{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "5rmpybwysXGV"
      },
      "source": [
        "````{admonition} Copyright 2020 The TensorFlow Authors.\n",
        "```\n",
        "#@title Licensed under the Apache License, Version 2.0 (the \"License\");\n",
        "# you may not use this file except in compliance with the License.\n",
        "# You may obtain a copy of the License at\n",
        "#\n",
        "# https://www.apache.org/licenses/LICENSE-2.0\n",
        "#\n",
        "# Unless required by applicable law or agreed to in writing, software\n",
        "# distributed under the License is distributed on an \"AS IS\" BASIS,\n",
        "# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.\n",
        "# See the License for the specific language governing permissions and\n",
        "# limitations under the License.\n",
        "```\n",
        "````"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "hrXv0rU9sIma"
      },
      "source": [
        "# TensorFlow 基础知识"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7S0BwJ_8sLu7"
      },
      "source": [
        "<table class=\"tfo-notebook-buttons\" align=\"left\">\n",
        "  <td>     <a target=\"_blank\" href=\"https://tensorflow.google.cn/guide/basics\" class=\"_active_edit_href\"><img src=\"https://tensorflow.google.cn/images/tf_logo_32px.png\">转到TensorFlow.org</a>\n",
        "</td>\n",
        "  <td><a target=\"_blank\" href=\"https://colab.research.google.com/github/tensorflow/docs-l10n/blob/master/site/zh-cn/guide/basics.ipynb\"><img src=\"https://tensorflow.google.cn/images/colab_logo_32px.png\">在 Google Colab 中运行</a></td>\n",
        "  <td><a target=\"_blank\" href=\"https://github.com/tensorflow/docs-l10n/blob/master/site/zh-cn/guide/basics.ipynb\">     <img src=\"https://tensorflow.google.cn/images/GitHub-Mark-32px.png\">     在 GitHub 上查看源代码</a></td>\n",
        "  <td>     <a href=\"https://storage.googleapis.com/tensorflow_docs/docs-l10n/site/zh-cn/guide/basics.ipynb\"><img src=\"https://tensorflow.google.cn/images/download_logo_32px.png\">下载笔记本</a>   </td>\n",
        "</table>\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "iJyZUDbzBTIG"
      },
      "source": [
        "本指南提供*TensorFlow 基础知识*的快速概览。本文档的每个部分都是对一个大主题的概述——您可以在每个部分的末尾找到指向完整指南的链接。\n",
        "\n",
        "TensorFlow 是一个端到端的机器学习平台。它支持以下内容：\n",
        "\n",
        "- 基于多维数组的数值计算（类似于<a href=\"https://numpy.org/\" class=\"external\">NumPy</a> 。）\n",
        "- GPU 和分布式处理\n",
        "- 自动微分\n",
        "- 模型构造、训练和导出\n",
        "- 及更多内容"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 28,
      "metadata": {},
      "outputs": [],
      "source": [
        "import os\n",
        "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '3'  # 设置日志级别为ERROR，以减少警告信息\n",
        "# 禁用 Gemini 的底层库（gRPC 和 Abseil）在初始化日志警告\n",
        "os.environ[\"GRPC_VERBOSITY\"] = \"ERROR\"\n",
        "os.environ[\"GLOG_minloglevel\"] = \"3\"  # 0: INFO, 1: WARNING, 2: ERROR, 3: FATAL\n",
        "os.environ[\"GLOG_minloglevel\"] = \"true\"\n",
        "import logging\n",
        "import tensorflow as tf\n",
        "tf.get_logger().setLevel(logging.ERROR)\n",
        "tf.compat.v1.logging.set_verbosity(tf.compat.v1.logging.ERROR)\n",
        "!export TF_FORCE_GPU_ALLOW_GROWTH=true\n",
        "from pathlib import Path\n",
        "\n",
        "temp_dir = Path(\".temp\")\n",
        "temp_dir.mkdir(parents=True, exist_ok=True)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "gvLegMMvBZYg"
      },
      "source": [
        "## 张量\n",
        "\n",
        "TensorFlow 对<code>tf.Tensor</code>对象表示的多维数组或<em>张量</em>进行运算。下面一个二维张量："
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 2,
      "metadata": {
        "id": "6ZqX5RnbBS1f"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "tf.Tensor(\n",
            "[[1. 2. 3.]\n",
            " [4. 5. 6.]], shape=(2, 3), dtype=float32)\n",
            "(2, 3)\n",
            "<dtype: 'float32'>\n"
          ]
        }
      ],
      "source": [
        "import tensorflow as tf\n",
        "\n",
        "x = tf.constant([[1., 2., 3.],\n",
        "                 [4., 5., 6.]])\n",
        "\n",
        "print(x)\n",
        "print(x.shape)\n",
        "print(x.dtype)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "k-AOMqevQGN4"
      },
      "source": [
        "`tf.Tensor`最重要的属性是它的`shape`(形状)和`dtype`(数据类型) ：\n",
        "\n",
        "- `Tensor.shape` ：表示张量每个轴上的大小。\n",
        "- `Tensor.dtype` ：表示张量中所有元素的类型。"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "bUkKeNWZCIJO"
      },
      "source": [
        "TensorFlow 实现了对张量的标准数学运算，以及许多机器学习的专用操作。\n",
        "\n",
        "例如："
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 3,
      "metadata": {
        "id": "BM7xXNDsBfN5"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "<tf.Tensor: shape=(2, 3), dtype=float32, numpy=\n",
              "array([[ 2.,  4.,  6.],\n",
              "       [ 8., 10., 12.]], dtype=float32)>"
            ]
          },
          "execution_count": 3,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "x + x"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 4,
      "metadata": {
        "id": "ZLGqscTxB61v"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "<tf.Tensor: shape=(2, 3), dtype=float32, numpy=\n",
              "array([[ 5., 10., 15.],\n",
              "       [20., 25., 30.]], dtype=float32)>"
            ]
          },
          "execution_count": 4,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "5 * x"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 5,
      "metadata": {
        "id": "2ImJHd8VfnWq"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "<tf.Tensor: shape=(2, 2), dtype=float32, numpy=\n",
              "array([[14., 32.],\n",
              "       [32., 77.]], dtype=float32)>"
            ]
          },
          "execution_count": 5,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "x @ tf.transpose(x)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 6,
      "metadata": {
        "id": "U9JZD6TYCZWu"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "<tf.Tensor: shape=(6, 3), dtype=float32, numpy=\n",
              "array([[1., 2., 3.],\n",
              "       [4., 5., 6.],\n",
              "       [1., 2., 3.],\n",
              "       [4., 5., 6.],\n",
              "       [1., 2., 3.],\n",
              "       [4., 5., 6.]], dtype=float32)>"
            ]
          },
          "execution_count": 6,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "tf.concat([x, x, x], axis=0)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 7,
      "metadata": {
        "id": "seGBLeD9P_PI"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "<tf.Tensor: shape=(2, 3), dtype=float32, numpy=\n",
              "array([[0.09003057, 0.24472848, 0.6652409 ],\n",
              "       [0.09003057, 0.24472848, 0.6652409 ]], dtype=float32)>"
            ]
          },
          "execution_count": 7,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "tf.nn.softmax(x, axis=-1)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 8,
      "metadata": {
        "id": "YZNZRv1ECjf8"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "<tf.Tensor: shape=(), dtype=float32, numpy=21.0>"
            ]
          },
          "execution_count": 8,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "tf.reduce_sum(x)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "TNHnIjOVLJfA"
      },
      "source": [
        "注：通常，在 TensorFlow 函数需要 `Tensor` 作为输入的任何地方，该函数也将接受可使用 `tf.convert_to_tensor` 转换为 `Tensor` 的任何内容。请参见下面的示例。"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 9,
      "metadata": {
        "id": "i_XKgjDsL4GE"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "<tf.Tensor: shape=(3,), dtype=int32, numpy=array([1, 2, 3], dtype=int32)>"
            ]
          },
          "execution_count": 9,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "tf.convert_to_tensor([1,2,3])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 10,
      "metadata": {
        "id": "wTBt-JUqLJDJ"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "<tf.Tensor: shape=(), dtype=int32, numpy=6>"
            ]
          },
          "execution_count": 10,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "tf.reduce_sum([1,2,3])"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "8-mi5031DVxz"
      },
      "source": [
        "在 CPU 上运行大型计算可能会很慢。配置正确的 TensorFlow 可以使用 GPU 等加速硬件非常快速地运算。"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 11,
