAh9LIBa4wYhk1heVg0ahvQy9bT1iutmlvr1gh4xOWCGCdMpzOICIfbW4AY47BD1w0AdI8AqjQh6WYk/pnTnYFJiZQWceeU8PjwWFrc397c4Hh8wu6ww7/+l/8nhk2P4/ERvMwy/w30wSzQhoHEAMpWkqapDI5iZu2zC+3ZYQTPCnkY+i/B6VUPOxNd4prHVIXkb9VctgUO1xgZgVbN8T6QVhU676XlwOVyFgpNA2xW5xk1RVFiYvnXey9bBqaInBk5OaVuM77++mv8+7/2a/jxj3+MEHr89m//J9hstrhcztjvDzgcDpqXlCXX9z36YcQ4bjEMG3RhEH9H6zKdCwidFaQEbLdbPD4+qhkL+uNLFkHoQVLXYH31rQRwnhekZcJymZAzsNlssd3foBtGOHLowohhFKyu7wfllAlOxiQNZUiZIilleCddlRat1np4/yBFzpTx9u07AIzj6YSf//T/w9t33wC5MmRSkw4yANY0EKEgTTVKbP1vM5VqvURIFRVQbr+Bx60gty8u5xKBJUfwdzdiLk3STSjag0v+Ch/aXtJeWnKO1ERj1Vdp4QOwJVBbTZjVxNae+CBfsZqc4X2P169f4wc/+AG+971fxqtXXyHGiB/+8IcAgMfHB9zf3yFn0g2+ZIOGvh9U4HoMg2xlaL0tQif5Sds7oBS0KHXIgoO+78E5SstR9c1MCFOKOD09ImuE67seruuw3W4Rl4i+7zAOW5CTfmyAdOsmLUiOyyxlgNY3TAFfy6kejwJZKNcBr776Ep0nMDu8e/s1Hsy/bQK3MuEqYLaHaPs+NRJR57LpIWfaFiTt7plLXrOVj7U2QwkUHItFYlKczAa43MS1c9b8+WEmwAnD0TVOY6MNeaW1Kqq2GhDVbjlG9OOAlBLmWfaChKZ7xrHHi08+wePDA4L3+M53v4dXr1/hJz/5CQ6HPe7v7/H49ASpiL/Fu3dvEEJfiIkAME1zacFgq3EYBgTl3u92O3z11VcAS/9aoxrlnLDMM4a+R6c9yELwWOYF87xgmU+4nM7YHu4Q1JSmVN2GJUWMWjXf9z2YHChYxEagUCnUrG3hjcNPRHh8fIR3DsfTAzwxTqcnzAvhZz/9aWlQcz13IiBcdpHhRqCcqrZrZdJOuL0t0Xku1gyt49/ITdFsmcEkgBexlOiHFgv62AXl/aS+mLYSV0VFJmSwOvGmSqi95aK91vvyWOK6njBimYGuG9EFj2WRSGoct9jtdgCkUPabb77B669f4/nze9DtAT/7+c9we7vH+/fvsN/f4pe/9yu4vb3Fw8MDcmZsNmPpvTHPF+x2O0yXC+YpYxxHbMaxMDFevHiBd9+8LR3AnXNwTEDM6Dah0L5TSjg+PuFyesTx/XskIgw5wvktuqEvnLigHblzivBezKdXDI5DJSfID5cKrr7vcTmfkeMCoozz6YTpeML5fBZe2c0O4zjgzdv3FUpgmYGiW4zsQVQ3sWV12htFUQpcCpuiTJGaylqVtI5R168SPGRB+rNQcOEPu/F3vy1H2fpRBAh8QfWzahahuNm1/2VmtNZB2rnqNdrftUdaFt56TAtSygofiAN7OZ/w+PgOD49vcZkuWJYF3/veL4m/0w04Xy44nU64ubnD3d0dmKW/xzhucDgcAAjKv91uxeHX9u2mgb33iIsQI512CUIWIuRm3OrnImTLdMbTu7c4PR2xORzQDSO22x2cC9jttzAAehgGhK5H1w8qfIqGs6XYjDJuboRM6nQ+43R8BIgQ5wVvv3mL4+lJ22wlvP36FY7HR0B3lGHmZlOPOk9mKhmAbwDkNqMBQBWFzXULrTSTVGxpmTU50jUWSv1izqIBPzCX13nHD/OQKMLVKiytM/rosbXesUGOG4HMmsMEaW8G4kJKDJ1TQU04Hh8RZ+my44LHOIxgZvziF7/A559/B8OwVSIi4auvvsTd3R0Oh5uSjzSn3ftQNNf5fC69a5kZXeiwHUe8fnzEsESkGMHBYdwMSDlhmhRwvUw4PT3izTffaKRlNO1Om/QBISgDl4FRgwdyTgiQKrgSUFiqS1nnIByfnvDmm6+Rk9RVztOEuMyYLhfkvofXtltOAc6cU3VFqve9mjNSv0o6kiu151vy1UV+PoKdrgROXR1OXHw9+7ZX31JKGr4FumgF68P3U0F4bYOAlS9/da5vy41a5AJyYAj6zSQ0mGVZMAySSzwenzBNE6Z5klrJrgcUUT6fLvji51/g8fEB2+0e4yjFwO/fP0gvjf0ejhzO5wne99hsdri5vUPX94gpYZ6WgoY7J/uhbzcb6f8xTcrrkwwCmBHnC85Pj3j36jWOTycE3b9J+GKyglOccJkuSJnRDaNwx7yHJ9UmQInWBRLZlGBlWRY8PDzgdHzCNF3w+PCA8/mIfuhwe3OAd8oMQU2d2XxbbziTASMxgrnuyKxJchOKVrNRnRxYE+MSbV5/JrgXYJT3bP4fo5X40Dri9nJF33C1Z8oIbWLOqpWMZt3CHrRuwtICj62WI9LmvoqT2Xcla5UxzxP6occyZ7x//4RxHOBzxvl4KoPQdUm1gPhgz5+9EBD4dMTbt2+x2Wyw2W0xXRjYRAMAACAASURBVC4oG7eGgMPhgMtlQlwV8wZsthvE+YA337zG5fSEzThimaUvRF5mnC8nPLx9g3dv3oDh0fUDuq6XHeugfiY5jJstNts9hn5Qs6T4l42Rc3CsewfA/LIFaVmwGUfE7Rbv378vhAKQUpFYC7tZuV0aHBUyQLP8GY2P1rzaiHL9gQpR6Vq8Ljppj9XZhyhmLgwduzJJT1LdLEK/fG0eqWBYH8aGBOl10VKJr7VVG0xYaC6pCl/OVo5TX8HaJdm1pumC8+UJtzfPMQxeOyGmwngNIWAcc9lmhpnx+PiIYeiVuDjgdD5jS2K6LpezbqMTVNBuC8Mh5yTtEzrBzY4PHZ7ev8fQD+jCVFb9Mk94enjA+XTC7fPP4EIPBimo24sPuNlisztgs9kJ6OoJHLk8t9RpxtU4pZRwuVyQkyT/j8cT5nlG13d4fDzhyy+/xKBa03mUBoHWcrNObitEXAMDVN+zgGfG5bcx/wiyYNmD1haywhnm4ki9wjo1VdpfiEpdS3WVRhR6bRHs2nAfNTAkLW9fMxY+uFtYiVUd2MwMjgui9aJXsxUzMC8iTM4TLpczDocbEHV4eHgsLAnZFifCeunnzFiWGXFZcLhx8JsNuhBwOYtwySCydgZCCQxiipLO0Q6PNGTc3Nzh8eEBp8cnBN/p5BIulxNOxyeAgXG7xW5/UC0mK3oYR+wPtwjdRuGRiJRzSUY776/GqI22xUIsy4J5nqSvyDLjcr6As+R3N5sNOGfM86y+5GW1yLMKV3HdCYahoiqJxkyCm62eGmFqcTRu5KMIIn8ok00QZf8Gs9lESpVlgB2tJF5sNRVzVhx/mOBLSPOtDmR5iWSnXBO4bR7UFpNUMdamcd57pLjg4eE9AJaO0OOI4/EowGgnLZus96xp5fPppJx/MWeyJU3Gu7cPSCljt9tjWWYMvbA+mDM4ZilS6Trs9gfc3z/H5XxGill3skt4eP8el9MJN7fPcHt3i8PNDTYHiSqHccTh9g59LzlKzhGcZ+TM6Met7lQs+F8hHTT7VHnvcdKO313f43x6QpoXbDcbTJctlkUYvMu8aL3DiCd6Wk2+gbBOhaf4XwUjs+9qNOmoNrMoNcIqld+OWKBVba1glU/1mWRrH3PAnROMY/VFO9laWrFahR8PGGr0KL5OXR65nIP4Y6V0DsxRHFmy0jVhORARNhs5/2effYZpWtCFXqnbWvy7LBIxdh3gCDFlDIMrxSVm2t69eyvbTjOw324hIYwFNBk3fY+UIv76Zz8FUsKyJCyXC9K8YLfd4/75C2wON3ChwzBssNvtBMLwQTRJToXLXFoneF/6skI3tyAXVj1WrUp+vlwwT7IRxPH4iGm66BaFsrNeSgmn8wmWQpLuPS2o3sAKjda69pGLWInGWWk8RnVlrrVWUQp2jvK9KnhiLhthMA21NnltUFpvuNwaMYQRK0vm2qmvEi59tYr6BqTB2gcQCZV/QeLLSKX4IhgTEZ6ennBW87fbHXC4uVGz1+Ozzz/HZZImfpvNBptxRPBek9wD5mWBCx6sfo8g6+LU932HtETs9htpn5AZ/dhjv93h8viAaZkxn89gIuwOdxj3B/TjCOc9NpsdtrsbDP0oUSgyEjJ8FzTzoJVHQuQv40zEpfLbWnfJziUJ8zSLWZym4oeacMQ54u7uFm/evpLNOJpqMzQTfQ0rtfNSP7gSngKGXn2X1n53y/m341pGsb3CCphrfK+aZFVHEusGHVbQK8G9L8LxMZ+sOvK1Jx998J32Exsg2SHEOhXKKg+lt6xRffphwG/8xm9KS08VqNNJnGbvHPrdTgHRDqQFuDGKtksp4fh0xOl0xN3NrWi2mOGD7Iw7jjvc3j3H49u3eP/mLZgZh2fPsdlusRm3GIcNbm5vG0eeYWVyFsVaQxZQU7CDjMiyQSxUe/kQMC8LLtMFzjt0Qfb1bNtNzfMs/Lkl4hdffLESMEAS0wlqlRiyHQ7MfFbXZJ2K0oQ2Sb2SkT9bX5FJG01cuUQlnWRRKK+VSxGyazt6XYMpDrIMhpU7WSQi3fnwwWpoz/eBr9b4eoABkKbUuFRXmyEVOo3DMI5gRmFFbDYbjOOIt2/e4E/+5N/g/v4Znj1/gc1mi77rpJo7yF7eMWc8vnsHZsb9/T3GzYicMtIiG4TFeSnpk5xZ0ki+wxICdre3uHv+AvPxCegCdocbbA57RN2CL3SdBkTaToAIXd9LwxbloVnKhyATb8lmWKUVhHXBWgaYYsS8SG40KQO3FY6n0xHsMvreYZ4SUmwCMX1lw69MWaj/nJsArQhDgSFaa6IChuqft/ewNo3Q7bWpLKamdZRNtNOaOl4JRjV1GkGi+lAfN68fvlYarvy/lUqbIIs8paOOlK9JC9AUxVxst9uyEG5vb/Hll69wOOzw5u03SAkYxy1ubm4xaDt2R4RpmmRPJ23vdLlccHNzh6H3mBSpzxvRymJWR0H6M2PoByycsb3Z4/nLl5gzY9jvEboRz+6fY9xu4IMkxUMIGDf9mvVK1YTIaqojaC3MM0To5un/r+zNenVL0juvX8Qa32lPZ8p0le02nS64cKsNtGU3FgIkREstcYXgAyDRXHKHEHf+BH0D4o4b+gsg07gFTXULWYBayCqpJzurypWZdTJPnmkP77imiODiiYgVa+13n2pWaufZ+33XECviiWd+/k/H9u6B5nDkl998BRjKqhSC82k/19fX9H1P2xzJnKXte0LTL2lvLRwrEPE5IJXUzTDqalMCmmoyaqLfPVpbPBaaziYMKxz5/OT5YYxJ0kQSjpf+HWX2YwPhLBdjSpDOBYMgiMxgfYXdZpJOHJIbb4zlzZu3rJbrWBjS911Miy7LmsvLC19tnVFWC5xzPqt0wPQtzcGQ5YrSux9QAsacKYk3aiRp8LjTLFcbTvUOrGNzsaG+WLParFksF6xWS+lc54H6qqJEZWNhTnizUuuY9pw2KnNGwGSkQUbGYAeWqwXGdBwOe3b7PcuyIluvORwOrNYLTqdj7EkV1k6YgOTyj41l07lnoqifk2LnfpeVdskyj7lk0YCDWNmUujEAcu0jQs66R7V4gM/eVF6pZyTnWeOe+Y5IBzPfNfiJP8/5ZKdrrSl0Rms6yWD1le7WSieSH/7wN3j37h0oR98bfvSvfcHLl5/zwx/+kBcvXuJ8ZdJgpCQs89DnNzfPGAbD/e17Pt6+ZbO5gqTuUynBEBsRtMVvtlytOC2XdMagc2nVkxeFbxNURk5WFAUOqTay1pJriYTkfoPkWkdL0jqpa0QJV+v7jvuHW47HA0VVUZiCTBfs9we2D/e4TLFer+n6XohyGHwlVAx+ECu17BhLTu33YC3OIQjmR6o2zZd7GsOWI4SuQnFySmh59IeoKeeZcqFQnW2T8yKOMqHaPu2fGAb5iMBScpqJZsAnERKhDJTKefn8BR8/fvRVQTl/8Ad/wHffvuN3f/evc319xTBYvvnla06nryTPfjBsLi6FGyV1k8ZblHW9oFosePbspcQuffWOzjPysqQuSwrfzMKUBRlw2m3JMs2qrtlcXXNxc8NqvY5iOGTQWmulX4DOyHUGHh0xJELawUSrDW9sGWPpO9HD+rbjdNhz+/GjiErTczjsWGUllOIGMdbSNEfKouRopKKe1FtqR9AV0aG9mLaPiWO+HoH4JlzwzPrNC1VSuglHSE/P5yc/JgJkgMHiiGatjoO3aoQnEkIZ+w89RWDTz9Jyee3hRxUGiTN++PDOE3fJfn/kH/7DH7NcLrm9+0CWKYqi5NVnn/Pq1WcY4zjsd6wvLlmt1pRlQdM0UR9zzvkM2SVtfcJpJ8Ho45FSZ1S6BONwOijEmtK3AmyXK4p6wc3z5yw3F+RFQV0vYkldSMdGCzcUi3z0fYleliDm+Dl2dkTuaU4NXduSaU1zPCIOZMNAT60KMRqM4XRqafuOXOcYF/qCTudXQeyxeW5t5wSSisG56JzTxzxZ8ZGO598HQKdUe84SVKSWoIJAXMqnrITU6zOyPbVQ0pebvuT0ZUPSnrOW0ne8zXJHlkm/oT/8w7/JZrMi8+Gd+/v7WC52f3/H9c0NVzc3XF5eij/MjRCeFxcXaK05nU589/23HPsTTom1ul6vOTYN2+29RBl8TySVa7QCMxhy72Kxvp4zizUCIziK1prSuy10lpElKkLqQ0o3oRkMh/2et2/fMgw9i8WC5Uq6rzSHI9pHSA67Pc5ZTu0B5STc1tthann62G/0dSV4I/M1CFZlkIVxbUB6TgW/mH2sr8+Nw3RN5+udi06gAjlNBqCUio7CAHHpxJ8xmrUxAHEeyecp1jxR9JNBhoJXrXIUUOUF7WAZBsswtHz55ZcAscgjtk7Mc9qm4fbje+E6bcNiWUvPJSftdLqu4+JiTdu2PHv2bNQZ8pIsK1guxcemihLjIFeKHMXgwDUd9thQ+nTt3BNWURSx74D21qwjVJiPLpgsWZBAbMHBut/v+f77N+x3DxyOO5rTSWARnMNYwewoi4KLiwtUnnF3dyfwDVYiBsZ65Tuuj8ywVWA8xwx95Z1z+BqPGBbM3GOGYHGTZl4qaFVKPTp3Qi9naCCPH7jHOllKjeEn7lil4iDP0dGc9U5dGFMCnLhRvP6nfAMr0anwhF6w2+1YrVYR4Uc62hpevHhBVQoBBFA9MwzoRY32pW1lWVJVBXVd03uId0HgGWICZN/3FLnGdA6nMzCWw3bL9v1bOBzIVmsKB4VP2VFuzAsLPin8wtqAXasVQ99H5Tms3dAPHI9H3r9/x/v37zgeDz4w3rHf79FWesIHp7H058zZb3eiR6bZF951YZy0pkaBco5cJdIoegOShfDjDR878HBRTzOIc6JxLmZT+gl965hzshHekrgzzpbD4fOGmA7qnI434XDxtcZ7OxeUT+dLxRR5UVHa0vvFLri7u4sYrCsPvb5er2iO0mOgLEva04n3797yKvuc5aJmdbEi1BcEcZyXBUM/YJRglEkrRMtiUeOsoR96MqVo7h94+9Of0334SK2A0wnbdth+IC8LcTzm4nS1IAgODt8QQ/kExgFjXKSu0AvUDIa7+3t5p+bEYXfPw919zOI1w8But6UoCxSOsq5ojG++oRSD6Udry1p6z2msCuVuoD1wxcRfpqa/Kq1wxiVR6LG+EiX1Acq5uEmeUn/Cv2lIyTlHHqLykZjCQIPXn2AUT8lCB9tZ7nTWJfG0qPSXJUaRPM9jrwZvsXMonyUb8+U9t3LW0p4alstltOyur68pfBnbfrdjsVhwsbmiH6Cuc99lxaMPaU1V1oQmB8MgsO1d00hP877l+PCR73/xNd/85Cd88dmvoxcVtmno7u4oS+k5pWrjLUopGraDla54ynn4+VZqSK24Q3pjpAMwjqbpuL/9yMPdB+7ubrm7vaNrW9FFfU+jxXKBRkWYBG0MJx+bdV6UKKVQTlE4FcFdrPJoh747n/J6VQRLUS627fZTP6o8fqMov7ZPKjyJrhbpIqnfDTQQnbEST5vKWLGAxM80fuW8pfdJ+nnC4pCAcDooYeJjLFPr0C88yHkxhbvOt4EeBjItNY3VqqI9nsA4qucBhl2ILNPCCZrTnrzM0CzIsoLGDDhr0XlOrivQCmOFIxhrMV3L/v6B1z/7c777+c/45l/+OVdlQffsFawWHLb3/Pk/+wmf/9Zv8a///u+zXn/h1QUl/Q36AYWjGzoO2w90hwN9e2LoWtr+xLHt6TsweYF1mcRY25Zh6NBKewirls16zWl/jKB2ocNJ6Gh89HMUw2BBSQ/uKBi5GW4Ui8F9khhcIfXcr7y4Vmbr+ZRPMyxkKAJOXR+B2CKRSW3dWLFCGGSkKJU4a30fxJmiN1cK3YTTBVYZ7i6/T4PlAelwfGHnBJs/t+IiCJ2Gy7Kk6zrZ9Ywm86k5CkexFodB4WiPDR+GlmOz49Q2KKe5efaCerWiqteU1YJM5VjlaE4Nr7/+BW++/gqz2/JsWVHrgsV6Q1Ev6bqGm3XNzbKkcAZlvb7jdaauazgdHth/+I6vfvrPefv65wxdR6nFD9cNhu19z15VXL38jKpacmxa0bH8a5dFQds0NK0Uuzy7vqEoC6qy4nZ7R0jsVEpJFoyfVitMC+20XypR3qUWUs4LQMIZTPLawpqFloapz+wphT6sYkqIqRsr/J10JJheEW/qd8aoFap48lxEprI6wBr5p8ddNz7usSHg6dFbsp5YrSHLxXrsu4GiyH0mhnjwVZbhkGzYh4eeqpGyuPVyydCvcMby7Vdf8v3rn3F/+5E6L8mLmvXNcy6ubsgXG158/htcbS558fnnODLWV1fkyyXqeKS8vuFqfclivSErS26uLtlcrqmvLikuNlAW0tXXCYG9f/stb776kq/+2T+hKhSHu/cYNLbQoDKyIsPSsb3b0fSW569eMRjHbrtDKQQxsqromobr62vJji0KirxAQXQeC7wm3sEqs6gT5V5Eo5suGaPHQgjAT7pYVnJlUiAUCOdTDvWUbp4KUSXW5RPyz4k+NIrIVPzJS+mzrBRv4Xj0+MCqHw1i5HwqIcYoLhHlucgrwYhA0o5TjIosz2makw8LwWJZo5SjbY58ePNL7r7/Baf9vSzIckG3u+f97TuGAXRVU29e8vz5c/7w3/sPuLx6QbVYcvXiJae2ZXl5yavnr6g3S8o8p6wqLl7esLy8pt5comKPT8f+7o7/+x/9mPtv/4JV3tPj2D7cs3U5L68vWOYa6zS950RSFOIwfUcWikS04nQ80DYNRR/CVFLy1rYNQ98LIUS/WyAiyYgJmrPoack6IPqZTop3A9/IRCny9/CEOGvIla7RxGsAk9K7sLYp55u0vXnCAIkefO/pmBGhGndL8pDRZeEf6AloSswjoU10AiS8lOdhJhy51gxeHJSFpDZvNpuIsBMcocvFAuUcx/2e+7tb7t/+kuH0gOl7UBn7U0OeaapyiR0M7z9+oFcfePtmTVbk/Lv//t+iWix59vIV275noTOWVy+oNitKpSiLnGq1obq4olhtkmb3ho/v3vJPf/JnPKsadKE5tSfA8ZffvGO1XJJZi3KKbHGJbnbUVU17bNnuH6ICZYaeoe0EKVtrbGYpioyu7yT12hqs95HNJcPcMy+czbua/H94MZhiwdqg6CeGX7DKIgdL3R6MTEL4znlPQrh//ujbuOwpWw0P0zEhb/pC8t0505ZI0fHps6fM2bDzInPMRyJ0XysqyZQ1+OokAcALfipJ8jOcjh2H3Z43b14zHB/A9FjTg87YnlqeX12Q5QNt39O0LftW0K9/9uVf8G//3t9kWT9jc3mNdlBnOav1FYvlgiLPqIqccrmgvLgkq2uUIjYsNdZgzMB2t6dVlrY5Ua4qPm73FMsL8rqmMxl1dcnpzS2XVxX7/c6rBNLwwjqL04hhUo6Ae23TgBO9VKvpRo6LHtZJxw9jvv+o47sggoQBMNKPcqPREKIxcU1mrqrJ+s4419ytMRWXgYDTgcNYocSY7hEGoVRy/ZnDYWM9XuBqU3Y6Eu04ZpU8Q9oWt23L1WpDUeQcT0dBmvaWIkqyHJxzDH3H6XBgt9tyOh0ZmhbTd4BFF3Boe1a9Iy8Np94wDNB0bayLPOweuLm6pqqXVK8WVGXO5fqCuhKIzyLPycuCoq7QRRGbIjgcg3PcPH/Jm5+9IaOnOx3RTcvQD1xefY7OV2AVXSOo24fDnsP+QFmXaJ+eXtcLqrLEOseiqqXiXXt91DeISGHgx4UNKrOQTnAJCYcb1zVwtmjgxTK2QGzeGIg463IVmpgMPZV8KjUyx3VP1jQK3pRuxxMt3jyJBubIrbxnPgEcPhdITTx8BJ/LSMf2EdU/PuQNRqTGnLIsKDPpgdQ2jVRma03Xtmwfttze3nI8nei6jlPT0nQDx7anawesBWM11y//Kn/1R7/DyVf9hPEfDwf6XtrLlFVFsVigqoJ8WVOuFyI2lwtUnkVQXwm8Gy6ur/mtL36b3Wng9qFh1zje3+7ZLNcYu6AbNDrP+Bf/4p9LZfhux2B6SSfKcuqyRuc5WVmyXC0pFzXlopZMhqwg05mU3qm5g4EJR5p6B8YlkJ9QKebAWd9dzsYTx/WwnoEkBPMkIxmjOqlXIfw+0ckeXaxAjfTtjYCUy4SM/PPFI/N7OxUyD8K0pOVwakLA49SJ/y4tpLi4uIBemmEFDNmmOXI8njjs9zzcP3BqTnTdUeAtLTgn7atzrSmqFVcvfoOvv/qXDL4Bl1QBCdpi057IckXet1i3EqyyqqIMfjUkqE0AkvESaL1Z84Pf/E1WF9e8+e6AHQaavqOocv7ff/J/cXF1zXZ3z4d3b3nx4jnG9CyXK6q6pioFsM9lCucMmfK6sBKITmsMPY6ri0vpw0Q7yQmLxOXG+SRwncDFZht6HvKLn3mwlehUDT/ApHhEefGaOGXPMZpHOf7pMQYX3LjcbmZlBMvlE97Z1B0SPGaBiFLxKIPUj69VjqIqYk7YsqjReeEhB3Lvn+o4Hg/s9jt2+x1t12LtIP4/J4HiwTiyQnBZnev4xS9+gbMpFlnO8bTncNhirbSRdk6gp6y1WGPpkHQeB2CtVE8jISJjHZurK/7N3/t3ePcnf8zDfsdgLE174KuvfkqVS4/Oq6tLtEggiVhUNeViSb1Y4pwR67LtMNaPvR9oTidQisVqSV5WcNqPk2SdePbjShHHKPaEiwq9bxcRvaHzdUuV/LjGEIt/z+ndKTz/OTdHhI6aU3QgiuQPEY2JwyWYsFEvmw+WVEENiY+ph8w9Oi+8WIh/iYsiA1VEULk8LyiritVy5Yt2HW0res7Dwz1tJ5gXkufu6xKCra8y2q7lm69/QXs80htDXS/ZXF5y/eyGV69eje4TpSgrzWK5IityjDUS4/SIPDjJzbdGiLkfGozrqcslD9t7tvs9u8MRO0jYqm97bm5uKOsKnedcXV2xXK1QOqOsakDRtCdOxwPHw5Guk35Kdhjo2obe9KAVD4c9p/1WjATnXUhq9IuJ33yiXHsCJH6vZmsUFzVeor0hQTQS1BMO2XPrnh4T6zKlxDGsNPKywL4nro6E4pk9IFB3QDYcX2M62OnA5GmRVTsFThpjBeJru45FvYhJgsPQczgc6DoJzWjlkQt9y2iHKMYWx9D3NO39BIF6sV5Q+CxaEb0niiJnuZTuKCISBvrW0Fmpz6SqIv6G9UaItZY8K7h4fsVf/xt/gzfff49SOafjMaYjkUke3mK5JMsLcMSiE+csQ9czdD12MFFfE2IvaQ8HHu7u2axW3GstWLH4KQpiUYW1cnFzpzOrAWeJmCOTyAwuKEbEcFW6nv7fR7UDTK3dR66UlBjm0fN04RUjmJrczDfkenTu9MG5N83l0qAfTF9+dCw6nJM6S2nLF86T9jjGGJq24XQ88u7dO7KiIFMZRV7FFyyKQpL+PAZGABpGZbFeU6kxmbEoClbLjU8ZGhiGjiyTFoNaCSS8wvrGKwJUZ8xAN3T0wwDGd4/z6UhlKbj9Lz//NX733/o9lssl5CNi9sXlBZuLCxGTZYFzoogPduB4OtCcBBo0NCoz1sYClaIoMG3HerkYlQ7fv8jqqSRK8igeOUqzbIwznuM8aQzyXOb0nKDOZeekBDy6MM4Qixq19DhglcQvo+3qH2rPKPwBEnPU40ZuNuec4fcMcE7gBFACjtIPrZTzH0xMFvzyyy9R2lEvqli+b62hyHP6oZfmWn0vldiZJsvEaSsWoUzgarWiqguqsuDyUpTq1WrDZnPBcn3Bol5RVQt0kWOHHtf3GOvIbQ7OYLzrQQizBJWByuj6nl/74Q/44kc/4rPdZ+RagJaVUtS1QBo458jKjLzIGdqO7nhkaAXmvTc9RZFTlxVt30XO2w49l3lNWVUYO/hGrYmbIir5BkXA3ZiQ0CMjMV2zKWd7QmcjKPvT6+bFKZPY5VPBT6/d+oGHfth4fSwMZprZel4EPn2kFo7zD7BSCeHrFJVPw5FMU+erk6QgpCYvffWPESW6riq6tqMfOvI8k8yMvIgoOYOxZLmmLHNWq3XkMPv9js9/7Qes12vBtFivqJfLCE6sM4E2CKVhEu5CeiIhWR86K+J7V/UClRU8f/EqjknmRfSdYRioFzVVXcu4OmnqKmVyVuo5M82paWjbVoD8tJaEy75js7mgaY7e+iNo+H4+z0kWn+WctBdMCeT/13oF8lJT+o26qjfwHjtjOUMkDpQSgLbw/Xzgkwck7ou5eZyGmeS64JtKXlRrSfsNRO/1vNDjUfn43tBL5Y9kLuRgQZfyLBGVS5yzNL4V4PF45Oj1oiDe63rpRYeAm2w2G3GPXF2yubhitZKayrquKYoctKZXCrTg2o7IkFb6JumMoqjiGOt6xdXNMz6+f8sGadwqRcaZj1BI5XjXtrSNYJ9prSXbohbxfzqdUL4KPaAbZVnG6dSwWl3y7t2bZO2IndqCeiMAeTZahUoprEl0pYTzpFXlT3kcxs+DZTqSQPzOeR9bNsLW5yQcZE6xstBIOpmz0jwUJv4ylMch9bLbJlwxWJ9hYeUaF38CgcV7Od9VNtEDskwqyPO8RFmHGwRV0RjDYlFTVxXKCcla71Iwg2C9VtWCqlzgHBG//9QELC9fwFuMXeaur6+Fs1UVq81FrCYvcmlaX+TSogetsUrgMTWC25rlBQEfTSlBD8p0QVEtIlz6crn0fcSFk+VFgfE6nrNORH5dg5Kw2cXVJaYf2O+k3eJ2uyXPc/YPD1z7lkIxhjmz8FPfpdc6HonJqHdlCu3xXefEFfSzVDcLKk9IOp2cHwjOSphNa00eAGoVko+Uzah3EmaC2HRTCGn0M9sZ55rL5jjAcJ+gt87keJZpD1ziLS7rdQqd49Tgi2F9gwrlGAbn/VkCP3A6nVgul9w8f+Yry1uKU0me5RwOB8q8IisylptNxD8zFnRWsNlcsllfslpvyFVGmRViyuuMDCjz53l7wQAAIABJREFUQtB+nOB8WW1RWUbhe4fj58ci3PfofVl5VlAuar59/YbNZoN1jmW1oD2esE5qLkPfdGsceVWiy1Kq4ruO9cUldjDU1YLtfkdeFrHxxHyO07l3sR3J4yNVzK0JEuX8eROnb8IQzqpZSvhoyhnzwGVCC+GQPzYONHU1jNQclfiZ8zQd/NTJmrhE1LgF0oE653zrGnF5hE5zee4tmKwkd6GeUDGiCSnv1ReCK8uS0+kUA+fPX7ygPTWoTMeMja43bC4v0VrzzetfioGSZdS19EJaLBYUZSkdSLSEsIwZpH+UcmIVKk2eFRD7hWsvtyzN8ch+v4/Pe//+jViaKIq6Fl2ulWKSFLu/qhcs12ssju2wZX15QaYc94c70GJhCiRpEUW9c85XlT1WfZ7St6OijuMRBJRSE+711D2fcsIaZyff57gRFVkx6kfzgSnGbIPg01HRIJi+4LmXii9HqslNj+CykEEHApRcKqUyikIQpofBsqxFDFk7oJRD5QoGwZYIxbyC2y/+MV3kXK+W0i3DKUrvcmjbFuVguV6R5wncQFVRFYXgS1iLynMPGep7USpJQsxVJn2RUJKRiuF0OPDhwzse7m65vrmk8QmItx/vpa6Smq4bMB53Lcu075xiY/isWtQ8f/4CZy0fPn7k1ec/oDmeyPMtt7cf6bqOYRir+IMMC3pRIPhzHC79W5DLg/ya6tOTyM4jo+4x25tbmBHVZ/JARawhHIlLRasuqO7yIDxBTCuJ5w8NfPhcOUI6qJGww3fhRTxKo8DGAErMf2NZrzeAELu1BquUoP8MI1ic1pq8KLh5tqFtpJtJWQrM+nq95uc//5k4MwdDXVcitvJM+gkpyDOpotJZRuaEyzonuK9RT8GJnuYMtu85Hk98eP+eTDnubu99H6aSly9fev3Pcv9wJx3lnCMri8Rnt6Jc1BR1BUh2yW9dXrHb7rh8vuJweJCZtM6rrh4O1LfGdgGWKi6Fm+hT6dwrJRvDBoh8v7zKK1xzhhOiC7Em/ZxK5C3MlBLy1H8fW/oyFvY6Z6V6RQVROorO0dIYfS9nOVjQ1ZRGYRgLgRUuKn1T5+94r8cW7DAMLBYeCHggpr9Y6wTa0/uP2rZjtVqz2VyKTja03Nw8RyklZXDDwGq1xNiOspQGErkvb9NKyuek9jPzSX4+8yRTqDyj8NCczovIIO6lS55Yg0PbMpiBqq6iCNoftr5MTlKVClt4rIyCopTwWblYUJU1CrEyF4sVg7Uc90cPU3DyhC4IR87ZEAAIU+43+MhRUm6kUWCsz4iN7k7PE9RkvsNmVSpZ/4Ro09y/tBW2iuLSk905mY0SBD6HpPpor4elOltKHOdcIZPzwps8ITSjv8yOkYTUAnVO+TaF4gWv6xqXZThjqarST/RA73H5Q3PV3XbrcSuqGOsUF8GBsqxEr7GK6+sb4Xw+rSZ0oRE4euXLy4iuDxWrxIXItV8YB2JFOst2u+X6+tqLMem00pxO4MNFfouitNRjtn1PYSwlkqCIEzVltVrG/qHKeagp5yMReFcFespd3WPHa0qFYRXiPo9LMg0pzUXndF29RGJa7ebCekPM4HlSFwsma/g5M9on3mJOuMFtkTGl56nzMFig6fVKST78crmMVlXwfwWOkfrbAjaFlNf1HhLUxy8TAlVKMVjLan3J5vLKg6eU/p00QTyPKeSjdR2yhjMtYTPj+5y3fUffdRwPR+7u7iOcZ5aJx785nQTJZzBSmheIwoY+7tJnqe86mqalHwYRn0VO25xEF9aawVfAj2G68f3nBDb3f0UHbqYEdUiNgDCfwkVJ12jyuwdPDJAWkeNpD5sVrUcXFjwRVXGUMtUxOS15gFJTCv7VhDYPrsakk1FP0KNV69wIZeScpFsHQsmyjDIv6LqWLCsk7WfQOCuhlqIssdYx9ANDPrBab1j4Rl6996NdXIjP7dWrz6gqARkehkHgCooc64SDZglHj1htns05JyLc9J1HCDrgzEBZL7i5usRaK4DDnnv1fU/fCWhf7v10fS/PVM0p4vxLM9icIstoTtLyxxnDqW1Y5LlYurR+A+iR65OypunaZJmOXCp4MZUWnXYknKlUCrcI6zy3Ji3SMzTUe4y+Mq/4Kzw8AL4IdEYSQQ1TSMcvZTX4pqg6dV+4x6QWZPnsw/MkmRC1vIRwEeFGMtjj8YDWGZeXl2PQOMvFI48QoxRbSDfcoiglqK+170ISxOSJi6tL8iIXzAmdcXlxjXHSEXcYBvqupapr6sWSIpdAtgvjRGFdQEk3dE3LyfdB+v77N+y29yyWNb0v22vblvv7e5/BmzFYS9O19G1LboI3Pydv2iiOcaGOoaRrWprmRNe2klqEo+9ajBmt+nNKfdCV030exLNW05rKiGjrRq4NoiYEl0RM0grcypfjOZy05skyjDUxTTs8dlZIMqXUeDOXdtcJIjWR9yqw4PN+mblpO6U5L4aYsXNSbjbqZql7QmuNygVSyjkwRvxkg8nJs4xlvaCuKgY7UFYFh+MJa6SnZd91DMNAURR89tln4qsyA8fdFlNVNFpTVRVFWXrdR4g+05nMhzc2zGDY77Yc9zuatuX+/pb2JIS6qGvatvGcVwyLvhVcscEa+mEQ562xDF2HKXJsprF5gaksxlppj22ddAg+Hr2eJvFbrRXDzMKb6ExKalgneo6b/APMQoLB6PMKqQ26KL4uIL2XaP++nXUoNgnPG5lJPlnURHk7G7sKBBb/8j6ZWCowdUmcUxTHjo4TSpvoFlFsh+u1Bie9ubVSNE1D5qHU21NDUYRuuZIZUngHap7nLFaSbaq0ou8Ni8UiwpQvlkuf0p2R5wX73Y7Oh4CsNTx7/oLr568kZ0yNprkDqUxqLUPXcn93R9OcOB6lOMQO0kJHedSgtm3JskyI8XDAOsNgZMdL+EnT9y1dN7aQxjnv7pAGEf3QowgNvBTGGrKsGOdMpYW8yttZdmJvRcnkVaOg+Y8MIRGRCk8oclJqTYb11cEsJYhH76hKfWskAXJhmdLRQk9JYLyxynz5uwKnUWrw33oUwVmAbO5jic/RiiwdnH/haeUynigF91WmMyPPcuEuHhag8EB3YiGK78gOYBgoi4Le9AI1VVRYqwhhkrZtWdYluZPelA7D7e1HmVCluLhY45xYiWVf43VpjJXuwYICJG0Jj8eDIFS3HW3Txo1iuo7jYYsxA6dTw8PDPYfDDq0zBp+5myl88yuH8cAtxhhMP9B1LfUik9BT29A3TSzoLYqCQXmDRo2iLDzb+6N8LBUBx2PUvUJE0PrMjMC9wvIF6ZP5TSXK/ZQeLGMFWvJF4jtQI5HNyAnOEJko/BJOEU4TKlymTevPhZPmv0flcHJ+2CVyRtgF2jsZwzlSOV5EKy/PC5QX17I4HZ1pWS6X5IVMUVUt6LoG5wx9P7BaLSiKjLY5YbtBHKk2KNagdc7zFzdYaySdRiuqaikcxPWYQZqc9l1H355om5amldbNpu942G/puxbtHB9v31EUJU3TYEwffWh5XtD1Q/Aa4nyYqG0bdJ6j0BRFjnOGrrPs9wdpcpbnDMZQLxaYZojoR37WHhFL1EXma+OIoMjJxMd5njvk01qOqE6hJtelngEcETXokU6WFiKknweymHwe6TE8+HHxZ6rfhahC9P6nFuzsBcc0msSBqMUyK+vKZ2Es/LkiFoZh4Hg4kOca56SUTKHo+5ZTs+X27o7nz19we/eRqytpqtq1Ldv9lt3unsViyXK5pioXHI8n6vpItX2g71oWyw5rxL1gnaVrWo6HPcfT0ddP7mjbA19//XP2+61EHbDsDlvKYkGRFxwPB8DRtkcGU5CpgsFI92DnJH2pbzuUOuGsIysynJbwldKK9WbNYb/HOUdRlZitB/KzAiwzEpishYs72YtOiGtkgkhUkrsf0rGt8TpyoCq/9lrpGEmIDe9tYATTDI40RAlJtVIUZ1ZjNeiUSPzgXOKCcASFPdQhz8zelESTHRSsjriJVBoHnVwUrd5QpZTirIYWMFYPOGdojgfJO9M6wkc5Z6nrBUPf8O03X6OznNevv6KqJPFP5xrbdrx9+8ZzuBXmqsOuLtjvdxSlpPjUdc397R3ff/+Gqq64uBC3hOl77u8+cDwcuL+/ZX/c8v79G7ruJK1rEIeqXThaEJdFnmPsIBhmmXDiYRCv/TBY+sFQOCdgxdbiBoPKJH37/tTSto3g2TqJ6RoTeaFHDNfR7xj2sMCfK088YT29cTi3Sj1CQFDbwrqFXqLES0b6GIZh6kWwI0dzzo0dScZD/EEq4TbS4Iq4S7R3so1iTuR+GrkPx5yInwRnSYjTWhtjg4ELBujN1WoTuVjww/SdQGWuqhrtu4OE3P48z/jlN1/z7u171qs1h9MRlUnLHJRFO9jfP2CBrjtKxGAYMEo4Tp6XPtPB8u23v6QsS549e0ldS3Ow7f0tD/d3nE4H7h7uOOz29EPDMPTRUYxzHE8Hqqqi78Xo2O124KBciPO378VAKMvSB80zhr7HlUJ4+8Oe4/EoE5Up8lISMNEZDJKxq9Bjy6C5GhIozo1xSqle8MZYYmGm6zH6yZJqdESPdHZMuz7XYjz8/mSOvw3mrIeknLs1VLRm0nJ4dXagE0JidFeMg5kSZXp2+Lvve1bLDafTifXFhqZpKPOCssjoMYClaVv0oDHWsLnYcH1zydt33/H9999y/eyG3XbHZrPGOoNSjv12i8XQ9SIKu1646ul0ZLd/4OOH91hrOJ4O4CRjt+97vnv9NZvNpaA62oGH+zu6rqNrjlJNPvR0doiTH5TtsixjaKiqKu+E7ajyGrxDV5zAhedUjlNzoh0G30JRCppRmmpRe67uCcovuohJO5lPHVwSPgN1DAPJEbwcc33aU2dUoazPssVzRq3URLJEqzRIIGfI0NMucXLypM5lItfD5wK2m4MyOBfUu9F8TY9JBCHcM7g+xOyJ7FdcB36CnFQVye/yAm3fcXWxIDQc1Sj22x3WDgymR1pWi2iRziM9b99+z3q9Ybu7Y7/fcTzsyPKMuipomoN/d7EujYNTc2AYOpr26EFQPMYrTmCwnEVhGdqj73oyeMeoQdkEuz4Uz/gkg2EYIoGlDRm6vkGhqCspLGm7jrquxc0ClFWBVRJGW9RLmuaE1orVciUWvg5QEsR1JOFgo6Iv5rFkbPh5tWJtat+2O9KVcyLyAtPzPkrtjYigOkke2tPxHgdYNROXMYnN+/e1ClEAX4AQsIW8LFcJ1aYWzORBiesiOurGLO5EJ9CR0MYvfXhNKd9yz7E/HLnYaGkkUZaUdUHXWDBSwaS05uLyksViwffvBLZ86ASKabO5wjnY7ba0zX4U/fFQo2thGHBNurPDFnMe+txKMa0b431hUsOGxHlHpd/AwzAImMrge2nmgtajnODXSi6cJCQWuRS3OAVlLbUGImIztMpB+bRxI6qNsUOsaEqXYBRjEue0niOLH1JgHlLEzNATJDDD0F4w81VPwpntyCH9mwf1yvqXd85RaFEVzhb3aseMOkcLLgg85wzYDKVDhcr0OBtc9btMnMl+wR6dH1VGFEEHlDGYvqd3jtPBSeFtUdB4fH4BJcl58fwZL1684OPte95+/x1de8IaEU99d+J4OrBeX7Dfd964GBXY8GhjTLTGIGXOKu4x00+La5RSArfOqDiP3oFRd1Fa0/dNvCZwrcNxx3KxIq8qDs0J5xQXaoNTiv5hB8ButxMu/UyyOpTrcVZqHrRKE+cf85a5L8skhAKp159oMYbrQhc7zSgenQvtFGVe7JxiPOFansAni/mW6UyRQBQQiGIUhWPFt4urEl4s6CbRYAhca7zbuWHgCLDuEmBeLBR5pkT/6Toe7rYoJQXEVVVxc3PDer3hcNzz+vU3viOckUZXStH51J++v0MpF+GmwljHkIqbTJoeMZPiqK1yk/kI40Ux1kF47hWgnowxwrm887PrOyqko/Cp63BtxmUlqd+mN5Jl4hzWuyMWiwVa51xdXvCwvZsSzkyvTY2t+VoIs5FVTgt8nRsXPFwf3slZN0mrJuF+lsdAxlGHd4+/e7zQqTHAzM8VFYHRCkxR18JOUXh2m9zp3G5Lj7SeQHBiM/bHhs6nuGgVsCpyikzz/PlzFqslRVXy+rvv0JnUVpqhQynHMHTgDOvNkizkh50R7fEzR0QfjM5NraZczf8TutYG8Yl1qLAgRS4N5xHYzOA5D07Uru+wTvo9te2Rh4d7uqFHlZIa1HatwBEYy6pecHmxZrPZsNvvGUKjW2/yWWP9z8iV03dyzid3hjI5CxmZdPe2/kVgkvOvGN1OwdqPnNk7ZPXsObJ+IwPS52KUjw8Xf6Ll4n0x1uNNSNB41FvCAOOOcONncRCPnWPTBbcW54w008pyqYHMq4ly23U9KhNguqurK/b7LXkmBRf397e+PmBsNaMcUXeZlIAlOqXym2QcO37nhk3nCO6coEcqpSS5EfFFheul4FXjjOBcODumC4naIWAxINZn15+4f7gFrCRlOlHOV+sF6/Way6srrLW0XTPOl0rXZZQyzuIJTsUf5dHsJOFZxXLAOOfnLMzkkHUTdd8GI+5XHDpeOLtRuNn4XQC8m+JlRBC85B7nOMTZ41cQeHh2UZSJjqQEn8v7wayz6Dxjvd7Qtid2u3sOh3vevPkGZoq5UpLmEyp8Ju/uHj/3iVE9+i3kujkHBjA6EZmAweEyja4KBqSrnvHNHAJnCFZjnmu6/sT7D2/QueLZs2cUi4pyuWBxsWaxWHB/f8/pdPB6pPMWb+A00znXITnPje8l8+E3gE7dUlMQlrnP89w6nuNe8/N0+uHcnTElUplIpZikdkCqqofPpkrxtDD0Uzslfdnxvjqk3RSF37E+CdBjiD17dsN6veT1t7/ku+9ec3d7K7j6g+8WMgwM1sS6gLOTlIzJJsUx+J/5xClvWYnPbXJHjP87Q4FOuKhX9FOuGQqScU70NZUJfGfvePP2Daf2xOXFJRebDZnS7Pd7bu9u6fvOtzSUZRQ9yUsclY4mmGp4w8SXF44yBZ1nolfpqd4dUsnjXJxZw/lannPm5pOTgt7uhzfZs1HGzxfBL5ArgMFjLWTxBYPITAeUfiafh6kIdsN4HghUkzHS/AEUTTOQ59Lq78WLV7x48ZJvv/2Gd+/eYMxocaXvVpdVTNd+lEgZxuTHpXRi7s84dHBNjAGRaeImDpSR6wYc+CYXDgnuD2YAX9GjtEKFOlLjQFm6U0NZV6jMcTzs+OoXP2Vzfc1nr37AcrEmz3LJ7GjbOI5QPR/aM4eEgaByhPx750SnDrHFsKGDby8YAUqpMe/fGwMpQ5nYg7O1RIWsmnHNn0S/9pd783i6NSTa7+LCqDDvEwTa4KSdWjvnFjgxSJMjBGwH2vbIarVhGHy1UpaxXqz4/Nc+58Xz5xyOW15/+zXGdhLATvCzwq4L3Xmf2nmPiOmTIjMd5uw8NQLxBnz8kHbdNI3sbC0GwEioI5fQmcRks6JAKuRb7u/e83B3z3p9wc3Nc5qTOKDHAXt9y6+t85p66qOM6tvoZo8iNFqCSmGHscAl3n68eHKPcQrG9XWJGyS8UwwrqUCBo96OKLVjUnZw4IVqbe0VQOVz3zOVCbdxEhZB20RZHh86Z7XnxehoLGgUz66usUp7/LCSly9ecX1zye7hjp/+9M9p2iN2MNgZF5tYOVo/Qo5Ox4Wa+oHi50yJMA2djdkKYS5kGQYcuVMUWcaAnbRmthA97LJBBTlJa/E3SYmbIysrHBpjeqzpMNtOIKb6o+dSwfzVPp1aQnzBIInz7adfnO0eXCWxPqPbxhMYQGhqrj0IzrljHkIEJj7WQEp5aAjlPPmlojLlMKluL/9qr6N5i8ULfUn/CBM6srewiIqpbyYw4GARzRmDtVLZ07cDl8+fsVzWlHlNvVjw7etvef36LzkethJMDpZgckRfzoy7hX8nnPUJO2QSmmFqEEm4iWT3y+dGQeYcWZ4lld7MGumGe4zzgJKGrcoKBKizjsVqzWAFdfFweEBlo9XrjWyi+1dJXcTEpaASdICEe8ZNrBRE+M5kUyk+SWBnK5tcQmhKKCk/nk5c+HL+cL9MjUSR3FW+tkwJIjQYcBZrBGbKKcETk401uuqiqqnSyVWM8UsIYhLCLrNgBw6nPepO050EN/bu7iMPu1uG/pSgND6ekLkYTI/UIAmOZOHACRmMdEXgEtGHl/aV8Up1ODNTYkG2fSfEPL6dF2nEs128N15fk3sEmKfmGDI4BixGqr5DKMuOye9WhVqI8d2VUpErhdlWWlwqcV7sjPMpFR3o8aJw7kRXOz+v8XPg1HbkbTdMb5hO+hPHxHUBkaOl3FBeViZOB9EQCPWMCEr/nj/fDgNNe6BpD2gPu2TsAMpghl7EN0nG5r+qPjUbw1wXUeF/bnp+fI5XiAOXSLmzvHe6UCp+HgnKz4UKYmN86iP+cTqdgFm1djSQxhT2MH+Td1PEtJw0EP4py/Dc4VIjINHnHh3RioLddo/6a3/td9z7d9+B7/wBoAKYmkoGHDhMsksmd0smSOkkVMGUnY4GqVT/BLl8ziR2Tgp1M6XJihxZJBtZdRC9+B7cqSiYT/ZcwZ1P9OS5KQOPYxxFUzw/6KR+XlJis4zvpJWPa0Yuo2JBTchscCGd3Y5pU+cWMxBZ+lkMwk+IIaxZoh5YEZc6jHc215NVDGOdcK4wx4nide5QgFEobVldvUT/0R/9kedAkm8UQiBeY/cPCH+HO0giY6pgBt+W0inRnXkRJ5wver7PHClxWGsZrPFB8BB77JOe3MGnF1b/X21Hzr3c8Wd+TpTxyeufG6/yxBmI2evkyvnFjVPqU5Ydo6XnnOf63veGwC/JfRN/5ExEhR8TfG3p3MkficEhYHeBRub3COeE93j0roE7RDr49KGUIssz/uv/5r9FdV3j/spv/joO6V4bM3oYHxQGdt7jL19GjqBtmG0/y1NO4bUNJIKQiV0wY+/nxJ3oCY9FqR5pC+vxW89NQuQqcz3s3POS1nqPRDqjjylwssn3UYebX5feXlYy3DvF309vpxMuGfTBxyhIT6sGKUGNitvj8+JcBO75ieNXbeD0yIqcP/1//inaGMvDw97HHT91ozF294j6k3PkJ2P0k83wZtXkH3+PqXh79ORgYifoNOO5oZJdPVrY9Pq5u+RTz5tfMx/nJz+b6XXhcEh1kFNMuM5kDPML45g//T5PjS2It5GLMoYxSRQdP+ZfRWDhub9K5xWJYPjw8Tb0Zs/5n/7e35PGBcE3FexsNR381Nj0u9CzdNBeVCrEBB1BctMGEzKA0aM052JPWYlaa5RT6CRxRJ6qQz+J8cMnJki+dhOdazIxyXNVYvafu96F7zzUeNRZA+efZZomdx45RphjRtGqJs9/rEvG+ROFLt7/3GHdY+swUlvgBwoPjXp+3ubqxKeel85dnuf8d//D/yjrNgyD2x/2vHr1kpurDcqNuPtxcsOYfF5TKmak6AQgA9Wf3f1Tl0XYP9mox50Z5Pw+WmuUR42xyscCtaZwAhI87x4bxzcnEjW+07miliA2Ytbn7JTUskw3iAo6mX9Fh4tl/UqldYvyb7rxglsDiI5apZQEvpWUqU3m3DlC1nuKpv6rvAL+JL87x039FP5++H4yN7PP5vOrMylmef3dW/7iL79juVyglVIs6gX/6Mf/mDzPI4TQ9GHTB04WUzLw4jnzh0s671ykjO87f4Hz8zLdRcGPEx2QZyYjXZQ50Qbl+9wx4WBndm/qFpiOL5mCZLHTBXTpZel4kudrYERzmeJUPFIZmKsuqdR5LFLDs0KbnLnxM5mfT6gKT/0uFxvINP/z3//fqBcLlPKyR2vNF198Qdv00/OTyfG3TNj2mHgorzvG0gIRyAAE/nJCaH5FRjcIj154TjRxssPETgc6cTjOOVpYnOm4plGAcJz77Nx9g0UZudjkmeO7PtWHaLqQ09oGZZ2EOhJ15ewmPKOrpeMLP3P3jMuUB/R7vPmAR5LsKc527rldZ7jfNvz2j/6NKCm0UgKAdn19zT/+P/+UrKwIzQykBxGoYAkqifAbM8SNqHyWpVYFipwAPRR3nxKnqbAHhUOjEWytwDVmI47IfCNxmshhFeJzyoA8gl2Jr11bb20mOz/VZwDxdH9CpXCM6M2oMen83CKLWZOEspwDZ3HKRE6DIjZOGPUxvKhTk7CWiGoHGnSehLvU+H3kkOPHca4DZPx0OmdqByp4qyLHns9Reo+nJIwi2QxOnPNWOerlgv/l7/8D1uvNONZQffLHf/zH/PjH/we///t/CDpjcAMOSV1On+McSWzMxgcmj37yJUczOdzrU9YdXlwBZL47riccK2nGzkbjPyrdYRhzEZmKr5C6HbqzzYcfcdfcWJ3zqQlP58Z7wcaXiCJrfDERm6HHUwhKu3HsWsX3BaK3fjo/0zjvfGxPGVBBDeCJ78M5n7JeRY+TphlKCTh/lgvMwt/5L/8rfv3X/0rcPEoplPWwe6EIdbfb87/+yT/g7/wX/zmff3ZD15xQaDJdxBePCruC0ecV9pYTX9n49sF9Sywp9Tj4RMV/KpY9aLsPgQQEv0EakeqxDF9uH6p0xiiD1Uwcneeq2s8eilGEzDrUTquyZ0cCY+nSsT1aQ+UzZp1wMT+HgUPLTDzmRIGxphJC+05vKrG2J/rfI72aR5+nx68yGs4TuWyGvKx4/+EDf/fv/vf8h3/rb1N5qNSU6wIC85jnBVdXF/xn/+l/wp/92U/44rd/h/cftuz30s9bdJuJ+hrmLnn440HFFw07yCu1cunTfhe5JnDHQCwu4S7nOao8ZpzMlP3PRcHcQIg3SM+ZjelJc16F/6mz8Fti1QqBqSc5yePFn4835RIp4YXKqLR24VPK/fz+c3fFOZ0unGeMYX888v5uyw9+4wv+5H//U/6jv/0fe2UfGL3MW1veAAAAB0lEQVRw/H+b2Lc3Rg46LgAAAABJRU5ErkJggg==" y="-21.499943"/>
-   </g>
-   <g id="text_1">
-    <!-- Train Image -->
-    <defs>
-     <path d="M -0.296875 72.90625 
-L 61.375 72.90625 
-L 61.375 64.59375 
-L 35.5 64.59375 
-L 35.5 0 
-L 25.59375 0 
-L 25.59375 64.59375 
-L -0.296875 64.59375 
-z
-" id="DejaVuSans-84"/>
-     <path d="M 41.109375 46.296875 
-Q 39.59375 47.171875 37.8125 47.578125 
-Q 36.03125 48 33.890625 48 
-Q 26.265625 48 22.1875 43.046875 
-Q 18.109375 38.09375 18.109375 28.8125 
-L 18.109375 0 
-L 9.078125 0 
-L 9.078125 54.6875 
-L 18.109375 54.6875 
-L 18.109375 46.1875 
-Q 20.953125 51.171875 25.484375 53.578125 
-Q 30.03125 56 36.53125 56 
-Q 37.453125 56 38.578125 55.875 
-Q 39.703125 55.765625 41.0625 55.515625 
-z
-" id="DejaVuSans-114"/>
-     <path d="M 34.28125 27.484375 
-Q 23.390625 27.484375 19.1875 25 
-Q 14.984375 22.515625 14.984375 16.5 
-Q 14.984375 11.71875 18.140625 8.90625 
-Q 21.296875 6.109375 26.703125 6.109375 
-Q 34.1875 6.109375 38.703125 11.40625 
-Q 43.21875 16.703125 43.21875 25.484375 
-L 43.21875 27.484375 
-z
-M 52.203125 31.203125 
-L 52.203125 0 
-L 43.21875 0 
-L 43.21875 8.296875 
-Q 40.140625 3.328125 35.546875 0.953125 
-Q 30.953125 -1.421875 24.3125 -1.421875 
-Q 15.921875 -1.421875 10.953125 3.296875 
-Q 6 8.015625 6 15.921875 
-Q 6 25.140625 12.171875 29.828125 
-Q 18.359375 34.515625 30.609375 34.515625 
-L 43.21875 34.515625 
-L 43.21875 35.40625 
-Q 43.21875 41.609375 39.140625 45 
-Q 35.0625 48.390625 27.6875 48.390625 
-Q 23 48.390625 18.546875 47.265625 
-Q 14.109375 46.140625 10.015625 43.890625 
-L 10.015625 52.203125 
-Q 14.9375 54.109375 19.578125 55.046875 
-Q 24.21875 56 28.609375 56 
-Q 40.484375 56 46.34375 49.84375 
-Q 52.203125 43.703125 52.203125 31.203125 
-z
-" id="DejaVuSans-97"/>
-     <path d="M 9.421875 54.6875 
-L 18.40625 54.6875 
-L 18.40625 0 
-L 9.421875 0 
-z
-M 9.421875 75.984375 
-L 18.40625 75.984375 
-L 18.40625 64.59375 
-L 9.421875 64.59375 
-z
-" id="DejaVuSans-105"/>
-     <path d="M 54.890625 33.015625 
-L 54.890625 0 
-L 45.90625 0 
-L 45.90625 32.71875 
-Q 45.90625 40.484375 42.875 44.328125 
-Q 39.84375 48.1875 33.796875 48.1875 
-Q 26.515625 48.1875 22.3125 43.546875 
-Q 18.109375 38.921875 18.109375 30.90625 
-L 18.109375 0 
-L 9.078125 0 
-L 9.078125 54.6875 
-L 18.109375 54.6875 
-L 18.109375 46.1875 
-Q 21.34375 51.125 25.703125 53.5625 
-Q 30.078125 56 35.796875 56 
-Q 45.21875 56 50.046875 50.171875 
-Q 54.890625 44.34375 54.890625 33.015625 
-z
-" id="DejaVuSans-110"/>
-     <path id="DejaVuSans-32"/>
-     <path d="M 9.8125 72.90625 
-L 19.671875 72.90625 
-L 19.671875 0 
-L 9.8125 0 
-z
-" id="DejaVuSans-73"/>
-     <path d="M 52 44.1875 
-Q 55.375 50.25 60.0625 53.125 
-Q 64.75 56 71.09375 56 
-Q 79.640625 56 84.28125 50.015625 
-Q 88.921875 44.046875 88.921875 33.015625 
-L 88.921875 0 
-L 79.890625 0 
-L 79.890625 32.71875 
-Q 79.890625 40.578125 77.09375 44.375 
-Q 74.3125 48.1875 68.609375 48.1875 
-Q 61.625 48.1875 57.5625 43.546875 
-Q 53.515625 38.921875 53.515625 30.90625 
-L 53.515625 0 
-L 44.484375 0 
-L 44.484375 32.71875 
-Q 44.484375 40.625 41.703125 44.40625 
-Q 38.921875 48.1875 33.109375 48.1875 
-Q 26.21875 48.1875 22.15625 43.53125 
-Q 18.109375 38.875 18.109375 30.90625 
-L 18.109375 0 
-L 9.078125 0 
-L 9.078125 54.6875 
-L 18.109375 54.6875 
-L 18.109375 46.1875 
-Q 21.1875 51.21875 25.484375 53.609375 
-Q 29.78125 56 35.6875 56 
-Q 41.65625 56 45.828125 52.96875 
-Q 50 49.953125 52 44.1875 
-z
-" id="DejaVuSans-109"/>
-     <path d="M 45.40625 27.984375 
-Q 45.40625 37.75 41.375 43.109375 
-Q 37.359375 48.484375 30.078125 48.484375 
-Q 22.859375 48.484375 18.828125 43.109375 
-Q 14.796875 37.75 14.796875 27.984375 
-Q 14.796875 18.265625 18.828125 12.890625 
-Q 22.859375 7.515625 30.078125 7.515625 
-Q 37.359375 7.515625 41.375 12.890625 
-Q 45.40625 18.265625 45.40625 27.984375 
-z
-M 54.390625 6.78125 
-Q 54.390625 -7.171875 48.1875 -13.984375 
-Q 42 -20.796875 29.203125 -20.796875 
-Q 24.46875 -20.796875 20.265625 -20.09375 
-Q 16.0625 -19.390625 12.109375 -17.921875 
-L 12.109375 -9.1875 
-Q 16.0625 -11.328125 19.921875 -12.34375 
-Q 23.78125 -13.375 27.78125 -13.375 
-Q 36.625 -13.375 41.015625 -8.765625 
-Q 45.40625 -4.15625 45.40625 5.171875 
-L 45.40625 9.625 
-Q 42.625 4.78125 38.28125 2.390625 
-Q 33.9375 0 27.875 0 
-Q 17.828125 0 11.671875 7.65625 
-Q 5.515625 15.328125 5.515625 27.984375 
-Q 5.515625 40.671875 11.671875 48.328125 
-Q 17.828125 56 27.875 56 
-Q 33.9375 56 38.28125 53.609375 
-Q 42.625 51.21875 45.40625 46.390625 
-L 45.40625 54.6875 
-L 54.390625 54.6875 
-z
-" id="DejaVuSans-103"/>
-     <path d="M 56.203125 29.59375 
-L 56.203125 25.203125 
-L 14.890625 25.203125 
-Q 15.484375 15.921875 20.484375 11.0625 
-Q 25.484375 6.203125 34.421875 6.203125 
-Q 39.59375 6.203125 44.453125 7.46875 
-Q 49.3125 8.734375 54.109375 11.28125 
-L 54.109375 2.78125 
-Q 49.265625 0.734375 44.1875 -0.34375 
-Q 39.109375 -1.421875 33.890625 -1.421875 
-Q 20.796875 -1.421875 13.15625 6.1875 
-Q 5.515625 13.8125 5.515625 26.8125 
-Q 5.515625 40.234375 12.765625 48.109375 
-Q 20.015625 56 32.328125 56 
-Q 43.359375 56 49.78125 48.890625 
-Q 56.203125 41.796875 56.203125 29.59375 
-z
-M 47.21875 32.234375 
-Q 47.125 39.59375 43.09375 43.984375 
-Q 39.0625 48.390625 32.421875 48.390625 
-Q 24.90625 48.390625 20.390625 44.140625 
-Q 15.875 39.890625 15.1875 32.171875 
-z
-" id="DejaVuSans-101"/>
-    </defs>
-    <g transform="translate(48.199347 16.318125)scale(0.12 -0.12)">
-     <use xlink:href="#DejaVuSans-84"/>
-     <use x="46.333984" xlink:href="#DejaVuSans-114"/>
-     <use x="87.447266" xlink:href="#DejaVuSans-97"/>
-     <use x="148.726562" xlink:href="#DejaVuSans-105"/>
-     <use x="176.509766" xlink:href="#DejaVuSans-110"/>
-     <use x="239.888672" xlink:href="#DejaVuSans-32"/>
-     <use x="271.675781" xlink:href="#DejaVuSans-73"/>
-     <use x="301.167969" xlink:href="#DejaVuSans-109"/>
-     <use x="398.580078" xlink:href="#DejaVuSans-97"/>
-     <use x="459.859375" xlink:href="#DejaVuSans-103"/>
-     <use x="523.335938" xlink:href="#DejaVuSans-101"/>
-    </g>
-   </g>
-  </g>
-  <g id="axes_2">
-   <g clip-path="url(#pf02e2d733d)">
-    <image height="153" id="imageb081ed1ee7" transform="scale(1 -1)translate(0 -153)" width="153" x="189.818182" xlink:href="data:image/png;base64,
-iVBORw0KGgoAAAANSUhEUgAAAJkAAACZCAYAAAA8XJi6AAAABHNCSVQICAgIfAhkiAAADEVJREFUeJzt3V9MW2UfB/Bv2xUKsm44Fkt0Gpdtzo3ojM4MnasJ4mYUsmxqvMBsmYlx80+miboLY7hxGnahhkVNVDSMhKibYEC2MmAONkWQbR3IYIVRkilsDAoUO1tOe96LveN9m7bQQp/zPE/5fRKS9ZyTnm/gu/Ocnp4/usLCQhUAkpOTsW/fPoTz2Wef4dVXXw07j8xOTk4O6urqQqZXVlbCbrdzSMSOnncAEmzr1q1YuXIl7xhxNVWypKQknjlIAtMDgNlsxltvvRV2Aa/XC7fbrWmo+WBychLj4+O8Y2hCn56ejjfffDPszOvXr+Pzzz+PuK9GZq+xsREvvfRS2HlpaWkwGAwaJ2JH/8Ybb4Sd4fF48O2330YsIGEnPz8ft99+O+8YcRN2x9/j8aC0tBR79uzROg9JQGFL1tPTg927d2udhSQoOoQhqOXLl8NkMvGOERdUMkFZrVYsWrSId4y4CCnZ6OgoSkpKeGQhCSqkZCMjI/j00095ZJl32tvbceTIEd4xmAsqmdvtxnvvvccry7zT3d2No0eP8o7BXFDJPB4PysvLeWUhCYp2/AlzVDLOqqqqcPDgQd4xmKKScXb16lU4nU7eMZiikhHmqGQC27lzJxYvXsw7xpxRyQRmMpmg0+l4x5gzKhlhjkpGmAsqmd/v55VjXgsEAlBVlXcMZqZKNjY2llBnY8rk448/xgcffBB23oIFCzROE380XApuz549SEtL4x1jTqhkhDkqGWGOSiaBjIwMqY+X6QFAVVU0NzfzzjKv9ff3Y2BgIOy8HTt2SH2Fvx4AfD4ftmzZwjvLvPbVV1+hoqKCdwwmaLiUxJo1a6QdMqlkAmlsbMTFixfDzsvPz4deL+efS87UCeq7775DS0sL7xhxRyWTSE5ODu8Is0Ilk0h2draU+2VUMsEUFxfjzJkz0y7z3HPPYfv27RolmjsqmWBaWlrw999/R5xfUFCANWvWICsrCy+88IKGyWaPSiaZ5cuXT/377rvv5pgkelQyiRmNRuzatYt3jBlRyQS0a9euqA5l6HQ6LFmyRINEc0MlE9DQ0BC8Xi/vGHFDJZNcamqq8LddpZIlANHP0KCSCcpqteL8+fNRLbto0SKh7/E7VbJEum98Ioj16iWRvwnQAzce3jUyMgKTyTT1I/ommATT6XTCbiimtmRmsxnXr1+f+jl58iQWL16M1NRUnvnmtfHxcQQCgaiWzcjIwM6dO9kGmqWI+2QbNmyAy+XCoUOHYDabtcxE/uuxxx5DX18f7xhzNuOVo9u2bYPH40FhYeHUtMHBQfzzzz8sc5EEEtXlyQUFBSgoKJh6/c477+D48eNRrUBRFHR0dMwuHUFnZyfuuusuqa8k16mMb8IwMTGBp556Cv/++y/++OMPlqtKWAMDA7BYLDMud/nyZXz99dcaJIqN3uFwwOFwoLe3l8kK0tLS0NTUhJ9++glPP/00NmzYwGQ9iayurg6Kosy4XEpKipD3M9HdfAa5wWBAfn4+gBthWT2iuKurC7t378Yvv/zC5P0T1ejoaFSPwent7UVZWZkGiaI39enS7/ejoqICFRUVqKqqwoULF5iscPXq1SguLsYTTzzB5P2JeMIewnC73bDZbGhsbERXV1fcV5qVlYUDBw5g8+bNcX9vIh7D448/XhhuhtfrhdPpxODgIEZHR6GqKm699da4rdhisSArKwtOp5PZ/mAimZiYwJNPPjnjtZculyvq7zy1MuPn4uHhYfz222+4dOkSOjs7Y3pzvV6PZ555JuL8devW4dFHH4XNZovpfeejgwcP4sCBA1Ieyog68ZUrV3DlypWY3lyn00FRFGzdujXiMs8++yxaWlpQXV0d03sTeUQcLuNlaGgI/f396Ovrw+rVq0PmL126FOvXr4fD4UBPTw/LKNL79ddf8eKLL057xkVKSgoMBgP6+/s1TDY95tveQCCAvr4+GAwGXLt2DRaLJWQIXblyJTIzM1lHkd6JEydmPAUoJSUFS5cu1ShRdDQ7adHv9+Ovv/7CuXPnwg6N+/fvR25urlZxpLV27VreEWKm+Zmxfr8fbrc7ZLrFYpH+Brxa6O7unnGZe+65R6j7ZnA5/bqnpwc1NTU8Vj0vGI1GZGdnw2q18o4CgFPJAoEA2traUFtbGzS9vLxcmF+MyKI5v89gMGDTpk145JFHNEg0PW4XkgQCgZAnoCQnJ0t7ozctRXsun16vF+L3yTVBuE9KIvxSSHxx/Yu2trbixIkTQdPq6urw0EMPcUqUeER4ZpN831GQqKiqilOnTuH06dO8o4hZsoULF0Kv10d9pc58E+lpvqqqTt1Dw263o6GhQctYEXEvmc/nw+TkJIxG49S0hoYGWK1WnDp1iooWxvDwcNBrVVUxMTGBgYEBlJeXc0oVGfeSNTc3w2w2Izs7O2j6yZMnsWzZMly+fJlTMnmMj4/jk08+4R0jIvooR5gTomQulwsTExO8YxBGhChZa2srHA5HyPTc3NygfTUC5OXl8Y4QMyFKFklJSQkWLlzIO4ZQZHzIlzAl6+vrg8vlCpn+8ssvS3nKMfkfYUrW3t6OwcHBkOkffvghkpOTOSQi8SJMyQDg3LlzGBkZ4R1DWEVFRVJ+tytU4osXL2JsbIx3DGG9/vrrIef3+3y+kFOmREM7OxJTFAWHDx8O+8lcJEJtyQCgtrY25GuTmpoaYW9VyUsgEEBZWZnwBQMELNng4GDIgxI2bdok9I13eVBVVajL3qYjXMlI4qGSEeaELFlpaSkdykggQpbM6/WGnEc2NjZG+2WSErJk4dDzBOQlTcmIvKhkhDkqGWGOSkaYo5IR5qhkhDkqGWGOSkaYo5IR5qhkhDkqGWGOSkaYo5IR5qhkhDkqGWGOSkaYo5IR5qhkhDkqGWGOSkaYo5JJymAw4O233+YdIypUMonJcnNAKhlhjkomkcnJSd4RZoVKJhGz2Rzy+EYZUMkIc1QywhyVjDBHJSPMUckIc1QywhyVjDBHJSPMUckIc1QywpwcX+PP0d69e/Hggw/yjhGipaUFxcXFvGMwl/Al27t3L959911YLBbeUULk5OQgNzcXFRUV+Oabb3jHYSahS/baa69h3759uO2223hHCSszMxN5eXl44IEHoCgKDh06xDsSEwm7T/bKK6/g/fffF7Zg/++OO+5AUVERnn/+ed5RmEjILdmOHTuwf/9+pKen844SNYvFgiVLlvCOwURClWzLli0oKSnBLbfcArPZzDtOzD766CNcvXoVR44c4R0lroQdLr/88ku43e6ol9+4cSN++OEHZGZmSlkw4MZJiaWlpdi8eTPvKHEl7JbM5/NBVdWoll23bh3q6+uRlJQUcZmysjI4nc44pYuPe++9F9u3bw+alpqaiqqqKvj9fqxfvx4dHR1T81wul5QPlxW2ZLHQ6/XTFuz7779Hb2+vhomi09HRgaSkJOTl5QVNNxqNMBqNIQ+1N5lMWsaLG2GHy2itWLECbW1tEedXVlbiwoULGiaKzXRba7vdjhUrVkj/dDzpS5acnBx2uqqqOHbsGOx2u8aJYnP27FnYbLaIZXM4HAgEAsjIyAiZJ8tFJVIPl6tWrQraZwFuPABeURQ0NTXh999/55QsNs3NzUhKSoLVag0ZIm8aGhoKeu33+1FUVKRFvDmTtmTp6eno7u4Omd7e3o7KykoOieamsbERRqMR2dnZUu7cT0eq4fLOO+8EAOh0OixbtixkvqIo8Hg8WseKm/r6erS2tkJRFN5R4kqqLZnT6cTatWthMpnC7uz39vaitraWQ7L4sdlsWLBgAe6//34YjcaIy127dk3DVHMjVckA4M8//ww73ev1Ynh4WOM0bPz888/Q6XS47777whZNVVV88cUXHJLNjtDDZX9/f8gD78Px+Xxoa2vD8ePHNUiljerqapw/fz7s0Hnp0iUOiWZP6JL9+OOP8Hq9My43PDycUAW7qbq6GmfOnEFnZ2fQf7aysjKOqWIn/HBpt9vx8MMPR/xo7/V60dXVpXEq7Rw9ehTAjS//p9tHE5nwJbPZbJicnMTGjRtDjnz7fD40NTXh9OnTnNJp59ixY7wjzJrQw+VNDQ0NIUfEFUVBfX39vCiY7ITfkt1UU1MT9DoQCODs2bOc0pBYSFOy6b4EJ2KTYrgkcqOSEeaoZIQ5KhlhjkpGmKOSEeaoZIQ5Khlh7j+IobnQcdL/mQAAAABJRU5ErkJggg==" y="-21.499943"/>
-   </g>
-   <g id="text_2">
-    <!-- Label -->
-    <defs>
-     <path d="M 9.8125 72.90625 
-L 19.671875 72.90625 
-L 19.671875 8.296875 
-L 55.171875 8.296875 
-L 55.171875 0 
-L 9.8125 0 
-z
-" id="DejaVuSans-76"/>
-     <path d="M 48.6875 27.296875 
-Q 48.6875 37.203125 44.609375 42.84375 
-Q 40.53125 48.484375 33.40625 48.484375 
-Q 26.265625 48.484375 22.1875 42.84375 
-Q 18.109375 37.203125 18.109375 27.296875 
-Q 18.109375 17.390625 22.1875 11.75 
-Q 26.265625 6.109375 33.40625 6.109375 
-Q 40.53125 6.109375 44.609375 11.75 
-Q 48.6875 17.390625 48.6875 27.296875 
-z
-M 18.109375 46.390625 
-Q 20.953125 51.265625 25.265625 53.625 
-Q 29.59375 56 35.59375 56 
-Q 45.5625 56 51.78125 48.09375 
-Q 58.015625 40.1875 58.015625 27.296875 
-Q 58.015625 14.40625 51.78125 6.484375 
-Q 45.5625 -1.421875 35.59375 -1.421875 
-Q 29.59375 -1.421875 25.265625 0.953125 
-Q 20.953125 3.328125 18.109375 8.203125 
-L 18.109375 0 
-L 9.078125 0 
-L 9.078125 75.984375 
-L 18.109375 75.984375 
-z
-" id="DejaVuSans-98"/>
-     <path d="M 9.421875 75.984375 
-L 18.40625 75.984375 
-L 18.40625 0 
-L 9.421875 0 
-z
-" id="DejaVuSans-108"/>
-    </defs>
-    <g transform="translate(249.721278 16.318125)scale(0.12 -0.12)">
-     <use xlink:href="#DejaVuSans-76"/>
-     <use x="55.712891" xlink:href="#DejaVuSans-97"/>
-     <use x="116.992188" xlink:href="#DejaVuSans-98"/>
-     <use x="180.46875" xlink:href="#DejaVuSans-101"/>
-     <use x="241.992188" xlink:href="#DejaVuSans-108"/>
-    </g>
-   </g>
-  </g>
- </g>
- <defs>
-  <clipPath id="p58ad9a7e6d">
-   <rect height="152.181818" width="152.181818" x="7.2" y="22.318125"/>
-  </clipPath>
-  <clipPath id="pf02e2d733d">
-   <rect height="152.181818" width="152.181818" x="189.818182" y="22.318125"/>
-  </clipPath>
- </defs>
-</svg>