      "metadata": {
        "id": "m97Gv5H6Dz0G"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "TensorFlow **IS** using the GPU\n"
          ]
        }
      ],
      "source": [
        "if tf.config.list_physical_devices('GPU'):\n",
        "  print(\"TensorFlow **IS** using the GPU\")\n",
        "else:\n",
        "  print(\"TensorFlow **IS NOT** using the GPU\")"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ln2FkLOqMX92"
      },
      "source": [
        "详情参阅 [Tensor guide](tensor.ipynb) （张量指南）。"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "oVbomvMyEIVF"
      },
      "source": [
        "## 变量\n",
        "\n",
        "普通的`tf.Tensor`对象是不可变的。要在 TensorFlow 中存储模型权重（或其他可变状态），请使用`tf.Variable` 。"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 14,
      "metadata": {
        "id": "SO8_bP4UEzxS"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "<tf.Variable 'Variable:0' shape=(3,) dtype=float32, numpy=array([0., 0., 0.], dtype=float32)>"
            ]
          },
          "execution_count": 14,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "var = tf.Variable([0.0, 0.0, 0.0])\n",
        "var"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 15,
      "metadata": {
        "id": "aDLYFvu5FAFa"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "<tf.Variable 'UnreadVariable' shape=(3,) dtype=float32, numpy=array([1., 2., 3.], dtype=float32)>"
            ]
          },
          "execution_count": 15,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "var.assign([1, 2, 3])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 16,
      "metadata": {
        "id": "9EpiOmxXFDSS"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "<tf.Variable 'UnreadVariable' shape=(3,) dtype=float32, numpy=array([2., 3., 4.], dtype=float32)>"
            ]
          },
          "execution_count": 16,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "var.assign_add([1, 1, 1])"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "tlvTpi1CMedC"
      },
      "source": [
        "详见 [Variables guide](variable.ipynb) （变量指南）。"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "rG1Dhv2QFkV3"
      },
      "source": [
        "## 自动微分\n",
        "\n",
        "<a href=\"https://en.wikipedia.org/wiki/Gradient_descent\" class=\"external\"><em>Gradient descent</em></a>(梯度下降)及相关算法是现代机器学习的基础。\n",
        "\n",
        "为此，TensorFlow 实现了自动微分 (autodiff)，它使用微积分来计算梯度。通常用它来计算模型基于其权重的*误差*或*损失*的梯度。"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 17,
      "metadata": {
        "id": "cYKOi-z4GY9Y"
      },
      "outputs": [],
      "source": [
        "x = tf.Variable(1.0)\n",
        "\n",
        "def f(x):\n",
        "  y = x**2 + 2*x - 5\n",
        "  return y"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 18,
      "metadata": {
        "id": "IQz99cxMGoF_"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "<tf.Tensor: shape=(), dtype=float32, numpy=-2.0>"
            ]
          },
          "execution_count": 18,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "f(x)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ozLLop0cHeYl"
      },
      "source": [
        "在 `x = 1.0`, `y = f(x) = (1**2 + 2*1 - 5) = -2`.\n",
        "\n",
        "`y`的导数是 `y' = f'(x) = (2*x + 2) = 4` 。 TensorFlow 可以自动计算："
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 19,
      "metadata": {
        "id": "N02NfWpHGvw8"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "<tf.Tensor: shape=(), dtype=float32, numpy=4.0>"
            ]
          },
          "execution_count": 19,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "with tf.GradientTape() as tape:\n",
        "  y = f(x)\n",
        "\n",
        "g_x = tape.gradient(y, x)  # g(x) = dy/dx\n",
        "\n",
        "g_x"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "s-DVYJfcIRPd"
      },
      "source": [
        "这个简化的例子只对单个标量 ( `x` ) 求导，但 TensorFlow 可以同时计算任意数量的非标量张量的梯度。"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ECK3I9bUMk_r"
      },
      "source": [
        "详见 [Autodiff guide](autodiff.ipynb) （自动微分）。"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "VglUM4M3KhNz"
      },
      "source": [
        "## 图和 tf.function装饰器\n",
        "\n",
        " TensorFlow 可以像任何 Python 库一样以交互方式使用，同时还提供以下工具：\n",
        "\n",
        "- **性能优化**：加速训练和推理。\n",
        "- **导出**：保存训练好的模型。\n",
        "\n",
        "这就要用 `tf.function` 装饰器将纯 TensorFlow 代码与普通 Python 代码隔离开来。"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 20,
      "metadata": {
        "id": "VitACyZWKJD_"
      },
      "outputs": [],
      "source": [
        "@tf.function\n",
        "def my_func(x):\n",
        "  print('Tracing.\\n')\n",
        "  return tf.reduce_sum(x)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "fBYDh-huNUBZ"
      },
      "source": [
        "第一次运行由`tf.function`装饰的函数时，虽然它在 Python 中执行，但它会取得经由 TensorFlow 计算的完整优化图。"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 21,
      "metadata": {
        "id": "vkOFSEkoM1bd"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Tracing.\n",
            "\n"
          ]
        },
        {
          "data": {
            "text/plain": [
              "<tf.Tensor: shape=(), dtype=int32, numpy=6>"
            ]
          },
          "execution_count": 21,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "x = tf.constant([1, 2, 3])\n",
        "my_func(x)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "a3aWzt-rNsBa"
      },
      "source": [
        "在后续调用中，TensorFlow 仅执行优化图，跳过所有非 TensorFlow 步骤。注意下面的`my_func`不打印*Tracing*，因为`print`是 Python 函数，而不是 TensorFlow 函数。"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 22,
      "metadata": {
        "id": "23dMHWwwNIoa"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "<tf.Tensor: shape=(), dtype=int32, numpy=27>"
            ]
          },
          "execution_count": 22,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "x = tf.constant([10, 9, 8])\n",
        "my_func(x)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "nSeTti6zki0n"
      },
      "source": [
        "输入的*签名*（`shape`和`dtype` ）不同，就不能使用原来的计算图，要生成新计算图："
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 23,
      "metadata": {
        "id": "OWffqyhqlVPf"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Tracing.\n",
            "\n"
          ]
        },
        {
          "data": {
            "text/plain": [
              "<tf.Tensor: shape=(), dtype=float32, numpy=27.3>"
            ]
          },
          "execution_count": 23,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "x = tf.constant([10.0, 9.1, 8.2], dtype=tf.float32)\n",
        "my_func(x)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "UWknAA_zNTOa"
      },
      "source": [
        "这些取得的图提供了两个好处：\n",
        "\n",
        "- 在许多情况下，显著提升了执行速度（虽然在上面这个例子中无关紧要）。\n",
        "- 可以用`tf.saved_model`导出这些图，以便在[服务器](https://tensorflow.google.cn/tfx/serving/docker)或[移动设备](https://tensorflow.google.cn/lite/guide)等其他系统上运行，无需安装 Python。"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "hLUJ6f2eMsA8"
      },
      "source": [
        "详见 [Intro to graphs](intro_to_graphs.ipynb) （计算图的说明）."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "t_36xPDPPBqp"
      },
      "source": [
        "## 模块、层和模型"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "oDaT7kCpUgnJ"
      },
      "source": [
        "`tf.Module`是一个类，用于管理`tf.Variable`对象以及对它们进行操作的`tf.function`对象。 有了`tf.Module`类，才能支持下面两个重要特性：\n",
        "\n",
        "1. 可以用`tf.train.Checkpoint`保存和恢复变量的值。因为可以快速保存和恢复模型的状态，所以在训练期间很有用。\n",
        "2. 可以用`tf.saved_model`导入和导出<code>tf.Variable</code>值<em>和</em>`tf.function`图。这使得模型可以不依赖原来的 Python 程序独立运行。\n",
        "\n",
        "下面是导出简单 `tf.Module` 对象的完整例子："
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 24,
      "metadata": {
        "id": "1MqEcZOqPBDV"
      },
      "outputs": [],
      "source": [
        "class MyModule(tf.Module):\n",
        "  def __init__(self, value):\n",
        "    self.weight = tf.Variable(value)\n",
        "\n",
        "  @tf.function\n",
        "  def multiply(self, x):\n",
        "    return x * self.weight"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 25,
      "metadata": {
        "id": "la2G82HfVfU0"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "<tf.Tensor: shape=(3,), dtype=int32, numpy=array([3, 6, 9], dtype=int32)>"
            ]
          },
          "execution_count": 25,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "mod = MyModule(3)\n",