doc/paddle/tutorial/cv_case/mnist_lenet_classification/mnist_lenet_classification.ipynb

+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "# MNIST数据集使用LeNet进行图像分类\n",
+    "本示例教程演示如何在MNIST数据集上用LeNet进行图像分类。\n",
+    "手写数字的MNIST数据集，包含60,000个用于训练的示例和10,000个用于测试的示例。这些数字已经过尺寸标准化并位于图像中心，图像是固定大小(28x28像素)，其值为0到1。该数据集的官方地址为：http://yann.lecun.com/exdb/mnist/"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## 环境\n",
+    "本教程基于paddle-2.0-beta编写，如果您的环境不是本版本，请先安装paddle-2.0-beta版本。"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "2.0.0-beta0\n"
+     ]
+    }
+   ],
+   "source": [
+    "import paddle\n",
+    "print(paddle.__version__)\n",
+    "paddle.disable_static()\n",
+    "# 开启动态图"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## 加载数据集\n",
+    "我们使用飞桨自带的paddle.dataset完成mnist数据集的加载。"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 4,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "download training data and load training data\n",
+      "load finished\n"
+     ]
+    }
+   ],
+   "source": [
+    "print('download training data and load training data')\n",
+    "train_dataset = paddle.vision.datasets.MNIST(mode='train')\n",
+    "test_dataset = paddle.vision.datasets.MNIST(mode='test')\n",
+    "print('load finished')"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "取训练集中的一条数据看一下。"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 5,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "train_data0 label is: [5]\n"
+     ]
+    },
+    {
+     "data": {
+      "image/png": "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\n",
+      "text/plain": [
+       "<Figure size 144x144 with 1 Axes>"
+      ]
+     },
+     "metadata": {
+      "needs_background": "light"
+     },
+     "output_type": "display_data"
+    }
+   ],
+   "source": [
+    "import numpy as np\n",
+    "import matplotlib.pyplot as plt\n",
+    "train_data0, train_label_0 = train_dataset[0][0],train_dataset[0][1]\n",
+    "train_data0 = train_data0.reshape([28,28])\n",
+    "plt.figure(figsize=(2,2))\n",
+    "plt.imshow(train_data0, cmap=plt.cm.binary)\n",
+    "print('train_data0 label is: ' + str(train_label_0))"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## 组网\n",
+    "用paddle.nn下的API，如`Conv2d`、`MaxPool2d`、`Linear`完成LeNet的构建。"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 7,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import paddle\n",
+    "import paddle.nn.functional as F\n",
+    "class LeNet(paddle.nn.Layer):\n",
+    "    def __init__(self):\n",
+    "        super(LeNet, self).__init__()\n",
+    "        self.conv1 = paddle.nn.Conv2d(in_channels=1, out_channels=6, kernel_size=5, stride=1, padding=2)\n",
+    "        self.max_pool1 = paddle.nn.MaxPool2d(kernel_size=2,  stride=2)\n",
+    "        self.conv2 = paddle.nn.Conv2d(in_channels=6, out_channels=16, kernel_size=5, stride=1)\n",
+    "        self.max_pool2 = paddle.nn.MaxPool2d(kernel_size=2, stride=2)\n",
+    "        self.linear1 = paddle.nn.Linear(in_features=16*5*5, out_features=120)\n",
+    "        self.linear2 = paddle.nn.Linear(in_features=120, out_features=84)\n",
+    "        self.linear3 = paddle.nn.Linear(in_features=84, out_features=10)\n",
+    "\n",
+    "    def forward(self, x):\n",
+    "        x = self.conv1(x)\n",
+    "        x = F.relu(x)\n",
+    "        x = self.max_pool1(x)\n",
+    "        x = F.relu(x)\n",
+    "        x = self.conv2(x)\n",
+    "        x = self.max_pool2(x)\n",
+    "        x = paddle.flatten(x, start_axis=1,stop_axis=-1)\n",
+    "        x = self.linear1(x)\n",
+    "        x = F.relu(x)\n",
+    "        x = self.linear2(x)\n",
+    "        x = F.relu(x)\n",
+    "        x = self.linear3(x)\n",
+    "        x = F.softmax(x)\n",
+    "        return x"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## 训练方式一\n",
+    "组网后，开始对模型进行训练，先构建`train_loader`，加载训练数据，然后定义`train`函数，设置好损失函数后，按batch加载数据，完成模型的训练。"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 10,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "epoch: 0, batch_id: 0, loss is: [2.3037894], acc is: [0.140625]\n",
+      "epoch: 0, batch_id: 100, loss is: [1.6175328], acc is: [0.9375]\n",
+      "epoch: 0, batch_id: 200, loss is: [1.5388051], acc is: [0.96875]\n",
+      "epoch: 0, batch_id: 300, loss is: [1.5251061], acc is: [0.96875]\n",
+      "epoch: 0, batch_id: 400, loss is: [1.4678856], acc is: [1.]\n",
+      "epoch: 0, batch_id: 500, loss is: [1.4944503], acc is: [0.984375]\n",
+      "epoch: 0, batch_id: 600, loss is: [1.5365536], acc is: [0.96875]\n",
+      "epoch: 0, batch_id: 700, loss is: [1.4885054], acc is: [0.984375]\n",
+      "epoch: 0, batch_id: 800, loss is: [1.4872254], acc is: [0.984375]\n",
+      "epoch: 0, batch_id: 900, loss is: [1.4884174], acc is: [0.984375]\n",
+      "epoch: 1, batch_id: 0, loss is: [1.4776722], acc is: [1.]\n",
+      "epoch: 1, batch_id: 100, loss is: [1.4751343], acc is: [1.]\n",
+      "epoch: 1, batch_id: 200, loss is: [1.4772581], acc is: [1.]\n",
+      "epoch: 1, batch_id: 300, loss is: [1.4918218], acc is: [0.984375]\n",
+      "epoch: 1, batch_id: 400, loss is: [1.5038397], acc is: [0.96875]\n",
+      "epoch: 1, batch_id: 500, loss is: [1.5088196], acc is: [0.96875]\n",
+      "epoch: 1, batch_id: 600, loss is: [1.4961376], acc is: [0.984375]\n",
+      "epoch: 1, batch_id: 700, loss is: [1.4755756], acc is: [1.]\n",
+      "epoch: 1, batch_id: 800, loss is: [1.4921497], acc is: [0.984375]\n",
+      "epoch: 1, batch_id: 900, loss is: [1.4944404], acc is: [1.]\n"
+     ]
+    }
+   ],
+   "source": [
+    "import paddle\n",
+    "train_loader = paddle.io.DataLoader(train_dataset, places=paddle.CPUPlace(), batch_size=64, shuffle=True)\n",
+    "# 加载训练集 batch_size 设为 64\n",
+    "def train(model):\n",
+    "    model.train()\n",
+    "    epochs = 2\n",
+    "    optim = paddle.optimizer.Adam(learning_rate=0.001, parameters=model.parameters())\n",
+    "    # 用Adam作为优化函数\n",
+    "    for epoch in range(epochs):\n",
+    "        for batch_id, data in enumerate(train_loader()):\n",
+    "            x_data = data[0]\n",
+    "            y_data = data[1]\n",
+    "            predicts = model(x_data)\n",
+    "            loss = paddle.nn.functional.cross_entropy(predicts, y_data)\n",
+    "            # 计算损失\n",
+    "            acc = paddle.metric.accuracy(predicts, y_data, k=2)\n",
+    "            avg_loss = paddle.mean(loss)\n",
+    "            avg_acc = paddle.mean(acc)\n",
+    "            avg_loss.backward()\n",
+    "            if batch_id % 100 == 0:\n",
+    "                print(\"epoch: {}, batch_id: {}, loss is: {}, acc is: {}\".format(epoch, batch_id, avg_loss.numpy(), avg_acc.numpy()))\n",
+    "            optim.step()\n",
+    "            optim.clear_grad()\n",
+    "model = LeNet()\n",
+    "train(model)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### 对模型进行验证\n",
+    "训练完成后，需要验证模型的效果，此时，加载测试数据集，然后用训练好的模对测试集进行预测，计算损失与精度。"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 11,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "batch_id: 0, loss is: [1.4915928], acc is: [1.]\n",
+      "batch_id: 20, loss is: [1.4818308], acc is: [1.]\n",
+      "batch_id: 40, loss is: [1.5006062], acc is: [0.984375]\n",
+      "batch_id: 60, loss is: [1.521233], acc is: [1.]\n",
+      "batch_id: 80, loss is: [1.4772738], acc is: [1.]\n",
+      "batch_id: 100, loss is: [1.4755945], acc is: [1.]\n",
+      "batch_id: 120, loss is: [1.4746133], acc is: [1.]\n",
+      "batch_id: 140, loss is: [1.4786345], acc is: [1.]\n"
+     ]
+    }
+   ],
+   "source": [
+    "import paddle\n",
+    "test_loader = paddle.io.DataLoader(test_dataset, places=paddle.CPUPlace(), batch_size=64)\n",
+    "# 加载测试数据集\n",
+    "def test(model):\n",
+    "    model.eval()\n",
+    "    batch_size = 64\n",
+    "    for batch_id, data in enumerate(test_loader()):\n",
+    "        x_data = data[0]\n",
+    "        y_data = data[1]\n",
+    "        predicts = model(x_data)\n",
+    "        # 获取预测结果\n",
+    "        loss = paddle.nn.functional.cross_entropy(predicts, y_data)\n",
+    "        acc = paddle.metric.accuracy(predicts, y_data, k=2)\n",
+    "        avg_loss = paddle.mean(loss)\n",
+    "        avg_acc = paddle.mean(acc)\n",
+    "        avg_loss.backward()\n",
+    "        if batch_id % 20 == 0:\n",
+    "            print(\"batch_id: {}, loss is: {}, acc is: {}\".format(batch_id, avg_loss.numpy(), avg_acc.numpy()))\n",
+    "test(model)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### 训练方式一结束\n",
+    "以上就是训练方式一，通过这种方式，可以清楚的看到训练和测试中的每一步过程。但是，这种方式句法比较复杂。因此，我们提供了训练方式二，能够更加快速、高效的完成模型的训练与测试。"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## 3.训练方式二\n",
+    "通过paddle提供的`Model` 构建实例，使用封装好的训练与测试接口，快速完成模型训练与测试。"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 12,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import paddle\n",
+    "from paddle.static import InputSpec\n",
+    "from paddle.metric import Accuracy\n",
+    "inputs = InputSpec([None, 784], 'float32', 'x')\n",
+    "labels = InputSpec([None, 10], 'float32', 'x')\n",
+    "model = paddle.Model(LeNet(), inputs, labels)\n",
+    "optim = paddle.optimizer.Adam(learning_rate=0.001, parameters=model.parameters())\n",
+    "\n",
+    "model.prepare(\n",
+    "    optim,\n",
+    "    paddle.nn.loss.CrossEntropyLoss(),\n",
+    "    Accuracy(topk=(1, 2))\n",
+    "    )"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### 使用model.fit来训练模型"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 16,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Epoch 1/2\n",
+      "step 200/938 - loss: 1.5219 - acc_top1: 0.9829 - acc_top2: 0.9965 - 14ms/step\n",
+      "step 400/938 - loss: 1.4765 - acc_top1: 0.9825 - acc_top2: 0.9958 - 13ms/step\n",
+      "step 600/938 - loss: 1.4624 - acc_top1: 0.9823 - acc_top2: 0.9953 - 13ms/step\n",
+      "step 800/938 - loss: 1.4768 - acc_top1: 0.9829 - acc_top2: 0.9955 - 13ms/step\n",
+      "step 938/938 - loss: 1.4612 - acc_top1: 0.9836 - acc_top2: 0.9956 - 13ms/step\n",
+      "Epoch 2/2\n",
+      "step 200/938 - loss: 1.4705 - acc_top1: 0.9834 - acc_top2: 0.9959 - 13ms/step\n",
+      "step 400/938 - loss: 1.4620 - acc_top1: 0.9833 - acc_top2: 0.9960 - 13ms/step\n",
+      "step 600/938 - loss: 1.4613 - acc_top1: 0.9830 - acc_top2: 0.9960 - 13ms/step\n",
+      "step 800/938 - loss: 1.4763 - acc_top1: 0.9831 - acc_top2: 0.9960 - 13ms/step\n",
+      "step 938/938 - loss: 1.4924 - acc_top1: 0.9834 - acc_top2: 0.9959 - 13ms/step\n"
+     ]
+    }
+   ],
+   "source": [
+    "model.fit(train_dataset,\n",
+    "        epochs=2,\n",
+    "        batch_size=64,\n",
+    "        log_freq=200\n",
+    "        )"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### 使用model.evaluate来预测模型"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 17,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Eval begin...\n",
+      "step  20/157 - loss: 1.5246 - acc_top1: 0.9773 - acc_top2: 0.9969 - 6ms/step\n",
+      "step  40/157 - loss: 1.4622 - acc_top1: 0.9758 - acc_top2: 0.9961 - 6ms/step\n",
+      "step  60/157 - loss: 1.5241 - acc_top1: 0.9763 - acc_top2: 0.9951 - 6ms/step\n",
+      "step  80/157 - loss: 1.4612 - acc_top1: 0.9787 - acc_top2: 0.9959 - 6ms/step\n",
+      "step 100/157 - loss: 1.4612 - acc_top1: 0.9823 - acc_top2: 0.9967 - 5ms/step\n",
+      "step 120/157 - loss: 1.4612 - acc_top1: 0.9835 - acc_top2: 0.9966 - 5ms/step\n",
+      "step 140/157 - loss: 1.4612 - acc_top1: 0.9844 - acc_top2: 0.9969 - 5ms/step\n",
+      "step 157/157 - loss: 1.4612 - acc_top1: 0.9838 - acc_top2: 0.9966 - 5ms/step\n",
+      "Eval samples: 10000\n"
+     ]
+    },
+    {
+     "data": {
+      "text/plain": [
+       "{'loss': [1.4611504], 'acc_top1': 0.9838, 'acc_top2': 0.9966}"
+      ]
+     },
+     "execution_count": 17,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "model.evaluate(test_dataset, log_freq=20, batch_size=64)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### 训练方式二结束\n",
+    "以上就是训练方式二，可以快速、高效的完成网络模型训练与预测。"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## 总结\n"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "以上就是用LeNet对手写数字数据及MNIST进行分类。本示例提供了两种训练模型的方式，一种可以快速完成模型的组建与预测，非常适合新手用户上手。另一种则需要多个步骤来完成模型的训练，适合进阶用户使用。"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.7.3"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 4
+}

doc/paddle/tutorial/cv_case/mnist_lenet_classification/mnist_lenet_classification.rst

@@ -7,18 +7,19 @@ MNIST数据集使用LeNet进行图像分类
 环境
 ----
 
-本教程基于paddle-develop编写，如果您的环境不是本版本，请先安装paddle-develop版本。
+本教程基于paddle-2.0-beta编写，如果您的环境不是本版本，请先安装paddle-2.0-beta版本。
 
 .. code:: ipython3
 
     import paddle
     print(paddle.__version__)
     paddle.disable_static()
+    # 开启动态图
 
 
 .. parsed-literal::
 
-    0.0.0
+    2.0.0-beta0
 
 
 加载数据集
@@ -34,12 +35,6 @@ MNIST数据集使用LeNet进行图像分类
     print('load finished')
 
 
-.. parsed-literal::
-
-    /Library/Python/3.7/site-packages/ipykernel/ipkernel.py:287: DeprecationWarning: `should_run_async` will not call `transform_cell` automatically in the future. Please pass the result to `transformed_cell` argument and any exception that happen during thetransform in `preprocessing_exc_tuple` in IPython 7.17 and above.
-      and should_run_async(code)
-
-
 .. parsed-literal::
 
     download training data and load training data
@@ -64,12 +59,14 @@ MNIST数据集使用LeNet进行图像分类
     train_data0 label is: [5]
 
 
-.. image:: https://github.com/PaddlePaddle/FluidDoc/blob/develop/doc/paddle/user_guides/cv_case/image_classification/image/cifar.png?raw=true
 
-2.组网
-------
+.. image:: https://github.com/PaddlePaddle/FluidDoc/tree/develop/doc/paddle/tutorial/cv_case/image_segmentation/mnist_lenet_classification_files/mnist_lenet_classification_001.png?raw=true
+
 
-用paddle.nn下的API，如\ ``Conv2d``\ 、\ ``Pool2D``\ 、\ ``Linead``\ 完成LeNet的构建。
+组网
+----
+
+用paddle.nn下的API，如\ ``Conv2d``\ 、\ ``MaxPool2d``\ 、\ ``Linear``\ 完成LeNet的构建。
 
 .. code:: ipython3
 
@@ -93,7 +90,7 @@ MNIST数据集使用LeNet进行图像分类
             x = F.relu(x)
             x = self.conv2(x)
             x = self.max_pool2(x)
-            x = paddle.reshape(x, shape=[-1, 16*5*5])
+            x = paddle.flatten(x, start_axis=1,stop_axis=-1)
             x = self.linear1(x)
             x = F.relu(x)
             x = self.linear2(x)
@@ -102,22 +99,15 @@ MNIST数据集使用LeNet进行图像分类
             x = F.softmax(x)
             return x
 
-
-.. parsed-literal::
-
-    /Library/Python/3.7/site-packages/ipykernel/ipkernel.py:287: DeprecationWarning: `should_run_async` will not call `transform_cell` automatically in the future. Please pass the result to `transformed_cell` argument and any exception that happen during thetransform in `preprocessing_exc_tuple` in IPython 7.17 and above.
-      and should_run_async(code)
-
-
-3.训练方式一
-------------
+训练方式一
+----------
 
 组网后，开始对模型进行训练，先构建\ ``train_loader``\ ，加载训练数据，然后定义\ ``train``\ 函数，设置好损失函数后，按batch加载数据，完成模型的训练。
 
 .. code:: ipython3
 
     import paddle
-    train_loader = paddle.io.DataLoader(train_dataset, places=paddle.CPUPlace(), batch_size=64)
+    train_loader = paddle.io.DataLoader(train_dataset, places=paddle.CPUPlace(), batch_size=64, shuffle=True)
     # 加载训练集 batch_size 设为 64
     def train(model):
         model.train()
@@ -137,34 +127,34 @@ MNIST数据集使用LeNet进行图像分类
                 avg_loss.backward()
                 if batch_id % 100 == 0:
                     print("epoch: {}, batch_id: {}, loss is: {}, acc is: {}".format(epoch, batch_id, avg_loss.numpy(), avg_acc.numpy()))
-                optim.minimize(avg_loss)
-                model.clear_gradients()
+                optim.step()
+                optim.clear_grad()
     model = LeNet()
     train(model)
 
 
 .. parsed-literal::
 
-    epoch: 0, batch_id: 0, loss is: [2.3062382], acc is: [0.109375]
-    epoch: 0, batch_id: 100, loss is: [1.6826601], acc is: [0.84375]
-    epoch: 0, batch_id: 200, loss is: [1.685574], acc is: [0.796875]
-    epoch: 0, batch_id: 300, loss is: [1.5752499], acc is: [0.96875]
-    epoch: 0, batch_id: 400, loss is: [1.5006541], acc is: [1.]
-    epoch: 0, batch_id: 500, loss is: [1.5343401], acc is: [0.984375]
-    epoch: 0, batch_id: 600, loss is: [1.4875913], acc is: [0.984375]
-    epoch: 0, batch_id: 700, loss is: [1.5139006], acc is: [0.984375]
-    epoch: 0, batch_id: 800, loss is: [1.5227785], acc is: [0.984375]
-    epoch: 0, batch_id: 900, loss is: [1.4938308], acc is: [1.]
-    epoch: 1, batch_id: 0, loss is: [1.4826943], acc is: [0.984375]
-    epoch: 1, batch_id: 100, loss is: [1.4852213], acc is: [0.984375]
-    epoch: 1, batch_id: 200, loss is: [1.5008337], acc is: [1.]
-    epoch: 1, batch_id: 300, loss is: [1.505826], acc is: [1.]
-    epoch: 1, batch_id: 400, loss is: [1.4768786], acc is: [1.]
-    epoch: 1, batch_id: 500, loss is: [1.4950027], acc is: [0.984375]
-    epoch: 1, batch_id: 600, loss is: [1.4762383], acc is: [0.984375]
-    epoch: 1, batch_id: 700, loss is: [1.5276604], acc is: [0.96875]
-    epoch: 1, batch_id: 800, loss is: [1.4897399], acc is: [1.]
-    epoch: 1, batch_id: 900, loss is: [1.4927337], acc is: [1.]
+    epoch: 0, batch_id: 0, loss is: [2.3037894], acc is: [0.140625]
+    epoch: 0, batch_id: 100, loss is: [1.6175328], acc is: [0.9375]
+    epoch: 0, batch_id: 200, loss is: [1.5388051], acc is: [0.96875]
+    epoch: 0, batch_id: 300, loss is: [1.5251061], acc is: [0.96875]
+    epoch: 0, batch_id: 400, loss is: [1.4678856], acc is: [1.]
+    epoch: 0, batch_id: 500, loss is: [1.4944503], acc is: [0.984375]
+    epoch: 0, batch_id: 600, loss is: [1.5365536], acc is: [0.96875]
+    epoch: 0, batch_id: 700, loss is: [1.4885054], acc is: [0.984375]
+    epoch: 0, batch_id: 800, loss is: [1.4872254], acc is: [0.984375]
+    epoch: 0, batch_id: 900, loss is: [1.4884174], acc is: [0.984375]
+    epoch: 1, batch_id: 0, loss is: [1.4776722], acc is: [1.]
+    epoch: 1, batch_id: 100, loss is: [1.4751343], acc is: [1.]
+    epoch: 1, batch_id: 200, loss is: [1.4772581], acc is: [1.]
+    epoch: 1, batch_id: 300, loss is: [1.4918218], acc is: [0.984375]
+    epoch: 1, batch_id: 400, loss is: [1.5038397], acc is: [0.96875]
+    epoch: 1, batch_id: 500, loss is: [1.5088196], acc is: [0.96875]
+    epoch: 1, batch_id: 600, loss is: [1.4961376], acc is: [0.984375]
+    epoch: 1, batch_id: 700, loss is: [1.4755756], acc is: [1.]
+    epoch: 1, batch_id: 800, loss is: [1.4921497], acc is: [0.984375]
+    epoch: 1, batch_id: 900, loss is: [1.4944404], acc is: [1.]
 
 
 对模型进行验证
@@ -180,7 +170,7 @@ MNIST数据集使用LeNet进行图像分类
     def test(model):
         model.eval()
         batch_size = 64
-        for batch_id, data in enumerate(train_loader()):
+        for batch_id, data in enumerate(test_loader()):
             x_data = data[0]
             y_data = data[1]
             predicts = model(x_data)
@@ -190,23 +180,21 @@ MNIST数据集使用LeNet进行图像分类
             avg_loss = paddle.mean(loss)
             avg_acc = paddle.mean(acc)
             avg_loss.backward()
-            if batch_id % 100 == 0:
+            if batch_id % 20 == 0:
                 print("batch_id: {}, loss is: {}, acc is: {}".format(batch_id, avg_loss.numpy(), avg_acc.numpy()))
     test(model)
 
 
 .. parsed-literal::
 
-    batch_id: 0, loss is: [1.4630548], acc is: [1.]
-    batch_id: 100, loss is: [1.4789999], acc is: [0.984375]
-    batch_id: 200, loss is: [1.4621592], acc is: [1.]
-    batch_id: 300, loss is: [1.486401], acc is: [1.]
-    batch_id: 400, loss is: [1.4767764], acc is: [1.]
-    batch_id: 500, loss is: [1.4987783], acc is: [0.984375]
-    batch_id: 600, loss is: [1.4767168], acc is: [1.]
-    batch_id: 700, loss is: [1.4876428], acc is: [0.984375]
-    batch_id: 800, loss is: [1.4924926], acc is: [0.984375]
-    batch_id: 900, loss is: [1.4799261], acc is: [1.]
+    batch_id: 0, loss is: [1.4915928], acc is: [1.]
+    batch_id: 20, loss is: [1.4818308], acc is: [1.]
+    batch_id: 40, loss is: [1.5006062], acc is: [0.984375]
+    batch_id: 60, loss is: [1.521233], acc is: [1.]
+    batch_id: 80, loss is: [1.4772738], acc is: [1.]
+    batch_id: 100, loss is: [1.4755945], acc is: [1.]
+    batch_id: 120, loss is: [1.4746133], acc is: [1.]
+    batch_id: 140, loss is: [1.4786345], acc is: [1.]
 
 
 训练方式一结束
@@ -244,214 +232,24 @@ MNIST数据集使用LeNet进行图像分类
     model.fit(train_dataset,
             epochs=2,
             batch_size=64,
-            save_dir='mnist_checkpoint')
+            log_freq=200
+            )
 
 
 .. parsed-literal::
 
     Epoch 1/2
-    step  10/938 - loss: 2.2252 - acc_top1: 0.2547 - acc_top2: 0.4234 - 16ms/step
-
-
-.. parsed-literal::
-
-    /Library/Python/3.7/site-packages/paddle/fluid/layers/utils.py:76: DeprecationWarning: Using or importing the ABCs from 'collections' instead of from 'collections.abc' is deprecated, and in 3.8 it will stop working
-      return (isinstance(seq, collections.Sequence) and
-
-
-.. parsed-literal::
-
-    step  20/938 - loss: 1.9721 - acc_top1: 0.3664 - acc_top2: 0.5164 - 15ms/step
-    step  30/938 - loss: 1.8697 - acc_top1: 0.4464 - acc_top2: 0.5651 - 15ms/step
-    step  40/938 - loss: 1.8475 - acc_top1: 0.4859 - acc_top2: 0.5898 - 15ms/step
-    step  50/938 - loss: 1.8683 - acc_top1: 0.5256 - acc_top2: 0.6156 - 14ms/step
-    step  60/938 - loss: 1.8091 - acc_top1: 0.5437 - acc_top2: 0.6237 - 14ms/step
-    step  70/938 - loss: 1.7934 - acc_top1: 0.5607 - acc_top2: 0.6335 - 14ms/step
-    step  80/938 - loss: 1.7796 - acc_top1: 0.5760 - acc_top2: 0.6418 - 14ms/step
-    step  90/938 - loss: 1.8004 - acc_top1: 0.5868 - acc_top2: 0.6476 - 14ms/step
-    step 100/938 - loss: 1.7650 - acc_top1: 0.5972 - acc_top2: 0.6536 - 14ms/step
-    step 110/938 - loss: 1.7839 - acc_top1: 0.6033 - acc_top2: 0.6570 - 14ms/step
-    step 120/938 - loss: 1.8094 - acc_top1: 0.6087 - acc_top2: 0.6592 - 14ms/step
-    step 130/938 - loss: 1.8125 - acc_top1: 0.6153 - acc_top2: 0.6638 - 14ms/step
-    step 140/938 - loss: 1.7318 - acc_top1: 0.6217 - acc_top2: 0.6673 - 14ms/step
-    step 150/938 - loss: 1.8209 - acc_top1: 0.6267 - acc_top2: 0.6702 - 14ms/step
-    step 160/938 - loss: 1.7661 - acc_top1: 0.6308 - acc_top2: 0.6725 - 14ms/step
-    step 170/938 - loss: 1.7099 - acc_top1: 0.6341 - acc_top2: 0.6741 - 14ms/step
-    step 180/938 - loss: 1.8059 - acc_top1: 0.6363 - acc_top2: 0.6753 - 14ms/step
-    step 190/938 - loss: 1.7681 - acc_top1: 0.6400 - acc_top2: 0.6779 - 14ms/step
-    step 200/938 - loss: 1.8631 - acc_top1: 0.6430 - acc_top2: 0.6826 - 14ms/step
-    step 210/938 - loss: 1.6808 - acc_top1: 0.6479 - acc_top2: 0.6879 - 14ms/step
-    step 220/938 - loss: 1.5447 - acc_top1: 0.6558 - acc_top2: 0.6965 - 14ms/step
-    step 230/938 - loss: 1.6170 - acc_top1: 0.6641 - acc_top2: 0.7051 - 14ms/step
-    step 240/938 - loss: 1.6190 - acc_top1: 0.6719 - acc_top2: 0.7134 - 14ms/step
-    step 250/938 - loss: 1.5698 - acc_top1: 0.6794 - acc_top2: 0.7209 - 14ms/step
-    step 260/938 - loss: 1.6071 - acc_top1: 0.6869 - acc_top2: 0.7284 - 14ms/step
-    step 270/938 - loss: 1.5507 - acc_top1: 0.6939 - acc_top2: 0.7364 - 14ms/step
-    step 280/938 - loss: 1.5286 - acc_top1: 0.7023 - acc_top2: 0.7451 - 14ms/step
-    step 290/938 - loss: 1.5740 - acc_top1: 0.7098 - acc_top2: 0.7532 - 14ms/step
-    step 300/938 - loss: 1.5179 - acc_top1: 0.7172 - acc_top2: 0.7608 - 14ms/step
-    step 310/938 - loss: 1.5325 - acc_top1: 0.7240 - acc_top2: 0.7677 - 14ms/step
-    step 320/938 - loss: 1.4961 - acc_top1: 0.7305 - acc_top2: 0.7744 - 14ms/step
-    step 330/938 - loss: 1.5420 - acc_top1: 0.7369 - acc_top2: 0.7804 - 14ms/step
-    step 340/938 - loss: 1.5652 - acc_top1: 0.7427 - acc_top2: 0.7861 - 14ms/step
-    step 350/938 - loss: 1.5122 - acc_top1: 0.7484 - acc_top2: 0.7918 - 14ms/step
-    step 360/938 - loss: 1.5308 - acc_top1: 0.7544 - acc_top2: 0.7972 - 14ms/step
-    step 370/938 - loss: 1.5354 - acc_top1: 0.7596 - acc_top2: 0.8023 - 14ms/step
-    step 380/938 - loss: 1.5433 - acc_top1: 0.7645 - acc_top2: 0.8073 - 14ms/step
-    step 390/938 - loss: 1.5341 - acc_top1: 0.7693 - acc_top2: 0.8119 - 14ms/step
-    step 400/938 - loss: 1.4826 - acc_top1: 0.7740 - acc_top2: 0.8163 - 14ms/step
-    step 410/938 - loss: 1.4995 - acc_top1: 0.7785 - acc_top2: 0.8205 - 14ms/step
-    step 420/938 - loss: 1.5057 - acc_top1: 0.7827 - acc_top2: 0.8244 - 14ms/step
-    step 430/938 - loss: 1.4927 - acc_top1: 0.7866 - acc_top2: 0.8282 - 14ms/step
-    step 440/938 - loss: 1.5281 - acc_top1: 0.7902 - acc_top2: 0.8316 - 14ms/step
-    step 450/938 - loss: 1.5060 - acc_top1: 0.7936 - acc_top2: 0.8347 - 14ms/step
-    step 460/938 - loss: 1.5135 - acc_top1: 0.7968 - acc_top2: 0.8380 - 14ms/step
-    step 470/938 - loss: 1.5206 - acc_top1: 0.8004 - acc_top2: 0.8411 - 14ms/step
-    step 480/938 - loss: 1.4963 - acc_top1: 0.8039 - acc_top2: 0.8441 - 14ms/step
-    step 490/938 - loss: 1.4984 - acc_top1: 0.8071 - acc_top2: 0.8470 - 14ms/step
-    step 500/938 - loss: 1.4947 - acc_top1: 0.8101 - acc_top2: 0.8498 - 14ms/step
-    step 510/938 - loss: 1.4639 - acc_top1: 0.8130 - acc_top2: 0.8524 - 14ms/step
-    step 520/938 - loss: 1.4781 - acc_top1: 0.8158 - acc_top2: 0.8549 - 14ms/step
-    step 530/938 - loss: 1.4806 - acc_top1: 0.8187 - acc_top2: 0.8575 - 14ms/step
-    step 540/938 - loss: 1.4830 - acc_top1: 0.8214 - acc_top2: 0.8600 - 14ms/step
-    step 550/938 - loss: 1.4852 - acc_top1: 0.8239 - acc_top2: 0.8623 - 14ms/step
-    step 560/938 - loss: 1.5302 - acc_top1: 0.8263 - acc_top2: 0.8645 - 14ms/step
-    step 570/938 - loss: 1.5520 - acc_top1: 0.8286 - acc_top2: 0.8667 - 14ms/step
-    step 580/938 - loss: 1.4897 - acc_top1: 0.8305 - acc_top2: 0.8687 - 14ms/step
-    step 590/938 - loss: 1.4857 - acc_top1: 0.8328 - acc_top2: 0.8707 - 14ms/step
-    step 600/938 - loss: 1.5081 - acc_top1: 0.8351 - acc_top2: 0.8727 - 14ms/step
-    step 610/938 - loss: 1.5013 - acc_top1: 0.8373 - acc_top2: 0.8746 - 14ms/step
-    step 620/938 - loss: 1.4949 - acc_top1: 0.8395 - acc_top2: 0.8764 - 14ms/step
-    step 630/938 - loss: 1.4971 - acc_top1: 0.8412 - acc_top2: 0.8781 - 14ms/step
-    step 640/938 - loss: 1.4869 - acc_top1: 0.8434 - acc_top2: 0.8800 - 14ms/step
-    step 650/938 - loss: 1.5202 - acc_top1: 0.8450 - acc_top2: 0.8815 - 14ms/step
-    step 660/938 - loss: 1.5002 - acc_top1: 0.8468 - acc_top2: 0.8832 - 14ms/step
-    step 670/938 - loss: 1.5178 - acc_top1: 0.8487 - acc_top2: 0.8848 - 14ms/step
-    step 680/938 - loss: 1.4939 - acc_top1: 0.8504 - acc_top2: 0.8864 - 14ms/step
-    step 690/938 - loss: 1.4650 - acc_top1: 0.8520 - acc_top2: 0.8878 - 14ms/step
-    step 700/938 - loss: 1.4934 - acc_top1: 0.8537 - acc_top2: 0.8892 - 14ms/step
-    step 710/938 - loss: 1.5473 - acc_top1: 0.8552 - acc_top2: 0.8905 - 14ms/step
-    step 720/938 - loss: 1.4956 - acc_top1: 0.8568 - acc_top2: 0.8918 - 14ms/step
-    step 730/938 - loss: 1.4644 - acc_top1: 0.8583 - acc_top2: 0.8932 - 14ms/step
-    step 740/938 - loss: 1.4868 - acc_top1: 0.8598 - acc_top2: 0.8946 - 14ms/step
-    step 750/938 - loss: 1.5142 - acc_top1: 0.8613 - acc_top2: 0.8959 - 14ms/step
-    step 760/938 - loss: 1.4656 - acc_top1: 0.8628 - acc_top2: 0.8971 - 14ms/step
-    step 770/938 - loss: 1.5005 - acc_top1: 0.8641 - acc_top2: 0.8983 - 14ms/step
-    step 780/938 - loss: 1.5557 - acc_top1: 0.8653 - acc_top2: 0.8994 - 14ms/step
-    step 790/938 - loss: 1.4687 - acc_top1: 0.8666 - acc_top2: 0.9006 - 14ms/step
-    step 800/938 - loss: 1.4686 - acc_top1: 0.8680 - acc_top2: 0.9017 - 14ms/step
-    step 810/938 - loss: 1.5202 - acc_top1: 0.8693 - acc_top2: 0.9028 - 14ms/step
-    step 820/938 - loss: 1.4773 - acc_top1: 0.8705 - acc_top2: 0.9038 - 14ms/step
-    step 830/938 - loss: 1.4838 - acc_top1: 0.8717 - acc_top2: 0.9049 - 14ms/step
-    step 840/938 - loss: 1.4726 - acc_top1: 0.8728 - acc_top2: 0.9059 - 14ms/step
-    step 850/938 - loss: 1.4734 - acc_top1: 0.8741 - acc_top2: 0.9069 - 14ms/step
-    step 860/938 - loss: 1.4627 - acc_top1: 0.8752 - acc_top2: 0.9078 - 14ms/step
-    step 870/938 - loss: 1.4872 - acc_top1: 0.8763 - acc_top2: 0.9088 - 14ms/step
-    step 880/938 - loss: 1.4916 - acc_top1: 0.8773 - acc_top2: 0.9096 - 14ms/step
-    step 890/938 - loss: 1.4818 - acc_top1: 0.8784 - acc_top2: 0.9105 - 14ms/step
-    step 900/938 - loss: 1.4967 - acc_top1: 0.8794 - acc_top2: 0.9114 - 14ms/step
-    step 910/938 - loss: 1.4614 - acc_top1: 0.8804 - acc_top2: 0.9123 - 14ms/step
-    step 920/938 - loss: 1.4819 - acc_top1: 0.8815 - acc_top2: 0.9132 - 14ms/step
-    step 930/938 - loss: 1.5114 - acc_top1: 0.8824 - acc_top2: 0.9140 - 14ms/step
-    step 938/938 - loss: 1.4621 - acc_top1: 0.8832 - acc_top2: 0.9146 - 14ms/step
-    save checkpoint at /Users/chenlong/online_repo/book/paddle2.0_docs/image_classification/mnist_checkpoint/0
+    step 200/938 - loss: 1.5219 - acc_top1: 0.9829 - acc_top2: 0.9965 - 14ms/step
+    step 400/938 - loss: 1.4765 - acc_top1: 0.9825 - acc_top2: 0.9958 - 13ms/step
+    step 600/938 - loss: 1.4624 - acc_top1: 0.9823 - acc_top2: 0.9953 - 13ms/step
+    step 800/938 - loss: 1.4768 - acc_top1: 0.9829 - acc_top2: 0.9955 - 13ms/step
+    step 938/938 - loss: 1.4612 - acc_top1: 0.9836 - acc_top2: 0.9956 - 13ms/step
     Epoch 2/2
-    step  10/938 - loss: 1.5033 - acc_top1: 0.9734 - acc_top2: 0.9906 - 15ms/step
-    step  20/938 - loss: 1.4812 - acc_top1: 0.9734 - acc_top2: 0.9906 - 14ms/step
-    step  30/938 - loss: 1.4623 - acc_top1: 0.9714 - acc_top2: 0.9911 - 14ms/step
-    step  40/938 - loss: 1.4775 - acc_top1: 0.9711 - acc_top2: 0.9918 - 14ms/step
-    step  50/938 - loss: 1.4857 - acc_top1: 0.9712 - acc_top2: 0.9922 - 14ms/step
-    step  60/938 - loss: 1.4895 - acc_top1: 0.9695 - acc_top2: 0.9904 - 14ms/step
-    step  70/938 - loss: 1.4746 - acc_top1: 0.9708 - acc_top2: 0.9908 - 14ms/step
-    step  80/938 - loss: 1.4945 - acc_top1: 0.9719 - acc_top2: 0.9912 - 14ms/step
-    step  90/938 - loss: 1.4644 - acc_top1: 0.9722 - acc_top2: 0.9911 - 14ms/step
-    step 100/938 - loss: 1.4727 - acc_top1: 0.9722 - acc_top2: 0.9912 - 14ms/step
-    step 110/938 - loss: 1.4634 - acc_top1: 0.9720 - acc_top2: 0.9915 - 14ms/step
-    step 120/938 - loss: 1.4856 - acc_top1: 0.9730 - acc_top2: 0.9915 - 14ms/step
-    step 130/938 - loss: 1.4778 - acc_top1: 0.9736 - acc_top2: 0.9916 - 14ms/step
-    step 140/938 - loss: 1.4949 - acc_top1: 0.9730 - acc_top2: 0.9914 - 14ms/step
-    step 150/938 - loss: 1.4836 - acc_top1: 0.9726 - acc_top2: 0.9914 - 14ms/step
-    step 160/938 - loss: 1.5430 - acc_top1: 0.9725 - acc_top2: 0.9917 - 14ms/step
-    step 170/938 - loss: 1.4882 - acc_top1: 0.9722 - acc_top2: 0.9916 - 14ms/step
-    step 180/938 - loss: 1.4777 - acc_top1: 0.9721 - acc_top2: 0.9919 - 14ms/step
-    step 190/938 - loss: 1.4816 - acc_top1: 0.9723 - acc_top2: 0.9920 - 14ms/step
-    step 200/938 - loss: 1.4916 - acc_top1: 0.9730 - acc_top2: 0.9923 - 14ms/step
-    step 210/938 - loss: 1.5290 - acc_top1: 0.9734 - acc_top2: 0.9923 - 14ms/step
-    step 220/938 - loss: 1.5006 - acc_top1: 0.9736 - acc_top2: 0.9923 - 14ms/step
-    step 230/938 - loss: 1.5103 - acc_top1: 0.9737 - acc_top2: 0.9923 - 14ms/step
-    step 240/938 - loss: 1.4905 - acc_top1: 0.9733 - acc_top2: 0.9920 - 14ms/step
-    step 250/938 - loss: 1.5066 - acc_top1: 0.9734 - acc_top2: 0.9920 - 14ms/step
-    step 260/938 - loss: 1.4846 - acc_top1: 0.9736 - acc_top2: 0.9920 - 14ms/step
-    step 270/938 - loss: 1.4717 - acc_top1: 0.9738 - acc_top2: 0.9921 - 14ms/step
-    step 280/938 - loss: 1.4648 - acc_top1: 0.9742 - acc_top2: 0.9921 - 14ms/step
-    step 290/938 - loss: 1.4657 - acc_top1: 0.9745 - acc_top2: 0.9921 - 14ms/step
-    step 300/938 - loss: 1.4630 - acc_top1: 0.9744 - acc_top2: 0.9920 - 14ms/step
-    step 310/938 - loss: 1.5053 - acc_top1: 0.9742 - acc_top2: 0.9918 - 14ms/step
-    step 320/938 - loss: 1.4843 - acc_top1: 0.9745 - acc_top2: 0.9919 - 14ms/step
-    step 330/938 - loss: 1.4915 - acc_top1: 0.9745 - acc_top2: 0.9919 - 14ms/step
-    step 340/938 - loss: 1.5146 - acc_top1: 0.9745 - acc_top2: 0.9918 - 14ms/step
-    step 350/938 - loss: 1.4768 - acc_top1: 0.9742 - acc_top2: 0.9916 - 14ms/step
-    step 360/938 - loss: 1.4827 - acc_top1: 0.9743 - acc_top2: 0.9918 - 14ms/step
-    step 370/938 - loss: 1.5097 - acc_top1: 0.9740 - acc_top2: 0.9917 - 14ms/step
-    step 380/938 - loss: 1.5225 - acc_top1: 0.9739 - acc_top2: 0.9916 - 14ms/step
-    step 390/938 - loss: 1.4701 - acc_top1: 0.9740 - acc_top2: 0.9917 - 14ms/step
-    step 400/938 - loss: 1.4986 - acc_top1: 0.9741 - acc_top2: 0.9920 - 14ms/step
-    step 410/938 - loss: 1.5210 - acc_top1: 0.9740 - acc_top2: 0.9918 - 14ms/step
-    step 420/938 - loss: 1.4799 - acc_top1: 0.9740 - acc_top2: 0.9917 - 14ms/step
-    step 430/938 - loss: 1.4845 - acc_top1: 0.9744 - acc_top2: 0.9919 - 14ms/step
-    step 440/938 - loss: 1.4773 - acc_top1: 0.9741 - acc_top2: 0.9918 - 14ms/step
-    step 450/938 - loss: 1.4719 - acc_top1: 0.9743 - acc_top2: 0.9918 - 14ms/step
-    step 460/938 - loss: 1.4773 - acc_top1: 0.9742 - acc_top2: 0.9918 - 14ms/step
-    step 470/938 - loss: 1.4944 - acc_top1: 0.9741 - acc_top2: 0.9918 - 14ms/step
-    step 480/938 - loss: 1.4793 - acc_top1: 0.9743 - acc_top2: 0.9919 - 14ms/step
-    step 490/938 - loss: 1.4625 - acc_top1: 0.9746 - acc_top2: 0.9920 - 14ms/step
-    step 500/938 - loss: 1.4829 - acc_top1: 0.9745 - acc_top2: 0.9921 - 14ms/step
-    step 510/938 - loss: 1.4659 - acc_top1: 0.9747 - acc_top2: 0.9921 - 14ms/step
-    step 520/938 - loss: 1.4862 - acc_top1: 0.9743 - acc_top2: 0.9921 - 14ms/step
-    step 530/938 - loss: 1.5039 - acc_top1: 0.9742 - acc_top2: 0.9921 - 14ms/step
-    step 540/938 - loss: 1.5070 - acc_top1: 0.9740 - acc_top2: 0.9921 - 14ms/step
-    step 550/938 - loss: 1.5033 - acc_top1: 0.9740 - acc_top2: 0.9922 - 14ms/step
-    step 560/938 - loss: 1.4846 - acc_top1: 0.9741 - acc_top2: 0.9921 - 14ms/step
-    step 570/938 - loss: 1.4613 - acc_top1: 0.9741 - acc_top2: 0.9921 - 14ms/step
-    step 580/938 - loss: 1.4616 - acc_top1: 0.9743 - acc_top2: 0.9921 - 14ms/step
-    step 590/938 - loss: 1.4801 - acc_top1: 0.9745 - acc_top2: 0.9921 - 14ms/step
-    step 600/938 - loss: 1.4772 - acc_top1: 0.9746 - acc_top2: 0.9921 - 14ms/step
-    step 610/938 - loss: 1.4612 - acc_top1: 0.9746 - acc_top2: 0.9921 - 14ms/step
-    step 620/938 - loss: 1.4951 - acc_top1: 0.9746 - acc_top2: 0.9922 - 14ms/step
-    step 630/938 - loss: 1.4755 - acc_top1: 0.9747 - acc_top2: 0.9923 - 14ms/step
-    step 640/938 - loss: 1.5296 - acc_top1: 0.9749 - acc_top2: 0.9924 - 14ms/step
-    step 650/938 - loss: 1.5054 - acc_top1: 0.9748 - acc_top2: 0.9924 - 14ms/step
-    step 660/938 - loss: 1.4775 - acc_top1: 0.9749 - acc_top2: 0.9925 - 14ms/step
-    step 670/938 - loss: 1.4829 - acc_top1: 0.9749 - acc_top2: 0.9925 - 14ms/step
-    step 680/938 - loss: 1.4612 - acc_top1: 0.9750 - acc_top2: 0.9926 - 14ms/step
-    step 690/938 - loss: 1.4869 - acc_top1: 0.9751 - acc_top2: 0.9926 - 14ms/step
-    step 700/938 - loss: 1.4612 - acc_top1: 0.9752 - acc_top2: 0.9927 - 14ms/step
-    step 710/938 - loss: 1.5235 - acc_top1: 0.9752 - acc_top2: 0.9927 - 14ms/step
-    step 720/938 - loss: 1.5317 - acc_top1: 0.9752 - acc_top2: 0.9926 - 14ms/step
-    step 730/938 - loss: 1.4898 - acc_top1: 0.9751 - acc_top2: 0.9926 - 14ms/step
-    step 740/938 - loss: 1.4612 - acc_top1: 0.9753 - acc_top2: 0.9926 - 14ms/step
-    step 750/938 - loss: 1.4935 - acc_top1: 0.9752 - acc_top2: 0.9926 - 14ms/step
-    step 760/938 - loss: 1.5140 - acc_top1: 0.9749 - acc_top2: 0.9926 - 14ms/step
-    step 770/938 - loss: 1.4883 - acc_top1: 0.9748 - acc_top2: 0.9925 - 14ms/step
-    step 780/938 - loss: 1.4759 - acc_top1: 0.9748 - acc_top2: 0.9926 - 14ms/step
-    step 790/938 - loss: 1.4773 - acc_top1: 0.9750 - acc_top2: 0.9926 - 14ms/step
-    step 800/938 - loss: 1.4766 - acc_top1: 0.9750 - acc_top2: 0.9926 - 14ms/step
-    step 810/938 - loss: 1.5058 - acc_top1: 0.9750 - acc_top2: 0.9927 - 14ms/step
-    step 820/938 - loss: 1.4867 - acc_top1: 0.9749 - acc_top2: 0.9927 - 14ms/step
-    step 830/938 - loss: 1.4766 - acc_top1: 0.9748 - acc_top2: 0.9927 - 14ms/step
-    step 840/938 - loss: 1.4680 - acc_top1: 0.9747 - acc_top2: 0.9927 - 14ms/step
-    step 850/938 - loss: 1.4628 - acc_top1: 0.9746 - acc_top2: 0.9927 - 14ms/step
-    step 860/938 - loss: 1.5035 - acc_top1: 0.9747 - acc_top2: 0.9928 - 14ms/step
-    step 870/938 - loss: 1.4857 - acc_top1: 0.9748 - acc_top2: 0.9928 - 14ms/step
-    step 880/938 - loss: 1.4767 - acc_top1: 0.9748 - acc_top2: 0.9927 - 14ms/step
-    step 890/938 - loss: 1.4612 - acc_top1: 0.9750 - acc_top2: 0.9928 - 14ms/step
-    step 900/938 - loss: 1.4620 - acc_top1: 0.9751 - acc_top2: 0.9928 - 14ms/step
-    step 910/938 - loss: 1.4621 - acc_top1: 0.9751 - acc_top2: 0.9928 - 14ms/step
-    step 920/938 - loss: 1.4768 - acc_top1: 0.9751 - acc_top2: 0.9927 - 14ms/step
-    step 930/938 - loss: 1.4806 - acc_top1: 0.9752 - acc_top2: 0.9928 - 14ms/step
-    step 938/938 - loss: 1.4910 - acc_top1: 0.9752 - acc_top2: 0.9928 - 14ms/step
-    save checkpoint at /Users/chenlong/online_repo/book/paddle2.0_docs/image_classification/mnist_checkpoint/1
-    save checkpoint at /Users/chenlong/online_repo/book/paddle2.0_docs/image_classification/mnist_checkpoint/final
+    step 200/938 - loss: 1.4705 - acc_top1: 0.9834 - acc_top2: 0.9959 - 13ms/step
+    step 400/938 - loss: 1.4620 - acc_top1: 0.9833 - acc_top2: 0.9960 - 13ms/step
+    step 600/938 - loss: 1.4613 - acc_top1: 0.9830 - acc_top2: 0.9960 - 13ms/step
+    step 800/938 - loss: 1.4763 - acc_top1: 0.9831 - acc_top2: 0.9960 - 13ms/step
+    step 938/938 - loss: 1.4924 - acc_top1: 0.9834 - acc_top2: 0.9959 - 13ms/step
 
 
 使用model.evaluate来预测模型
@@ -459,28 +257,20 @@ MNIST数据集使用LeNet进行图像分类
 
 .. code:: ipython3
 
-    model.evaluate(test_dataset, batch_size=64)
+    model.evaluate(test_dataset, log_freq=20, batch_size=64)
 
 
 .. parsed-literal::
 
     Eval begin...
-    step  10/157 - loss: 1.5014 - acc_top1: 0.9766 - acc_top2: 0.9953 - 6ms/step
-    step  20/157 - loss: 1.5239 - acc_top1: 0.9742 - acc_top2: 0.9922 - 6ms/step
-    step  30/157 - loss: 1.4926 - acc_top1: 0.9740 - acc_top2: 0.9932 - 6ms/step
-    step  40/157 - loss: 1.4612 - acc_top1: 0.9734 - acc_top2: 0.9938 - 6ms/step
-    step  50/157 - loss: 1.4612 - acc_top1: 0.9719 - acc_top2: 0.9938 - 6ms/step
-    step  60/157 - loss: 1.5114 - acc_top1: 0.9721 - acc_top2: 0.9938 - 6ms/step
-    step  70/157 - loss: 1.4793 - acc_top1: 0.9696 - acc_top2: 0.9935 - 6ms/step
-    step  80/157 - loss: 1.4736 - acc_top1: 0.9695 - acc_top2: 0.9932 - 6ms/step
-    step  90/157 - loss: 1.4892 - acc_top1: 0.9720 - acc_top2: 0.9939 - 6ms/step
-    step 100/157 - loss: 1.4623 - acc_top1: 0.9738 - acc_top2: 0.9941 - 6ms/step
-    step 110/157 - loss: 1.4612 - acc_top1: 0.9737 - acc_top2: 0.9939 - 6ms/step
-    step 120/157 - loss: 1.4612 - acc_top1: 0.9746 - acc_top2: 0.9939 - 6ms/step
-    step 130/157 - loss: 1.4703 - acc_top1: 0.9757 - acc_top2: 0.9942 - 6ms/step
-    step 140/157 - loss: 1.4612 - acc_top1: 0.9771 - acc_top2: 0.9946 - 6ms/step
-    step 150/157 - loss: 1.4748 - acc_top1: 0.9782 - acc_top2: 0.9950 - 6ms/step
-    step 157/157 - loss: 1.4612 - acc_top1: 0.9770 - acc_top2: 0.9949 - 6ms/step
+    step  20/157 - loss: 1.5246 - acc_top1: 0.9773 - acc_top2: 0.9969 - 6ms/step
+    step  40/157 - loss: 1.4622 - acc_top1: 0.9758 - acc_top2: 0.9961 - 6ms/step
+    step  60/157 - loss: 1.5241 - acc_top1: 0.9763 - acc_top2: 0.9951 - 6ms/step
+    step  80/157 - loss: 1.4612 - acc_top1: 0.9787 - acc_top2: 0.9959 - 6ms/step
+    step 100/157 - loss: 1.4612 - acc_top1: 0.9823 - acc_top2: 0.9967 - 5ms/step
+    step 120/157 - loss: 1.4612 - acc_top1: 0.9835 - acc_top2: 0.9966 - 5ms/step
+    step 140/157 - loss: 1.4612 - acc_top1: 0.9844 - acc_top2: 0.9969 - 5ms/step
+    step 157/157 - loss: 1.4612 - acc_top1: 0.9838 - acc_top2: 0.9966 - 5ms/step
     Eval samples: 10000
 
 
@@ -488,7 +278,7 @@ MNIST数据集使用LeNet进行图像分类
 
 .. parsed-literal::
 
-    {'loss': [1.4611504], 'acc_top1': 0.977, 'acc_top2': 0.9949}
+    {'loss': [1.4611504], 'acc_top1': 0.9838, 'acc_top2': 0.9966}
 
 
 

doc/paddle/tutorial/nlp_case/imdb_bow_classification/imdb_bow_classification.ipynb

@@ -9,31 +9,28 @@
     "本示例教程演示如何在IMDB数据集上用简单的BOW网络完成文本分类的任务。\n",
     "\n",
     "IMDB数据集是一个对电影评论标注为正向评论与负向评论的数据集，共有25000条文本数据作为训练集，25000条文本数据作为测试集。\n",
-    "该数据集的官方地址为： http://ai.stanford.edu/~amaas/data/sentiment/\n",
-    "\n",
-    "- Warning: `paddle.dataset.imdb`先在是一个非常粗野的实现，后续需要有替代的方案。"
+    "该数据集的官方地址为： http://ai.stanford.edu/~amaas/data/sentiment/"
    ]
   },
   {
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# 环境设置\n",
+    "## 环境设置\n",
     "\n",
     "本示例基于飞桨开源框架2.0版本。"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 4,
+   "execution_count": 2,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "0.0.0\n",
-      "264e76cae6861ad9b1d4bcd8c3212f7a78c01e4d\n"
+      "2.0.0-beta0\n"
      ]
     }
    ],
@@ -42,22 +39,21 @@
     "import numpy as np\n",
     "\n",
     "paddle.disable_static()\n",
-    "print(paddle.__version__)\n",
-    "print(paddle.__git_commit__)\n"
+    "print(paddle.__version__)"
    ]
   },
   {
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# 加载数据\n",
+    "## 加载数据\n",
     "\n",
     "我们会使用`paddle.dataset`完成数据下载，构建字典和准备数据读取器。在飞桨2.0版本中，推荐使用padding的方式来对同一个batch中长度不一的数据进行补齐，所以在字典中，我们还会添加一个特殊的`<pad>`词，用来在后续对batch中较短的句子进行填充。"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 5,
+   "execution_count": 3,
    "metadata": {},
    "outputs": [
     {
@@ -78,7 +74,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 6,
+   "execution_count": 4,
    "metadata": {},
    "outputs": [
     {
@@ -119,14 +115,14 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# 参数设置\n",
+    "## 参数设置\n",
     "\n",
     "在这里我们设置一下词表大小，`embedding`的大小，batch_size，等等"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 7,
+   "execution_count": 22,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -157,7 +153,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 8,
+   "execution_count": 23,
    "metadata": {},
    "outputs": [
     {
@@ -183,14 +179,14 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# 用padding的方式对齐数据\n",
+    "## 用padding的方式对齐数据\n",
     "\n",
     "文本数据中，每一句话的长度都是不一样的，为了方便后续的神经网络的计算，常见的处理方式是把数据集中的数据都统一成同样长度的数据。这包括：对于较长的数据进行截断处理，对于较短的数据用特殊的词`<pad>`进行填充。接下来的代码会对数据集中的数据进行这样的处理。"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 9,
+   "execution_count": 24,
    "metadata": {},
    "outputs": [
     {
@@ -234,14 +230,14 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# 组建网络\n",
+    "## 组建网络\n",
     "\n",
     "本示例中，我们将会使用一个不考虑词的顺序的BOW的网络，在查找到每个词对应的embedding后，简单的取平均，作为一个句子的表示。然后用`Linear`进行线性变换。为了防止过拟合，我们还使用了`Dropout`。"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 11,
+   "execution_count": 25,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -264,24 +260,24 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# 开始模型的训练\n"
+    "## 开始模型的训练\n"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 13,
+   "execution_count": 26,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "epoch: 0, batch_id: 0, loss is: [0.6926701]\n",
-      "epoch: 0, batch_id: 500, loss is: [0.41248566]\n",
-      "[validation] accuracy/loss: 0.8505121469497681/0.3615057170391083\n",
-      "epoch: 1, batch_id: 0, loss is: [0.29521096]\n",
-      "epoch: 1, batch_id: 500, loss is: [0.2916747]\n",
-      "[validation] accuracy/loss: 0.86475670337677/0.3259459137916565\n"
+      "epoch: 0, batch_id: 0, loss is: [0.6918494]\n",
+      "epoch: 0, batch_id: 500, loss is: [0.33142853]\n",
+      "[validation] accuracy/loss: 0.8506321907043457/0.3620821535587311\n",
+      "epoch: 1, batch_id: 0, loss is: [0.37161]\n",
+      "epoch: 1, batch_id: 500, loss is: [0.2296829]\n",
+      "[validation] accuracy/loss: 0.8622759580612183/0.3286365270614624\n"
      ]
     }
    ],
@@ -311,8 +307,8 @@
     "            if batch_id % 500 == 0:\n",
     "                print(\"epoch: {}, batch_id: {}, loss is: {}\".format(epoch, batch_id, avg_loss.numpy()))\n",
     "            avg_loss.backward()\n",
-    "            opt.minimize(avg_loss)\n",
-    "            model.clear_gradients()\n",
+    "            opt.step()\n",
+    "            opt.clear_grad()\n",
     "\n",
     "        # evaluate model after one epoch\n",
     "        model.eval()\n",
@@ -345,17 +341,10 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# The End\n",
+    "## The End\n",
     "\n",
     "可以看到，在这个数据集上，经过两轮的迭代可以得到86%左右的准确率。你也可以通过调整网络结构和超参数，来获得更好的效果。"
    ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": []
   }
  ],
  "metadata": {
@@ -369,8 +358,20 @@
    "display_name": "Python 3",
    "language": "python",
    "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.7.3"
   }
  },
  "nbformat": 4,
- "nbformat_minor": 1
+ "nbformat_minor": 4
 }

doc/paddle/tutorial/nlp_case/imdb_bow_classification/imdb_bow_classification.rst

@@ -6,29 +6,23 @@ IMDB 数据集使用BOW网络的文本分类
 IMDB数据集是一个对电影评论标注为正向评论与负向评论的数据集，共有25000条文本数据作为训练集，25000条文本数据作为测试集。
 该数据集的官方地址为： http://ai.stanford.edu/~amaas/data/sentiment/
 
--  Warning:
-   ``paddle.dataset.imdb``\ 先在是一个非常粗野的实现，后续需要有替代的方案。
-
 环境设置
 --------
 
 本示例基于飞桨开源框架2.0版本。
 
-.. code:: 
+.. code:: ipython3
 
     import paddle
     import numpy as np
     
     paddle.disable_static()
     print(paddle.__version__)
-    print(paddle.__git_commit__)
-
 
 
 .. parsed-literal::
 
-    0.0.0
-    264e76cae6861ad9b1d4bcd8c3212f7a78c01e4d
+    2.0.0-beta0
 
 
 加载数据
@@ -36,7 +30,7 @@ IMDB数据集是一个对电影评论标注为正向评论与负向评论的数
 
 我们会使用\ ``paddle.dataset``\ 完成数据下载，构建字典和准备数据读取器。在飞桨2.0版本中，推荐使用padding的方式来对同一个batch中长度不一的数据进行补齐，所以在字典中，我们还会添加一个特殊的\ ``<pad>``\ 词，用来在后续对batch中较短的句子进行填充。
 
-.. code:: 
+.. code:: ipython3
 
     print("Loading IMDB word dict....")
     word_dict = paddle.dataset.imdb.word_dict()
@@ -51,7 +45,7 @@ IMDB数据集是一个对电影评论标注为正向评论与负向评论的数
     Loading IMDB word dict....
 