        "mod.multiply(tf.constant([1, 2, 3]))"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "GaSJX7zQXCm4"
      },
      "source": [
        "保存`Module` ："
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 29,
      "metadata": {
        "id": "1MlfbEMjVzG4"
      },
      "outputs": [],
      "source": [
        "save_path = temp_dir/'./saved'\n",
        "tf.saved_model.save(mod, save_path)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "LgfoftD4XGJW"
      },
      "source": [
        "保存的图独立于创建它的代码。您可以从 Python、其他语言绑定或[TensorFlow Serving](https://tensorflow.google.cn/tfx/serving/docker)加载保存的图。还可以通过转换，让它在[TensorFlow Lite](https://tensorflow.google.cn/lite/guide)或[TensorFlow JS](https://tensorflow.google.cn/js/guide)上运行。"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 30,
      "metadata": {
        "id": "pWuLOIKBWZYG"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "<tf.Tensor: shape=(3,), dtype=int32, numpy=array([3, 6, 9], dtype=int32)>"
            ]
          },
          "execution_count": 30,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "reloaded = tf.saved_model.load(save_path)\n",
        "reloaded.multiply(tf.constant([1, 2, 3]))"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "nxU6P1RGwHyC"
      },
      "source": [
        "建立在`tf.Module`上的`tf.keras.layers.Layer`类和`tf.keras.Model`类，为构建、训练和保存模型提供了更多的功能和便利。下一节中将展示其中一部分。"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "tQzt3yaWMzLf"
      },
      "source": [
        "详见 [Intro to modules](intro_to_modules.ipynb) （模块介绍）."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Rk1IEG5aav7X"
      },
      "source": [
        "## 训练循环\n",
        "\n",
        "现在用这些东西一起构建基本模型并从头开始训练。\n",
        "\n",
        "首先，生成一些示例数据，这是一些围绕二次曲线的松散的点形成的云："
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 32,
      "metadata": {
        "id": "VcuFr7KPRPzn"
      },
      "outputs": [],
      "source": [
        "import matplotlib\n",
        "from matplotlib import pyplot as plt\n",
        "\n",
        "matplotlib.rcParams['figure.figsize'] = [9, 6]"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 33,
      "metadata": {
        "id": "sXN9E_xf-GiP"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 900x600 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "x = tf.linspace(-2, 2, 201)\n",
        "x = tf.cast(x, tf.float32)\n",
        "\n",
        "def f(x):\n",
        "  y = x**2 + 2*x - 5\n",
        "  return y\n",
        "\n",
        "y = f(x) + tf.random.normal(shape=[201])\n",
        "\n",
        "plt.plot(x.numpy(), y.numpy(), '.', label='Data')\n",
        "plt.plot(x, f(x), label='Ground truth')\n",
        "plt.legend();"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "De5LldboSWcW"
      },
      "source": [
        "创建具有随机初始化权重和偏差的二次模型："
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 34,
      "metadata": {
        "id": "Pypd0GB4SRhf"
      },
      "outputs": [],
      "source": [
        "class Model(tf.Module):\n",
        "\n",
        "  def __init__(self):\n",
        "    # Randomly generate weight and bias terms\n",
        "    rand_init = tf.random.uniform(shape=[3], minval=0., maxval=5., seed=22)\n",
        "    # Initialize model parameters\n",
        "    self.w_q = tf.Variable(rand_init[0])\n",
        "    self.w_l = tf.Variable(rand_init[1])\n",
        "    self.b = tf.Variable(rand_init[2])\n",
        "  \n",
        "  @tf.function\n",
        "  def __call__(self, x):\n",
        "    # Quadratic Model : quadratic_weight * x^2 + linear_weight * x + bias\n",
        "    return self.w_q * (x**2) + self.w_l * x + self.b"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "36o7VjaesScg"
      },
      "source": [
        "首先，在训练前观察您的模型的性能："
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 35,
      "metadata": {
        "id": "GkwToC5BWV1c"
      },
      "outputs": [],
      "source": [
        "quad_model = Model()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 36,
      "metadata": {
        "id": "ReWhH40wTY5F"
      },
      "outputs": [],
      "source": [
        "def plot_preds(x, y, f, model, title):\n",
        "  plt.figure()\n",
        "  plt.plot(x, y, '.', label='Data')\n",
        "  plt.plot(x, f(x), label='Ground truth')\n",
        "  plt.plot(x, model(x), label='Predictions')\n",
        "  plt.title(title)\n",
        "  plt.legend()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 37,
      "metadata": {
        "id": "Y0JtXQat-nlk"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 900x600 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "plot_preds(x, y, f, quad_model, 'Before training')"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "hLzwD0-ascGf"
      },
      "source": [
        "现在，为您的模型定义损失：\n",
        "\n",
        "鉴于此模型的作用是预测连续值，因此均方误差 (MSE) 是损失函数的不错选择。给定一个预测向量 $\\hat{y}$ 和一个真实目标向量 $y$，MSE 被定义为预测值与基准值之间平方差的平均值。\n",
        "\n",
        "$MSE = \\frac{1}{m}\\sum_{i=1}^{m}(\\hat{y}_i -y_i)^2$"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 38,
      "metadata": {
        "id": "eCtJ1uuCseZd"
      },
      "outputs": [],
      "source": [
        "def mse_loss(y_pred, y):\n",
        "  return tf.reduce_mean(tf.square(y_pred - y))"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7EWyDu3zot2w"
      },
      "source": [
        "为模型编写一个基本训练循环。此循环将利用 MSE 损失函数及其相对于输入的梯度来迭代更新模型的参数。使用 mini-batch 进行训练可以提供内存效率和更快的收敛速度。`tf.data.Dataset` API 具有用于批处理和重排的实用函数。"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 39,
      "metadata": {
        "id": "8kX_-zily2Ia"
      },
      "outputs": [],
      "source": [
        "batch_size = 32\n",
        "dataset = tf.data.Dataset.from_tensor_slices((x, y))\n",
        "dataset = dataset.shuffle(buffer_size=x.shape[0]).batch(batch_size)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 40,
      "metadata": {
        "id": "nOaES5gyTDtG"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Mean squared error for step 0: 57.023\n",
            "Mean squared error for step 10: 9.908\n",
            "Mean squared error for step 20: 4.178\n",
            "Mean squared error for step 30: 2.158\n",
            "Mean squared error for step 40: 1.459\n",
            "Mean squared error for step 50: 1.218\n",
            "Mean squared error for step 60: 1.131\n",
            "Mean squared error for step 70: 1.102\n",
            "Mean squared error for step 80: 1.094\n",
            "Mean squared error for step 90: 1.091\n",
            "\n",
            "\n"
          ]
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 900x600 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "# Set training parameters\n",
        "epochs = 100\n",
        "learning_rate = 0.01\n",
        "losses = []\n",
        "\n",
        "# Format training loop\n",
        "for epoch in range(epochs):\n",
        "  for x_batch, y_batch in dataset:\n",
        "    with tf.GradientTape() as tape:\n",
        "      batch_loss = mse_loss(quad_model(x_batch), y_batch)\n",
        "    # Update parameters with respect to the gradient calculations\n",
        "    grads = tape.gradient(batch_loss, quad_model.variables)\n",
        "    for g, v in zip(grads, quad_model.variables):\n",
        "        v.assign_sub(learning_rate*g)\n",
        "  # Keep track of model loss per epoch\n",
        "  loss = mse_loss(quad_model(x), y)\n",
        "  losses.append(loss)\n",
        "  if epoch % 10 == 0:\n",
        "    print(f'Mean squared error for step {epoch}: {loss.numpy():0.3f}')\n",
        "\n",
        "# Plot model results\n",
        "print(\"\\n\")\n",
        "plt.plot(range(epochs), losses)\n",
        "plt.xlabel(\"Epoch\")\n",
        "plt.ylabel(\"Mean Squared Error (MSE)\")\n",
        "plt.title('MSE loss vs training iterations');"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "dW5B2TTRsvxE"
      },
      "source": [
        "现在，观察模型在训练后的性能："
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 41,
      "metadata": {
        "id": "Qcvzyg3eYLh8"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 900x600 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "plot_preds(x, y, f, quad_model, 'After training')"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "hbtmFJIXb6qm"
      },
      "source": [
        "成功了，但请记住，`tf.keras` 模块中提供了常见训练实用工具的实现。因此在您自己动手编写之前，请优先考虑使用现成的内容。首先，`Model.compile` 和 `Model.fit` 方法为您实现了训练循环："
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "cjx23MiztFmT"