 
-.. code:: 
+.. code:: ipython3
 
     # add a pad token to the dict for later padding the sequence
     word_dict['<pad>'] = len(word_dict)
@@ -88,7 +82,7 @@ IMDB数据集是一个对电影评论标注为正向评论与负向评论的数
 
 在这里我们设置一下词表大小，\ ``embedding``\ 的大小，batch_size，等等
 
-.. code:: 
+.. code:: ipython3
 
     vocab_size = len(word_dict)
     emb_size = 256
@@ -109,7 +103,7 @@ IMDB数据集是一个对电影评论标注为正向评论与负向评论的数
 
 在这里，取出一条数据打印出来看看，可以对数据有一个初步直观的印象。
 
-.. code:: 
+.. code:: ipython3
 
     # 取出来第一条数据看看样子。
     sent, label = next(train_reader())
@@ -127,11 +121,11 @@ IMDB数据集是一个对电影评论标注为正向评论与负向评论的数
 
 
 用padding的方式对齐数据
-----------------------------
+-----------------------
 
 文本数据中，每一句话的长度都是不一样的，为了方便后续的神经网络的计算，常见的处理方式是把数据集中的数据都统一成同样长度的数据。这包括：对于较长的数据进行截断处理，对于较短的数据用特殊的词\ ``<pad>``\ 进行填充。接下来的代码会对数据集中的数据进行这样的处理。
 
-.. code:: 
+.. code:: ipython3
 
     def create_padded_dataset(reader):
         padded_sents = []
@@ -172,7 +166,7 @@ IMDB数据集是一个对电影评论标注为正向评论与负向评论的数
 
 本示例中，我们将会使用一个不考虑词的顺序的BOW的网络，在查找到每个词对应的embedding后，简单的取平均，作为一个句子的表示。然后用\ ``Linear``\ 进行线性变换。为了防止过拟合，我们还使用了\ ``Dropout``\ 。
 
-.. code:: 
+.. code:: ipython3
 
     class MyNet(paddle.nn.Layer):
         def __init__(self):
@@ -191,7 +185,7 @@ IMDB数据集是一个对电影评论标注为正向评论与负向评论的数
 开始模型的训练
 --------------
 
-.. code:: 
+.. code:: ipython3
 
     def train(model):
         model.train()
@@ -218,8 +212,8 @@ IMDB数据集是一个对电影评论标注为正向评论与负向评论的数
                 if batch_id % 500 == 0:
                     print("epoch: {}, batch_id: {}, loss is: {}".format(epoch, batch_id, avg_loss.numpy()))
                 avg_loss.backward()
-                opt.minimize(avg_loss)
-                model.clear_gradients()
+                opt.step()
+                opt.clear_grad()
     
             # evaluate model after one epoch
             model.eval()
@@ -250,16 +244,15 @@ IMDB数据集是一个对电影评论标注为正向评论与负向评论的数
 
 .. parsed-literal::
 
-    epoch: 0, batch_id: 0, loss is: [0.6926701]
-    epoch: 0, batch_id: 500, loss is: [0.41248566]
-    [validation] accuracy/loss: 0.8505121469497681/0.3615057170391083
-    epoch: 1, batch_id: 0, loss is: [0.29521096]
-    epoch: 1, batch_id: 500, loss is: [0.2916747]
-    [validation] accuracy/loss: 0.86475670337677/0.3259459137916565
+    epoch: 0, batch_id: 0, loss is: [0.6918494]
+    epoch: 0, batch_id: 500, loss is: [0.33142853]
+    [validation] accuracy/loss: 0.8506321907043457/0.3620821535587311
+    epoch: 1, batch_id: 0, loss is: [0.37161]
+    epoch: 1, batch_id: 500, loss is: [0.2296829]
+    [validation] accuracy/loss: 0.8622759580612183/0.3286365270614624
 
 
 The End
---------
+-------
 
 可以看到，在这个数据集上，经过两轮的迭代可以得到86%左右的准确率。你也可以通过调整网络结构和超参数，来获得更好的效果。
-

doc/paddle/tutorial/nlp_case/n_gram_model/n_gram_model.ipynb

@@ -16,21 +16,21 @@
    "metadata": {},
    "source": [
     "## 环境\n",
-    "本教程基于paddle-develop编写，如果您的环境不是本版本，请先安装paddle-develop。"
+    "本教程基于paddle-2.0-beta编写，如果您的环境不是本版本，请先安装paddle-2.0-beta。"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 23,
+   "execution_count": 1,
    "metadata": {},
    "outputs": [
     {
      "data": {
       "text/plain": [
-       "'0.0.0'"
+       "'2.0.0-beta0'"
       ]
      },
-     "execution_count": 23,
+     "execution_count": 1,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -51,23 +51,22 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 24,
+   "execution_count": 3,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "--2020-09-09 14:58:26--  https://ocw.mit.edu/ans7870/6/6.006/s08/lecturenotes/files/t8.shakespeare.txt\n",
-      "正在解析主机 ocw.mit.edu (ocw.mit.edu)... 151.101.110.133\n",
-      "正在连接 ocw.mit.edu (ocw.mit.edu)|151.101.110.133|:443... 已连接。\n",
-      "已发出 HTTP 请求，正在等待回应... 200 OK\n",
+      "--2020-09-12 13:49:29--  https://ocw.mit.edu/ans7870/6/6.006/s08/lecturenotes/files/t8.shakespeare.txt\n",
+      "正在连接 172.19.57.45:3128... 已连接。\n",
+      "已发出 Proxy 请求，正在等待回应... 200 OK\n",
       "长度：5458199 (5.2M) [text/plain]\n",
       "正在保存至: “t8.shakespeare.txt”\n",
       "\n",
-      "t8.shakespeare.txt  100%[===================>]   5.21M  94.1KB/s  用时 70s       \n",
+      "t8.shakespeare.txt  100%[===================>]   5.21M  2.01MB/s  用时 2.6s      \n",
       "\n",
-      "2020-09-09 14:59:38 (75.7 KB/s) - 已保存 “t8.shakespeare.txt” [5458199/5458199])\n",
+      "2020-09-12 13:49:33 (2.01 MB/s) - 已保存 “t8.shakespeare.txt” [5458199/5458199])\n",
       "\n"
      ]
     }
@@ -197,7 +196,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 12,
+   "execution_count": 9,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -233,12 +232,13 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 13,
+   "execution_count": 30,
    "metadata": {},
    "outputs": [],
    "source": [
     "import paddle\n",
     "import numpy as np\n",
+    "import paddle.nn.functional as F\n",
     "hidden_size = 1024\n",
     "class NGramModel(paddle.nn.Layer):\n",
     "    def __init__(self, vocab_size, embedding_dim, context_size):\n",
@@ -251,7 +251,7 @@
     "        x = self.embedding(x)\n",
     "        x = paddle.reshape(x, [-1, context_size * embedding_dim])\n",
     "        x = self.linear1(x)\n",
-    "        x = paddle.nn.functional.relu(x)\n",
+    "        x = F.relu(x)\n",
     "        x = self.linear2(x)\n",
     "        return x"
    ]
@@ -265,33 +265,34 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 19,
+   "execution_count": 31,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "epoch: 0, batch_id: 0, loss is: [10.252193]\n",
-      "epoch: 0, batch_id: 500, loss is: [6.894636]\n",
-      "epoch: 0, batch_id: 1000, loss is: [6.849346]\n",
-      "epoch: 0, batch_id: 1500, loss is: [6.931605]\n",
-      "epoch: 0, batch_id: 2000, loss is: [6.6860313]\n",
-      "epoch: 0, batch_id: 2500, loss is: [6.2472367]\n",
-      "epoch: 0, batch_id: 3000, loss is: [6.8818874]\n",
-      "epoch: 0, batch_id: 3500, loss is: [6.941615]\n",
-      "epoch: 1, batch_id: 0, loss is: [6.3628616]\n",
-      "epoch: 1, batch_id: 500, loss is: [6.2065206]\n",
-      "epoch: 1, batch_id: 1000, loss is: [6.5334334]\n",
-      "epoch: 1, batch_id: 1500, loss is: [6.5788]\n",
-      "epoch: 1, batch_id: 2000, loss is: [6.352103]\n",
-      "epoch: 1, batch_id: 2500, loss is: [6.6272373]\n",
-      "epoch: 1, batch_id: 3000, loss is: [6.801074]\n",
-      "epoch: 1, batch_id: 3500, loss is: [6.2274427]\n"
+      "epoch: 0, batch_id: 0, loss is: [10.252176]\n",
+      "epoch: 0, batch_id: 500, loss is: [6.6429553]\n",
+      "epoch: 0, batch_id: 1000, loss is: [6.801544]\n",
+      "epoch: 0, batch_id: 1500, loss is: [6.7114644]\n",
+      "epoch: 0, batch_id: 2000, loss is: [6.628998]\n",
+      "epoch: 0, batch_id: 2500, loss is: [6.511376]\n",
+      "epoch: 0, batch_id: 3000, loss is: [6.878798]\n",
+      "epoch: 0, batch_id: 3500, loss is: [6.8752203]\n",
+      "epoch: 1, batch_id: 0, loss is: [6.5908413]\n",
+      "epoch: 1, batch_id: 500, loss is: [6.9765778]\n",
+      "epoch: 1, batch_id: 1000, loss is: [6.603841]\n",
+      "epoch: 1, batch_id: 1500, loss is: [6.9935036]\n",
+      "epoch: 1, batch_id: 2000, loss is: [6.751287]\n",
+      "epoch: 1, batch_id: 2500, loss is: [7.1222277]\n",
+      "epoch: 1, batch_id: 3000, loss is: [6.6431484]\n",
+      "epoch: 1, batch_id: 3500, loss is: [6.6024966]\n"
      ]
     }
    ],
    "source": [
+    "import paddle.nn.functional as F\n",
     "vocab_size = len(vocab)\n",
     "epochs = 2\n",
     "losses = []\n",
@@ -303,15 +304,15 @@
     "            x_data = data[0]\n",
     "            y_data = data[1]\n",
     "            predicts = model(x_data)\n",
-    "            y_data = paddle.reshape(y_data, ([-1, 1]))\n",
-    "            loss = paddle.nn.functional.softmax_with_cross_entropy(predicts, y_data)\n",
+    "            y_data = paddle.reshape(y_data, shape=[-1, 1])\n",
+    "            loss = F.softmax_with_cross_entropy(predicts, y_data)\n",
     "            avg_loss = paddle.mean(loss)\n",
     "            avg_loss.backward()\n",
     "            if batch_id % 500 == 0:\n",
     "                losses.append(avg_loss.numpy())\n",
     "                print(\"epoch: {}, batch_id: {}, loss is: {}\".format(epoch, batch_id, avg_loss.numpy())) \n",
-    "            optim.minimize(avg_loss)\n",
-    "            model.clear_gradients()\n",
+    "            optim.step()\n",
+    "            optim.clear_grad()\n",
     "model = NGramModel(vocab_size, embedding_dim, context_size)\n",
     "train(model)"
    ]
@@ -326,22 +327,22 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 20,
+   "execution_count": 32,
    "metadata": {},
    "outputs": [
     {
      "data": {
       "text/plain": [
-       "[<matplotlib.lines.Line2D at 0x14e27b3c8>]"
+       "[<matplotlib.lines.Line2D at 0x15c295cc0>]"
       ]
      },
-     "execution_count": 20,
+     "execution_count": 32,
      "metadata": {},
      "output_type": "execute_result"
     },
     {
      "data": {
-      "image/png": "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\n",
+      "image/png": "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\n",
       "text/plain": [
        "<Figure size 432x288 with 1 Axes>"
       ]
@@ -371,7 +372,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 22,
+   "execution_count": 36,
    "metadata": {},
    "outputs": [
     {

doc/paddle/tutorial/nlp_case/n_gram_model/n_gram_model.rst

@@ -10,7 +10,7 @@ trigram，以此类推。实际应用通常采用 bigram 和 trigram 进行计
 环境
 ----
 
-本教程基于paddle-develop编写，如果您的环境不是本版本，请先安装paddle-develop。
+本教程基于paddle-2.0-beta编写，如果您的环境不是本版本，请先安装paddle-2.0-beta。
 
 .. code:: ipython3
 
@@ -22,7 +22,7 @@ trigram，以此类推。实际应用通常采用 bigram 和 trigram 进行计
 
 .. parsed-literal::
 
-    '0.0.0'
+    '2.0.0-beta0'
 
 
 
@@ -39,16 +39,15 @@ context_size设为2，意味着是trigram。embedding_dim设为256。
 
 .. parsed-literal::
 
-    --2020-09-09 14:58:26--  https://ocw.mit.edu/ans7870/6/6.006/s08/lecturenotes/files/t8.shakespeare.txt
-    正在解析主机 ocw.mit.edu (ocw.mit.edu)... 151.101.110.133
-    正在连接 ocw.mit.edu (ocw.mit.edu)|151.101.110.133|:443... 已连接。
-    已发出 HTTP 请求，正在等待回应... 200 OK
+    --2020-09-12 13:49:29--  https://ocw.mit.edu/ans7870/6/6.006/s08/lecturenotes/files/t8.shakespeare.txt
+    正在连接 172.19.57.45:3128... 已连接。
+    已发出 Proxy 请求，正在等待回应... 200 OK
     长度：5458199 (5.2M) [text/plain]
     正在保存至: “t8.shakespeare.txt”
     
-    t8.shakespeare.txt  100%[===================>]   5.21M  94.1KB/s  用时 70s       
+    t8.shakespeare.txt  100%[===================>]   5.21M  2.01MB/s  用时 2.6s      
     
-    2020-09-09 14:59:38 (75.7 KB/s) - 已保存 “t8.shakespeare.txt” [5458199/5458199])
+    2020-09-12 13:49:33 (2.01 MB/s) - 已保存 “t8.shakespeare.txt” [5458199/5458199])
     
 
 
@@ -164,6 +163,7 @@ context_size设为2，意味着是trigram。embedding_dim设为256。
 
     import paddle
     import numpy as np
+    import paddle.nn.functional as F
     hidden_size = 1024
     class NGramModel(paddle.nn.Layer):
         def __init__(self, vocab_size, embedding_dim, context_size):
@@ -176,7 +176,7 @@ context_size设为2，意味着是trigram。embedding_dim设为256。
             x = self.embedding(x)
             x = paddle.reshape(x, [-1, context_size * embedding_dim])
             x = self.linear1(x)
-            x = paddle.nn.functional.relu(x)
+            x = F.relu(x)
             x = self.linear2(x)
             return x
 
@@ -185,6 +185,7 @@ context_size设为2，意味着是trigram。embedding_dim设为256。
 
 .. code:: ipython3
 
+    import paddle.nn.functional as F
     vocab_size = len(vocab)
     epochs = 2
     losses = []
@@ -196,37 +197,37 @@ context_size设为2，意味着是trigram。embedding_dim设为256。
                 x_data = data[0]
                 y_data = data[1]
                 predicts = model(x_data)
-                y_data = paddle.reshape(y_data, ([-1, 1]))
-                loss = paddle.nn.functional.softmax_with_cross_entropy(predicts, y_data)
+                y_data = paddle.reshape(y_data, shape=[-1, 1])
+                loss = F.softmax_with_cross_entropy(predicts, y_data)
                 avg_loss = paddle.mean(loss)
                 avg_loss.backward()
                 if batch_id % 500 == 0:
                     losses.append(avg_loss.numpy())
                     print("epoch: {}, batch_id: {}, loss is: {}".format(epoch, batch_id, avg_loss.numpy())) 
-                optim.minimize(avg_loss)
-                model.clear_gradients()
+                optim.step()
+                optim.clear_grad()
     model = NGramModel(vocab_size, embedding_dim, context_size)
     train(model)
 
 
 .. parsed-literal::
 
-    epoch: 0, batch_id: 0, loss is: [10.252193]
-    epoch: 0, batch_id: 500, loss is: [6.894636]
-    epoch: 0, batch_id: 1000, loss is: [6.849346]
-    epoch: 0, batch_id: 1500, loss is: [6.931605]
-    epoch: 0, batch_id: 2000, loss is: [6.6860313]
-    epoch: 0, batch_id: 2500, loss is: [6.2472367]
-    epoch: 0, batch_id: 3000, loss is: [6.8818874]
-    epoch: 0, batch_id: 3500, loss is: [6.941615]
-    epoch: 1, batch_id: 0, loss is: [6.3628616]
-    epoch: 1, batch_id: 500, loss is: [6.2065206]
-    epoch: 1, batch_id: 1000, loss is: [6.5334334]
-    epoch: 1, batch_id: 1500, loss is: [6.5788]
-    epoch: 1, batch_id: 2000, loss is: [6.352103]
-    epoch: 1, batch_id: 2500, loss is: [6.6272373]
-    epoch: 1, batch_id: 3000, loss is: [6.801074]
-    epoch: 1, batch_id: 3500, loss is: [6.2274427]
+    epoch: 0, batch_id: 0, loss is: [10.252176]
+    epoch: 0, batch_id: 500, loss is: [6.6429553]
+    epoch: 0, batch_id: 1000, loss is: [6.801544]
+    epoch: 0, batch_id: 1500, loss is: [6.7114644]
+    epoch: 0, batch_id: 2000, loss is: [6.628998]
+    epoch: 0, batch_id: 2500, loss is: [6.511376]
+    epoch: 0, batch_id: 3000, loss is: [6.878798]
+    epoch: 0, batch_id: 3500, loss is: [6.8752203]
+    epoch: 1, batch_id: 0, loss is: [6.5908413]
+    epoch: 1, batch_id: 500, loss is: [6.9765778]
+    epoch: 1, batch_id: 1000, loss is: [6.603841]
+    epoch: 1, batch_id: 1500, loss is: [6.9935036]
+    epoch: 1, batch_id: 2000, loss is: [6.751287]
+    epoch: 1, batch_id: 2500, loss is: [7.1222277]
+    epoch: 1, batch_id: 3000, loss is: [6.6431484]
+    epoch: 1, batch_id: 3500, loss is: [6.6024966]
 
 
 打印loss下降曲线
@@ -248,12 +249,12 @@ context_size设为2，意味着是trigram。embedding_dim设为256。
 
 .. parsed-literal::
 
-    [<matplotlib.lines.Line2D at 0x14e27b3c8>]
+    [<matplotlib.lines.Line2D at 0x15c295cc0>]
 
 
 
 
-.. image:: n_gram_model_files/n_gram_model_19_1.png
+.. image:: https://github.com/PaddlePaddle/FluidDoc/tree/develop/doc/paddle/tutorial/nlp_case/n_gram_model/n_gram_model_files/n_gram_model_001.png?raw=true
 
 
 预测

doc/paddle/tutorial/nlp_case/seq2seq_with_attention/seq2seq_with_attention.ipynb

@@ -13,7 +13,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# 环境设置\n",
+    "## 环境设置\n",
     "\n",
     "本示例教程基于飞桨2.0-beta版本。"
    ]
@@ -27,8 +27,7 @@
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "0.0.0\n",
-      "89af2088b6e74bdfeef2d4d78e08461ed2aafee5\n"
+      "2.0.0-beta0\n"
      ]
     }
    ],
@@ -39,15 +38,14 @@
     "import numpy as np\n",
     "\n",
     "paddle.disable_static()\n",
-    "print(paddle.__version__)\n",
-    "print(paddle.__git_commit__)"
+    "print(paddle.__version__)"
    ]
   },
   {
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# 下载数据集\n",
+    "## 下载数据集\n",
     "\n",
     "我们将使用 [http://www.manythings.org/anki/](http://www.manythings.org/anki/) 提供的中英文的英汉句对作为数据集，来完成本任务。该数据集含有23610个中英文双语的句对。"
    ]
@@ -61,16 +59,16 @@
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "--2020-09-04 16:13:35--  https://www.manythings.org/anki/cmn-eng.zip\n",
-      "Resolving www.manythings.org (www.manythings.org)... 104.24.109.196, 172.67.173.198, 2606:4700:3037::6818:6cc4, ...\n",
-      "Connecting to www.manythings.org (www.manythings.org)|104.24.109.196|:443... connected.\n",
+      "--2020-09-10 16:17:25--  https://www.manythings.org/anki/cmn-eng.zip\n",
+      "Resolving www.manythings.org (www.manythings.org)... 2606:4700:3033::6818:6dc4, 2606:4700:3036::ac43:adc6, 2606:4700:3037::6818:6cc4, ...\n",
+      "Connecting to www.manythings.org (www.manythings.org)|2606:4700:3033::6818:6dc4|:443... connected.\n",
       "HTTP request sent, awaiting response... 200 OK\n",
       "Length: 1030722 (1007K) [application/zip]\n",
       "Saving to: ‘cmn-eng.zip’\n",
       "\n",
-      "cmn-eng.zip         100%[===================>]   1007K   520KB/s    in 1.9s    \n",
+      "cmn-eng.zip         100%[===================>]   1007K  91.2KB/s    in 11s     \n",
       "\n",
-      "2020-09-04 16:13:38 (520 KB/s) - ‘cmn-eng.zip’ saved [1030722/1030722]\n",
+      "2020-09-10 16:17:38 (91.2 KB/s) - ‘cmn-eng.zip’ saved [1030722/1030722]\n",
       "\n",
       "Archive:  cmn-eng.zip\n",
       "  inflating: cmn.txt                 \n",
@@ -91,7 +89,7 @@
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "   23610 cmn.txt\r\n"
+      "   23610 cmn.txt\n"
      ]
     }
    ],
@@ -103,7 +101,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# 构建双语句对的数据结构\n",
+    "## 构建双语句对的数据结构\n",
     "\n",
     "接下来我们通过处理下载下来的双语句对的文本文件，将双语句对读入到python的数据结构中。这里做了如下的处理。\n",
     "\n",
@@ -169,7 +167,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "#  创建词表\n",
+    "##  创建词表\n",
     "\n",
     "接下来我们分别创建中英文的词表，这两份词表会用来将英文和中文的句子转换为词的ID构成的序列。词表中还加入了如下三个特殊的词：\n",
     "- `<pad>`: 用来对较短的句子进行填充。\n",
@@ -220,7 +218,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# 创建padding过的数据集\n",
+    "## 创建padding过的数据集\n",
     "\n",
     "接下来根据词表，我们将会创建一份实际的用于训练的用numpy array组织起来的数据集。\n",
     "- 所有的句子都通过`<pad>`补充成为了长度相同的句子。\n",
@@ -271,7 +269,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# 创建网络\n",
+    "## 创建网络\n",
     "\n",
     "我们将会创建一个Encoder-AttentionDecoder架构的模型结构用来完成机器翻译任务。\n",
     "首先我们将设置一些必要的网络结构中用到的参数。"
@@ -296,7 +294,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# Encoder部分\n",
+    "## Encoder部分\n",
     "\n",
     "在编码器的部分，我们通过查找完Embedding之后接一个LSTM的方式构建一个对源语言编码的网络。飞桨的RNN系列的API，除了LSTM之外，还提供了SimleRNN, GRU供使用，同时，还可以使用反向RNN，双向RNN，多层RNN等形式。也可以通过`dropout`参数设置是否对多层RNN的中间层进行`dropout`处理，来防止过拟合。\n",
     "\n",
@@ -328,7 +326,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# AttentionDecoder部分\n",
+    "## AttentionDecoder部分\n",
     "\n",
     "在解码器部分，我们通过一个带有注意力机制的LSTM来完成解码。\n",
     "\n",
@@ -402,7 +400,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# 训练模型\n",
+    "## 训练模型\n",
     "\n",
     "接下来我们开始训练模型。\n",
     "\n",
@@ -421,65 +419,65 @@
      "output_type": "stream",
      "text": [
       "epoch:0\n",
-      "iter 0, loss:[7.6194725]\n",
-      "iter 200, loss:[3.4147663]\n",
+      "iter 0, loss:[7.620109]\n",
+      "iter 200, loss:[2.9760551]\n",
       "epoch:1\n",
-      "iter 0, loss:[3.0931656]\n",
-      "iter 200, loss:[2.7543137]\n",
+      "iter 0, loss:[2.9679596]\n",
+      "iter 200, loss:[3.161064]\n",
       "epoch:2\n",
-      "iter 0, loss:[2.8413522]\n",
-      "iter 200, loss:[2.340513]\n",
+      "iter 0, loss:[2.7516625]\n",
+      "iter 200, loss:[2.9755423]\n",
       "epoch:3\n",
-      "iter 0, loss:[2.597812]\n",
-      "iter 200, loss:[2.5552855]\n",
+      "iter 0, loss:[2.7249248]\n",
+      "iter 200, loss:[2.3419888]\n",
       "epoch:4\n",
-      "iter 0, loss:[2.0783448]\n",
-      "iter 200, loss:[2.4544785]\n",
+      "iter 0, loss:[2.3236473]\n",
+      "iter 200, loss:[2.3453429]\n",
       "epoch:5\n",
-      "iter 0, loss:[1.8709135]\n",
-      "iter 200, loss:[1.8736631]\n",
+      "iter 0, loss:[2.1926975]\n",
+      "iter 200, loss:[2.1977856]\n",
       "epoch:6\n",
-      "iter 0, loss:[1.9589291]\n",
-      "iter 200, loss:[2.119414]\n",
+      "iter 0, loss:[2.014393]\n",
+      "iter 200, loss:[2.1863418]\n",
       "epoch:7\n",
-      "iter 0, loss:[1.5829577]\n",
-      "iter 200, loss:[1.6002902]\n",
+      "iter 0, loss:[1.8619595]\n",
+      "iter 200, loss:[1.8904227]\n",
       "epoch:8\n",
-      "iter 0, loss:[1.6022769]\n",
-      "iter 200, loss:[1.52694]\n",
+      "iter 0, loss:[1.5901132]\n",
+      "iter 200, loss:[1.7812968]\n",
       "epoch:9\n",
-      "iter 0, loss:[1.3616685]\n",
-      "iter 200, loss:[1.5420443]\n",
+      "iter 0, loss:[1.341565]\n",
+      "iter 200, loss:[1.4957166]\n",
       "epoch:10\n",
-      "iter 0, loss:[1.0397792]\n",
-      "iter 200, loss:[1.2458231]\n",
+      "iter 0, loss:[1.2202356]\n",
+      "iter 200, loss:[1.3485341]\n",
       "epoch:11\n",
-      "iter 0, loss:[1.2107158]\n",
-      "iter 200, loss:[1.426417]\n",
+      "iter 0, loss:[1.1035374]\n",
+      "iter 200, loss:[1.2871654]\n",
       "epoch:12\n",
-      "iter 0, loss:[1.1840894]\n",
-      "iter 200, loss:[1.0999664]\n",
+      "iter 0, loss:[1.194801]\n",
+      "iter 200, loss:[1.0479954]\n",
       "epoch:13\n",
-      "iter 0, loss:[1.0968472]\n",
-      "iter 200, loss:[0.8149167]\n",
+      "iter 0, loss:[1.0022258]\n",
+      "iter 200, loss:[1.0899843]\n",
       "epoch:14\n",
-      "iter 0, loss:[0.95585203]\n",
-      "iter 200, loss:[1.0070628]\n",
+      "iter 0, loss:[0.93466896]\n",
+      "iter 200, loss:[0.99347967]\n",
       "epoch:15\n",
-      "iter 0, loss:[0.89463925]\n",
-      "iter 200, loss:[0.8288595]\n",
+      "iter 0, loss:[0.83665943]\n",
+      "iter 200, loss:[0.9594004]\n",
       "epoch:16\n",
-      "iter 0, loss:[0.5672495]\n",
-      "iter 200, loss:[0.7317069]\n",
+      "iter 0, loss:[0.78929776]\n",
+      "iter 200, loss:[0.945769]\n",
       "epoch:17\n",
-      "iter 0, loss:[0.76785177]\n",
-      "iter 200, loss:[0.5319323]\n",
+      "iter 0, loss:[0.62574965]\n",
+      "iter 200, loss:[0.6308163]\n",
       "epoch:18\n",
-      "iter 0, loss:[0.5250005]\n",
-      "iter 200, loss:[0.4182841]\n",
+      "iter 0, loss:[0.63433456]\n",
+      "iter 200, loss:[0.6287957]\n",
       "epoch:19\n",
-      "iter 0, loss:[0.52320284]\n",
-      "iter 200, loss:[0.47618982]\n"
+      "iter 0, loss:[0.54270047]\n",
+      "iter 200, loss:[0.72688276]\n"
      ]
     }
    ],
@@ -527,16 +525,15 @@
     "            print(\"iter {}, loss:{}\".format(iteration, loss.numpy()))\n",
     "\n",
     "        loss.backward()\n",
-    "        opt.minimize(loss)\n",
-    "        encoder.clear_gradients()\n",
-    "        atten_decoder.clear_gradients()"
+    "        opt.step()\n",
+    "        opt.clear_grad()"
    ]
   },
   {
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# 使用模型进行机器翻译\n",
+    "## 使用模型进行机器翻译\n",
     "\n",
     "根据你所使用的计算设备的不同，上面的训练过程可能需要不等的时间。（在一台Mac笔记本上，大约耗时15~20分钟）\n",
     "完成上面的模型训练之后，我们可以得到一个能够从英文翻译成中文的机器翻译模型。接下来我们通过一个greedy search来实现使用该模型完成实际的机器翻译。（实际的任务中，你可能需要用beam search算法来提升效果）"
@@ -544,43 +541,43 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 18,
+   "execution_count": 12,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "i agree with him\n",
-      "true: 我同意他。\n",
-      "pred: 我同意他。\n",
-      "i think i ll take a bath tonight\n",
-      "true: 我想我今晚會洗澡。\n",
-      "pred: 我想我今晚會洗澡。\n",
-      "he asked for a drink of water\n",
-      "true: 他要了水喝。\n",
-      "pred: 他喝了一杯水。\n",
-      "i began running\n",
-      "true: 我開始跑。\n",
-      "pred: 我開始跑。\n",
-      "i m sick\n",
-      "true: 我生病了。\n",
-      "pred: 我生病了。\n",
-      "you had better go to the dentist s\n",
-      "true: 你最好去看牙醫。\n",
-      "pred: 你最好去看牙醫。\n",
-      "we went for a walk in the forest\n",
-      "true: 我们去了林中散步。\n",
-      "pred: 我們去公园散步。\n",
-      "you ve arrived very early\n",
-      "true: 你來得很早。\n",
-      "pred: 你去早个。\n",
-      "he pretended not to be listening\n",
-      "true: 他裝作沒在聽。\n",
-      "pred: 他假装聽到它。\n",
-      "he always wanted to study japanese\n",
-      "true: 他一直想學日語。\n",
-      "pred: 他一直想學日語。\n"
+      "i want to study french\n",
+      "true: 我要学法语。\n",
+      "pred: 我要学法语。\n",
+      "i didn t know that he was there\n",
+      "true: 我不知道他在那裡。\n",
+      "pred: 我不知道他在那裡。\n",
+      "i called tom\n",
+      "true: 我給湯姆打了電話。\n",
+      "pred: 我看見湯姆了。\n",
+      "he is getting along with his employees\n",
+      "true: 他和他的員工相處。\n",
+      "pred: 他和他的員工相處。\n",
+      "we raced toward the fire\n",
+      "true: 我們急忙跑向火。\n",
+      "pred: 我們住在美國。\n",
+      "i ran away in a hurry\n",
+      "true: 我趕快跑走了。\n",
+      "pred: 我在班里是最高。\n",
+      "he cut the envelope open\n",
+      "true: 他裁開了那個信封。\n",
+      "pred: 他裁開了信封。\n",
+      "he s shorter than tom\n",
+      "true: 他比湯姆矮。\n",
+      "pred: 他比湯姆矮。\n",
+      "i ve just started playing tennis\n",
+      "true: 我剛開始打網球。\n",
+      "pred: 我剛去打網球。\n",
+      "i need to go home\n",
+      "true: 我该回家了。\n",
+      "pred: 我该回家了。\n"
      ]
     }
    ],
@@ -628,17 +625,10 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# The End\n",
+    "## The End\n",
     "\n",
     "你还可以通过变换网络结构，调整数据集，尝试不同的参数的方式来进一步提升本示例当中的机器翻译的效果。同时，也可以尝试在其他的类似的任务中用飞桨来完成实际的实践。"
    ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": []
   }
  ],
  "metadata": {
@@ -657,7 +647,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.7.7"
+   "version": "3.7.3"
   }
  },
  "nbformat": 4,

doc/paddle/tutorial/nlp_case/seq2seq_with_attention/seq2seq_with_attention.rst

@@ -4,7 +4,7 @@
 本示例教程介绍如何使用飞桨完成一个机器翻译任务。我们将会使用飞桨提供的LSTM的API，组建一个\ ``sequence to sequence with attention``\ 的机器翻译的模型，并在示例的数据集上完成从英文翻译成中文的机器翻译。
 
 环境设置
----------
+--------
 
 本示例教程基于飞桨2.0-beta版本。
 
@@ -17,17 +17,15 @@
     
     paddle.disable_static()
     print(paddle.__version__)
-    print(paddle.__git_commit__)
 
 
 .. parsed-literal::
 
-    0.0.0
-    89af2088b6e74bdfeef2d4d78e08461ed2aafee5
+    2.0.0-beta0
 
 
 下载数据集
-------------
+----------
 
 我们将使用 http://www.manythings.org/anki/
 提供的中英文的英汉句对作为数据集，来完成本任务。该数据集含有23610个中英文双语的句对。
@@ -39,16 +37,16 @@
 
 .. parsed-literal::
 
-    --2020-09-04 16:13:35--  https://www.manythings.org/anki/cmn-eng.zip
-    Resolving www.manythings.org (www.manythings.org)... 104.24.109.196, 172.67.173.198, 2606:4700:3037::6818:6cc4, ...
-    Connecting to www.manythings.org (www.manythings.org)|104.24.109.196|:443... connected.
+    --2020-09-10 16:17:25--  https://www.manythings.org/anki/cmn-eng.zip
+    Resolving www.manythings.org (www.manythings.org)... 2606:4700:3033::6818:6dc4, 2606:4700:3036::ac43:adc6, 2606:4700:3037::6818:6cc4, ...
+    Connecting to www.manythings.org (www.manythings.org)|2606:4700:3033::6818:6dc4|:443... connected.
     HTTP request sent, awaiting response... 200 OK
     Length: 1030722 (1007K) [application/zip]
     Saving to: ‘cmn-eng.zip’
     
-    cmn-eng.zip         100%[===================>]   1007K   520KB/s    in 1.9s    
+    cmn-eng.zip         100%[===================>]   1007K  91.2KB/s    in 11s     
     
-    2020-09-04 16:13:38 (520 KB/s) - ‘cmn-eng.zip’ saved [1030722/1030722]
+    2020-09-10 16:17:38 (91.2 KB/s) - ‘cmn-eng.zip’ saved [1030722/1030722]
     
     Archive:  cmn-eng.zip
       inflating: cmn.txt                 
@@ -62,11 +60,11 @@
 
 .. parsed-literal::
 
-       23610 cmn.txt
+       23610 cmn.txt
 
 
 构建双语句对的数据结构
--------------------------
+----------------------
 
 接下来我们通过处理下载下来的双语句对的文本文件，将双语句对读入到python的数据结构中。这里做了如下的处理。
 
@@ -116,7 +114,7 @@
 
 
 创建词表
-----------
+--------
 
 接下来我们分别创建中英文的词表，这两份词表会用来将英文和中文的句子转换为词的ID构成的序列。词表中还加入了如下三个特殊的词：
 - ``<pad>``: 用来对较短的句子进行填充。 - ``<bos>``: “begin of
@@ -157,7 +155,7 @@ Note:
 
 
 创建padding过的数据集
------------------------------
+---------------------
 
 接下来根据词表，我们将会创建一份实际的用于训练的用numpy
 array组织起来的数据集。 -
@@ -198,7 +196,7 @@ array组织起来的数据集。 -
 
 
 创建网络
----------
+--------
 
 我们将会创建一个Encoder-AttentionDecoder架构的模型结构用来完成机器翻译任务。
 首先我们将设置一些必要的网络结构中用到的参数。
@@ -214,7 +212,7 @@ array组织起来的数据集。 -
     batch_size = 16
 
 Encoder部分
-----------------
+-----------
 
 在编码器的部分，我们通过查找完Embedding之后接一个LSTM的方式构建一个对源语言编码的网络。飞桨的RNN系列的API，除了LSTM之外，还提供了SimleRNN,
 GRU供使用，同时，还可以使用反向RNN，双向RNN，多层RNN等形式。也可以通过\ ``dropout``\ 参数设置是否对多层RNN的中间层进行\ ``dropout``\ 处理，来防止过拟合。
@@ -239,7 +237,7 @@ LSTMCell等API更灵活的创建单步的RNN计算，甚至通过继承RNNCellBa
             return x
 
 AttentionDecoder部分
-------------------------
+--------------------
 
 在解码器部分，我们通过一个带有注意力机制的LSTM来完成解码。
 
@@ -358,77 +356,76 @@ AttentionDecoder部分
                 print("iter {}, loss:{}".format(iteration, loss.numpy()))
     
             loss.backward()
-            opt.minimize(loss)
-            encoder.clear_gradients()
-            atten_decoder.clear_gradients()
+            opt.step()
+            opt.clear_grad()
 
 
 .. parsed-literal::
 
     epoch:0
-    iter 0, loss:[7.6194725]
-    iter 200, loss:[3.4147663]
+    iter 0, loss:[7.620109]
+    iter 200, loss:[2.9760551]
     epoch:1
-    iter 0, loss:[3.0931656]
-    iter 200, loss:[2.7543137]
+    iter 0, loss:[2.9679596]
+    iter 200, loss:[3.161064]
     epoch:2
-    iter 0, loss:[2.8413522]
-    iter 200, loss:[2.340513]
+    iter 0, loss:[2.7516625]
+    iter 200, loss:[2.9755423]
     epoch:3
-    iter 0, loss:[2.597812]
-    iter 200, loss:[2.5552855]
+    iter 0, loss:[2.7249248]
+    iter 200, loss:[2.3419888]
     epoch:4
-    iter 0, loss:[2.0783448]
-    iter 200, loss:[2.4544785]
+    iter 0, loss:[2.3236473]
+    iter 200, loss:[2.3453429]
     epoch:5
-    iter 0, loss:[1.8709135]
-    iter 200, loss:[1.8736631]
+    iter 0, loss:[2.1926975]
+    iter 200, loss:[2.1977856]
     epoch:6
-    iter 0, loss:[1.9589291]
-    iter 200, loss:[2.119414]
+    iter 0, loss:[2.014393]
+    iter 200, loss:[2.1863418]
     epoch:7
-    iter 0, loss:[1.5829577]
-    iter 200, loss:[1.6002902]
+    iter 0, loss:[1.8619595]
+    iter 200, loss:[1.8904227]
     epoch:8
-    iter 0, loss:[1.6022769]
-    iter 200, loss:[1.52694]
+    iter 0, loss:[1.5901132]
+    iter 200, loss:[1.7812968]
     epoch:9
-    iter 0, loss:[1.3616685]
-    iter 200, loss:[1.5420443]
+    iter 0, loss:[1.341565]
+    iter 200, loss:[1.4957166]
     epoch:10
-    iter 0, loss:[1.0397792]
-    iter 200, loss:[1.2458231]
+    iter 0, loss:[1.2202356]
+    iter 200, loss:[1.3485341]
     epoch:11
-    iter 0, loss:[1.2107158]
-    iter 200, loss:[1.426417]
+    iter 0, loss:[1.1035374]
+    iter 200, loss:[1.2871654]
     epoch:12
-    iter 0, loss:[1.1840894]
-    iter 200, loss:[1.0999664]
+    iter 0, loss:[1.194801]
+    iter 200, loss:[1.0479954]
     epoch:13
-    iter 0, loss:[1.0968472]
-    iter 200, loss:[0.8149167]
+    iter 0, loss:[1.0022258]
+    iter 200, loss:[1.0899843]
     epoch:14
-    iter 0, loss:[0.95585203]
-    iter 200, loss:[1.0070628]
+    iter 0, loss:[0.93466896]
+    iter 200, loss:[0.99347967]
     epoch:15
-    iter 0, loss:[0.89463925]
-    iter 200, loss:[0.8288595]
+    iter 0, loss:[0.83665943]
+    iter 200, loss:[0.9594004]
     epoch:16
-    iter 0, loss:[0.5672495]
-    iter 200, loss:[0.7317069]
+    iter 0, loss:[0.78929776]
+    iter 200, loss:[0.945769]
     epoch:17
-    iter 0, loss:[0.76785177]
-    iter 200, loss:[0.5319323]
+    iter 0, loss:[0.62574965]
+    iter 200, loss:[0.6308163]
     epoch:18
-    iter 0, loss:[0.5250005]
-    iter 200, loss:[0.4182841]
+    iter 0, loss:[0.63433456]
+    iter 200, loss:[0.6287957]
     epoch:19
-    iter 0, loss:[0.52320284]
-    iter 200, loss:[0.47618982]
+    iter 0, loss:[0.54270047]
+    iter 200, loss:[0.72688276]
 
 
 使用模型进行机器翻译
------------------------
+--------------------
 
 根据你所使用的计算设备的不同，上面的训练过程可能需要不等的时间。（在一台Mac笔记本上，大约耗时15~20分钟）
 完成上面的模型训练之后，我们可以得到一个能够从英文翻译成中文的机器翻译模型。接下来我们通过一个greedy
@@ -478,40 +475,39 @@ search算法来提升效果）
 
 .. parsed-literal::
 
-    i agree with him
-    true: 我同意他。
-    pred: 我同意他。
-    i think i ll take a bath tonight
-    true: 我想我今晚會洗澡。
-    pred: 我想我今晚會洗澡。
-    he asked for a drink of water
-    true: 他要了水喝。
-    pred: 他喝了一杯水。
-    i began running
-    true: 我開始跑。
-    pred: 我開始跑。
-    i m sick
-    true: 我生病了。
-    pred: 我生病了。
-    you had better go to the dentist s
-    true: 你最好去看牙醫。
-    pred: 你最好去看牙醫。
-    we went for a walk in the forest
-    true: 我们去了林中散步。
-    pred: 我們去公园散步。
-    you ve arrived very early
-    true: 你來得很早。
-    pred: 你去早个。
-    he pretended not to be listening
-    true: 他裝作沒在聽。
-    pred: 他假装聽到它。
-    he always wanted to study japanese
-    true: 他一直想學日語。
-    pred: 他一直想學日語。
+    i want to study french
+    true: 我要学法语。
+    pred: 我要学法语。
+    i didn t know that he was there
+    true: 我不知道他在那裡。
+    pred: 我不知道他在那裡。
+    i called tom
+    true: 我給湯姆打了電話。
+    pred: 我看見湯姆了。
+    he is getting along with his employees
+    true: 他和他的員工相處。
+    pred: 他和他的員工相處。
+    we raced toward the fire
+    true: 我們急忙跑向火。
+    pred: 我們住在美國。
+    i ran away in a hurry
+    true: 我趕快跑走了。
+    pred: 我在班里是最高。
+    he cut the envelope open
+    true: 他裁開了那個信封。
+    pred: 他裁開了信封。
+    he s shorter than tom
+    true: 他比湯姆矮。
+    pred: 他比湯姆矮。
+    i ve just started playing tennis
+    true: 我剛開始打網球。
+    pred: 我剛去打網球。
+    i need to go home
+    true: 我该回家了。
+    pred: 我该回家了。
 
 
 The End
 -------
 
 你还可以通过变换网络结构，调整数据集，尝试不同的参数的方式来进一步提升本示例当中的机器翻译的效果。同时，也可以尝试在其他的类似的任务中用飞桨来完成实际的实践。
-

doc/paddle/tutorial/quick_start/dynamic_graph/dynamic_graph.ipynb

@@ -15,7 +15,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# 设置环境\n",
+    "## 设置环境\n",
     "\n",
     "我们将使用飞桨2.0beta版本，并确认已经开启了动态图模式。"
    ]
@@ -29,8 +29,7 @@
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "0.0.0\n",
-      "89af2088b6e74bdfeef2d4d78e08461ed2aafee5\n"
+      "2.0.0-beta0\n"
      ]
     }
    ],
@@ -40,15 +39,14 @@
     "import numpy as np\n",
     "\n",
     "paddle.disable_static()\n",
-    "print(paddle.__version__)\n",
-    "print(paddle.__git_commit__)\n"
+    "print(paddle.__version__)"
    ]
   },
   {
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# 基本用法\n",
+    "## 基本用法\n",
     "\n",
     "在动态图模式下，您可以直接运行一个飞桨提供的API，它会立刻返回结果到python。不再需要首先创建一个计算图，然后再给定数据去运行。"
    ]
@@ -62,16 +60,16 @@
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "[[-0.49341336 -0.8112665 ]\n",
-      " [ 0.8929015   0.24661176]\n",
-      " [-0.64440054 -0.7945008 ]\n",
-      " [-0.07345356  1.3641853 ]]\n",
+      "[[ 1.5645729  -0.74514765]\n",
+      " [-0.01248     0.68240154]\n",
+      " [ 0.11316949 -1.6579045 ]\n",
+      " [-0.1425675  -1.0153968 ]]\n",
       "[1. 2.]\n",
-      "[[0.5065867  1.1887336 ]\n",
-      " [1.8929014  2.2466118 ]\n",
-      " [0.35559946 1.2054992 ]\n",
-      " [0.92654645 3.3641853 ]]\n",
-      "[-2.1159463  1.386125  -2.2334023  2.654917 ]\n"
+      "[[2.5645728  1.2548523 ]\n",
+      " [0.98752    2.6824017 ]\n",
+      " [1.1131694  0.3420955 ]\n",
+      " [0.8574325  0.98460317]]\n",
+      "[ 0.07427764  1.352323   -3.2026396  -2.173361  ]\n"
      ]
     }
    ],
@@ -93,14 +91,14 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# 使用python的控制流\n",
+    "## 使用python的控制流\n",
     "\n",
     "动态图模式下，您可以使用python的条件判断和循环，这类控制语句来执行神经网络的计算。（不再需要`cond`, `loop`这类OP)\n"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 4,
+   "execution_count": 5,
    "metadata": {},
    "outputs": [
     {
@@ -108,12 +106,12 @@
      "output_type": "stream",
      "text": [
       "0 +> [5 6 7]\n",
-      "1 +> [5 7 9]\n",
+      "1 -> [-3 -3 -3]\n",
       "2 +> [ 5  9 15]\n",
       "3 -> [-3  3 21]\n",
-      "4 -> [-3 11 75]\n",
-      "5 +> [  5  37 249]\n",
-      "6 +> [  5  69 735]\n",
+      "4 +> [ 5 21 87]\n",
+      "5 -> [ -3  27 237]\n",
+      "6 -> [ -3  59 723]\n",
       "7 -> [  -3  123 2181]\n",
       "8 +> [   5  261 6567]\n",
       "9 +> [    5   517 19689]\n"
@@ -138,7 +136,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# 构建更加灵活的网络：控制流\n",
+    "## 构建更加灵活的网络：控制流\n",
     "\n",
     "- 使用动态图可以用来创建更加灵活的网络，比如根据控制流选择不同的分支网络，和方便的构建权重共享的网络。接下来我们来看一个具体的例子，在这个例子中，第二个线性变换只有0.5的可能性会运行。\n",
     "- 在sequence to sequence with attention的机器翻译的示例中，你会看到更实际的使用动态图构建RNN类的网络带来的灵活性。\n"
@@ -146,7 +144,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 5,
+   "execution_count": 6,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -172,28 +170,28 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 5,
+   "execution_count": 7,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "0 [2.0915627]\n",
-      "200 [0.67530334]\n",
-      "400 [0.52042854]\n",
-      "600 [0.28010666]\n",
-      "800 [0.09739777]\n",
-      "1000 [0.09307177]\n",
-      "1200 [0.04252927]\n",
-      "1400 [0.03095707]\n",
-      "1600 [0.03022156]\n",
-      "1800 [0.01616007]\n",
-      "2000 [0.01069116]\n",
-      "2200 [0.0055158]\n",
-      "2400 [0.00195092]\n",
-      "2600 [0.00101116]\n",
-      "2800 [0.00192219]\n"
+      "0 [1.3384138]\n",
+      "200 [0.7855983]\n",
+      "400 [0.59084535]\n",
+      "600 [0.30849028]\n",
+      "800 [0.26992702]\n",
+      "1000 [0.03990713]\n",
+      "1200 [0.07111286]\n",
+      "1400 [0.01177792]\n",
+      "1600 [0.03160322]\n",
+      "1800 [0.02757282]\n",
+      "2000 [0.00916022]\n",
+      "2200 [0.00217024]\n",
+      "2400 [0.00186833]\n",
+      "2600 [0.00101926]\n",
+      "2800 [0.0009654]\n"
      ]
     }
    ],
@@ -220,39 +218,39 @@
     "        print(t, loss.numpy())\n",
     "\n",
     "    loss.backward()\n",
-    "    optimizer.minimize(loss)\n",
-    "    model.clear_gradients()"
+    "    optimizer.step()\n",
+    "    optimizer.clear_grad()"
    ]
   },
   {
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# 构建更加灵活的网络：共享权重\n",
+    "## 构建更加灵活的网络：共享权重\n",
     "\n",
-    "- 使用动态图还可以更加方便的创建共享权重的网络，下面的示例展示了一个共享了权重的简单的AutoEncoder的示例。\n",
+    "- 使用动态图还可以更加方便的创建共享权重的网络，下面的示例展示了一个共享了权重的简单的AutoEncoder。\n",
     "- 你也可以参考图像搜索的示例看到共享参数权重的更实际的使用。"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 7,
+   "execution_count": 8,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "step: 0, loss: [0.37666085]\n",
-      "step: 1, loss: [0.3063845]\n",
-      "step: 2, loss: [0.2647248]\n",
-      "step: 3, loss: [0.23831272]\n",
-      "step: 4, loss: [0.21714918]\n",
-      "step: 5, loss: [0.1955545]\n",
-      "step: 6, loss: [0.17261818]\n",
-      "step: 7, loss: [0.15009595]\n",
-      "step: 8, loss: [0.13051331]\n",
-      "step: 9, loss: [0.11537809]\n"
+      "step: 0, loss: [0.33474904]\n",
+      "step: 1, loss: [0.31669515]\n",
+      "step: 2, loss: [0.29729688]\n",
+      "step: 3, loss: [0.27288628]\n",
+      "step: 4, loss: [0.24694422]\n",
+      "step: 5, loss: [0.2203041]\n",
+      "step: 6, loss: [0.19171436]\n",
+      "step: 7, loss: [0.16213782]\n",
+      "step: 8, loss: [0.13443354]\n",
+      "step: 9, loss: [0.11170781]\n"
      ]
     }
    ],
@@ -270,15 +268,15 @@
     "    loss = loss_fn(outputs, inputs)\n",
     "    loss.backward()\n",
     "    print(\"step: {}, loss: {}\".format(i, loss.numpy()))\n",
-    "    optimizer.minimize(loss)\n",
-    "    linear.clear_gradients()"
+    "    optimizer.step()\n",
+    "    optimizer.clear_grad()"
    ]
   },
   {
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# The end\n",
+    "## The end\n",
     "\n",
     "可以看到使用动态图带来了更灵活易用的方式来组网和训练。"
    ]
@@ -300,7 +298,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.7.7"
+   "version": "3.7.3"
   }
  },
  "nbformat": 4,

doc/paddle/tutorial/quick_start/dynamic_graph/dynamic_graph.rst

@@ -18,14 +18,11 @@
     
     paddle.disable_static()
     print(paddle.__version__)
-    print(paddle.__git_commit__)
-
 
 
 .. parsed-literal::
 
-    0.0.0
-    89af2088b6e74bdfeef2d4d78e08461ed2aafee5
+    2.0.0-beta0
 
 
 基本用法
@@ -50,16 +47,16 @@
 
 .. parsed-literal::
 
-    [[-0.49341336 -0.8112665 ]
-     [ 0.8929015   0.24661176]
-     [-0.64440054 -0.7945008 ]
-     [-0.07345356  1.3641853 ]]
+    [[ 1.5645729  -0.74514765]
+     [-0.01248     0.68240154]
+     [ 0.11316949 -1.6579045 ]
+     [-0.1425675  -1.0153968 ]]
     [1. 2.]
-    [[0.5065867  1.1887336 ]
-     [1.8929014  2.2466118 ]
-     [0.35559946 1.2054992 ]
-     [0.92654645 3.3641853 ]]
-    [-2.1159463  1.386125  -2.2334023  2.654917 ]
+    [[2.5645728  1.2548523 ]
+     [0.98752    2.6824017 ]
+     [1.1131694  0.3420955 ]
+     [0.8574325  0.98460317]]
+    [ 0.07427764  1.352323   -3.2026396  -2.173361  ]
 
 
 使用python的控制流
@@ -87,19 +84,19 @@
 .. parsed-literal::
 
     0 +> [5 6 7]
-    1 +> [5 7 9]
+    1 -> [-3 -3 -3]
     2 +> [ 5  9 15]
     3 -> [-3  3 21]
-    4 -> [-3 11 75]
-    5 +> [  5  37 249]
-    6 +> [  5  69 735]
+    4 +> [ 5 21 87]
+    5 -> [ -3  27 237]
+    6 -> [ -3  59 723]
     7 -> [  -3  123 2181]
     8 +> [   5  261 6567]
     9 +> [    5   517 19689]
 
 
 构建更加灵活的网络：控制流
--------------------------------
+--------------------------
 
 -  使用动态图可以用来创建更加灵活的网络，比如根据控制流选择不同的分支网络，和方便的构建权重共享的网络。接下来我们来看一个具体的例子，在这个例子中，第二个线性变换只有0.5的可能性会运行。
 -  在sequence to sequence with
@@ -150,33 +147,33 @@
             print(t, loss.numpy())
     
         loss.backward()
-        optimizer.minimize(loss)
-        model.clear_gradients()
+        optimizer.step()
+        optimizer.clear_grad()
 
 
 .. parsed-literal::
 
-    0 [2.0915627]
-    200 [0.67530334]
-    400 [0.52042854]
-    600 [0.28010666]
-    800 [0.09739777]
-    1000 [0.09307177]
-    1200 [0.04252927]
-    1400 [0.03095707]
-    1600 [0.03022156]
-    1800 [0.01616007]
-    2000 [0.01069116]
-    2200 [0.0055158]
-    2400 [0.00195092]
-    2600 [0.00101116]
-    2800 [0.00192219]
+    0 [1.3384138]
+    200 [0.7855983]
+    400 [0.59084535]
+    600 [0.30849028]
+    800 [0.26992702]
+    1000 [0.03990713]
+    1200 [0.07111286]
+    1400 [0.01177792]
+    1600 [0.03160322]
+    1800 [0.02757282]
+    2000 [0.00916022]
+    2200 [0.00217024]
+    2400 [0.00186833]
+    2600 [0.00101926]
+    2800 [0.0009654]
 
 
 构建更加灵活的网络：共享权重
----------------------------------
+----------------------------
 
--  使用动态图还可以更加方便的创建共享权重的网络，下面的示例展示了一个共享了权重的简单的AutoEncoder的示例。
+-  使用动态图还可以更加方便的创建共享权重的网络，下面的示例展示了一个共享了权重的简单的AutoEncoder。
 -  你也可以参考图像搜索的示例看到共享参数权重的更实际的使用。
 
 .. code:: ipython3
@@ -194,25 +191,25 @@
         loss = loss_fn(outputs, inputs)
         loss.backward()
         print("step: {}, loss: {}".format(i, loss.numpy()))
-        optimizer.minimize(loss)
-        linear.clear_gradients()
+        optimizer.step()
+        optimizer.clear_grad()
 
 
 .. parsed-literal::
 
-    step: 0, loss: [0.37666085]
-    step: 1, loss: [0.3063845]
-    step: 2, loss: [0.2647248]
-    step: 3, loss: [0.23831272]
-    step: 4, loss: [0.21714918]
-    step: 5, loss: [0.1955545]
-    step: 6, loss: [0.17261818]
-    step: 7, loss: [0.15009595]
-    step: 8, loss: [0.13051331]
-    step: 9, loss: [0.11537809]
+    step: 0, loss: [0.33474904]
+    step: 1, loss: [0.31669515]
+    step: 2, loss: [0.29729688]
+    step: 3, loss: [0.27288628]
+    step: 4, loss: [0.24694422]
+    step: 5, loss: [0.2203041]
+    step: 6, loss: [0.19171436]
+    step: 7, loss: [0.16213782]
+    step: 8, loss: [0.13443354]
+    step: 9, loss: [0.11170781]
 
 
 The end
---------
+-------
 
 可以看到使用动态图带来了更灵活易用的方式来组网和训练。

doc/paddle/tutorial/quick_start/getting_started/getting_started.ipynb

@@ -31,16 +31,16 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 4,
+   "execution_count": 1,
    "metadata": {},
    "outputs": [
     {
      "data": {
       "text/plain": [
-       "'0.0.0'"
+       "'2.0.0-beta0'"
       ]
      },
-     "execution_count": 4,
+     "execution_count": 1,
      "metadata": {},
      "output_type": "execute_result"
     }

doc/paddle/tutorial/quick_start/getting_started/getting_started.rst

@@ -26,7 +26,7 @@
 
 .. parsed-literal::
 
-    '0.0.0'
+    '2.0.0-beta0'
 
 
 

doc/paddle/tutorial/quick_start/hello_paddle/hello_paddle.ipynb

@@ -18,7 +18,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# 普通程序跟机器学习程序的逻辑区别\n",
+    "## 普通程序跟机器学习程序的逻辑区别\n",
     "\n",
     "作为一名开发者，你最熟悉的开始学习一门编程语言，或者一个深度学习框架的方式，可能是通过一个hello, world程序。\n",
     "\n",
@@ -37,7 +37,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 24,
+   "execution_count": 22,
    "metadata": {},
    "outputs": [
     {
@@ -80,7 +80,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# 导入飞桨\n",
+    "## 导入飞桨\n",
     "\n",
     "为了能够使用飞桨，我们需要先用python的`import`语句导入飞桨`paddle`。\n",
     "同时，为了能够更好的对数组进行计算和处理，我们也还需要导入`numpy`。\n",
@@ -90,28 +90,28 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 25,
+   "execution_count": 3,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "paddle version 0.0.0\n"
+      "paddle 2.0.0-beta0\n"
      ]
     }
    ],
    "source": [
     "import paddle\n",
     "paddle.disable_static()\n",
-    "print(\"paddle version \" + paddle.__version__)"
+    "print(\"paddle \" + paddle.__version__)"
    ]
   },
   {
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# 准备数据\n",
+    "## 准备数据\n",
     "\n",
     "在这个机器学习任务中，我们已经知道了乘客的行驶里程`distance_travelled`，和对应的，这些乘客的总费用`total_fee`。\n",
     "通常情况下，在机器学习任务中，像`distance_travelled`这样的输入值，一般被称为`x`（或者特征`feature`），像`total_fee`这样的输出值，一般被称为`y`（或者标签`label`)。\n",
@@ -121,7 +121,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 26,
+   "execution_count": 4,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -133,7 +133,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# 用飞桨定义模型的计算\n",
+    "## 用飞桨定义模型的计算\n",
     "\n",
     "使用飞桨定义模型的计算的过程，本质上，是我们用python，通过飞桨提供的API，来告诉飞桨我们的计算规则的过程。回顾一下，我们想要通过飞桨用机器学习方法，从数据当中学习出来如下公式当中的`w`和`b`。这样在未来，给定`x`时就可以估算出来`y`值（估算出来的`y`记为`y_predict`）\n",
     "\n",
@@ -150,7 +150,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 27,
+   "execution_count": 5,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -161,21 +161,21 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# 准备好运行飞桨\n",
+    "## 准备好运行飞桨\n",
     "\n",
     "机器（计算机）在一开始的时候会随便猜`w`和`b`，我们先看看机器猜的怎么样。你应该可以看到，这时候的`w`是一个随机值，`b`是0.0，这是飞桨的初始化策略，也是这个领域常用的初始化策略。（如果你愿意，也可以采用其他的初始化的方式，今后你也会看到，选择不同的初始化策略也是对于做好深度学习任务来说很重要的一点）。"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 28,
+   "execution_count": 6,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "w before optimize: -1.7107375860214233\n",
+      "w before optimize: -1.696260690689087\n",
       "b before optimize: 0.0\n"
      ]
     }
@@ -192,7 +192,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# 告诉飞桨怎么样学习\n",
+    "## 告诉飞桨怎么样学习\n",
     "\n",
     "前面我们定义好了神经网络（尽管是一个最简单的神经网络），我们还需要告诉飞桨，怎么样去**学习**，从而能得到参数`w`和`b`。\n",
     "\n",
@@ -205,7 +205,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 29,
+   "execution_count": 7,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -217,26 +217,26 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# 运行优化算法\n",
+    "## 运行优化算法\n",
     "\n",
     "接下来，我们让飞桨运行一下这个优化算法，这会是一个前面介绍过的逐步调整参数的过程，你应该可以看到loss值（衡量`y`和`y_predict`的差距的`loss`)在不断的降低。"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 30,
+   "execution_count": 8,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "epoch 0 loss [2107.3943]\n",
-      "epoch 1000 loss [7.8432994]\n",
-      "epoch 2000 loss [1.7537074]\n",
-      "epoch 3000 loss [0.39211753]\n",
-      "epoch 4000 loss [0.08767726]\n",
-      "finished training， loss [0.01963376]\n"
+      "epoch 0 loss [2094.069]\n",
+      "epoch 1000 loss [7.8451133]\n",
+      "epoch 2000 loss [1.7541145]\n",
+      "epoch 3000 loss [0.39221546]\n",
+      "epoch 4000 loss [0.08769739]\n",
+      "finished training， loss [0.0196382]\n"
      ]
     }
    ],
@@ -246,8 +246,8 @@
     "    y_predict = linear(x_data)\n",
     "    loss = mse_loss(y_predict, y_data)\n",
     "    loss.backward()\n",
-    "    sgd_optimizer.minimize(loss)\n",
-    "    linear.clear_gradients()\n",
+    "    sgd_optimizer.step()\n",
+    "    sgd_optimizer.clear_grad()\n",
     "    \n",
     "    if i%1000 == 0:\n",
     "        print(\"epoch {} loss {}\".format(i, loss.numpy()))\n",
@@ -259,22 +259,22 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# 机器学习出来的参数\n",
+    "## 机器学习出来的参数\n",
     "\n",
     "经过了这样的对参数`w`和`b`的调整（**学习**)，我们再通过下面的程序，来看看现在的参数变成了多少。你应该会发现`w`变成了很接近2.0的一个值，`b`变成了接近10.0的一个值。虽然并不是正好的2和10，但却是从数据当中学习出来的还不错的模型的参数，可以在未来的时候，用从这批数据当中学习到的参数来预估了。（如果你愿意，也可以通过让机器多学习一段时间，从而得到更加接近2.0和10.0的参数值。)"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 31,
+   "execution_count": 9,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "w after optimize: 2.017843246459961\n",
-      "b after optimize: 9.771851539611816\n"
+      "w after optimize: 2.0178451538085938\n",
+      "b after optimize: 9.771825790405273\n"
      ]
     }
    ],
@@ -290,14 +290,14 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# hello paddle\n",
+    "## hello paddle\n",
     "\n",
     "通过这个小示例，希望你已经初步了解了飞桨，能在接下来随着对飞桨的更多学习，来解决实际遇到的问题。"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 32,
+   "execution_count": 10,
    "metadata": {},
    "outputs": [
     {
@@ -335,9 +335,9 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.7.7"
+   "version": "3.7.3"
   }
  },
  "nbformat": 4,
- "nbformat_minor": 1
+ "nbformat_minor": 4
 }

doc/paddle/tutorial/quick_start/hello_paddle/hello_paddle.rst

@@ -51,7 +51,7 @@ world程序。
 接下来，我们看看用飞桨如何实现这个hello, world级别的机器学习程序。
 
 导入飞桨
----------
+--------
 
 为了能够使用飞桨，我们需要先用python的\ ``import``\ 语句导入飞桨\ ``paddle``\ 。
 同时，为了能够更好的对数组进行计算和处理，我们也还需要导入\ ``numpy``\ 。
@@ -62,16 +62,16 @@ world程序。
 
     import paddle
     paddle.disable_static()
-    print("paddle version " + paddle.__version__)
+    print("paddle " + paddle.__version__)
 