      },
      "source": [
        "首先，使用 `tf.keras.Sequential` 在 Keras 中创建序贯模型。最简单的 Keras 层之一是密集层，可以使用 `tf.keras.layers.Dense` 进行实例化。密集层能够学习 $\\mathrm{Y} = \\mathrm{W}\\mathrm{X} + \\vec{b}$ 形式的多维线性关系。要学习 $w_1x^2 + w_2x + b$ 形式的非线性方程，密集层的输入应当是一个以 $x^2$ 和 $x$ 为特征的数据矩阵。lambda 层 `tf.keras.layers.Lambda` 可用于执行这种堆叠转换。"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 42,
      "metadata": {
        "id": "5rt8HP2TZhEM"
      },
      "outputs": [],
      "source": [
        "new_model = tf.keras.Sequential([\n",
        "    tf.keras.layers.Lambda(lambda x: tf.stack([x, x**2], axis=1)),\n",
        "    tf.keras.layers.Dense(units=1, kernel_initializer=tf.random.normal)])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 44,
      "metadata": {
        "id": "73kCo1BtP3rQ"
      },
      "outputs": [],
      "source": [
        "new_model.compile(\n",
        "    loss=tf.keras.losses.MSE,\n",
        "    optimizer=tf.keras.optimizers.SGD(learning_rate=0.01))\n",
        "\n",
        "history = new_model.fit(x, y,\n",
        "                        epochs=100,\n",
        "                        batch_size=32,\n",
        "                        verbose=0)\n",
        "\n",
        "new_model.save(temp_dir/'./my_new_model.keras')"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "u3q5d1SzvzTq"
      },
      "source": [
        "训练后观察 Keras 模型的性能："
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 45,
      "metadata": {
        "id": "Mo7zRV7XZjv7"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 900x600 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "plt.plot(history.history['loss'])\n",
        "plt.xlabel('Epoch')\n",
        "plt.ylim([0, max(plt.ylim())])\n",
        "plt.ylabel('Loss [Mean Squared Error]')\n",
        "plt.title('Keras training progress');"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 46,
      "metadata": {
        "id": "bB44a9YsvnfK"
      },
      "outputs": [
        {
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAvQAAAIQCAYAAADuCfqCAAAAP3RFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjkuMS5wb3N0MSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8kixA/AAAACXBIWXMAAA9hAAAPYQGoP6dpAACUlklEQVR4nOzdd3xT1f/H8VeS7g20pQUKZe8heyiCylBxoYLiQv3hAP2quMAFuMCNE9fXjYIDXF9BmS4QAVH2lFJGoZSRQneT+/sjtlI66Egz38/How9ocnPvyW3Svu/J55xjMgzDQEREREREvJLZ3Q0QEREREZHqU6AXEREREfFiCvQiIiIiIl5MgV5ERERExIsp0IuIiIiIeDEFehERERERL6ZALyIiIiLixRToRURERES8mAK9iIiIiIgXU6AXESnDhx9+SJs2bQgMDCQmJsbdzak17733HiaTiZSUlCo/dunSpZhMJpYuXer0domISOUp0IuI33nttdcwmUz06tWrzPs3b97M6NGjad68OW+99RZvvvkm2dnZTJ482WXhdcCAAZhMplN+TZ482SXt8RajR48mIiKi1O1r164lNjaW5OTkal28iIh4MpNhGIa7GyEi4kr9+vVj3759pKSksG3bNlq0aFHi/tdff51bb721xH0ZGRnExcUxadIkl4ToBQsWcODAgeLvV65cyUsvvcQDDzxA27Zti2/v1KkTnTp1qvZxbDYbBQUFBAcHYzKZqvRYu91Ofn4+QUFBmM2e0T80evRoPv/8c44fP1582/r16xk4cCBhYWEsXbqUpk2burGFIiLOF+DuBoiIuNLOnTtZtmwZc+bM4eabb2bmzJlMmjSpxDbp6ekALim1ycrKIjw8vNTtgwYNKvF9SEgIL730EoMGDWLAgAFV3l95LBYLFoul0tufyGw2ExISUq3HusqGDRs466yzCA0NZcmSJU4J81U9xyIitc0zulRERFxk5syZ1KlTh/PPP5/LLruMmTNnlrg/OTm5OODHxcVhMpkYPXo0cXFxAEyZMqXMcpfNmzdz2WWXUbduXUJCQujevTtff/11iX0X1av/+OOPjB07lvj4eBo1alTt5zJ58mRMJhMbN25k1KhR1KlTh9NPPx1wlJiMHj2aZs2aERISQkJCAjfccAOHDh0qs00nlqEkJyczbNgwfvnlF3r27ElISAjNmjXjgw8+KPHYsmroBwwYQIcOHdi4cWNxr3jDhg15+umnS7V/165dXHjhhYSHhxMfH89dd93F999/X2qf2dnZbN68mYyMjCqdn02bNnH22WcTHBzMkiVLaNasWYn7V6xYwdChQ4mOjiYsLIwzzzyTX3/9tcQ2zjjHx44d48477yQ5OZng4GDi4+MZNGgQf/zxR5Wej4hIedRDLyJ+ZebMmQwfPpygoCCuvPJKZsyYwcqVK+nRowcA06dP54MPPmDu3LnMmDGDiIgIOnbsSO/evbn11lu55JJLGD58OEBxqcuGDRvo168fDRs2ZMKECYSHh/Ppp59y8cUX88UXX3DJJZeUaMPYsWOJi4vjkUceISsrq8bP6fLLL6dly5Y8+eSTFFVRLliwgL///pvrr7+ehIQENmzYwJtvvsmGDRv47bffTlles337di677DJuvPFGrrvuOt555x1Gjx5Nt27daN++fYWPPXLkCEOHDmX48OGMGDGCzz//nPvvv5+OHTty7rnnAo5e7rPOOou0tDTuuOMOEhIS+Pjjj1myZEmp/f3+++8MHDiwSuVOW7Zs4ayzziIgIIAlS5bQvHnzEvcvXryYc889l27dujFp0iTMZjPvvvsuZ511Fj///DM9e/YssX1NzvEtt9zC559/zm233Ua7du04dOgQv/zyC5s2baJr166Vej4iIhUyRET8xKpVqwzAWLBggWEYhmG3241GjRoZd9xxR4ntJk2aZADGwYMHi287ePCgARiTJk0qtd+zzz7b6Nixo5Gbm1t8m91uN/r27Wu0bNmy+LZ3333XAIzTTz/dKCwsrFLbP/vsMwMwlixZUqqdV155Zants7OzS932ySefGIDx008/lWrTzp07i29r0qRJqe3S09ON4OBg4+677y6+bcmSJaXadOaZZxqA8cEHHxTflpeXZyQkJBiXXnpp8W3PPfecARhffvll8W05OTlGmzZtSu2z6DhlnfuTXXfddUZgYKCRmJhoNGjQwNi6dWupbex2u9GyZUtjyJAhht1uL749OzvbaNq0qTFo0KDi25xxjqOjo41x48adsu0iItWlkhsR8RszZ86kfv36DBw4EACTycTIkSOZNWsWNputWvs8fPgwixcvZsSIERw7doyMjAwyMjI4dOgQQ4YMYdu2bezdu7fEY8aMGVPtuvWy3HLLLaVuCw0NLf5/bm4uGRkZ9O7dG6BSpR7t2rXjjDPOKP4+Li6O1q1b8/fff5/ysREREVx99dXF3wcFBdGzZ88Sj50/fz4NGzbkwgsvLL4tJCSEMWPGlNrfgAEDMAyj0r3zNpuNjIwM6tatS2xsbKn7//zzT7Zt28aoUaM4dOhQ8c8sKyuLs88+m59++gm73V7iMTU5xzExMaxYsYJ9+/ZVqv0iIlWlQC8ifsFmszFr1iwGDhzIzp072b59O9u3b6dXr14cOHCARYsWVWu/27dvxzAMHn74YeLi4kp8FdXiFw2yLeLsWVbK2t/hw4e54447qF+/PqGhocTFxRVvZ7VaT7nPxo0bl7qtTp06HDly5JSPbdSoUamSnpMfu2vXLpo3b15qu5NnHKqO0NBQPvjgAzZu3Mj5559fqqxp27ZtAFx33XWlfmZvv/02eXl5pc5RTc7x008/zfr160lKSqJnz55Mnjy5UhdGIiKVpRp6EfELixcvJi0tjVmzZjFr1qxS98+cOZPBgwdXeb9FPbn33HMPQ4YMKXObk0PqiT27zlDW/kaMGMGyZcu499576dKlCxEREdjtdoYOHVqq97ks5X2CYFRipuOaPNZZrrjiCo4cOcLYsWMZPnw433zzDUFBQcC/P7NnnnmGLl26lPn4k+eyr8k5HjFiBGeccQZz587lhx9+4JlnnuGpp55izpw5xWMKRERqQoFeRPzCzJkziY+P59VXXy1135w5c5g7dy6vv/56uWG7vEGkRTOnBAYGcs455zivwTVw5MgRFi1axJQpU3jkkUeKby/qmfYETZo0YePGjRiGUeLcbt++3WnHuPXWWzl8+DAPPfQQV199NbNmzcJsNhcPkI2Kiqr2z6yq5zgxMZGxY8cyduxY0tPT6dq1K0888YQCvYg4hUpuRMTn5eTkMGfOHIYNG8Zll11W6uu2227j2LFjpaaZPFFYWBgAR48eLXF7fHw8AwYM4I033iAtLa3U4w4ePOjU51IZRT3kJ/eIT58+3eVtKc+QIUPYu3dviXOem5vLW2+9VWrb6k5bCfDggw9y11138dlnn3HzzTcD0K1bN5o3b86zzz5bYgGqIpX5mVX2HNtstlLlO/Hx8TRo0IC8vLyqPBURkXKph15EfN7XX3/NsWPHSgzAPFHv3r2Ji4tj5syZjBw5ssxtQkNDadeuHbNnz6ZVq1bUrVuXDh060KFDB1599VVOP/10OnbsyJgxY2jWrBkHDhxg+fLl7Nmzh7/++qs2n14pUVFR9O/fn6effpqCggIaNmzIDz/8wM6dO13ajorcfPPNvPLKK1x55ZXccccdJCYmMnPmzOKFqk7sta/OtJUneu655zhy5Ahvv/02devW5amnnuLtt9/m3HPPpX379lx//fU0bNiQvXv3smTJEqKiovjmm28q3Gdlz/GxY8do1KgRl112GZ07dyYiIoKFCxeycuVKnnvuuSo/FxGRsijQi4jPKwqKJ6++WsRsNnP++eczc+bMUosCnejtt9/m9ttv56677iI/P59JkybRoUMH2rVrx6pVq5gyZQrvvfcehw4dIj4+ntNOO61EOYYrffzxx9x+++28+uqrGIbB4MGDmTdvHg0aNHBLe04WERHB4sWLuf3223nxxReJiIjg2muvpW/fvlx66aVOXYHWZDLx9ttvc/ToUZ5++mnq1KnDhAkTWL58OY899hivvPIKx48fJyEhgV69ehX35J9KZc5xWFgYY8eO5YcffmDOnDnY7XZatGjBa6+9xq233uq05ygi/s1kuHKUkoiISAWmT5/OXXfdxZ49e2jYsKG7myMi4hUU6EVExC1ycnJKzeV+2mmnYbPZ2Lp1qxtbJiLiXVRyIyIibjF8+HAaN25Mly5dsFqtfPTRR2zevJmZM2e6u2kiIl5FgV5ERNxiyJAhvP3228ycORObzUa7du2YNWtWuQOTRUSkbCq5ERERERHxYpqHXkRERETEiynQi4iIiIh4Ma+vobfb7ezbt4/IyMhyl2YXEREREfEmhmFw7NgxGjRogNlccR+81wf6ffv2kZSU5O5miIiIiIg43e7du2nUqFGF23h9oI+MjAQcTzYqKsrNrRERERERqbnMzEySkpKKs25FvD7QF5XZREVFKdCLiIiIiE+pTEm5BsWKiIiIiHgxBXoRERERES+mQC8iIiIi4sW8voa+smw2GwUFBe5uhniowMBALBaLu5shIiIiUmU+H+gNw2D//v0cPXrU3U0RDxcTE0NCQoLWMxARERGv4vOBvijMx8fHExYWprAmpRiGQXZ2Nunp6QAkJia6uUUiIiIilefTgd5msxWH+Xr16rm7OeLBQkNDAUhPTyc+Pl7lNyIiIuI1fHpQbFHNfFhYmJtbIt6g6HWisRYiIiLiTXw60BdRmY1Uhl4nIiIi4o38ItCLiIiIiPgqBXoRERERES+mQO+hRo8ejclkwmQyERgYSP369Rk0aBDvvPMOdru90vt57733iImJqb2GioiIiIhbKdB7sKFDh5KWlkZKSgrz5s1j4MCB3HHHHQwbNozCwkJ3N09EREREPIACfSWlWXNYtiODNGuOy44ZHBxMQkICDRs2pGvXrjzwwAN89dVXzJs3j/feew+A559/no4dOxIeHk5SUhJjx47l+PHjACxdupTrr78eq9Va3Ns/efJkAD788EO6d+9OZGQkCQkJjBo1qngedhERERHxHgr0lTB7ZSr9pi1m1Fsr6DdtMbNXprqtLWeddRadO3dmzpw5AJjNZl566SU2bNjA+++/z+LFi7nvvvsA6Nu3L9OnTycqKoq0tDTS0tK45557AMfUjI899hh//fUXX375JSkpKYwePdpdT0tEREREqsmnF5ZyhjRrDhPnrMNuOL63G/DAnPX0bxVHYnSoW9rUpk0b1q5dC8Cdd95ZfHtycjKPP/44t9xyC6+99hpBQUFER0djMplISEgosY8bbrih+P/NmjXjpZdeokePHhw/fpyIiAiXPA8RERHxPWnWHHZmZNE0NtxtWcnfKNCfws6MrOIwX8RmGKRkZLvtRWoYRvGc6QsXLmTq1Kls3ryZzMxMCgsLyc3NJTs7u8IFtVavXs3kyZP566+/OHLkSPFA29TUVNq1a+eS5yEiIiK+ZfbK1OKOULMJpg7vyMgejd3dLJ+nkptTaBobjvmk9YYsJhPJse5bfXbTpk00bdqUlJQUhg0bRqdOnfjiiy9YvXo1r776KgD5+fnlPj4rK4shQ4YQFRXFzJkzWblyJXPnzj3l40RERETKU15VgyvHH/orBfpTSIwOZerwjlj+6RG3mEw8ObyD23rnFy9ezLp167j00ktZvXo1drud5557jt69e9OqVSv27dtXYvugoCBsNluJ2zZv3syhQ4eYNm0aZ5xxBm3atNGAWBEREamRiqoapHap5KYSRvZoTP9WcaRkZJMcG+ayMJ+Xl8f+/fux2WwcOHCA+fPnM3XqVIYNG8a1117L+vXrKSgo4OWXX+aCCy7g119/5fXXXy+xj+TkZI4fP86iRYvo3LkzYWFhNG7cmKCgIF5++WVuueUW1q9fz2OPPeaS5yQiIiK+qaiq4cRQ7+6qBn+hHvpKSowOpU/zei7tmZ8/fz6JiYkkJyczdOhQlixZwksvvcRXX32FxWKhc+fOPP/88zz11FN06NCBmTNnMnXq1BL76Nu3L7fccgsjR44kLi6Op59+mri4ON577z0+++wz2rVrx7Rp03j22Wdd9rxERETE93haVYM/MRmGYZx6M8+VmZlJdHQ0VquVqKioEvfl5uayc+dOmjZtSkhIiJtaKN5CrxcREZGaS7PmuLyqwRdVlHFPppIbEREREXGaxOhQBXkXU8mNiIiIiIgXU6AXEREREfFiCvQiIiIiIl5MgV5ERERExIsp0IuIiIiIT0qz5rBsR0b1V6v1kskgNcuNiIiIiPic2StTmThnHXYDzCaYOrwjI3s0rvwOCvPgkyuh5xhofW7tNdQJ1EMvIiIiIj4lzZpTHObBsXrtA3PWV76n3jDg27tgxyL48lbItdZeY51AgV5EREREfMrOjKziMF/EZhikZGRXbge/vQZ/zgSTGS57B0