 
 .. parsed-literal::
 
-    paddle version 0.0.0
+    paddle 2.0.0-beta0
 
 
 准备数据
----------
+--------
 
 在这个机器学习任务中，我们已经知道了乘客的行驶里程\ ``distance_travelled``\ ，和对应的，这些乘客的总费用\ ``total_fee``\ 。
 通常情况下，在机器学习任务中，像\ ``distance_travelled``\ 这样的输入值，一般被称为\ ``x``\ （或者特征\ ``feature``\ ），像\ ``total_fee``\ 这样的输出值，一般被称为\ ``y``\ （或者标签\ ``label``)。
@@ -104,7 +104,7 @@ world程序。
     linear = paddle.nn.Linear(in_features=1, out_features=1)
 
 准备好运行飞桨
-----------------
+--------------
 
 机器（计算机）在一开始的时候会随便猜\ ``w``\ 和\ ``b``\ ，我们先看看机器猜的怎么样。你应该可以看到，这时候的\ ``w``\ 是一个随机值，\ ``b``\ 是0.0，这是飞桨的初始化策略，也是这个领域常用的初始化策略。（如果你愿意，也可以采用其他的初始化的方式，今后你也会看到，选择不同的初始化策略也是对于做好深度学习任务来说很重要的一点）。
 
@@ -119,12 +119,12 @@ world程序。
 
 .. parsed-literal::
 
-    w before optimize: -1.7107375860214233
+    w before optimize: -1.696260690689087
     b before optimize: 0.0
 
 
 告诉飞桨怎么样学习
---------------------
+------------------
 
 前面我们定义好了神经网络（尽管是一个最简单的神经网络），我们还需要告诉飞桨，怎么样去\ **学习**\ ，从而能得到参数\ ``w``\ 和\ ``b``\ 。
 
@@ -143,7 +143,7 @@ descent)作为优化算法（传给\ ``paddle.optimizer.SGD``\ 的参数\ ``lear
     sgd_optimizer = paddle.optimizer.SGD(learning_rate=0.001, parameters = linear.parameters())
 
 运行优化算法
----------------
+------------
 
 接下来，我们让飞桨运行一下这个优化算法，这会是一个前面介绍过的逐步调整参数的过程，你应该可以看到loss值（衡量\ ``y``\ 和\ ``y_predict``\ 的差距的\ ``loss``)在不断的降低。
 
@@ -154,8 +154,8 @@ descent)作为优化算法（传给\ ``paddle.optimizer.SGD``\ 的参数\ ``lear
         y_predict = linear(x_data)
         loss = mse_loss(y_predict, y_data)
         loss.backward()
-        sgd_optimizer.minimize(loss)
-        linear.clear_gradients()
+        sgd_optimizer.step()
+        sgd_optimizer.clear_grad()
         
         if i%1000 == 0:
             print("epoch {} loss {}".format(i, loss.numpy()))
@@ -165,16 +165,16 @@ descent)作为优化算法（传给\ ``paddle.optimizer.SGD``\ 的参数\ ``lear
 
 .. parsed-literal::
 
-    epoch 0 loss [2107.3943]
-    epoch 1000 loss [7.8432994]
-    epoch 2000 loss [1.7537074]
-    epoch 3000 loss [0.39211753]
-    epoch 4000 loss [0.08767726]
-    finished training， loss [0.01963376]
+    epoch 0 loss [2094.069]
+    epoch 1000 loss [7.8451133]
+    epoch 2000 loss [1.7541145]
+    epoch 3000 loss [0.39221546]
+    epoch 4000 loss [0.08769739]
+    finished training， loss [0.0196382]
 
 
 机器学习出来的参数
--------------------
+------------------
 
 经过了这样的对参数\ ``w``\ 和\ ``b``\ 的调整（\ **学习**)，我们再通过下面的程序，来看看现在的参数变成了多少。你应该会发现\ ``w``\ 变成了很接近2.0的一个值，\ ``b``\ 变成了接近10.0的一个值。虽然并不是正好的2和10，但却是从数据当中学习出来的还不错的模型的参数，可以在未来的时候，用从这批数据当中学习到的参数来预估了。（如果你愿意，也可以通过让机器多学习一段时间，从而得到更加接近2.0和10.0的参数值。)
 
@@ -190,12 +190,12 @@ descent)作为优化算法（传给\ ``paddle.optimizer.SGD``\ 的参数\ ``lear
 
 .. parsed-literal::
 
-    w after optimize: 2.017843246459961
-    b after optimize: 9.771851539611816
+    w after optimize: 2.0178451538085938
+    b after optimize: 9.771825790405273
 
 
 hello paddle
----------------
+------------
 
 通过这个小示例，希望你已经初步了解了飞桨，能在接下来随着对飞桨的更多学习，来解决实际遇到的问题。
 

doc/paddle/tutorial/quick_start/high_level_api/high_level_api.ipynb

 {
- "metadata": {
-  "language_info": {
-   "codemirror_mode": {
-    "name": "ipython",
-    "version": 3
-   },
-   "file_extension": ".py",
-   "mimetype": "text/x-python",
-   "name": "python",
-   "nbconvert_exporter": "python",
-   "pygments_lexer": "ipython3",
-   "version": "3.7.4-final"
-  },
-  "orig_nbformat": 2,
-  "kernelspec": {
-   "name": "python37464bitc4da1ac836094043840bff631bedbf7f",
-   "display_name": "Python 3.7.4 64-bit"
-  }
- },
- "nbformat": 4,
- "nbformat_minor": 2,
  "cells": [
   {
    "cell_type": "markdown",
@@ -57,16 +36,18 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 4,
+   "execution_count": 1,
    "metadata": {},
    "outputs": [
     {
-     "output_type": "execute_result",
      "data": {
-      "text/plain": "'0.0.0'"
+      "text/plain": [
+       "'2.0.0-beta0'"
+      ]
      },
+     "execution_count": 1,
      "metadata": {},
-     "execution_count": 4
+     "output_type": "execute_result"
     }
    ],
    "source": [
@@ -91,10 +72,7 @@
     "* 如何进行模型的组网。\n",
     "* 高层API进行模型训练的相关API使用。\n",
     "* 如何在fit接口满足需求的时候进行自定义，使用基础API来完成训练。\n",
-    "* 如何使用多卡来加速训练。\n",
-    "\n",
-    "其他端到端的示例教程：\n",
-    "* TBD"
+    "* 如何使用多卡来加速训练。"
    ]
   },
   {
@@ -112,22 +90,23 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 17,
+   "execution_count": 5,
    "metadata": {
     "tags": []
    },
    "outputs": [
     {
-     "output_type": "execute_result",
-     "data": {
-      "text/plain": "['DatasetFolder',\n 'ImageFolder',\n 'MNIST',\n 'Flowers',\n 'Cifar10',\n 'Cifar100',\n 'VOC2012']"
-     },
-     "metadata": {},
-     "execution_count": 17
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "视觉相关数据集： ['DatasetFolder', 'ImageFolder', 'MNIST', 'Flowers', 'Cifar10', 'Cifar100', 'VOC2012']\n",
+      "自然语言相关数据集： ['Conll05st', 'Imdb', 'Imikolov', 'Movielens', 'MovieReviews', 'UCIHousing', 'WMT14', 'WMT16']\n"
+     ]
     }
    ],
    "source": [
-    "paddle.vision.datasets.__all__"
+    "print('视觉相关数据集：', paddle.vision.datasets.__all__)\n",
+    "print('自然语言相关数据集：', paddle.text.datasets.__all__)"
    ]
   },
   {
@@ -143,7 +122,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "# 测试数据集\n",
+    "# 训练数据集\n",
     "train_dataset = vision.datasets.MNIST(mode='train')\n",
     "\n",
     "# 验证数据集\n",
@@ -167,9 +146,20 @@
    },
    "outputs": [
     {
-     "output_type": "stream",
      "name": "stdout",
-     "text": "=============train dataset=============\ntraindata1 label1\ntraindata2 label2\ntraindata3 label3\ntraindata4 label4\n=============evaluation dataset=============\ntestdata1 label1\ntestdata2 label2\ntestdata3 label3\ntestdata4 label4\n"
+     "output_type": "stream",
+     "text": [
+      "=============train dataset=============\n",
+      "traindata1 label1\n",
+      "traindata2 label2\n",
+      "traindata3 label3\n",
+      "traindata4 label4\n",
+      "=============evaluation dataset=============\n",
+      "testdata1 label1\n",
+      "testdata2 label2\n",
+      "testdata3 label3\n",
+      "testdata4 label4\n"
+     ]
     }
    ],
    "source": [
@@ -458,9 +448,9 @@
    "source": [
     "## 5. 模型训练\n",
     "\n",
-    "使用`paddle.Model`封装成模型类后进行训练非常的简洁方便，我们可以直接通过调用`Model.fit`就可以完成训练过程。\n",
+    "网络结构通过`paddle.Model`接口封装成模型类后进行执行操作非常的简洁方便，可以直接通过调用`Model.fit`就可以完成训练过程。\n",
     "\n",
-    "在使用`Model.fit`接口启动训练前，我们先通过`Model.prepare`接口来对训练进行提前的配置准备工作，包括设置模型优化器，Loss计算方法，精度计算方法等。\n",
+    "使用`Model.fit`接口启动训练前，我们先通过`Model.prepare`接口来对训练进行提前的配置准备工作，包括设置模型优化器，Loss计算方法，精度计算方法等。\n",
     "\n"
    ]
   },
@@ -553,13 +543,269 @@
     "python -m paddle.distributed.launch train.py"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### 5.3 自定义Loss\n",
+    "\n",
+    "有时我们会遇到特定任务的Loss计算方式在框架既有的Loss接口中不存在，或算法不符合自己的需求，那么期望能够自己来进行Loss的自定义，我们这里就会讲解介绍一下如何进行Loss的自定义操作，首先来看下面的代码：\n",
+    "\n",
+    "```python\n",
+    "class SelfDefineLoss(paddle.nn.Layer):\n",
+    "    \"\"\"\n",
+    "    1. 继承paddle.nn.Layer\n",
+    "    \"\"\"\n",
+    "    def __init__(self):\n",
+    "        \"\"\"\n",
+    "        2. 构造函数根据自己的实际算法需求和使用需求进行参数定义即可\n",
+    "        \"\"\"\n",
+    "        super(SelfDefineLoss, self).__init__()\n",
+    "\n",
+    "    def forward(self, input, label):\n",
+    "        \"\"\"\n",
+    "        3. 实现forward函数，forward在调用时会传递两个参数：input和label\n",
+    "            - input：单个或批次训练数据经过模型前向计算输出结果\n",
+    "            - label：单个或批次训练数据对应的标签数据\n",
+    "\n",
+    "            接口返回值是一个Tensor，根据自定义的逻辑加和或计算均值后的损失\n",
+    "        \"\"\"\n",
+    "        # 使用Paddle中相关API自定义的计算逻辑\n",
+    "        # output = xxxxx\n",
+    "        # return output\n",
+    "```\n",
+    "\n",
+    "那么了解完代码层面如果编写自定义代码后我们看一个实际的例子，下面是在图像分割示例代码中写的一个自定义Loss，当时主要是想使用自定义的softmax计算维度。\n",
+    "\n",
+    "```python\n",
+    "class SoftmaxWithCrossEntropy(paddle.nn.Layer):\n",
+    "    def __init__(self):\n",
+    "        super(SoftmaxWithCrossEntropy, self).__init__()\n",
+    "\n",
+    "    def forward(self, input, label):\n",
+    "        loss = F.softmax_with_cross_entropy(input, \n",
+    "                                            label, \n",
+    "                                            return_softmax=False,\n",
+    "                                            axis=1)\n",
+    "        return paddle.mean(loss)\n",
+    "```"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### 5.4 自定义Metric\n",
+    "\n",
+    "和Loss一样，如果遇到一些想要做个性化实现的操作时，我们也可以来通过框架完成自定义的评估计算方法，具体的实现方式如下：\n",
+    "\n",
+    "```python\n",
+    "class SelfDefineMetric(paddle.metric.Metric):\n",
+    "    \"\"\"\n",
+    "    1. 继承paddle.metric.Metric\n",
+    "    \"\"\"\n",
+    "    def __init__(self):\n",
+    "        \"\"\"\n",
+    "        2. 构造函数实现，自定义参数即可\n",
+    "        \"\"\"\n",
+    "        super(SelfDefineMetric, self).__init__()\n",
+    "\n",
+    "    def name(self):\n",
+    "        \"\"\"\n",
+    "        3. 实现name方法，返回定义的评估指标名字\n",
+    "        \"\"\"\n",
+    "        return '自定义评价指标的名字'\n",
+    "\n",
+    "    def compute(self, ...)\n",
+    "        \"\"\"\n",
+    "        4. 本步骤可以省略，实现compute方法，这个方法主要用于`update`的加速，可以在这个方法中调用一些paddle实现好的Tensor计算API，编译到模型网络中一起使用低层C++ OP计算。\n",
+    "        \"\"\"\n",
+    "\n",
+    "        return 自己想要返回的数据，会做为update的参数传入。\n",
+    "\n",
+    "    def update(self, ...):\n",
+    "        \"\"\"\n",
+    "        5. 实现update方法，用于单个batch训练时进行评估指标计算。\n",
+    "        - 当`compute`类函数未实现时，会将模型的计算输出和标签数据的展平作为`update`的参数传入。\n",
+    "        - 当`compute`类函数做了实现时，会将compute的返回结果作为`update`的参数传入。\n",
+    "        \"\"\"\n",
+    "        return acc value\n",
+    "    \n",
+    "    def accumulate(self):\n",
+    "        \"\"\"\n",
+    "        6. 实现accumulate方法，返回历史batch训练积累后计算得到的评价指标值。\n",
+    "        每次`update`调用时进行数据积累，`accumulate`计算时对积累的所有数据进行计算并返回。\n",
+    "        结算结果会在`fit`接口的训练日志中呈现。\n",
+    "        \"\"\"\n",
+    "        # 利用update中积累的成员变量数据进行计算后返回\n",
+    "        return accumulated acc value\n",
+    "\n",
+    "    def reset(self):\n",
+    "        \"\"\"\n",
+    "        7. 实现reset方法，每个Epoch结束后进行评估指标的重置，这样下个Epoch可以重新进行计算。\n",
+    "        \"\"\"\n",
+    "        # do reset action\n",
+    "```\n",
+    "\n",
+    "我们看一个框架中的具体例子，这个是框架中已提供的一个评估指标计算接口，这里就是按照上述说明中的实现方法进行了相关类继承和成员函数实现。\n",
+    "\n",
+    "```python\n",
+    "from paddle.metric import Metric\n",
+    "\n",
+    "\n",
+    "class Precision(Metric):\n",
+    "    \"\"\"\n",
+    "    Precision (also called positive predictive value) is the fraction of\n",
+    "    relevant instances among the retrieved instances. Refer to\n",
+    "    https://en.wikipedia.org/wiki/Evaluation_of_binary_classifiers\n",
+    "\n",
+    "    Noted that this class manages the precision score only for binary\n",
+    "    classification task.\n",
+    "    \n",
+    "    ......\n",
+    "\n",
+    "    \"\"\"\n",
+    "\n",
+    "    def __init__(self, name='precision', *args, **kwargs):\n",
+    "        super(Precision, self).__init__(*args, **kwargs)\n",
+    "        self.tp = 0  # true positive\n",
+    "        self.fp = 0  # false positive\n",
+    "        self._name = name\n",
+    "\n",
+    "    def update(self, preds, labels):\n",
+    "        \"\"\"\n",
+    "        Update the states based on the current mini-batch prediction results.\n",
+    "\n",
+    "        Args:\n",
+    "            preds (numpy.ndarray): The prediction result, usually the output\n",
+    "                of two-class sigmoid function. It should be a vector (column\n",
+    "                vector or row vector) with data type: 'float64' or 'float32'.\n",
+    "            labels (numpy.ndarray): The ground truth (labels),\n",
+    "                the shape should keep the same as preds.\n",
+    "                The data type is 'int32' or 'int64'.\n",
+    "        \"\"\"\n",
+    "        if isinstance(preds, paddle.Tensor):\n",
+    "            preds = preds.numpy()\n",
+    "        elif not _is_numpy_(preds):\n",
+    "            raise ValueError(\"The 'preds' must be a numpy ndarray or Tensor.\")\n",
+    "\n",
+    "        if isinstance(labels, paddle.Tensor):\n",
+    "            labels = labels.numpy()\n",
+    "        elif not _is_numpy_(labels):\n",
+    "            raise ValueError(\"The 'labels' must be a numpy ndarray or Tensor.\")\n",
+    "\n",
+    "        sample_num = labels.shape[0]\n",
+    "        preds = np.floor(preds + 0.5).astype(\"int32\")\n",
+    "\n",
+    "        for i in range(sample_num):\n",
+    "            pred = preds[i]\n",
+    "            label = labels[i]\n",
+    "            if pred == 1:\n",
+    "                if pred == label:\n",
+    "                    self.tp += 1\n",
+    "                else:\n",
+    "                    self.fp += 1\n",
+    "\n",
+    "    def reset(self):\n",
+    "        \"\"\"\n",
+    "        Resets all of the metric state.\n",
+    "        \"\"\"\n",
+    "        self.tp = 0\n",
+    "        self.fp = 0\n",
+    "\n",
+    "    def accumulate(self):\n",
+    "        \"\"\"\n",
+    "        Calculate the final precision.\n",
+    "\n",
+    "        Returns:\n",
+    "            A scaler float: results of the calculated precision.\n",
+    "        \"\"\"\n",
+    "        ap = self.tp + self.fp\n",
+    "        return float(self.tp) / ap if ap != 0 else .0\n",
+    "\n",
+    "    def name(self):\n",
+    "        \"\"\"\n",
+    "        Returns metric name\n",
+    "        \"\"\"\n",
+    "        return self._name\n",
+    "```"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### 5.5 自定义Callback\n",
+    "\n",
+    "`fit`接口的callback参数支持我们传一个Callback类实例，用来在每轮训练和每个batch训练前后进行调用，可以通过callback收集到训练过程中的一些数据和参数，或者实现一些自定义操作。\n",
+    "\n",
+    "```python\n",
+    "class SelfDefineCallback(paddle.callbacks.Callback):\n",
+    "    \"\"\"\n",
+    "    1. 继承paddle.callbacks.Callback\n",
+    "    2. 按照自己的需求实现以下类成员方法：\n",
+    "        def on_train_begin(self, logs=None)                 训练开始前，`Model.fit`接口中调用\n",
+    "        def on_train_end(self, logs=None)                   训练结束后，`Model.fit`接口中调用\n",
+    "        def on_eval_begin(self, logs=None)                  评估开始前，`Model.evaluate`接口调用\n",
+    "        def on_eval_end(self, logs=None)                    评估结束后，`Model.evaluate`接口调用\n",
+    "        def on_test_begin(self, logs=None)                  预测测试开始前，`Model.predict`接口中调用\n",
+    "        def on_test_end(self, logs=None)                    预测测试结束后，`Model.predict`接口中调用 \n",
+    "        def on_epoch_begin(self, epoch, logs=None)          每轮训练开始前，`Model.fit`接口中调用 \n",
+    "        def on_epoch_end(self, epoch, logs=None)            每轮训练结束后，`Model.fit`接口中调用 \n",
+    "        def on_train_batch_begin(self, step, logs=None)     单个Batch训练开始前，`Model.fit`和`Model.train_batch`接口中调用\n",
+    "        def on_train_batch_end(self, step, logs=None)       单个Batch训练结束后，`Model.fit`和`Model.train_batch`接口中调用\n",
+    "        def on_eval_batch_begin(self, step, logs=None)      单个Batch评估开始前，`Model.evalute`和`Model.eval_batch`接口中调用\n",
+    "        def on_eval_batch_end(self, step, logs=None)        单个Batch评估结束后，`Model.evalute`和`Model.eval_batch`接口中调用\n",
+    "        def on_test_batch_begin(self, step, logs=None)      单个Batch预测测试开始前，`Model.predict`和`Model.test_batch`接口中调用\n",
+    "        def on_test_batch_end(self, step, logs=None)        单个Batch预测测试结束后，`Model.predict`和`Model.test_batch`接口中调用\n",
+    "    \"\"\"\n",
+    "    def __init__(self):\n",
+    "        super(SelfDefineCallback, self).__init__()\n",
+    "\n",
+    "    按照需求定义自己的类成员方法\n",
+    "```\n",
+    "\n",
+    "我们看一个框架中的实际例子，这是一个框架自带的ModelCheckpoint回调函数，方便用户在fit训练模型时自动存储每轮训练得到的模型。\n",
+    "\n",
+    "```python\n",
+    "class ModelCheckpoint(Callback):\n",
+    "    def __init__(self, save_freq=1, save_dir=None):\n",
+    "        self.save_freq = save_freq\n",
+    "        self.save_dir = save_dir\n",
+    "\n",
+    "    def on_epoch_begin(self, epoch=None, logs=None):\n",
+    "        self.epoch = epoch\n",
+    "\n",
+    "    def _is_save(self):\n",
+    "        return self.model and self.save_dir and ParallelEnv().local_rank == 0\n",
+    "\n",
+    "    def on_epoch_end(self, epoch, logs=None):\n",
+    "        if self._is_save() and self.epoch % self.save_freq == 0:\n",
+    "            path = '{}/{}'.format(self.save_dir, epoch)\n",
+    "            print('save checkpoint at {}'.format(os.path.abspath(path)))\n",
+    "            self.model.save(path)\n",
+    "\n",
+    "    def on_train_end(self, logs=None):\n",
+    "        if self._is_save():\n",
+    "            path = '{}/final'.format(self.save_dir)\n",
+    "            print('save checkpoint at {}'.format(os.path.abspath(path)))\n",
+    "            self.model.save(path)\n",
+    "\n",
+    "```"
+   ]
+  },
   {
    "cell_type": "markdown",
    "metadata": {},
    "source": [
     "## 6. 模型评估\n",
     "\n",
-    "对于训练好的模型进行评估操作可以使用`evaluate`接口来实现。"
+    "对于训练好的模型进行评估操作可以使用`evaluate`接口来实现，事先定义好用于评估使用的数据集后，可以简单的调用`evaluate`接口即可完成模型评估操作，结束后根据prepare中loss和metric的定义来进行相关评估结果计算返回。\n",
+    "\n",
+    "返回格式是一个字典：\n",
+    "* 只包含loss，`{'loss': xxx}`\n",
+    "* 包含loss和一个评估指标，`{'loss': xxx, 'metric name': xxx}`\n",
+    "* 包含loss和多个评估指标，`{'loss': xxx, 'metric name': xxx, 'metric name': xxx}`"
    ]
   },
   {
@@ -577,7 +823,13 @@
    "source": [
     "## 7. 模型预测\n",
     "\n",
-    "高层API中提供`predict`接口，支持用户使用测试数据来完成模型的预测。"
+    "高层API中提供了`predict`接口来方便用户对训练好的模型进行预测验证，只需要基于训练好的模型将需要进行预测测试的数据放到接口中进行计算即可，接口会将经过模型计算得到的预测结果进行返回。\n",
+    "\n",
+    "返回格式是一个list，元素数目对应模型的输出数目：\n",
+    "* 模型是单一输出：[(numpy_ndarray_1, numpy_ndarray_2, ..., numpy_ndarray_n)]\n",
+    "* 模型是多输出：[(numpy_ndarray_1, numpy_ndarray_2, ..., numpy_ndarray_n), (numpy_ndarray_1, numpy_ndarray_2, ..., numpy_ndarray_n), ...]\n",
+    "\n",
+    "numpy_ndarray_n是对应原始数据经过模型计算后得到的预测数据，数目对应预测数据集的数目。"
    ]
   },
   {
@@ -589,6 +841,23 @@
     "pred_result = model.predict(val_dataset)"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### 7.1 使用多卡进行预测\n",
+    "\n",
+    "有时我们需要进行预测验证的数据较多，单卡无法满足我们的时间诉求，那么`predict`接口也为用户支持实现了使用多卡模式来运行。\n",
+    "\n",
+    "使用起来也是超级简单，无需修改代码程序，只需要使用launch来启动对应的预测脚本即可。\n",
+    "\n",
+    "```bash\n",
+    "$ python3 -m paddle.distributed.launch infer.py\n",
+    "```\n",
+    "\n",
+    "infer.py里面就是包含model.predict的代码程序。"
+   ]
+  },
   {
    "cell_type": "markdown",
    "metadata": {},
@@ -597,7 +866,7 @@
     "\n",
     "### 8.1 模型存储\n",
     "\n",
-    "模型训练和验证达到我们的预期后，可以使用`save`接口来将我们的模型保存下来，用于后续模型的Fine-tuning或推理部署。"
+    "模型训练和验证达到我们的预期后，可以使用`save`接口来将我们的模型保存下来，用于后续模型的Fine-tuning（接口参数training=True）或推理部署（接口参数training=False）。"
    ]
   },
   {
@@ -619,5 +888,26 @@
     "有了用于推理部署的模型，就可以使用推理部署框架来完成预测服务部署，具体可以参见：[预测部署](https://www.paddlepaddle.org.cn/documentation/docs/zh/advanced_guide/inference_deployment/index_cn.html)， 包括服务端部署、移动端部署和模型压缩。"
    ]
   }
- ]
-}
+ ],
+ "metadata": {
\ No newline at end of file
+  "kernelspec": {
+   "display_name": "Python 3.7.4 64-bit",
+   "language": "python",
+   "name": "python37464bitc4da1ac836094043840bff631bedbf7f"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.7.4"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 4
+}

doc/paddle/tutorial/quick_start/high_level_api/high_level_api.rst

@@ -45,7 +45,7 @@ paddle即可使用相关高层API，如：paddle.Model、视觉领域paddle.visi
 
 .. parsed-literal::
 
-    '0.0.0'
+    '2.0.0-beta0'
 
 
 
@@ -62,8 +62,6 @@ paddle即可使用相关高层API，如：paddle.Model、视觉领域paddle.visi
 -  如何在fit接口满足需求的时候进行自定义，使用基础API来完成训练。
 -  如何使用多卡来加速训练。
 
-其他端到端的示例教程： \* TBD
-
 3. 数据集定义、加载和数据预处理
 -------------------------------
 
@@ -76,28 +74,21 @@ paddle即可使用相关高层API，如：paddle.Model、视觉领域paddle.visi
 
 .. code:: ipython3
 
-    paddle.vision.datasets.__all__
-
-
+    print('视觉相关数据集：', paddle.vision.datasets.__all__)
+    print('自然语言相关数据集：', paddle.text.datasets.__all__)
 
 
 .. parsed-literal::
 
-    ['DatasetFolder',
-     'ImageFolder',
-     'MNIST',
-     'Flowers',
-     'Cifar10',
-     'Cifar100',
-     'VOC2012']
-
+    视觉相关数据集： ['DatasetFolder', 'ImageFolder', 'MNIST', 'Flowers', 'Cifar10', 'Cifar100', 'VOC2012']
+    自然语言相关数据集： ['Conll05st', 'Imdb', 'Imikolov', 'Movielens', 'MovieReviews', 'UCIHousing', 'WMT14', 'WMT16']
 
 
 这里我们是加载一个手写数字识别的数据集，用\ ``mode``\ 来标识是训练数据还是测试数据集。数据集接口会自动从远端下载数据集到本机缓存目录\ ``~/.cache/paddle/dataset``\ 。
 
 .. code:: ipython3
 
-    # 测试数据集
+    # 训练数据集
     train_dataset = vision.datasets.MNIST(mode='train')
     
     # 验证数据集
@@ -340,9 +331,9 @@ paddle即可使用相关高层API，如：paddle.Model、视觉领域paddle.visi
 5. 模型训练
 -----------
 
-使用\ ``paddle.Model``\ 封装成模型类后进行训练非常的简洁方便，我们可以直接通过调用\ ``Model.fit``\ 就可以完成训练过程。
+网络结构通过\ ``paddle.Model``\ 接口封装成模型类后进行执行操作非常的简洁方便，可以直接通过调用\ ``Model.fit``\ 就可以完成训练过程。
 
-在使用\ ``Model.fit``\ 接口启动训练前，我们先通过\ ``Model.prepare``\ 接口来对训练进行提前的配置准备工作，包括设置模型优化器，Loss计算方法，精度计算方法等。
+使用\ ``Model.fit``\ 接口启动训练前，我们先通过\ ``Model.prepare``\ 接口来对训练进行提前的配置准备工作，包括设置模型优化器，Loss计算方法，精度计算方法等。
 
 .. code:: ipython3
 
@@ -398,10 +389,252 @@ paddle即可使用相关高层API，如：paddle.Model、视觉领域paddle.visi
     # train.py里面包含的就是单机单卡代码
     python -m paddle.distributed.launch train.py
 
+5.3 自定义Loss
+~~~~~~~~~~~~~~
+
+有时我们会遇到特定任务的Loss计算方式在框架既有的Loss接口中不存在，或算法不符合自己的需求，那么期望能够自己来进行Loss的自定义，我们这里就会讲解介绍一下如何进行Loss的自定义操作，首先来看下面的代码：
+
+.. code:: python
+
+   class SelfDefineLoss(paddle.nn.Layer):
+       """
+       1. 继承paddle.nn.Layer
+       """
+       def __init__(self):
+           """
+           2. 构造函数根据自己的实际算法需求和使用需求进行参数定义即可
+           """
+           super(SelfDefineLoss, self).__init__()
+
+       def forward(self, input, label):
+           """
+           3. 实现forward函数，forward在调用时会传递两个参数：input和label
+               - input：单个或批次训练数据经过模型前向计算输出结果
+               - label：单个或批次训练数据对应的标签数据
+
+               接口返回值是一个Tensor，根据自定义的逻辑加和或计算均值后的损失
+           """
+           # 使用Paddle中相关API自定义的计算逻辑
+           # output = xxxxx
+           # return output
+
+那么了解完代码层面如果编写自定义代码后我们看一个实际的例子，下面是在图像分割示例代码中写的一个自定义Loss，当时主要是想使用自定义的softmax计算维度。
+
+.. code:: python
+
+   class SoftmaxWithCrossEntropy(paddle.nn.Layer):
+       def __init__(self):
+           super(SoftmaxWithCrossEntropy, self).__init__()
+
+       def forward(self, input, label):
+           loss = F.softmax_with_cross_entropy(input, 
+                                               label, 
+                                               return_softmax=False,
+                                               axis=1)
+           return paddle.mean(loss)
+
+5.4 自定义Metric
+~~~~~~~~~~~~~~~~
+
+和Loss一样，如果遇到一些想要做个性化实现的操作时，我们也可以来通过框架完成自定义的评估计算方法，具体的实现方式如下：
+
+.. code:: python
+
+   class SelfDefineMetric(paddle.metric.Metric):
+       """
+       1. 继承paddle.metric.Metric
+       """
+       def __init__(self):
+           """
+           2. 构造函数实现，自定义参数即可
+           """
+           super(SelfDefineMetric, self).__init__()
+
+       def name(self):
+           """
+           3. 实现name方法，返回定义的评估指标名字
+           """
+           return '自定义评价指标的名字'
+
+       def compute(self, ...)
+           """
+           4. 本步骤可以省略，实现compute方法，这个方法主要用于`update`的加速，可以在这个方法中调用一些paddle实现好的Tensor计算API，编译到模型网络中一起使用低层C++ OP计算。
+           """
+
+           return 自己想要返回的数据，会做为update的参数传入。
+
+       def update(self, ...):
+           """
+           5. 实现update方法，用于单个batch训练时进行评估指标计算。
+           - 当`compute`类函数未实现时，会将模型的计算输出和标签数据的展平作为`update`的参数传入。
+           - 当`compute`类函数做了实现时，会将compute的返回结果作为`update`的参数传入。
+           """
+           return acc value
+       
+       def accumulate(self):
+           """
+           6. 实现accumulate方法，返回历史batch训练积累后计算得到的评价指标值。
+           每次`update`调用时进行数据积累，`accumulate`计算时对积累的所有数据进行计算并返回。
+           结算结果会在`fit`接口的训练日志中呈现。
+           """
+           # 利用update中积累的成员变量数据进行计算后返回
+           return accumulated acc value
+
+       def reset(self):
+           """
+           7. 实现reset方法，每个Epoch结束后进行评估指标的重置，这样下个Epoch可以重新进行计算。
+           """
+           # do reset action
+
+我们看一个框架中的具体例子，这个是框架中已提供的一个评估指标计算接口，这里就是按照上述说明中的实现方法进行了相关类继承和成员函数实现。
+
+.. code:: python
+
+   from paddle.metric import Metric
+
+
+   class Precision(Metric):
+       """
+       Precision (also called positive predictive value) is the fraction of
+       relevant instances among the retrieved instances. Refer to
+       https://en.wikipedia.org/wiki/Evaluation_of_binary_classifiers
+
+       Noted that this class manages the precision score only for binary
+       classification task.
+       
+       ......
+
+       """
+
+       def __init__(self, name='precision', *args, **kwargs):
+           super(Precision, self).__init__(*args, **kwargs)
+           self.tp = 0  # true positive
+           self.fp = 0  # false positive
+           self._name = name
+
+       def update(self, preds, labels):
+           """
+           Update the states based on the current mini-batch prediction results.
+
+           Args:
+               preds (numpy.ndarray): The prediction result, usually the output
+                   of two-class sigmoid function. It should be a vector (column
+                   vector or row vector) with data type: 'float64' or 'float32'.
+               labels (numpy.ndarray): The ground truth (labels),
+                   the shape should keep the same as preds.
+                   The data type is 'int32' or 'int64'.
+           """
+           if isinstance(preds, paddle.Tensor):
+               preds = preds.numpy()
+           elif not _is_numpy_(preds):
+               raise ValueError("The 'preds' must be a numpy ndarray or Tensor.")
+
+           if isinstance(labels, paddle.Tensor):
+               labels = labels.numpy()
+           elif not _is_numpy_(labels):
+               raise ValueError("The 'labels' must be a numpy ndarray or Tensor.")
+
+           sample_num = labels.shape[0]
+           preds = np.floor(preds + 0.5).astype("int32")
+
+           for i in range(sample_num):
+               pred = preds[i]
+               label = labels[i]
+               if pred == 1:
+                   if pred == label:
+                       self.tp += 1
+                   else:
+                       self.fp += 1
+
+       def reset(self):
+           """
+           Resets all of the metric state.
+           """
+           self.tp = 0
+           self.fp = 0
+
+       def accumulate(self):
+           """
+           Calculate the final precision.
+
+           Returns:
+               A scaler float: results of the calculated precision.
+           """
+           ap = self.tp + self.fp
+           return float(self.tp) / ap if ap != 0 else .0
+
+       def name(self):
+           """
+           Returns metric name
+           """
+           return self._name
+
+5.5 自定义Callback
+~~~~~~~~~~~~~~~~~~
+
+``fit``\ 接口的callback参数支持我们传一个Callback类实例，用来在每轮训练和每个batch训练前后进行调用，可以通过callback收集到训练过程中的一些数据和参数，或者实现一些自定义操作。
+
+.. code:: python
+
+   class SelfDefineCallback(paddle.callbacks.Callback):
+       """
+       1. 继承paddle.callbacks.Callback
+       2. 按照自己的需求实现以下类成员方法：
+           def on_train_begin(self, logs=None)                 训练开始前，`Model.fit`接口中调用
+           def on_train_end(self, logs=None)                   训练结束后，`Model.fit`接口中调用
+           def on_eval_begin(self, logs=None)                  评估开始前，`Model.evaluate`接口调用
+           def on_eval_end(self, logs=None)                    评估结束后，`Model.evaluate`接口调用
+           def on_test_begin(self, logs=None)                  预测测试开始前，`Model.predict`接口中调用
+           def on_test_end(self, logs=None)                    预测测试结束后，`Model.predict`接口中调用 
+           def on_epoch_begin(self, epoch, logs=None)          每轮训练开始前，`Model.fit`接口中调用 
+           def on_epoch_end(self, epoch, logs=None)            每轮训练结束后，`Model.fit`接口中调用 
+           def on_train_batch_begin(self, step, logs=None)     单个Batch训练开始前，`Model.fit`和`Model.train_batch`接口中调用
+           def on_train_batch_end(self, step, logs=None)       单个Batch训练结束后，`Model.fit`和`Model.train_batch`接口中调用
+           def on_eval_batch_begin(self, step, logs=None)      单个Batch评估开始前，`Model.evalute`和`Model.eval_batch`接口中调用
+           def on_eval_batch_end(self, step, logs=None)        单个Batch评估结束后，`Model.evalute`和`Model.eval_batch`接口中调用
+           def on_test_batch_begin(self, step, logs=None)      单个Batch预测测试开始前，`Model.predict`和`Model.test_batch`接口中调用
+           def on_test_batch_end(self, step, logs=None)        单个Batch预测测试结束后，`Model.predict`和`Model.test_batch`接口中调用
+       """
+       def __init__(self):
+           super(SelfDefineCallback, self).__init__()
+
+       按照需求定义自己的类成员方法
+
+我们看一个框架中的实际例子，这是一个框架自带的ModelCheckpoint回调函数，方便用户在fit训练模型时自动存储每轮训练得到的模型。
+
+.. code:: python
+
+   class ModelCheckpoint(Callback):
+       def __init__(self, save_freq=1, save_dir=None):
+           self.save_freq = save_freq
+           self.save_dir = save_dir
+
+       def on_epoch_begin(self, epoch=None, logs=None):
+           self.epoch = epoch
+
+       def _is_save(self):
+           return self.model and self.save_dir and ParallelEnv().local_rank == 0
+
+       def on_epoch_end(self, epoch, logs=None):
+           if self._is_save() and self.epoch % self.save_freq == 0:
+               path = '{}/{}'.format(self.save_dir, epoch)
+               print('save checkpoint at {}'.format(os.path.abspath(path)))
+               self.model.save(path)
+
+       def on_train_end(self, logs=None):
+           if self._is_save():
+               path = '{}/final'.format(self.save_dir)
+               print('save checkpoint at {}'.format(os.path.abspath(path)))
+               self.model.save(path)
+
 6. 模型评估
 -----------
 