Kind9IJ1KgF6eaPHkyXbp0cXczGDBgAHfeeae7myEiIiJu0DQ2HLOp5G0Wk4nk2LBTP3j7QvjhIcf/hzwJzc9yfgOdTIHeQ+3fv5877riDFi1aEBISQv369enXrx8zZswgO7uSV5ceaOnSpZhMJo4ePeqR+xMRERHvlxgdytThHbGYHKneYjLx5PAOp17BNmMbfHYDGHY47WrodYsLWltzGhTrgf7++2/69etHTEwMTz75JB07diQ4OJh169bx5ptv0rBhQy688MIyH1tQUEBgYKCLW+x8+fn5BAUFubsZIiIi4qVG9mhM/1ZxpGRkkxwbduown30YPh4JeVZI6gXnPw8mU8WP8RDqofdAY8eOJSAggFWrVjFixAjatm1Ls2bNuOiii/jf//7HBRdcULytyWRixowZXHjhhYSHh/PEE08AMGPGDJo3b05QUBCtW7fmww8/LH5MSkoKJpOJP//8s/i2o0ePYjKZWLp0KfBvz/eiRYvo3r07YWFh9O3bly1btpRo67Rp06hfvz6RkZHceOON5Obmlvu8UlJSGDhwIAB16tTBZDIxevRowFEic9ttt3HnnXcSGxvLkCFDTtnOivYHYLfbue+++6hbty4JCQlMnjy5sj8CERER8QGJ0aH0aV7v1GHeVgCfjYbDOyA6CUbOhIBgl7TRGfwr0BsG5Ge556uS85geOnSIH374gXHjxhEeHl7mNqaTrhYnT57MJZdcwrp167jhhhuYO3cud9xxB3fffTfr16/n5ptv5vrrr2fJkiVVPmUPPvggzz33HKtWrSIgIIAbbrih+L5PP/2UyZMn8+STT7Jq1SoSExN57bXXyt1XUlISX3zxBQBbtmwhLS2NF198sfj+999/n6CgIH799Vdef/31U7atMvsLDw9nxYoVPP300zz66KMsWLCgyudAREREfJhhwLz7YOePEBgOV86CiDh3t6pK/KvkpiAbnmzgnmM/sA+Cyg7oJ9q+fTuGYdC6desSt8fGxhb3fo8bN46nnnqq+L5Ro0Zx/fXXF39/5ZVXMnr0aMaOHQvA+PHj+e2333j22WeLe7Qr64knnuDMM88EYMKECZx//vnk5uYSEhLC9OnTufHGG7nxxhsBePzxx1m4cGG5vfQWi4W6desCEB8fT0xMTIn7W7ZsydNPP138fUpKSoVtO9X+OnXqxKRJk4r3/corr7Bo0SIGDRpUqecuIiIifuD3t2DVO4AJLn0bEjq4u0VV5l899F7s999/588//6R9+/bk5eWVuK979+4lvt+0aRP9+vUrcVu/fv3YtGlTlY/bqVOn4v8nJiYCkJ6eXnycXr16ldi+T58+VT5GkW7dulX7sWU5se3gaH9R20VERETYvgjmT3D8/5zJ0OY8tzanuvyrhz4wzNFT7q5jV0KLFi0wmUylatWbNWsGQGho6Rqw8kpzymM2O67jjBPKgAoKCsrc9sQBtkWlPna7vUrHq6yTn0dV2lmWkwcHm0ymWmu7iIiIVF+aNYedGVk0jQ0/db27sxzcCp9dD4YNOo+Cfne45ri1wL966E0mR9mLO74qOUq6Xr16DBo0iFdeeYWsrKxqPc22bdvy66+/lrjt119/pV27dgDExTnqwtLS0orvP3HgaVWOs2LFihK3/fbbbxU+pmjmGpvNdsr9V6adVdmfiIiIeJ7ZK1PpN20xo95aQb9pi5m9MrX2D5p9GD4pmtGmN1ww3WtmtCmLf/XQe4nXXnuNfv360b17dyZPnkynTp0wm82sXLmSzZs3n7I05d5772XEiBGcdtppnHPOOXzzzTfMmTOHhQsXAo5e/t69ezNt2jSaNm1Keno6Dz30UJXbeccddzB69Gi6d+9Ov379mDlzJhs2bCj+NKEsTZo0wWQy8e2333LeeecRGhpKREREmdtWpp1V2Z+IiIh4ljRrDhPnrCte1dVuwANz1tO/VVzt9dTbCuDTa+Hw3xDdGEZ+5FUz2pTFv3rovUTz5s1Zs2YN55xzDhMnTqRz5850796dl19+mXvuuYfHHnuswsdffPHFvPjiizz77LO0b9+eN954g3fffZcBAwYUb/POO+9QWFhIt27duPPOO3n88cer3M6RI0fy8MMPc99999GtWzd27drFrbfeWuFjGjZsyJQpU5gwYQL169fntttuq3D7U7WzqvsTERERz7EzI6s4zBexGQYpGVVbRDPNmsOyHRmkWXMq3tAw4Lt7IeVnCIqAUd43o01ZTIZRyfkUPVRmZibR0dFYrVaioqJK3Jebm8vOnTtp2rQpISEhbmqheAu9XkRERFwrzZpDv2mLS4R6i8nELxMGVrqHfvbK1OJefrMJpg7vyMgejcveeMUbjikqMcGVn0Drcyvc9/qM9bSu25pAs+sX7awo455MPfQiIiIi4haJ0aFMHd4Ryz/16xaTiSeHd6h0mC+vZKfMnvrtC/+d0WbQo6cM83+m/8no+aMZt3Ac2QVV+8TA1VRDLyIiIiJuM7JHY/q3iiMlI5vk2LAq1c5XVLJTYj8Ht8BnN4Bhhy5XQ9/bK9zv7szd/Gfxf8iz5RFoCSTIElSVp+RyCvQiIiIi4laJ0aHVGgTbNDYcs4lSJTvJsSdMF559GD7+Z0abxn1g2PMVzmhzNPcoty66lSN5R2hbty3P9H+GALNnR2aV3IiIiIiIVzplyU5hvmNGmyM7IebUM9rk2fL4z5L/sCtzF4nhibx69quEVXItIXfy7MsNEREREZEKlFuyYxgw74QZba6cDeGx5e7Hbth58JcHWZO+hsjASF47+zXiwrxjBhwFehERERHxamWW7Cx/BVa/B5jg0v9C/XYV7mP6H9P5PuV7AswBvDDwBVrUaVFr7XU2ldyIiIiIiG/Z9A388LDj/4Mfh9ZDK9z80y2f8u76dwGY0ncKvRJ71XYLnUqBXkRERER8x97V8MUYwIAe/wd9xlW4+U97fuKJFU8AMLbLWC5sfqELGulcCvQiIiIi4huOpsInV0JhDrQYBEOfqnBGm/UZ67nnx3uwG3Yuan4Rt3S6xYWNdR4Fej83evRoLr744uLvBwwYwJ133lmjfTpjHyIiIiJVkmt1TE95/ADU7wCXvQOW8oeL7s7czbhF48gpzKFvg75M6jsJUwXh35Mp0Huo0aNHYzKZMJlMBAUF0aJFCx599FEKCwtr9bhz5szhscceq9S2S5cuxWQycfTo0WrvQ0RERKTGbAXw2WhI3wgRCTBqNoRElbv54dzD3LLwFg7nHqZt3bY8P+B5As2BgGP12WU7MspebdZDaZYbDzZ06FDeffdd8vLy+O677xg3bhyBgYFMnDixxHb5+fkEBTlnBbO6det6xD5ERETEt6VZc9iZkUXT2PASM9SUd3u5DAO+uxd2LIbAMA5e+D7bMkJoSk6Zj88uyOa2RbeReiyVhhENee2c1wgPDAdg9spUJs5Zh90AswmmDu/IyB6Nnfaca4t66D1YcHAwCQkJNGnShFtvvZVzzjmHr7/+urhM5oknnqBBgwa0bt0agN27dzNixAhiYmKoW7cuF110ESkpKcX7s9lsjB8/npiYGOrVq8d9992HYZRcL/nkcpm8vDzuv/9+kpKSCA4OpkWLFvz3v/8lJSWFgQMHAlCnTh1MJhOjR48ucx9Hjhzh2muvpU6dOoSFhXHuueeybdu24vvfe+89YmJi+P7772nbti0REREMHTqUtLS04m2WLl1Kz549CQ8PJyYmhn79+rFr1y4nnWkRERH/5upe6dkrU+k3bTGj3lpBv2mLmb0ytcLbK7TsZVj9LmDi587T6PXuoTIfn2bN4eftB/jPovGsy1hHeEAUj/aaTmxobPH9RWEeHKvPPjBnvVf01PtVD71hGOQUuueHEhoQWuO6rNDQUA4dOgTAokWLiIqKYsGCBQAUFBQwZMgQ+vTpw88//0xAQACPP/44Q4cOZe3atQQFBfHcc8/x3nvv8c4779C2bVuee+455s6dy1lnnVXuMa+99lqWL1/OSy+9ROfOndm5cycZGRkkJSXxxRdfcOmll7JlyxaioqIIDS37Knr06NFs27aNr7/+mqioKO6//37OO+88Nm7cSGCg4+Ot7Oxsnn32WT788EPMZjNXX30199xzDzNnzqSwsJCLL76YMWPG8Mknn5Cfn8/vv//utXVuIiIinsTVvdLlBec2CZFl3t6/VVz5PfUbv4YFjwBgPXMK1/0QW+bjf9p6kIlz1hJYfy5BdX7HsAeQvv0qrli/g6nDwxjZozE7M7KKH1vEZhikZGRX7pMCN3J7oJ86dSpz5sxh8+bNhIaG0rdvX5566qniXmdnyinModfH7plXdMWoFdVeOtgwDBYtWsT333/P7bffzsGDBwkPD+ftt98uLrX56KOPsNvtvP3228VB99133yUmJoalS5cyePBgpk+fzsSJExk+fDgAr7/+Ot9//325x926dSuffvopCxYs4JxzzgGgWbNmxfcXldbEx8cTExNT5j6Kgvyvv/5K3759AZg5cyZJSUl8+eWXXH755YDjguT111+nefPmANx22208+uijAGRmZmK1Whk2bFjx/W3btq36iRQREZESygvXFYboGhxrZ0YWh7PyywzOK1OOVC1Q71kNc27CMT3lGDYkjcJu/F7q8atTjjBxzjoC6i12hHnDRO7eK7HnNAFg4hfrCA8OIKlOKGYTJdpgMZlIjq1efnMltwf6H3/8kXHjxtGjRw8KCwt54IEHGDx4MBs3biQ8PNzdzXOrb7/9loiICAoKCrDb7YwaNYrJkyczbtw4OnbsWKJu/q+//mL79u1ERkaW2Edubi47duzAarWSlpZGr17/XtAEBATQvXv3UmU3Rf78808sFgtnnnlmtZ/Dpk2bCAgIKHHcevXq0bp1azZt2lR8W1hYWHFYB0hMTCQ9PR1wXDiMHj2aIUOGMGjQIM455xxGjBhBYmJitdslIiIiuKxX+sRPAUw4vk48rMVkokdyncoH6qOp8MkVjukpWw6GodNoerygzMdjAnPUSoLjHFUNefsvovB4++Jt7MBtH6/BbIJLTmvIl2v2YTMMLCYTTw7v4PG98+ABgX7+/Pklvn/vvfeIj49n9erV9O/f36nHCg0IZcWoFU7dZ1WOXVUDBw5kxowZBAUF0aBBAwIC/v1xnXyxc/z4cbp168bMmTNL7ScuLq7qDYZyS2hqQ1HpTRGTyVTiQuPdd9/lP//5D/Pnz2f27Nk89NBDLFiwgN69e7usjSIiIr6maWx4rfdKn/wpgIEj0Bcdtyg4d06qw9ThHXlgzvqKA3WuFWaOgKx0qN+xeHrKxOiAMh9vC9lESOJcAPIyBlBwtOzsYDfgyzX7mDO2D9n5dpJjw7wizIMHBPqTWa1WoHZmSjGZTNUue3GH8PBwWrRoUaltu3btyuzZs4mPjycqquxpmhITE1mxYkXxhVJhYSGrV6+ma9euZW7fsWNH7HY7P/74Y3HJzYmKPiGw2Wzltqtt27YUFhayYsWK4pKbQ4cOsWXLFtq1a1ep51bktNNO47TTTmPixIn06dOHjz/+WIFeRESkBhKjQysXomugrE8BDODlK06jXkRwieA8skdj+reKIyUju+xAbSuAT6+Dg5sgMtExPWXwv9UJJz/+cMHfXP/9REwmO4XWruQfHIIJwOSYHOdkNsMgO99On+b1nPb8XcGjAr3dbufOO++kX79+dOjQocxt8vLyyMvLK/4+MzPTVc3zaFdddRXPPPMMF110EY8++iiNGjVi165dzJkzh/vuu49GjRpxxx13MG3aNFq2bEmbNm14/vnnS80hf6Lk5GSuu+46brjhhuJBsbt27SI9PZ0RI0bQpEkTTCYT3377Leeddx6hoaFERESU2EfLli256KKLGDNmDG+88QaRkZFMmDCBhg0bctFFF1Xque3cuZM333yTCy+8kAYNGrBlyxa2bdvGtddeW5NTJiIiIlQiRNdQeZ8CdEuuU+axEqNDy26DYcD/7oa/l0BgOFw5C6Iblvv43cd2M3bRWHIKc+iT2IeHLnyWvYcLij99WJ1yhP/MWuOVNfMn86hpK8eNG8f69euZNWtWudtMnTqV6Ojo4q+kpCQXttBzhYWF8dNPP9G4cWOGDx9O27ZtufHGG8nNzS3usb/77ru55ppruO666+jTpw+RkZFccsklFe53xowZXHbZZYwdO5Y2bdowZswYsrKyAGjYsCFTpkxhwoQJ1K9fn9tuu63Mfbz77rt069aNYcOG0adPHwzD4LvvvitVZlPRc9u8eTOXXnoprVq14qabbmLcuHHcfPPNVThDIiIiUp7E6FD6NK9XKyUmRZ8CWP6ZtKPanwIsewn+eB9MZrjsv9CgS7mbHs49zK0Lby1eOOqFgS/QuE5U8XNMjA5lWOcGzmmXBzAZ5Y2IdLHbbruNr776ip9++ommTZuWu11ZPfRJSUlYrdZSpSa5ubns3LmTpk2bEhISUmttF9+g14uIiEjtSbPmVP9TgHWfwxc3Ov4/dBr0vrXcTXMKc/i/7/+PtRlraRDegI/O+4i4sPLHE9aoXbUoMzOT6OjoMjPuydxecmMYBrfffjtz585l6dKlFYZ5cCy2FBwc7KLWiYiIiIgzlFtKcyo7f4Yv/wnwvcdWGOYL7YXc9+N9rM1YS3RwNDMGzagwzNeoXR7E7YF+3LhxfPzxx3z11VdERkayf/9+AKKjo106y4qIiIiIeJj0TTDrKrDlQ7uLYPAT5W5qGAaP/fYYS/csJdgSzMtnvUyz6Gblbu9L3F5DP2PGDKxWKwMGDCAxMbH4a/bs2e5umoiIiIi4S+Y++OgyyLNC4z5wyZtgLj+6vrTmJeZsm4PZZOap/k9xWvxpLmyse7m9h95DSvhFRERExFPkZsLMyyFzD9RrCVd8DIHlj2/7YMMHvL3ubQAm9ZnE2Y3PdlVLPYLbe+hdQRcNUhl6nYiIiHiAwnz49Bo4sB7C4+HqzyGs/PWJvtnxDc+segaAO7rewfCWw13VUo/h04G+aFrE7OxsN7dEvEHR66Sy02mKiIiIkxkGfPMf+HupY675qz6FOsnlbv7znp955NdHALi67dXc2OFG17TTw7i95KY2WSwWYmJiSE9PBxzzmZv+mWtUpIhhGGRnZ5Oenk5MTAwWi8XdTRIREfFPS56Avz4BkwVGvA8Nyq+D/zP9T8YvHU+hUcj5zc7n3h73+m3O8+lAD5CQkABQHOpFyhMTE1P8ehEREREXW/Uu/OQoneGC6dByULmbbj+ynXGLxpFry+X0hqfzWL/HMJt8uvCkQj4f6E0mE4mJicTHx1NQUODu5oiHCgwMVM+8iIiIu2yZD/8b7/j/mROg67Xlbpp2PI2bF95MZn4mneI68dyZzxFo9u9yWZ8P9EUsFosCm4iIiIin2bsaPr8eDDt0uRoGTCh30yO5R7h54c2kZ6fTLLoZr571KmGBYS5srGfy388mRERERMS9Dv8NM0dAQTY0P9tRalNOHXx2QTbjFo1jp3UnCeEJvDHoDWJCYlzaXE+lQC8iIiIirpd1yLFwVHYGJHRyDIK1lF06U2Ar4K6ld7EuYx0xwTG8MegNEsJrf9xbmjWHZTsySLPm1PqxasJvSm5ERERExEMU5MAnV8DhHRDdGK76DIIjy9zUbth58NcHWbZvGaEBobx29ms0i25W602cvTKViXPWYTfAbIKpwzsyskfjWj9udaiHXkRERERcx1YIn98Ae36HkBjHwlGRZfe2G4bBU78/xbyd8wgwBTB9wHQ6xnWs9SamWXOKwzyA3YAH5qz32J56BXoRERERcQ3DgG/vgC3fQUAIXPkJxLUud/O31r3Fx5s/BuCJ05+gb8O+Lmnmzoys4jBfxGYYpGR45mKlCvQiIiIi4hqLHoU1H4HJDJe9A03KD+ifbP6El9e8DMCEnhM4r9l5rmolTWPDMZ80NtdiMpEc65kz6ijQi4iIiEjt+20G/PK84//DpkOb88vd9Jsd3/DkiicBuLnTzVzV9ioXNPBfidGhTB3eEcs/M+5YTCaeHN6BxOhQl7ajsjQoVkRERERq17rPYf4/88uf9TB0u67cTRelLuLhXx8G4Kq2VzGuyzhXtLCUkT0a079VHCkZ2STHhnlsmAcFehERERGpTdsXwdxbHP/veTOccXe5my7ft5x7f7wXm2HjouYXcV+P+zCVMy+9KyRGh3p0kC+ikhsRERERqR17V8Psa8BeAB0uhaHTyl046s/0P7ljyR0U2AsY1GQQk/tOxmxSVK0MnSURERERcb6MbTDzcijIgmYD4OLXwVx29NxyeAtjF40lpzCHvg36Mu2MaQSYVUhSWQr0IiIiIuJcmWnw4XDIPgQNToORH0FAUJmbplhTuGnBTRzLP8Zp8afxwoAXCLKUva2UTYFeRERERJwn5yh8dClYU6FucxhV/iqwacfTGLNgDIdzD9OmbhteOfsVwgI9c2pIT6ZALyIiIiLOUZADn1wB6Rsgoj5cMwci4srcNCMng5sW3MT+rP0kRyXz+jmvExUU5eIG+wYFehERERGpOVshfH4DpC6H4