-对于训练好的模型进行评估操作可以使用\ ``evaluate``\ 接口来实现。
+对于训练好的模型进行评估操作可以使用\ ``evaluate``\ 接口来实现，事先定义好用于评估使用的数据集后，可以简单的调用\ ``evaluate``\ 接口即可完成模型评估操作，结束后根据prepare中loss和metric的定义来进行相关评估结果计算返回。
+
+返回格式是一个字典： \* 只包含loss，\ ``{'loss': xxx}`` \*
+包含loss和一个评估指标，\ ``{'loss': xxx, 'metric name': xxx}`` \*
+包含loss和多个评估指标，\ ``{'loss': xxx, 'metric name': xxx, 'metric name': xxx}``
 
 .. code:: ipython3
 
@@ -410,19 +643,40 @@ paddle即可使用相关高层API，如：paddle.Model、视觉领域paddle.visi
 7. 模型预测
 -----------
 
-高层API中提供\ ``predict``\ 接口，支持用户使用测试数据来完成模型的预测。
+高层API中提供了\ ``predict``\ 接口来方便用户对训练好的模型进行预测验证，只需要基于训练好的模型将需要进行预测测试的数据放到接口中进行计算即可，接口会将经过模型计算得到的预测结果进行返回。
+
+返回格式是一个list，元素数目对应模型的输出数目： \*
+模型是单一输出：[(numpy_ndarray_1, numpy_ndarray_2, …, numpy_ndarray_n)]
+\* 模型是多输出：[(numpy_ndarray_1, numpy_ndarray_2, …,
+numpy_ndarray_n), (numpy_ndarray_1, numpy_ndarray_2, …,
+numpy_ndarray_n), …]
+
+numpy_ndarray_n是对应原始数据经过模型计算后得到的预测数据，数目对应预测数据集的数目。
 
 .. code:: ipython3
 
     pred_result = model.predict(val_dataset)
 
+7.1 使用多卡进行预测
+~~~~~~~~~~~~~~~~~~~~
+
+有时我们需要进行预测验证的数据较多，单卡无法满足我们的时间诉求，那么\ ``predict``\ 接口也为用户支持实现了使用多卡模式来运行。
+
+使用起来也是超级简单，无需修改代码程序，只需要使用launch来启动对应的预测脚本即可。
+
+.. code:: bash
+
+   $ python3 -m paddle.distributed.launch infer.py
+
+infer.py里面就是包含model.predict的代码程序。
+
 8. 模型部署
 -----------
 
 8.1 模型存储
 ~~~~~~~~~~~~
 
-模型训练和验证达到我们的预期后，可以使用\ ``save``\ 接口来将我们的模型保存下来，用于后续模型的Fine-tuning或推理部署。
+模型训练和验证达到我们的预期后，可以使用\ ``save``\ 接口来将我们的模型保存下来，用于后续模型的Fine-tuning（接口参数training=True）或推理部署（接口参数training=False）。
 
 .. code:: ipython3
 

doc/paddle/tutorial/quick_start/linear_regression/linear_regression.rst

 线性回归
 ========
 
-NOTE:
-本示例教程依然在开发中，目前是基于2.0beta版本（由于2.0beta没有正式发版，在用最新develop版whl包下载的paddle）。
+NOTE: 本示例教程是基于2.0beta版本开发
 
 简要介绍
 --------
@@ -10,20 +9,35 @@ NOTE:
 经典的线性回归模型主要用来预测一些存在着线性关系的数据集。回归模型可以理解为：存在一个点集，用一条曲线去拟合它分布的过程。如果拟合曲线是一条直线，则称为线性回归。如果是一条二次曲线，则被称为二次回归。线性回归是回归模型中最简单的一种。
 本示例简要介绍如何用飞桨开源框架，实现波士顿房价预测。其思路是，假设uci-housing数据集中的房子属性和房价之间的关系可以被属性间的线性组合描述。在模型训练阶段，让假设的预测结果和真实值之间的误差越来越小。在模型预测阶段，预测器会读取训练好的模型，对从未遇见过的房子属性进行房价预测。
 
-环境设置
---------
+数据集介绍
+----------
 
-本示例基于飞桨开源框架2.0版本。
+本示例采用uci-housing数据集，这是经典线性回归的数据集。数据集共7084条数据，可以拆分成506行,每行14列。前13列用来描述房屋的各种信息，最后一列为该类房屋价格中位数。
+
+前13列用来描述房屋的各种信息
+
+.. figure:: https://ai-studio-static-online.cdn.bcebos.com/c19602ce74284e3b9a50422f8dc37c0c1c79cf5cd8424994b6a6b073dcb7c057
+   :alt: avatar
+
+   avatar
+
+训练方式一
+----------
+
+环境设置
+~~~~~~~~
 
 .. code:: ipython3
 
     import paddle
     import numpy as np
     import os
+    import matplotlib
     import matplotlib.pyplot as plt
     import pandas as pd
     import seaborn as sns
     
+    paddle.disable_static()
     paddle.__version__
 
 
@@ -31,164 +45,143 @@ NOTE:
 
 .. parsed-literal::
 
-    '0.0.0'
-
-
-
-数据集
-------
+    '2.0.0-beta0'
 
-本示例采用uci-housing数据集，这是经典线性回归的数据集。数据集共506行,每行14列。前13列用来描述房屋的各种信息，最后一列为该类房屋价格中位数。飞桨提供了读取uci_housing训练集和测试集的接口，分别为paddle.dataset.uci_housing.train()和paddle.dataset.uci_housing.test()。
 
-前13列用来描述房屋的各种信息
 
-.. figure:: https://ai-studio-static-online.cdn.bcebos.com/c19602ce74284e3b9a50422f8dc37c0c1c79cf5cd8424994b6a6b073dcb7c057
-   :alt: avatar
+数据处理
+~~~~~~~~
 
-   avatar
+.. code:: ipython3
 
-下面我们来浏览一下数据是什么样子的：
+    #下载数据
+    #!wget https://archive.ics.uci.edu/ml/machine-learning-databases/housing/housing.data -O housing.data 
 
 .. code:: ipython3
 
-    import matplotlib.pyplot as plt
-    import matplotlib
-    
-    train_data=paddle.dataset.uci_housing.train()
-    sample_data=next(train_data())
-    print(sample_data[0])
+    # 从文件导入数据
+    datafile = './housing.data'
+    housing_data = np.fromfile(datafile, sep=' ')
     
-    # 画图看特征间的关系,主要是变量两两之间的关系（线性或非线性，有无明显较为相关关系）
     feature_names = ['CRIM', 'ZN', 'INDUS', 'CHAS', 'NOX', 'RM', 'AGE','DIS', 'RAD', 'TAX', 'PTRATIO', 'B', 'LSTAT', 'MEDV']
     feature_num = len(feature_names)
-    features_np=np.array([x[0] for x in train_data()],np.float32)
-    labels_np=np.array([x[1] for x in train_data()],np.float32)
-    data_np=np.c_[features_np,labels_np]
-    df=pd.DataFrame(data_np,columns=feature_names)
-    
+    # 将原始数据进行Reshape，变成[N, 14]这样的形状
+    housing_data = housing_data.reshape([housing_data.shape[0] // feature_num, feature_num])
+
+.. code:: ipython3
+
+    # 画图看特征间的关系,主要是变量两两之间的关系（线性或非线性，有无明显较为相关关系）
+    features_np = np.array([x[:13] for x in housing_data], np.float32)
+    labels_np = np.array([x[-1] for x in housing_data], np.float32)
+    data_np = np.c_[features_np, labels_np]
+    df = pd.DataFrame(data_np, columns=feature_names)
     matplotlib.use('TkAgg')
     %matplotlib inline
-    sns.pairplot(df.dropna())
-    
+    sns.pairplot(df.dropna(), y_vars=feature_names[-1], x_vars=feature_names[:])
     plt.show()
-    
 
 
 
-.. parsed-literal::
-
-    [-0.0405441   0.06636364 -0.32356227 -0.06916996 -0.03435197  0.05563625
-     -0.03475696  0.02682186 -0.37171335 -0.21419304 -0.33569506  0.10143217
-     -0.21172912]
-
+.. image:: https://github.com/PaddlePaddle/FluidDoc/tree/develop/doc/paddle/tutorial/quick_start/linear_regression/linear_regression_files/linear_regression_001.png?raw=true
 
 
-.. image:: linear_regression_files/linear_regression_6_1.png
-
-
-上图中，对角线上是各属性的直方图，非对角线上的是两个不同属性之间的相关图。
-从图中我们可以看出，RM（每栋房平均客房数）、LSTAT（低收入人群占比）、与房价成明显的相关关系、NOX（一氧化碳浓度）和DIS（与波士顿就业中心距离）成明显相关关系等。
-
 .. code:: ipython3
 
     # 相关性分析
-    fig, ax = plt.subplots(figsize=(15,15)) 
-    ax=sns.heatmap(df.corr(), cbar=True, annot=True)
-    ax.set_ylim([14, 0])
+    fig, ax = plt.subplots(figsize=(15, 1)) 
+    corr_data = df.corr().iloc[-1]
+    corr_data = np.asarray(corr_data).reshape(1, 14)
+    ax = sns.heatmap(corr_data, cbar=True, annot=True)
     plt.show()
 
 
 
-.. image:: linear_regression_files/linear_regression_8_0.png
+.. image:: https://github.com/PaddlePaddle/FluidDoc/tree/develop/doc/paddle/tutorial/quick_start/linear_regression/linear_regression_files/linear_regression_002.png?raw=true    
 
 
-**数据归一化处理**
-
-下图为大家展示各属性的取值范围分布：
+**数据归一化处理**\  下图为大家展示各属性的取值范围分布：
 
 .. code:: ipython3
 
-    sns.boxplot(data=df.iloc[:,0:13])
+    sns.boxplot(data=df.iloc[:, 0:13])
 
 
 
 
 .. parsed-literal::
 
-    <matplotlib.axes._subplots.AxesSubplot at 0x1a3adcb410>
+    <matplotlib.axes._subplots.AxesSubplot at 0x1a3e2b4e50>
 
 
 
 
-.. image:: linear_regression_files/linear_regression_11_1.png
+.. image:: https://github.com/PaddlePaddle/FluidDoc/tree/develop/doc/paddle/tutorial/quick_start/linear_regression/linear_regression_files/linear_regression_003.png?raw=true
 
 
-做归一化（或 Feature scaling）至少有以下3个理由：
+从上图看出，我们各属性的数值范围差异太大，甚至不能够在一个画布上充分的展示各属性具体的最大、最小值以及异常值等。下面我们进行归一化。
 
--  过大或过小的数值范围会导致计算时的浮点上溢或下溢。
+做归一化（或 Feature scaling）至少有以下2个理由：
 
+-  过大或过小的数值范围会导致计算时的浮点上溢或下溢。
 -  不同的数值范围会导致不同属性对模型的重要性不同（至少在训练的初始阶段如此），而这个隐含的假设常常是不合理的。这会对优化的过程造成困难，使训练时间大大的加长.
 
--  很多的机器学习技巧/模型（例如L1，L2正则项，向量空间模型-Vector Space
-   Model）都基于这样的假设：所有的属性取值都差不多是以0为均值且取值范围相近的。
-
 .. code:: ipython3
 
-    features_max=[]
-    features_min=[]
-    features_avg=[]
-    for i in range(13):
-      i_feature_max=max([data[1][0][i] for data in enumerate(train_data())])
-      features_max.append(i_feature_max)
-      i_feature_min=min([data[1][0][i] for data in enumerate(train_data())])
-      features_min.append(i_feature_min)
-      i_feature_avg=sum([data[1][0][i] for data in enumerate(train_data())])/506
-      features_avg.append(i_feature_avg)
-
+    features_max = housing_data.max(axis=0)
+    features_min = housing_data.min(axis=0)
+    features_avg = housing_data.sum(axis=0) / housing_data.shape[0]
 
 .. code:: ipython3
 
-    BATCH_SIZE=20
+    BATCH_SIZE = 20
     def feature_norm(input):
-      f_size=input.shape[0]
-      output_features=np.zeros((f_size,13),np.float32)
-      for batch_id in range(f_size):
-        for index in range(13):
-          output_features[batch_id][index]=(input[batch_id][index]-features_avg[index])/(features_max[index]-features_min[index])
-      return output_features 
-    
+        f_size = input.shape
+        output_features = np.zeros(f_size, np.float32)
+        for batch_id in range(f_size[0]):
+            for index in range(13):
+                output_features[batch_id][index] = (input[batch_id][index] - features_avg[index]) / (features_max[index] - features_min[index])
+        return output_features 
 
+.. code:: ipython3
 
-定义绘制训练过程的损失值变化趋势的方法draw_train_process
+    #只对属性进行归一化
+    housing_features = feature_norm(housing_data[:, :13])
+    # print(feature_trian.shape)
+    housing_data = np.c_[housing_features, housing_data[:, -1]].astype(np.float32)
+    # print(training_data[0])
 
 .. code:: ipython3
 
-    global iter
-    iter=0
-    iters=[]
-    train_costs=[]
-    
-    def draw_train_process(iters,train_costs):
-        plt.title("training cost" ,fontsize=24)
-        plt.xlabel("iter", fontsize=14)
-        plt.ylabel("cost", fontsize=14)
-        plt.plot(iters, train_costs,color='red',label='training cost')
-        plt.show()
-        
+    #归一化后的train_data,我们看下各属性的情况
+    features_np = np.array([x[:13] for x in housing_data],np.float32)
+    labels_np = np.array([x[-1] for x in housing_data],np.float32)
+    data_np = np.c_[features_np, labels_np]
+    df = pd.DataFrame(data_np, columns=feature_names)
+    sns.boxplot(data=df.iloc[:, 0:13])
+
 
-**数据提供器**
 
-下面我们分别定义了用于训练和测试的数据提供器。提供器每次读入一个大小为BATCH_SIZE的数据批次。如果您希望加一些随机性，它可以同时定义一个批次大小和一个缓存大小。这样的话，每次数据提供器会从缓存中随机读取批次大小那么多的数据。
+
+.. parsed-literal::
+
+    <matplotlib.axes._subplots.AxesSubplot at 0x1a3e4cd4d0>
+
+
+
+
+.. image:: https://github.com/PaddlePaddle/FluidDoc/tree/develop/doc/paddle/tutorial/quick_start/linear_regression/linear_regression_files/linear_regression_004.png?raw=true
+
 
 .. code:: ipython3
 
-    BATCH_SIZE=20
-    BUF_SIZE=500
-    
-    train_reader=paddle.batch(paddle.reader.shuffle(paddle.dataset.uci_housing.train(),buf_size=BUF_SIZE),batch_size=BATCH_SIZE)
+    #将训练数据集和测试数据集按照8:2的比例分开
+    ratio = 0.8
+    offset = int(housing_data.shape[0] * ratio)
+    train_data = housing_data[:offset]
+    test_data = housing_data[offset:]
 
 模型配置
---------
+~~~~~~~~
 
 线性回归就是一个从输入到输出的简单的全连接层。
 
@@ -197,16 +190,30 @@ NOTE:
 .. code:: ipython3
 
     class Regressor(paddle.nn.Layer):
-      def __init__(self):
-        super(Regressor,self).__init__()
-        self.fc=paddle.nn.Linear(13,1,None)
+        def __init__(self):
+            super(Regressor, self).__init__()
+            self.fc = paddle.nn.Linear(13, 1,)
+    
+        def forward(self, inputs):
+            pred = self.fc(inputs)
+            return pred
+
+定义绘制训练过程的损失值变化趋势的方法draw_train_process
+
+.. code:: ipython3
+
+    train_nums = []
+    train_costs = []
     
-      def forward(self,inputs):
-        pred=self.fc(inputs)
-        return pred
+    def draw_train_process(iters, train_costs):
+        plt.title("training cost", fontsize=24)
+        plt.xlabel("iter", fontsize=14)
+        plt.ylabel("cost", fontsize=14)
+        plt.plot(iters, train_costs, color='red', label='training cost')
+        plt.show()
 
 模型训练
----------
+~~~~~~~~
 
 下面为大家展示模型训练的代码。
 这里用到的是线性回归模型最常用的损失函数–均方误差（MSE），用来衡量模型预测的房价和真实房价的差异。
@@ -214,136 +221,128 @@ NOTE:
 
 .. code:: ipython3
 
-    y_preds=[]
-    labels_list=[]
-    def train(model):
-      print('start training ... ')
-      model.train()
-      EPOCH_NUM=500
-      optimizer=paddle.optimizer.SGD(learning_rate=0.001, parameters = model.parameters())
-      iter=0
+    import paddle.nn.functional as F 
+    y_preds = []
+    labels_list = []
     
-      for epoch_id in range(EPOCH_NUM):
-        train_cost=0
-        for batch_id,data in enumerate(train_reader()):
-            features_np=np.array([x[0] for x in data],np.float32)
-            labels_np=np.array([x[1] for x in data],np.float32)
-            features=paddle.to_variable(feature_norm(features_np))
-            labels=paddle.to_variable(labels_np)
-            #前向计算
-            y_pred=model(features)
-            
-            cost=paddle.nn.functional.square_error_cost(y_pred,label=labels)
-            avg_cost=paddle.mean(cost)
-            train_cost = [avg_cost.numpy()]
-            #反向传播
-            avg_cost.backward()
-            #最小化loss，更新参数
-            opts=optimizer.minimize(avg_cost)
-            # 清除梯度
-            model.clear_gradients()
-            if batch_id%30==0 and epoch_id%30==0:
-                print("Pass:%d,Cost:%0.5f"%(epoch_id,train_cost[0][0]))
+    def train(model):
+        print('start training ... ')
+        # 开启模型训练模式
+        model.train()
+        EPOCH_NUM = 500
+        train_num = 0
+        optimizer = paddle.optimizer.SGD(learning_rate=0.001, parameters=model.parameters())
+        for epoch_id in range(EPOCH_NUM):
+            # 在每轮迭代开始之前，将训练数据的顺序随机的打乱
+            np.random.shuffle(train_data)
+            # 将训练数据进行拆分，每个batch包含20条数据
+            mini_batches = [train_data[k:k+BATCH_SIZE] for k in range(0, len(train_data), BATCH_SIZE)]
+            for batch_id, data in enumerate(mini_batches):
+                features_np = np.array(data[:, :13], np.float32)
+                labels_np = np.array(data[:, -1:], np.float32)
+                features = paddle.to_tensor(features_np)
+                labels = paddle.to_tensor(labels_np)
+                #前向计算
+                y_pred = model(features)
+                cost = F.mse_loss(y_pred, label=labels)
+                train_cost = cost.numpy()[0]
+                #反向传播
+                cost.backward()
+                #最小化loss，更新参数
+                optimizer.step()
+                # 清除梯度
+                optimizer.clear_grad()
+                
+                if batch_id%30 == 0 and epoch_id%50 == 0:
+                    print("Pass:%d,Cost:%0.5f"%(epoch_id, train_cost))
     
-            iter=iter+BATCH_SIZE
-            iters.append(iter)
-            train_costs.append(train_cost[0][0])
+                train_num = train_num + BATCH_SIZE
+                train_nums.append(train_num)
+                train_costs.append(train_cost)
             
-                
-              
-    paddle.disable_static()
     model = Regressor()
     train(model)
-    
-
 
 
 .. parsed-literal::
 
     start training ... 
-    Pass:0,Cost:531.75244
-    Pass:30,Cost:61.10927
-    Pass:60,Cost:22.68571
-    Pass:90,Cost:34.80560
-    Pass:120,Cost:78.28358
-    Pass:150,Cost:124.95644
-    Pass:180,Cost:91.88014
-    Pass:210,Cost:15.23689
-    Pass:240,Cost:34.86035
-    Pass:270,Cost:54.76824
-    Pass:300,Cost:65.88247
-    Pass:330,Cost:41.25426
-    Pass:360,Cost:64.10200
-    Pass:390,Cost:77.11707
-    Pass:420,Cost:20.80456
-    Pass:450,Cost:29.80167
-    Pass:480,Cost:41.59278
+    Pass:0,Cost:740.21814
+    Pass:50,Cost:36.40338
+    Pass:100,Cost:86.01823
+    Pass:150,Cost:50.86654
+    Pass:200,Cost:31.14208
+    Pass:250,Cost:20.54596
+    Pass:300,Cost:22.30817
+    Pass:350,Cost:24.18756
+    Pass:400,Cost:22.22965
+    Pass:450,Cost:39.25978
 
 
 .. code:: ipython3
 
     matplotlib.use('TkAgg')
     %matplotlib inline
-    draw_train_process(iters,train_costs)
+    draw_train_process(train_nums, train_costs)
 
 
 
-.. image:: linear_regression_files/linear_regression_23_0.png
+.. image:: https://github.com/PaddlePaddle/FluidDoc/tree/develop/doc/paddle/tutorial/quick_start/linear_regression/linear_regression_files/linear_regression_005.png?raw=true
 
 
 可以从上图看出，随着训练轮次的增加，损失在呈降低趋势。但由于每次仅基于少量样本更新参数和计算损失，所以损失下降曲线会出现震荡。
 
 模型预测
-----------
+~~~~~~~~
 
 .. code:: ipython3
 
     #获取预测数据
-    INFER_BATCH_SIZE=100
-    infer_reader=paddle.batch(paddle.dataset.uci_housing.test(),batch_size=INFER_BATCH_SIZE)
-    infer_data = next(infer_reader())
-    infer_features_np = np.array([data[0] for data in infer_data]).astype("float32")
-    infer_labels_np= np.array([data[1] for data in infer_data]).astype("float32")
+    INFER_BATCH_SIZE = 100
     
-    infer_features=paddle.to_variable(feature_norm(infer_features_np))
-    infer_labels=paddle.to_variable(infer_labels_np)
-    fetch_list=model(infer_features).numpy()
+    infer_features_np = np.array([data[:13] for data in test_data]).astype("float32")
+    infer_labels_np = np.array([data[-1] for data in test_data]).astype("float32")
     
-    sum_cost=0
+    infer_features = paddle.to_tensor(infer_features_np)
+    infer_labels = paddle.to_tensor(infer_labels_np)
+    fetch_list = model(infer_features)
+    
+    sum_cost = 0
     for i in range(INFER_BATCH_SIZE):
-        infer_result=fetch_list[i][0]
-        ground_truth=infer_labels.numpy()[i]
-        if i%10==0:
-            print("No.%d: infer result is %.2f,ground truth is %.2f" % (i, infer_result,ground_truth))
-        cost=np.power(infer_result-ground_truth,2)
-        sum_cost+=cost
-    print("平均误差为:",sum_cost/INFER_BATCH_SIZE)
+        infer_result = fetch_list[i][0]
+        ground_truth = infer_labels[i]
+        if i % 10 == 0:
+            print("No.%d: infer result is %.2f,ground truth is %.2f" % (i, infer_result, ground_truth))
+        cost = paddle.pow(infer_result - ground_truth, 2)
+        sum_cost += cost
+    mean_loss = sum_cost / INFER_BATCH_SIZE
+    print("Mean loss is:", mean_loss.numpy())
 
 
 .. parsed-literal::
 
-    No.0: infer result is 12.20,ground truth is 8.50
-    No.10: infer result is 5.65,ground truth is 7.00
-    No.20: infer result is 14.87,ground truth is 11.70
-    No.30: infer result is 16.60,ground truth is 11.70
-    No.40: infer result is 13.71,ground truth is 10.80
-    No.50: infer result is 16.11,ground truth is 14.90
-    No.60: infer result is 18.78,ground truth is 21.40
-    No.70: infer result is 15.53,ground truth is 13.80
-    No.80: infer result is 18.10,ground truth is 20.60
-    No.90: infer result is 21.39,ground truth is 24.50
-    平均误差为: [12.917107]
+    No.0: infer result is 12.15,ground truth is 8.50
+    No.10: infer result is 5.21,ground truth is 7.00
+    No.20: infer result is 14.32,ground truth is 11.70
+    No.30: infer result is 16.11,ground truth is 11.70
+    No.40: infer result is 13.42,ground truth is 10.80
+    No.50: infer result is 15.50,ground truth is 14.90
+    No.60: infer result is 18.81,ground truth is 21.40
+    No.70: infer result is 15.42,ground truth is 13.80
+    No.80: infer result is 18.16,ground truth is 20.60
+    No.90: infer result is 21.48,ground truth is 24.50
+    Mean loss is: [12.195988]
 
 
 .. code:: ipython3
 
-    def plot_pred_ground(pred, groud):
+    def plot_pred_ground(pred, ground):
         plt.figure()   
-        plt.title("Predication v.s. Ground", fontsize=24)
-        plt.xlabel("groud price(unit:$1000)", fontsize=14)
+        plt.title("Predication v.s. Ground truth", fontsize=24)
+        plt.xlabel("ground truth price(unit:$1000)", fontsize=14)
         plt.ylabel("predict price", fontsize=14)
-        plt.scatter(pred, groud, alpha=0.5)  #  scatter:散点图,alpha:"透明度"
-        plt.plot(groud, groud, c='red')
+        plt.scatter(ground, pred, alpha=0.5)  #  scatter:散点图,alpha:"透明度"
+        plt.plot(ground, ground, c='red')
         plt.show()
 
 .. code:: ipython3
@@ -352,7 +351,78 @@ NOTE:
 
 
 
-.. image:: linear_regression_files/linear_regression_28_0.png
+.. image:: https://github.com/PaddlePaddle/FluidDoc/tree/develop/doc/paddle/tutorial/quick_start/linear_regression/linear_regression_files/linear_regression_001.png?raw=true
 
 
 上图可以看出，我们训练出来的模型的预测结果与真实结果是较为接近的。
+
+训练方式二
+----------
+
+我们也可以用我们的高层API来做线性回归训练，高层API相较于底层API更加的简洁方便。
+
+.. code:: ipython3
+
+    import paddle
+    paddle.disable_static()
+    paddle.set_default_dtype("float64")
+    
+    #step1:用高层API定义数据集，无需进行数据处理等，高层API为您一条龙搞定
+    train_dataset = paddle.text.datasets.UCIHousing(mode='train')
+    eval_dataset = paddle.text.datasets.UCIHousing(mode='test')
+    
+    #step2:定义模型
+    class UCIHousing(paddle.nn.Layer):
+        def __init__(self):
+            super(UCIHousing, self).__init__()
+            self.fc = paddle.nn.Linear(13, 1, None)
+    
+        def forward(self, input):
+            pred = self.fc(input)
+            return pred
+    
+    #step3:训练模型
+    model = paddle.Model(UCIHousing())
+    model.prepare(paddle.optimizer.Adam(parameters=model.parameters()),
+                  paddle.nn.loss.MSELoss())
+    model.fit(train_dataset, eval_dataset, epochs=5, batch_size=8, log_freq=20)
+
+
+.. parsed-literal::
+
+    Epoch 1/5
+    step 20/51 - loss: 520.8663 - 1ms/step
+    step 40/51 - loss: 611.7135 - 1ms/step
+    step 51/51 - loss: 620.0662 - 1ms/step
+    Eval begin...
+    step 13/13 - loss: 389.7871 - 1ms/step
+    Eval samples: 102
+    Epoch 2/5
+    step 20/51 - loss: 867.4678 - 3ms/step
+    step 40/51 - loss: 1081.1701 - 2ms/step
+    step 51/51 - loss: 420.8705 - 2ms/step
+    Eval begin...
+    step 13/13 - loss: 387.2432 - 1ms/step
+    Eval samples: 102
+    Epoch 3/5
+    step 20/51 - loss: 810.1555 - 2ms/step
+    step 40/51 - loss: 840.3570 - 2ms/step
+    step 51/51 - loss: 421.0806 - 2ms/step
+    Eval begin...
+    step 13/13 - loss: 384.7417 - 693us/step
+    Eval samples: 102
+    Epoch 4/5
+    step 20/51 - loss: 647.1215 - 1ms/step
+    step 40/51 - loss: 682.9673 - 1ms/step
+    step 51/51 - loss: 422.0570 - 1ms/step
+    Eval begin...
+    step 13/13 - loss: 382.2546 - 591us/step
+    Eval samples: 102
+    Epoch 5/5
+    step 20/51 - loss: 713.3719 - 1ms/step
+    step 40/51 - loss: 567.0962 - 1ms/step
+    step 51/51 - loss: 456.8702 - 1ms/step
+    Eval begin...
+    step 13/13 - loss: 379.7527 - 985us/step
+    Eval samples: 102
+

doc/paddle/tutorial/quick_start/save_model/save_model.ipynb

@@ -14,19 +14,19 @@
    "metadata": {},
    "source": [
     "## 环境\n",
-    "本教程基于paddle-develop编写，如果您的环境不是本版本，请先安装paddle-develop版本。"
+    "本教程基于paddle-2.0Beta编写，如果您的环境不是此版本，请先安装paddle-2.0Beta版本，使用命令：pip3 install paddlepaddle==2.0Beta。"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 9,
+   "execution_count": 1,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "0.0.0\n"
+      "2.0.0-beta0\n"
      ]
     }
    ],

doc/paddle/tutorial/quick_start/save_model/save_model.rst

@@ -7,7 +7,8 @@
 环境
 ----
 
-本教程基于paddle-develop编写，如果您的环境不是本版本，请先安装paddle-develop版本。
+本教程基于paddle-2.0Beta编写，如果您的环境不是此版本，请先安装paddle-2.0Beta版本，使用命令：pip3
+install paddlepaddle==2.0Beta。
 
 .. code:: ipython3
 
@@ -25,7 +26,7 @@
 
 .. parsed-literal::
 
-    0.0.0
+    2.0.0-beta0
 
 
 数据集