Gi4eg7USS5z08z8TG5ZcAspmSkkhify1uC3qBdaz7Xt9SEK9CIiIiJSM4YB394JW74DSzBc+QkkdChz0+yCbMYtHMeWI1uoF1KPtwa/RUJ4gmvb62MU6EVERESkZhY/Bms+BJMZLnsHkvuVuVm+LZ87l9zJnwf/JDIokjcGvUGTqCYubqzvUaAXERERkepb9gr8/Jzj/8NegLbDytys0F7IfT/dx/K05YQGhDLjnBm0rtvahQ31XQr0IiIiIh4kzZrDsh0ZpFlz3N2UU/vjA/jhQcf/z3oIuo0uczO7YeeRXx9hUeoiAs2BvHTWS3SO61ztw3rVOXIBzdgvIiIi4iFmr0xl4px12A0wm2Dq8I6M7NHY3c0q24Yv4Zs7HP/vezuccU+Zm9kNO48uf5Rv/v4Gi8nCs2c+S+/E3tU+rFedIxdRD72IiIiIB0iz5hQHVQC7AQ/MWe+ZvdDbF8IX/weGHbpeC4MeA5Op1GaGYTDt92l8se0LzCYz086YxlmNz6r2Yb3qHLmQAr2IiIiIB9iZkVUcVIvYDIOUjGz3NKg8u5bDrKvBXgDtL4Fh08sN88+vfp5PNn+CCROP9XuMoU2H1ujQXnOOXEyBXkRERMQDNI0Nx3xSLraYTCTHunbl1Arr09P+go9HQGEOtBgEl7wJZkuZ+3n1z1d5b8N7ADzc52EubH5hjdvmKefI0yjQi4iIiHiAxOhQpg7viOWf3m6LycSTwzuQGB3qsjbMXplKv2mLGfXWCvpNW8zslan/3pmxDT4cDnmZ0LgvjPgAAoLK3M9ba9/ijbVvADCh5wQub3W5U9rnCefIE5kMwzBOvZnnyszMJDo6GqvVSlSUlgsWERER75ZmzSElI5vk2LBSQTXNmsPOjCyaxoaXuK+826t63H7TFpcoabGYTPwyYSCJRga8MxQy90BiZ7juGwiJLnM/7294n2dXPQvA3d3uZnSH0dVqz6naWt458hVVybia5UZERETEgyRGh5YZUsub3cVZs76UV5++d/cuEpdc5Qjzsa3g6jnlhvlPNn9SHObHdRlXK2Eeyj9H/kolNyIiIiIerrzZXf7afcRps76UVZ8eY8qm8483wKHtEN0YrvkSwmPLfPycbXN4csWTAIzpOIabO91c5TZI9SjQi4iIiHi48nrPV6YccdqsLyfXp0eY8liY8AqBBzdAeDxc+yVENyzzsd/+/S2Tl00G4Np213L7abdjKmPmG6kdKrkRERER8XBFvecn17f3SK5T5u3VnfVlZI/G9G8Vx64DRzjt11sJ3vWno7zmmrlQr3mZj/k+5Xse/OVBDAxGth7JPd3vcVqYd8bYAH+gHnoRERGRGqpwqkcnKG92l85JdZw+60tiRCC919xP8K6lEBgGV30OCR3K3HZJ6hIm/DQBu2FneMvhPNDrAaeF+Qpn3JESNMuNiIiISA04a1BqZZQ3u4vTZn2x2+Hr2+HPj8ASBKNmQ/OyV3b9de+v3L74dgrsBZzf7Hye6PcElnLmpK+qCmfc8ZOees1yIyIiIuIC5Q1W7d8qrlaCZ3mzuzhl1hfDgO/ucYR5kxkue6fcMP9b2m/cseQOCuwFDGoyiMf7Pe60MA8VrwjrL4G+KlRyIyIiIlJNFQVPr2IY8MNDsOq/gAkueQPaXlDmpivSVnD7otvJs+UxIGkAT/V/igCzc/uItSJs1SjQi4iIiFSTzwTPxY/D8lcc/7/wJeg0oszNVqSt4LZFt5Fry+XMRmfy3JnPEWgOdHpztCJs1ajkRkRERKSaioLnA3PWYzMM7wyePz4DPzsWg+K8Z6HrtWVudmKY79+oP88PeJ4gS1CtNatoxh1fXxHWGRToRURERGrAq4PnspdhyeOO/w9+HHqOKXOz39N+LxHmXxjwQq2G+SJaEbZyFOhFREREasgrg+fvbznq5gEGPgR9by9zs5X7VzJu0Thybbmc0fAMl4V5qTzV0IuIiIj4m9XvO2a0ATjjbjjz3jI3OzHMn97wdF4YqDDviRToRURERPzJX7Phmzsc/+89Ds56uMzNisJ8TmEOpzc8nekDpxNsCXZhQ6WyFOhFRERE/MWGL+HLWwADut8IQ56AMlZ2PTHM92vYT2HewynQi4iIiPiDLfPgixvBsEOXqx0z2pQR5lftX1UizL848EWFeQ+nQC8iIiLi67YvhE+vBXshdLzcMde8uXQMXLV/FWMXjXWE+QYK895CgV5ERETEl+38GWZdBbZ8x+qvF78OZkupzVYfWF0c5vs26MuLZynMewsFehERERFflfobfDwSCnOh5RC49B2wlJ61fPWB1dy68NZ/w7x65r2KAr2IiIiIL0pdAR9dCgVZ0GwAjPgAAkpPObly/8riMN8nsQ8vDnyRkIAQ17dXqk2BXkRERMTXFIX5/OOQfAZc8QkElg7py/YtY+zCscVh/qWzXlKY90IK9CIiIiJukGbNYdmODNKsOc7d8e7f/wnzxxxhftSnEBRWarOf9vzE7YtuJ9eWS/9G/Xn57JedHuZr7TlKCaWLqERERESkVs1emcrEOeuwG2A2wdThHRnZo3HNd7z7d/hw+AlhfnaZYX5R6iLu+fEeCu2FnN34bJ7p/wyBlsCaH/8EtfYcpRSP6KF/9dVXSU5OJiQkhF69evH777+7u0kiIiIitSLNmlMcdAHsBjwwZ33Ne7F3rywjzIeX2mx+ynzuXno3hfZChiYP5ZkznR/ma+05SpncHuhnz57N+PHjmTRpEn/88QedO3dmyJAhpKenu7tpIiIiIk63MyOrOOgWsRkGKRnZ1d/pnlXw0anD/Dc7vuH+n+7HZti4oNkFTD1jKoHmyof5ypbQ1MpzlHK5PdA///zzjBkzhuuvv5527drx+uuvExYWxjvvvOPupomIiIg4XdPYcMwnLdBqMZlIji1dGlMpe1bBh5dAXiY0Ob3cMD9321we/OVB7Iad4S2H81i/xwgwV776evbKVPpNW8yot1bQb9piZq9MLXdbpz9HqZBbA31+fj6rV6/mnHPOKb7NbDZzzjnnsHz58jIfk5eXR2ZmZokvEREREW+RGB3K1OEdsZgciddiMvHk8A4kRodWfWclwnw/uOrTMsP8p1s+5ZFlj2BgMLL1SCb1mYSljMWlylPVEhqnPkc5JbcOis3IyMBms1G/fv0St9evX5/NmzeX+ZipU6cyZcoUVzRPREREpFaM7NGY/q3iSMnIJjk2rJphfvVJYf6zMsP8Rxs/4qmVTwFwddurua/HfZhMplLblSXNmsPOjCwOZ+WXW0JTXtud8hylUrxulpuJEycyfvz44u8zMzNJSkpyY4tEREREqi4xOrT6IXfvCWG+cd9/pqYsHebfWf8OL6x+AYAbO9zIHV3vqHSYP3GWGhOOrxMzfWVKaGr0HKXS3BroY2NjsVgsHDhwoMTtBw4cICEhoczHBAcHExyspYhFRETET+1dDR9cAnlWR5i/6jMIjii12et/vc6rf74KwK2db+XWzrdWqWf+xBIbA0egN5sc5TYnl9AU9eQ3jQ1XgHcDtwb6oKAgunXrxqJFi7j44osBsNvtLFq0iNtuu82dTRMRERHxPCXCfJ8yw7xhGLy85mXeWvcWAP857T+M6TSmSocpa5YaA3j5itOoFxFcooRG8827n9tLbsaPH891111H9+7d6dmzJ9OnTycrK4vrr7/e3U0TERERF1Iv7ynsWQ0fnRjmPy8zzD+36jne3/g+APd0v4fr2l9X5UMVzVJzYqi3mEx0S65T4mdT3mDZ/q3i9DN0IbcH+pEjR3Lw4EEeeeQR9u/fT5cuXZg/f36pgbIiIiLiu9TLewqpv8FHlznmmU/qXWbPvM1u47HfHuOLbV8AMLHnREa1HVWtwxXNUvPAnPXYDKPcWWoqmm9egd51TIZhGKfezHNlZmYSHR2N1WolKirK3c0RERGRKkqz5tBv2uJSvcG/TBioUAiQ8gvMHAEFWY5Fo66cVSrMF9gLeODnB5ifMh+zyczkPpO5pOUlNT50mjWnwllq9LOrPVXJuG5fWEpERET8m1YVrcCOJY6e+YIsaDbQMZvNSWE+tzCXO5fcyfyU+QSYA3im/zNOCfPg6Knv07xeueFc8817BreX3IiIiIh/K69e2+9XFd22AGZdBbY8aDkYRnwIgSElNjmef5zbF9/OqgOrCLGE8MLAFzi94ekubabmm3c/9dCLiIiIW6mXtwybv4NZoxxhvvX5MPKjUmH+aO5RxvwwhlUHVhERGMHrg153eZgvcqqefKld6qEXERERt1Mv7wk2fAlf3Aj2Qmh3EVz6X7AEltjkYPZBblpwE9uPbicmOIbXB71O+3rt3dNecTsFehEREfEIWlUUWPsZzL0ZDBt0vBwufp204wXszLAWT+e559gexvwwhj3H9xAfGs+bg9+keUxzd7dc3EiBXkRERMQT/PkxfDkWMKDLVXDhy8xevbfEdJ53nx/NnLRJpGen0yiiEW8NfotGkY3c3XJxMwV6EREREXdb9S58exdgQNfrYNh00o7llVi0ieC9zNjyGKaALJpHN+fNwW8SHxbvzla7lBYeK58CvYiIiIg7LX8Vvn/A8f8eY+Dcp8FsLjGdpyU0hdCkdzFZ8mgS0Zp3h75FnZA67muzi2nhsYpplhsRERERdzAMWPrUv2G+73/gvGfA7IhnRdN5WsK3Etr4v5gsediymzL9zBl+FebTrDklPqmwG/DAnPWkWXPc2zAPokAvIiIi4mqGAQsehqVPOr4f+BAMehT+mboTHIOErzn7KKFJ72MyF2A73oYHuj5Hi9g4NzXaPbTw2Kmp5EZERETElex2+O5uWPWO4/shU6HP2FKbfbL5E+bufQqTyaB77ECmXPg4jetEubix7qeFx05NPfQiIiIirmIrhC9v/SfMm+CCl0qFecMwePXPV3lyxZMYGIxsPZK3z33BL8M8aOGxylAPvYiIiIgrFOY5Foza9A2YLDD8Teh4WYlNbHYbT6x4gs+2fgbA2M5juaXzLZhOKMXxR1p4rGIK9CIiIiK1LT8bZl8NOxaBJQgufx/anFdikzxbHhN/nsiCXQswYeKh3g8xovUINzXY82jhsfIp0IuIiIjUptxM+HgkpC6DwDC44mNoPrDEJsfzj/OfJf9h5f6VBJoDmXbGNAYnD3ZTg8XbKNCLiIiIOFnRIkjNwvNI+Ppq2PcHBEfBVZ9B494lts3IyeDWhbey+fBmwgPDeXHgi/RK7OWmlleeFnryHAr0IiIiIk5UtAhSvHGID4OmkWDeC6F14Zq50KBLiW13Z+7mpgU3sef4HuqG1GXGOTNoV6+dexpeBVroybNolhsRERERJylaBKkJaXwePIWW5r2kGXU5eHnpML/p0CaumXcNe47voWFEQz4890OvCPNa6MnzKNCLiIiIOMnOjCzaksJnQVNoZMrgb3sCl+VNYruRVGK7lftXcv3313Mo9xCt67Tmw3M/pHGUd/Rwa6Enz6OSGxEREREnaZ27lllBjxFpymG9PZnr8u/nqCmmxCJIC3ct5L6f7qPAXkC3+t14+ayXiQyKdGOrq0YLPXke9dCLiIiIOMOWedSbeyWRphxW2NtwZf5DHDXFlFgE6bOtn3H3j3dTYC/grKSzeGPQG9UO82nWHJbtyHB5qYsWevI86qEXERERqam/ZjtWgDVs0OpcGg95jTePGsWLIBmGwSt/vsKba98E4NKWl/JQ74cIMFcvirl7UKoWevIsCvQiIiIiNfHb6zD/fsf/O10BF71CoiWQxHqOmwpsBUxePpmvd3wNwM2dbmZcl3HVXv21vEGp/VvFuTRYa6Enz6FALyIiIlIdhgFLp8KPTzm+7z0WBj8B5n8rmo/lH2P80vH8lvYbFpOFh3s/zKWtLq3RYSsalKqA7Z8U6EVERESqym6DeffDyrcc3w98CPrfAyf0uh/IOsDYRWPZemQroQGhPHfmc5zR6IwaH1qDUuVkGhQrIiIiUhUFufD59f+EeROc9yyceW+JML/1yFau+u4qth7ZSr2Qerw79F2nhHnQoFQpTT30IiIiIpWVcxRmXQW7fgFLEFzyBnQYXmKTFWkruHPJnRwvOE7T6KbMOGcGDSMaOrUZGpQqJ1IPvYiIiEhlZKbBu+c5wnxQJIcu+Zhlof1LTBv57d/fcsvCWzhecJyu8V358NwPMdvqnnJ6yepMQZkYHUqf5vWqFObdNdWl1C710IuIiIicysGt8NFwsO6GiPp83+UVbp2Zj91YgdkET17SgeOhC3jxjxcBGJI8hCdOf4Iv/zhwyuklXTUFpbunupTaYzIMwzj1Zp4rMzOT6OhorFYrUVFR7m6OiIiI+JrdK+HjEZBzGOo2J/3ij+k9Y8cJg1JthCR8Q2Cd3wC4rt11jO8+ngOZefSbtrjU4NVfJgws7lVPs+acchtncNVxxHmqknFVciMiIiJSnq3fw/sXOMJ8w24cuPwrvtoV+G8wNuUT2uhDAuv8hgkTE3pO4J4e92A2mSucXrJIZbZxBlcdR9xDJTciIiIiZVnzEXz9H8fqry0G8Xnzx7nvpbXFwdhkOU5o0ntYQvdg2AN4pPcTXN72vOKHV2Z6SVdNQampLn2beuhFRERETmQY8NOz8NU4R5jvPIq0897hvq//LbMxBx0gLPk1R5gvDOP6ZtNKhHmo3PSSrpqCUlNd+jbV0IuIiIgUsRXCvPtg1X8d359+F2nd7+PbdWk88b/NAFjCthHaaCYmSy7RAQm8MOBlejRsU+4u06w5p5xesjLbOIOrjiM1V5WMq0AvIiIiAuzPyCBo7hjq7l0MmGDoNGZbziueGQYgMOZ3ghO+xGSyY8tOZs6lb9ImPtGt7RbfpEGxIiIiIlXw1c9/kP7SIOruXUyuEcgvXZ8nre11J4R5O8Hx3xGSOAeTyU6h9TQe7DpdYV48ggbFioiI+Jk0aw47M7JoGhuusgsgfedaui4cQZL5IIeNCP4v/x7+Wp7Ai8lHHGHelE9Iw1kERm4EoHedK5ly0Z00iNGAUvEMCvQiIiJ+RIsLnWTXMup8cgWBJisp9vqMLriPFCMRMMAAS2AmwQ3fxxK6F8NuIX//5Tx68V0efSGkCzb/o0AvIiJ+x18DT5o1p0Q9uN2AB+asp3+rOL86D8XWfwFzbyHQls8aewtuzL+HwzhqlS0mEzF1DlK/zZscK8zAXhhO/t5refzcYR59rnTB5p8U6EVExK/4c+CpaHEhTw6pTmcYsOxlWPCw4/s2w9iR/AjWr3aAYWAxmbhhUA53/3wT2YXZJEUkc3PrJ+iZ1NKjz5Mu2PyXAr2IiPgNfw88WlwI0o4ch/kTSNzyoeOGnjfD0KlcZrbQr21jdh7M4s/Mb3hj/XTshp1eCb14bsBzRAdHu7fhlaALNv+lWW5ERMRvVBR4/IEnLi6UZs1h2Y4M0qw5tX6sL5ZvZsPzFxSH+TVt74FznwKzBYDYiAB+SH+FGeuex27YGd5yODPOmeEVYR7+vWA7kb9dsPkr9dCLiIjfUA81jOzRmP6t4jxicSFXlj8d2LODNvNG0N6yizwjkLsKbuX7P7vxy9BcEqNDOZJ7hLuW3sXqA6sxm8zc3e1urml3DSaT6dQ79xBFF2wPzFmP7Z/SIXdfsIlrKNCLiIjfUOBxSIwOdftzdmn5074/iZk5gvrmA2QYUdyUP54/jFaA49OZ4/Y93L74dvYe30tEYARP93+aMxqd4dw2uIgnXbCJ6yjQi4iIX1Hg8Qy1Xe9dNJNRG+vP1J03luCCbLbaG3JDwb3sMeIBx6cz+wtXc+d3DzsGv0Ym8cpZr9AsplmNj+9Olb1g89fZnnyRAr2IiPgdT+ih9ne1Wf7kKOVZy/Xm7+gd8DGYDGg2kPUtniDtm12AgcUEF/TfwqTf3sXAoGdCT5478zliQmJqfHxPU1Zw9+fZnnyRAr2IiIi4XG2VP6VZc3h4zhoes7zPVQGLAPjYdjYDh73H8LpR9GnfjG3pR/ly94ss2P0/AEa0GsGEXhMINAfW+HlVpn2u7BUvK7j3bxXn17M9+SIFehEREXGL2ih/St23n7cDnqG/ZR12w8QThVfxX9u5fHKkgMS6EBSUzVvb7uPPg39iMVm4v+f9XNH6CpcMfnV1r3h54xRevLKLprf0MQr0IiIi4jZOLX86souuC0YQaNlKthHMfwpuY6G9W3Epz4ZDG7hzyZ3sz9pPZFAkz575LH0b9HXOsU/BHWsglDdOgX8uKPx5tidfo3noRURExPulroC3zybw8Fayg+O5omBScZh/cngHVh9axHXzrmN/1n6So5L5+LyPXRbmwT1rIJQ3L3235Doetx6B1Ix66EVERATw4llP1syEb+8EWz4kdCRs1Ke8YdQhJSObpLpBzNr+Bu///D4A/Rv1Z9oZ04gMinRpE92xBkJF4xQ025NvMRmGYZx6M8+VmZlJdHQ0VquVqKgodzdHRETEK3nlrCd2Gyx4BJa/4vi+7YVwyesQFA6ANc/KvT/ey/K05QCM6TiGcV3GYflnZVh3DFA9OVy74hynWXMU3L1QVTKuAr2IiIifS7Pm0G/a4lK9x79MGOi5ATDXCl/8H2z7wfH9mffDmRPA7Kgm3npkK3csvoM9x/cQGhDK4/0eZ3Dy4OKHu+sCRuFaKqsqGVclNyIiIn6uthd5crpDO+CTKyFjCwSEwMWvQYdLi+9esGsBD/7yIDmFOTSMaMiLA1+kdd3Wxfe7Y4BqEa2BILXBbYNiU1JSuPHGG2natCmhoaE0b96cSZMmkZ+f764miYiI+KXyBk965KwnO3+Ct892hPnIBnDD/OIwbzfsvPTHS4xfOp6cwhx6JfZi1vmzSoR5cM8AVZHa5LYe+s2bN2O323njjTdo0aIF69evZ8yYMWRlZfHss8+6q1kiIiJ+p7YWeXK6lf+FefeBvRAadoMrPobIBAAy8zN54OcH+HHPjwBc0+4axncbT4C5dNRxxwBVkdrkUTX0zzzzDDNmzODvv/+u9GNUQy8iIuIcHlvfbSuA+RNg5duO7zuOgAtfgkBHG7cd2cadS+4k9VgqQeYgJvWdxIXNLyx+eFmDX901QFWksry2ht5qtVK3bt0Kt8nLyyMvL6/4+8zMzNpuloiIiF/wyPrurEPw+WhHqQ0mOPsROP0u+GcO9e/+/o7JyyeTU5hDYngiLwx4gfax7YsfXt7gV03bKL7EYxaW2r59Oy+//DI333xzhdtNnTqV6Ojo4q+kpCQXtVBEREROlmbNYdmODNKsObWw87/gzQGOMB8UAVfMhDPGg8lEgb2Ap35/ivt/vp+cwhx6J/Zm9rDZJcJ8eYNfi9qaGB1Kn+b1FObF6zk90E+YMAGTyVTh1+bNm0s8Zu/evQwdOpTLL7+cMWPGVLj/iRMnYrVai792797t7KcgIiIilTB7ZSr9pi1m1Fsr6DdtMbNXpjpv5+s+h/8OAWsq1G0G/7cQ2pwPwMHsg/zf9//HR5s+AuD/Ov4fr5/zOnVC6pTYhQa/ir9wesnN3XffzejRoyvcplmzZsX/37dvHwMHDqRv3768+eabp9x/cHAwwcHBNW2miIiI1ECtTf1oK4RFk2HZy47vWwyCS9+CUEdYX5O+hruX3s3BnINEBEbw+OmPc3bjs8vclQa/ir9weqCPi4sjLi6uUtvu3buXgQMH0q1bN959913MZo+pABIREZEK1GTu+nJXaM0+DJ/fAH8vcXx/+ng46yEwWzAMg483f8yzK5+l0CikeXRzpg+cTnJ0crnH8ZrZe0RqyG2DYvfu3cuAAQNo0qQJzz77LAcPHiy+LyEhwV3NEhERqZJyw6mPq27vd7krtO5fD7NGwdFdEBjmWCyq/SUAZBdkM2X5FL7b+R0AQ5OHMqXvFMICT93TrsGv4g/cFugXLFjA9u3b2b59O40aNSpxnwfNpCkiIlKucsOpH6hO73d5ZTqDjWXUWXAXFGRDTBO48hOo7xjcutO6k/FLx7P96HYsJgvju43nmnbXYDKZyj1OWW2tapD31ws18U4eNQ99dWgeehERcYc0aw79pi0u1UP9y4SBfhUAqzJ3/bIdGYx6a0Xx92bs3BPwKWMDvnbc0GwgXPYOhDmmsJ6/cz6Tlk0iuzCbeiH1eObMZ+iR0KPWnksRf75QE8/htfPQi4iIeIua1JD7kqr0fp9YplOHTF4MfJX+lnWOO/v+B86eBJYA8m35PL3yaWZvmQ1A9/rdebr/08SFVW6MXk3U2mBfkVqkQC8iIlINmkGl6orKdGbN/ZJXAqfT0HSIQksoARe/Ah0vA2Dv8b3cvfRuNhzaAMCYjmMY22UsAWbXRBZdqIk30rQyIiIi1VAUTi3/1HL72gwqtbJglGEwkgXMCX3MEebrNCfgpsXFYX7p7qVc/s3lbDi0gejgaF49+1X+0/U/FYZ5Z7ez6ELtRLpQE0+nHnoREZFq8tUZVGqlhrwgB74dD399jAmgzTACLp4BIVEU2gt5ac1LvLv+XQA6xXbimTOfoUFEA5e3U1NdijfSoFgREREP5Y6ZVmplsO/hnfDpNbB/HZjMcM5kR828yUR6djr3/ngvf6T/AcBVba/i7m53E2gJdH07T9q/r12oiXfRoFgRES+nKfPEXTOtOL2GfMt8mHsT5FohPM4xi03T/gAs27uMib9M5HDuYcIDw5nSdwpDkoe4p50nqc5UlyLuokAvIuJhNGWeuHOmFWcN9k07cpzCxU+StO5Vxw2NesKI9yGqAQX2Al5d8yr/Xf9fAFrVacVzZz5X4aqvtdVOEV+gQbEiIh6kvCDn1IGJ4vEq6n2ubc4Y7PvlL2vY/sLQ4jC/rcmVMPp/ENWAfcf3cf3864vD/MjWI5l53swqhXlntVPEV6iHXkTEg2jKPAH39z7XZLDvoQ2L6b3gBhLMR8g2gplYcCPfbj2DX7JsbDy6iIeXPcyx/GNEBkYyue9kBicPrlLbTixH89VBySJVpUAvIuJB3B3kxDN4wkwrVa4ht9vh1xeou/hxTCY72+wNubXgDrYbjcCUz5MrnmRp2peAYxabp/o/RaPIRlVqU3nlaAry4u80y42IiIeZvTK1VJBTDb1/8pqZVrIOwdybYfsCAObYTufBghvIIQRT0EHCGn6COWQfANe3v57bu95OoLniWWxOVtuz2oh4Gs1yIyLixVRGIEW8YqaV1N/g8xsgcy8EhMB5z1BQeCb5czcQELmakMQvMZnzqRNchydOf4IzGp1R5UOkWXP4du0+laOJlEOBXkTEA3lFkBP/ZrfD8pdh4RQwbFCvBVz+PiR0YFhBFr8d/5GFu78DoGdCT6aeMZX4sPgqH+bEMpuTqRxNxEGBXkREpIp8aZ2Aaj2X7MPw5a2wdb7j+w6XwQXTITiStQfXcv9P97Pn+B7MJjO3dr6VMR3HYDFbqtW2isK8ZrURcVCgFxERqQJfWiegWs8ldQV8cSNYd4MlGM6dBt2ux2bYefuvN5jx1wxsho0G4Q2YesZUutbvWq22lVdmA/Dw+W05r1OiwrzIPxToRUREKsmdCz45W5Wfi90GvzwPS6Y6SmzqNoPL34PEzqQdT2PCzxP4I/0PAM5tei4P9X6IqKDqTVZxqjIbhXmRkhToRUREKsmX1gmo0nPJTIM5YyDlZ8f3HS+H85+HkCjm75zPo8sf5VjBMcIDw3mw14MMazYM0z8LPlWVymxEqk6BXkREpJJ8aZ2ASj+XLfMd9fI5hyEwDM57FrqMIqswmyd/eZCvd3wNQKe4Tkw7YxpJkUk1aldZFxqgMhuRipjd3QARERFvUbTgk+Wf3mdv7jE+5XMpzIN5E+CTkY4wn9ARbv4JTruKtRnruPyby/l6x9eYTWZu7nQz7w19jwB7LMt2ZJBmzal2u4ouNE6kMhuRimlhKRERkSrymgWfKqHM55KxHT6/HvavdXzf61YYNIUCs5m3173NG3+9gc2wkRieyLQzptG1flenDhbW4moiVcu4CvQiIiLiYBjw58fw3b1QkAWhdeHiGdB6KDutO3ng5wdYf2g9AOcmn8tDfRwDX2tjFVdfumgSqQ6tFCsiIiJVk3MUvrsH1n3m+D75DBj+JkZkIrM2f8Lzq54n15ZLZFAkD/V6iPOanVf80NoYLKzF1UQqT4FeRETE36X8AnNvccwtb7LAwIlw+ngO5GTwyMJbWLZvGQB9EvvwaL9HSQhPKPFwXxosLOKNFOhFRET8VWE+LH0SfpkOGFAnGYa/BUk9mbdzHo//9jiZ+ZkEW4K5q9tdXNnmSsym0vNpFA2wPbnuXT3sIq6hQC8iIuKPDm6FOf8HaX85vj/tahg6DSt2nvjxPualzAOgfb32PHnGkzSLblbh7kb2aEz/VnGqexdxAwV6ERERH5FmzWFnRhZNY8PLD9SGASvfhh8ehsIcCK0DF7wE7S7k172/8sivj5Cek47FZOGmTjcxptMYAs2BlTq+6t5F3EOBXkRExAdUatrI4+nw1TjY9oPj+2YD4eIZHA+J5Nllk/li2xcAJEcl8+TpT9IxrmOZx6rUhYOIuIwCvYiIiBeoKESnWXOKwzw4Bqc+MGc9/VvF/bvtlnnw1W2QnQGWYBj0KPS8iWX7f2PSD5PYn7UfgFFtRnFntzsJDSh9jJ0ZWazba+WpeZudMt+8iDiHAr2IiIiHO1Xve4XTRgYXwA8Pwh8fOO6o3wGGv0VW3SY8u+IxPt/6OQCNIhrxaL9H6ZHQo8Ljn6jMCwcRcbnSQ9WlVqVZc2q8LLaIiPiP8nrfT/w7UjRt5IksJhOtcv6AGf3+CfMm6HMbjFnMcpuVS766pDjMX9nmSr648Isyw/zJxz9Z0YWDiLiPeuhdyJnLYouIiH+ozKJNJ08bGW7KZ27rBdT7fKbjATFN4OLXyGp4Gs+vfJpPt34KQMOIhjzW77Eyg3xFxz+R5psXcT8FehepVH2jiIjISSq7aFPRtJEZm36l7YqHCUjZ4bij2/Uw+DFWHN7II18NZ1/WPgCuaH0Fd3W7i7DAisN4Wcc/sR2ab17E/RToXaQ2lsUWERHfV+lFmwrzSFz5NIm/TgfDDpEN4MKXOdakF8+vfr64vKZhREMe7fsoPRN7Vvv49w1tTadGMZpvXsRDKNC7iJbFFhGR6jp50SaAZTsy/p3xZv86mHsLHFjveECnK+DcaSw+uIYnvryY9Jx0x35aj2R8t/Gn7JU/1fEV4kU8iwK9i2hZbBERqYmiRZtOHI8VZCrki46/03H7G2AvgLBYGPYCGU37MnXF4/ywyzHffJOoJkzqM6nCWvnKHl9EPI8CvQuph0NERGrixPFY7U0pPB34Bu237nLc2WYYxvkv8NWBZTzz5UVk5mdiMVkY3X40t3S+hZCAEPc2XkRqjQK9i6mHQ0REqmtnRhYBRgG3B8zlVsvXBJjsHDEiyDj9UcJ6D2HK8odYnrYcgLZ12zKl7xTa1mvr5laLSG1ToBcREberaBVUd+7L07Qq2MK3QQ/QyrwXgG9tvZhSeC2j6+bz/tfDySnMIdgSzK2db+Xa9tcSaA50c4tFxBUU6EVExK2cuUaHz673kZ8NS54g9rfXiDXbyTCieKjgBhYGJdG8w1xeX78VgG71uzG5z2SSo5Pd214RcSmtFCsiIm5TmVVQ3bEvj5LyK7zeD5a/4piOstMVHBuzhLCzAghv9gr7crcSHhjOw70f5p0h7yjMi/gh9dCLiIjbOHONDk9f76PKpUB5x2DhFFj5luP7yAZwwXSWhgbzxG+3sj9rPwCDmwzm/p73Ex8WX4utFxFPpkAvIiJu48w1Ojx5vY8qlwJt/g6+uwcyHbXydL2W/af/h2l/vcqi1EWAY4GoB3o9QP9G/V3wDETEk6nkRkRESkiz5rBsR4ZLSlWK1uiwmEwANVqjw5n7cqYqlQJlpsHsa2DWlY4wH9OEwqs/58MWPblo3tUsSl1EgCmAGzrcwNyL5irMiwigHnoRETmBOwaVOnONDk9c76NSpUB2O6x+x1Fik5cJJgv0vZ0NHS5kyqqn2HR4EwCd4zrzSJ9HaFWnlYufhYh4MgV6EREByu9J7t8qrtaDsTPX6Kit9T6qOx3mKUuB0jfBN3fA7hWO7xt2wzr0CV7eu4jPFtyA3bATGRTJXd3u4tKWl2I26cN1ESlJgV5ERADPH1TqTjX55KKoFOiBOeuxGca/pUBhJlj8OPwyHewFEBSB/ayH+apuHC/8ej9H8o4AcF7T87i3x73EhsbW4jMUEW+mQC8iIoBnDyp1J2d8clGqFOjwSphxMRze4dig9Xls7HszT2x4i7Vb1gLQPLo5D/R6gJ6JPWvhWYmIL1GgFxHxEO5e4bTcnuQqtsXdz8PZnPXJRWJ0KIlmK/xwO6z7zHFjRALWwY/ycvY2Pl36HwwMwgLCGNtlLKPajtJKryJSKQr0IiIewFNWOK3poFJPeR7O5JRPLmyFjvnklzzpGPSKCXv3G/iqWTdeWPtqcXnNuU3P5Z7u9zhtTnlfu7gSkbKZDMMwTr2Z58rMzCQ6Ohqr1UpUVJS7myMiUmVp1hz6TVtcKjD+MmGgV4UwX3keZZm9MrXUJxeVvlDZ/Tt8Ox4OrHN837AbG864jSd3fMHajH/Lax7s/SA9Eno4tc2+dnEl4k+qknHVQy8i4ma+MhjVV55HWar1yUXWIVg4CdZ86Pg+JIaMM+/lJdsBvlwxuVbLa9w5Y5GIuJ4CvYiIm/nKYFRfeR7lqfR0mHY7rPkAFk6GHEcpTX7nUXzUpB1vbv6IrIIswDF7zd3d73Zaec2JfPniSkRK02S2IiJu5qkrnFaVrzyPGtn7B/x3kGNe+ZwjGPXbs/iCp7jY2MUL694kqyCL9vXa8+G5H/JU/6dqJczDvxdXJ/Kli6vyuHKVYxFP4hE19Hl5efTq1Yu//vqLNWvW0KVLl0o/VjX0IuIr0qw5HrXCaXX5yvOokuPpsGgKrJkJGBAUwba+t/B03i5+2+9YMCo2NJY7u97JBc0vcMniUDWq+/dCGjMgvsbraujvu+8+GjRowF9//eXupoiIuE1trXDqas58Hh4/S0thPvz+Bvz49D+z14C143BejW/Apzu/wGbYCDQHcm27axnTaQzhgeEua1pNZyzyJhozIP7O7YF+3rx5/PDDD3zxxRfMmzfP3c0REREP4fE9rtsWwvwJcGgbAAWJnZnVYTCvp84j8+9VAJzd+Gzu7n43SZFJbmniiRdX7rw4qu1ja8yA+Du3BvoDBw4wZswYvvzyS8LCfLuurzwe3/skIuIGHt3jemgHfP8AbJ0PgBEex4LuI5l+ZA27t88GoGWdlkzoMcFjVnl158WRK47t6wOyRU7FbYHeMAxGjx7NLbfcQvfu3UlJSanU4/Ly8sjLyyv+PjMzs5ZaWPs8vvdJRMRNPLLHNe8Y/PQs/PYa2PLBHMCfp43gWdNR/kr9GnDUyY/rMo6LW1xMgNntH4ID7r04ctWxnbXKsYi3cvpvmwkTJvDUU09VuM2mTZv44YcfOHbsGBMnTqzS/qdOncqUKVNq0kSXOFXPu0f3PomI1/G1T/s8qsfVbnPMJb/4CchKByC12RlMr9+ABft/ASA0IJTR7Uczuv1owgJd28ZT/ezdeXHkymP705gBkZM5PdDffffdjB49usJtmjVrxuLFi1m+fDnBwcEl7uvevTtXXXUV77//fpmPnThxIuPHjy/+PjMzk6Qk99QmlqcyPe8e2fskIl7JFz/t85ge1+0L4YeHIX0jAEfrNeWNVn2Ylf4bhft3YTaZuaTFJYztMrbWpqCsSGV+9u68OHL1sX1lYLlIVblt2srU1NQS5TL79u1jyJAhfP755/Tq1YtGjRpVaj+eNm1lZZc+9+Ul0kXEdXz9d4nbpsA8sBF+eAh2LAIgNySGjzsO5m3reo4VHAfg9IanM77beFrWaem6dp2gKj97d05h6W/TZ4o4i1dMW9m4cck3c0REBADNmzevdJj3RJXtefeY3icRD+JrZSOu4Ouf9rm8x/XYAVjyhKPExrBTYA7kyw5DeL1gH+kZvwHQuk5rxncfT98GfZ1++Kq8B6rys3dnOYpKYURqn2eM2PEhVfl4Ub/kpCz+Gmp9sWzEFTyq1tyb5WfD8lfhlxegIAs78EOrM3glqIBdx9YCkBieyLgu4xjWbBgWs8XpTajqe6CqP3t3lqOoFEakdnnESrE14WklN6CPF6X6/DXU+nrZSG3T75wasBXCX5/A0qmQuRcDWNaoIy/WiWLT8d0ARAbGcHWbG/m/zqMIsgTVSjOq+x7Qz17Ed3lFyY0vU8+7VIc/z3zkyrIRX/wERL9zqsEwYPO3sOgxyNgCwF91G/Niw2RWHk+B41aCzKEcP9CPfYfP4Nl1wdSz7a+1sFzd94B+9iICCvS1Rh8vSlX5ei10RVxVNuLLn4Dod04VpPwCCyfDnpUAbI2sxytN2rMkKwWOpxBoDuSCppfx4fzm2Asd47vs1O4Fdk3eA/rZi4jZ3Q0QEYeiP+gn8pda6KJB4haT4wTUxiDx8j4BSbPmOO0Y4uHS1sJHl8J758OelWwPjeTudn25NDacJVkpxVNQ/u+S/zEk8abiMF+k6AK7NrjiPSAivks99F7GF8sFapM3nS9/n/motksH/PkTEH9Q4Xv98N+ORaHWfw7A30EhvN60E/PzD2Dk7AFgcJPBjO0yluYxzR2PseW4fLCxymdEpLoU6F3AWaHSl8sFaoM3ni9//4Nem6UDmg3Gd5X7Xs/cBz8/B6vfA3shuwICeLFROxaZrNjz9wNwTuNzuKXzLbSu27rEPt11ga3yGRGpDs1yU8ucFSo1C0jVuPp8edMnAf5MM4L4nrLe6/VNVhb3/oPwtR+ALY/dARamJ7Tkh4AcMDk2bBXRi8cH3E3bem1Puf/auMDW7wwRORXNcuMhnDlricoFqsaV58sbPwnwV/7+CQj4XpA88b1el0xuDviGay0LCF2TT2pAAP9t0oavzLnYcNS+Fx5rQ17GOfyZl0TMOcmn3H9t9Jjrd4aIOJsCfS1yZqhUuUDVuOp8+fNUk97Kn0safDFINo0Np67pGDda/sdoy/eEm/LYERjAGwkt+T4gH3tRkD/eiryD52DPLXq+7ukQ0e8MEakNmuWmFjlz1hLNgFA1rjpfFV20iXiS2p7lJ82aw7IdGa6dNSjnKIl/vMCK8PGMC/ia1GA7N8Y15pJGDZgXkIcdgzMansELZ7xN3p4bTgjz7usQ0e8MEakN6qGvRc4eVKVygapxxfnSJyfiqU4uranNMjSX9/xnH4YVb8CKGZBrZWNwEK8nNuWXAFvxJuc0PocxncbQrl47AKYOLz1+wh2/Q/U7Q0RqgwbFukBtDaoSz6CBllLbqlr3XlbA7t8qrlYGirt0APrxg7D8FVj5Nkb+cVaFBPNGXCIrAuwAmE1mhiYP5f86/h8t67Qss62e8LtYvzNEpDKqknEV6EWcwFOCgvieqvZ+VxSwf9p60OlBctmODEa9taLU7Z+M6U2f5vVqtO9imftg2cuw6l1shTksCQvl3dgE1locPfIBpgCGNR/GjR1uJDk62TnHrGX6nSEip6JZblzI12aMkOrx54GWUnuqM4CyotKa2ihDq9USkiO74NfpsOYj8uz5fB0Rzvt1k9lltgM2gsxBXNziYm7oeAMNIxrW/HgupN8ZIuJMCvQ14IszRriLr14Y+erzEteoTt37qQK2s4NkrSzAlLEdfnke1s7Gio3PIiP5qE4DDpnsgJ3IoEiuaH0F5zQcztHjIZht4U57PiIi3kiBvpo09Zjz+OqFka8+L3Gd6vR+u2OFU6f1/O9Z5eiR3/Qt+y1mPoyJ5PPoaLIxADv1w+pzbbtrubTVpXz75yGGTf/LI99fupAXEVdToK8mLfTkHL56YeSrz0tcq7rh3B0zYlW7599uh20/wLKXYNevbAwK5KPYusyLDKcQAIMWMS24ocMNDG06lEBzoEe/v3QhLyLuoEBfTd469Zin9Rw568LIV5+XSHXDucfXaBfmw7rPYNlL2A5uZmlYKB8m1md1SHDxJj0SenB9++s5veHpmEz/Lurhqe8vT77QEBHfpkBfTe74WLumPLHnyBkXRr76vDyBp10o+SuPD+dVkZsJq9+D32Zw7HgacyMj+DipIXsDLIBjxprByYO5pt01dIjtUOYuPPX95akXGiLi+xToa8CbFnry1J6jml4Y+erz8gSeeKEkXuzILvj9TfjjA3bbspkZFcncxo3I/mc57ejgaEa0GsHI1iOpH16/wl156vurrAsNM3AoK480a47b2ycivkuBvoa8pefMk3uOanJh5KvPy9089UJJvIxhQOpy+O017Jv/x4rgQD6JjmRpWAzGPxU0zaKbcXW7qxnWbBihAZV/bXni++vkCw0TYAC3fbxGF8UiUqsU6P2Ep35EXaS6F0a++rzK4sryF0++UHIHlR5VUWEerJ8Dv71GZvo6voqI4NMG9UkJCize5PSGp3NN22vo06BPifr4qvDEDpWiC43VKUf4z6w1uigWEZdQoPcTnvoRdU356vM6mavKX4qCa3iQxaMvlFxJpUdVcDwdVr0DK//LpoIjzI6K5H9Jjcj9p6wmPDCcYc2GMarNKJrFNHNzY2tPYnQodSN0USwirqNA70c88SNqZ/DV51WkNspfyupxPjm4XnJaQ75cs8+nL5QqkmbNYVXKYZUenYphwN7VsPJt8tZ/wQ8hAcyKimRtSGLxJi1iWnBF6ysY1nwY4YH+sQiUp396KCK+RYHez3jiR9TO4KvPC5xf/lJWj3P/VnGlguuXa/YxZ2wfsvPtPnmhVJETz9HJ1Mv6j/wsWPc5rHybXRkb+SIygi8bxnHE8u9sNYOaDGJkm5F0je9aZlmNL5cyVeXTQ18+DyLiGgr0NWG3wd9LodkAMFvc3RrxUc7s6Suvt//FK7uUedGQnW+nT/N6NWi99zn5HJ3M73tZD26Blf8l969ZLAwo4IvICFYlNSi+u35YfS5vdTmXtrqU2NDYcnfjD6VMlfn00B/Og4jUPgX6mti+CD6+HKKT4LSrHV/RjdzdqlLU++PdnDlOoLzefv4JEyoPKPscFfHH0iPAsQjU5m9h1Tts2fsbX0RG8G39SI5ZzACYMdOvYT8ubXkpZyadSYC54j8t/jSLUkWfHvrTeRCR2qVAXxNZ6eSFxBBs3Q1Lp8KPT0GLQdDtOmg5BCzuP73q/fENzhonUF5vf7fkOn4xuLgyyptL/OVRp9G1SR3/OieHdsCaD8n662PmcZwvIiNY3+jf2vjE8EQuaXkJl7S4hITwhErvVrMoOeg8iIizuD9xerEtTXowunEig6PO5MKDezlt1yrM276Hbd9DRAKcdhV0vRbqJLulfer98S3VGSdw8qczFfX2+/rg4soq7xyd36nBqR/sC/KzYeNX2Nd8wMoDq/k6IpwFdcPIMTtKrwJMAQxsPJDLWl5Gr8ReWKpRbuiJA0bd8UmmJ54HEfFOJsMwyvlw2TtkZmYSHR2N1WolKirKpcee8ecMXvvrteLvG4bW5wJLDBek/EnjYwf/3bDZQEewb30eBIa4rH3LdmQw6q0VpW7/ZExvv6uL9kcVfTqTZs3x++B+Kn51jgwD9q2BNR+yc+Mcvgk2+CYinP0B//b5NI1K5tJWlzGs2TDqhdb898fslamlLprc9emhOz/J9KTzICKepSoZV4G+BuyGnT8O/MHXO77mh10/kFWQVXzfaRGNueB4DkN2/UFUUfdLSDR0uBQ6j4JG3aGai6lUVpo1h37TFpfq/fllwkDfDyh+Tj97z+VRY1qyD8PaT7Gu+YD52bv4OiKctSHBxXdHBkYwtOm5XNj8QjrHda72AlDl8YSLJk94r3jCeRARz1OVjKuSmxowm8x0T+hO94TuTOw1kSWpS/j6769Zvm85a46nsgaY1rQZA4MTuOBACn0P7SVw1TuOhVfqtYTOVzi+amkgrb8suiSlqTbXM3nEmJbCPNj6Pfl/fcIve37i2/AQloaFUhBWFwDLPwNcL2xxEQOSBhBsCT7FDqvPE6ab9YT3iiecBxHxbuqhrwXp2el89/d3fLXjK7Yf3V58e3RAGOeYIjh33za6H7fiqDw1QdP+0GUUtL0Agpy/6Ip6f/yPJ/Q6ehJP6BV368/EMCD1Nwr/+oTft3/LvCCDRWFhxbPUALSJackFLS7mvGbnVTjdpK/Re0VEPJVKbjyEYRhsObKFr7Z/xfyU+WTkZBTfFxsQzpACE0PTttM5Lx8TQFAEtBkGHS9zzG1vCXRX08UHqDbXwSN6xXHTmJZDO7D/+Ql/bZzNdxzjh/AwDlv+HcQaH1yXc5tfwAXNL6B13da104aTeMLF1cn0XhERT6RA74FsdhurDqxi3s55LNi1gMz8zOL7GljCGZqVzbmH9tE6v8AR7kPrQruLHOG+cV8wm8vdtzN44h9ZqfnPxd8/nfGk3leXtSVzH8b6uWzeOJt5WbuYHxFG2gmDW2MCwhnc7DzObXoeXet3xWyq3d8tJ6rpxVVt/p7y9/eKiHgeBXoPV2ArYHnacubtnMfi1MVkF2YX39fEEsbZmVYGHT1E+/x/eu4jG0CH4Y6vBl2dPpi2rD+y/VvFVfsPpy4OnMNTepZPxZN/3p4201Ot9QQfO4B9w5es2zibhcd2sDA8lD2B/37CF2EO5qwm53Bu82H0SuxFoNn1n/7V9ILGW94PIiLOokDvRXIKc/h5z8/M2zmPn/b8RL49v/i+BFMw5xzL5JzMo3TJy3PU3Ndp6pgpp92FkNCpxuG+rD+yJhy7rc4fTmdfHPgrT+pZrkh1Q5arLgI88Tw6rSc4K4PCDXNZs3E2CzK3syg8lPQTeuJDTAH0b9CX81oO5/RGp9fq4NbKqMnFlSf+HEVEaptmufEioQGhDE4ezODkwWQVZPHznp9ZsGsBP+/9mf2FOXwUEcxHEfWJKDQzKPs4Q7PT6PHzswT+/CzENHEMpG13ETTsXq2ynLJmeDBwjKGDqi1GVdZCVhO+WFftiwN/cnLA9YSZN06luguXubKn1RNneqrRjCbH0ynY9A0rNs5i4bEdLA4L4YjFAtGRAISbA+nfoB+DWlxIvwb9CAv0nAWKarKIkje8H0RE3EmB3oOEB4YztOlQhjYdSm5hLv/bvpQHf/iYgIhNHA/IZW5UGHOjwojEzOlZ2ZyZdZDTV7xK9PJXHCvTth0GbS+EJv3AUrkfbVl/ZE9WmT+cadYcvl27z2kXB5XhyaUeVVHepxqevoJkdUKWO1Yv9voVcA/t4MiGL/hl+zcszdnHsrAQjpvNEOmYESvaHMzAhmcwqNUl9E7sTZAlyM0NLltNLq5cvaKqr/xuERH/oUDvoUICQkgM7E7uPhtQiCX8bwIi1xMQuZFjAceZFx7CvPAQLECXvAIGHD9O/z/fo+nKtzGF1oU250Hr86HZmRVOhXnyH1kz/4TwE7Y51R/OEwPpqTirV81X6mnLC7i/TBjocT3LJ6tOyHJXT6tXzfNtGBj7/uTv9R+zNHUxP9oz+Ss4GLvFBBGOcxtrCeXsRgM4p/VwutfvToDZO36VV/fiypWftPjK7xYR8S/e8VfAT/0bmAKwZbXCltWKwgOX8MaYWNYf+Y0f9/zI9qPbWR0cyOrgOjxHHRoX2umfdZwBmz6j65qPCAwIccxz32oItBwCMUmljnPyH9mfth6s9B/OkwPpiapzcVAZtdHL664euYoCrqf3LFcnZLm6p9VrFOaRv/NHVm+czU/7V7DUUugY1BoEEAJA6+BYzmwyiAEtLqB9bHuXzk7jTNW9uHLF+8EdnyCJiDiDAr0HKzswdeScZo05h97c2e1O9hzbw497fuTH3T+y8sBKUgMK+Sg6io+iowgzoGd2Nn33L6PfzsU0/t/dUL8DtBrq+GrYFcyW4mMV/cGqyh/OsgIpwMPnt+W8TolVujioLGf38rqzR+5UAdfTe5arGrI8sabdXYwjqeza+Bm/7vyeZcd3sTI4gByzGUJMQCBBmOgZ2YwzW17ImU3PJTEi0d1Ndrvafj+oVl9EvJVmufEClZ0VI6sgi2X7lvHj7h/5ee/PHM49XOL+RgWF9M3JoW9OLr1ycokIrQctB0Hzsx0LWUXEVattp5p9wtnzOztzxgtPmD3DHxe18cs5v20FHNu5lBUbZ/Fr+h8sJ5e9gSX7VGJNQZwR25kz21xOn6QzPWpQqz/whN8HIiJFNG2lYDfsbD68mWX7lvHr3l/5M/1PCo3C4vsDDINOeXn0zcmlZ04uHfLyCUzoBM3Pcnw17g0BlZvmzh2B1FnH9JR5ymsScDWAz3mcei4Ng/yDW1i38VN+3/szy7P2sDbIgu2EqWYDDegamkDfpAH0a30preq2xuTkdSa8kTtf0/54gS0inkmBXkrJKshi5f6V/Lr3V5btW0bqsdQS94fa7ZyWm0eP3Dx65ubSzmYhILnfvwE/rk2Fc967o8fVGcf09h45DeBznsqey4rCZkHmPjZsmM3vqYv5/dhO/rIY5J40nWyyOZR+9TrSt/WldG88QL3wJ/GE17RffoIkIh5Hgd4H1HYP1e5ju1m+bzm/pf3Gqv2rOJJ3pMT94XY7XXPz6JGbS8+cPNoExmBJ7gfJp0PyGRDbyukr1rqLt/bIefvFiCep7Lk8OWw+NawRbUJ+5/eUBaw8upU/zAWOOvgT1DXM9AxvTM/GZ9Kv7UgaRJUemC4Oek2LiPxLC0t5OVf0UCVFJpHUOokRrUdgN+xsP7qdlftX8nva76w6sIrM/Ex+Dgvl5zDHH9Ewu51Oh37htL2L6bIoj84BkYQ3Lgr4p5+yB9+TefpsMuXRAD7nqcy5TLPm8MyXC+kevoKosK0cC8vg6a32f3vgAwDMRBsmeoQm0LNBX3q2uYxmse1VRlNJek2LiFSPAr2Hqcq0ac7qxTebzLSq04pWdVpxVdursBt2thzewsr9K1m5fyWrD6ziWMFxfgsN5bdQx3HMhkGro7/RZcWPnPZjHl1M4SQ26o2pcW9I6gmJnStdg+8JPH02mbJoCkjnKftcQrhtE98umc+a/Sv5I3sf+S0NNpUI52Yi7dAtOJae9bvTs82ltGzQ02unlHQ3vaZFRKpHJTceprKDNF1ZZ2qz29h+dDv/XbmYb7YswxKaginoaKnt4gsL6ZCXT4e8fNoX2mlfpw3RSf8E/EY9IUrT7jmbt5YL1ZaaXOR+8OsqZi35hOjQzdhCD5AeksshS+me9foFBvVyoinIbsqe7G7MvesmGsaUv3ibVI1e0yIiDqqh92KVnQbS1XWmJx/TFJBJYNgurjjDxtaja9l8ZDM2w17qcY0LCmj/T8jvEFiHNgldCWvUExqcBgmdIDiiVtrrT7xlAF9tjwupykVudt5xNqYsYMOupaw/tIH1uQfZYy79+g0wDNoagXSJbMJpDfqQVtCXKfOPKWzWMm95TYuI1CbV0LuJMwJLZRbeKa/OdHXKEepG1E5gOvmYRmEU+ZkdObdBbx49ox7ZBdlsPLSRDYc2sD5jPevT17An+wCpgYGkBgYyL8LRg2nJWk3y+t9o9UcBrfMLaBMSR+vYjsQ26gmJXSCxEwRHFh/HHdPXeds0kN5QLlTbnyiVV6p2RstYLJajbNm1hK37VrDlyBY256STSgH2E0tn/qmQaVJo0D6oDh3qtKZDo9Np03IYoeGxJY41qIvzwqa3vdZcxRte0yIinkQ99E7ijMBy4h93oNzQUFYPvQnHmNTaDExV/VTgaO7RfwP+wT/ZcHAtB/Mzy9y2rs1G6/x8WucV0Cq4Lq3qtuVYYVNmbAhlky2J/aZ6TB3eySXz27t7yjxf44pPlJbtyGDU278QErSPBiEbiQzZiS3kINaQXI6ay/4VF19oo4MlnA6RybRP7En7lsOIjnXd4G691kREpCIquXExZwSWqv5xP7HO1AwY/3xV9/iVbWNNa1vTs9PZfHgzW49sZUv6WrYc2sCunIPYKf0yNBsGjQoLaZZfQKMCCMmPpk+TjrRP6kl4Yheo3w5C6zjp2WnKvKqoSs+ysxfvOp5/nL8PbeLvvb/x98F1/J2ZwvbcDPYZ+RhlhHGzYZBcaKe1JYLWkUm0jutE6yZnEpfUDyyBVT5+ZVV0jvRaExGRU1HJjYvVdKq1qsxsU+TEqRYPZeVx28drqn38ynLG9I7xYfHEh8XTv1H/4ttyCnPYcXQHmw9vZsuBP9lw4C+2Z+0hx1RYXLLjUMDbOX/A1j+ov7GQZgUFNDECSQqJJSkyicb1WtMw4TRC4tpBnSZVDmuaMq9yqnrxWZ2ZS3IKc9iTuZvdGRvZnb6W3Ue3s+v4Xv7OO0w6hWU/yGQiymajSb5B3fww2sQ0Z2Dr02mefDYhcW3B7LqZZ051jlz9Wivv4kIlPyIivkGB3glqOtVadf+4F9WZpllzXDbVW23UtoYGhNIhtgMdYjsw29qT/65eh90wMAUcwxyUTmDwfqKDUwgO3o85/BiHjDwOBARwICCA5QAchayjkLUOUj8nvrCQxoU2ksyhxWE/IaYpCfXaEBffAUudZAgqPSuJpsw7tepcfJY1LuTxS9qC5Shr9m8k7fBW9mRsZPfRv0nN2seeAivpRkGF7YgrLKSZDZoFRtMsshHN67alaYOeFEZ2YmdOOMlx7guoZZ2jiV+sIzw4gG5N6pAYHerS11p5Fxcq+RER8R1uD/T/+9//ePTRR1m7di0hISGceeaZfPnll+5uVpVUZiBrRWr6x72mx/cUJYOQCaMwClthFLbsFhSYzigu8cnMz+Tvo3+zM2MjqQfXkXp0B7uz9rO7wMpx7KQHBJAeEMAq7GBPB2s6WFfDLrAYBvE2Gwl2EwmWMBKCY0gIr09iVBPio5sy9cxInvuxgAwjBpPJUq3z6Mu9npW5+My35XMo5xCHcg+RfmwP+w9tZb/1by7ovpt92RkcNo4zbVM+T26q+FiRNjuNCwtJIoik4GiSwhvSLLYdTRv0JCqxK0TEl1nvnuCsJ1tNZZ0jO3Dbx2tKBGdXvGfLuwBrkxBZ5QuzE/fpq69vERFv5dZA/8UXXzBmzBiefPJJzjrrLAoLC1m/fr07m1RtNSlHcUYg99bVTk9UVhACePj8tpzXKbH4OUUFRdElvgtd4ruU2M4wDI7mHWW3dRe7D64n9eA69lh3sif7APsLjpFuFFBoMpEWEEAaAHlQeACsB8C6Fnb/s6M2kGizE2uY+H59EKs3h1EvKJp6ofWoF1af6LB4oiISiI5sQFRUI6IjkwgJclx8+WKvZ25hLtY8K9Z8K8c4SGjkX4SYjxIWcJjggKNYAjKZ8fsLTF2exSFbDpnYKrXfAMOgfqGNRJuNBqZgkoLr0DiiIUl1WtI4vhPR9TtCTBMIDCkOkfGx4UR5+Gu7rAv0IicGZ1e8Z8u7AFuZcqRanwpW5/WtCwARkdrntkGxhYWFJCcnM2XKFG688cZq78cTBsU6i7/PvVzbAwVtdhsZORnsP7KdtIyNHDiynbTM3ezPSSct30qGPY/D2CmsxiwnwYZBpGEiuNBEkM1CgD0Qkz0Asz2Ijg3iiA6JICwogrCgKEKDIgkLiSYkOIagwFACA0IJDAwjMDCMgMBwAgOCCDQFEmgJJMAcUKlVR+2GnUJ7IQX2AseXraDk94W55BfkkFtwjJzco2TnZZKdn0l23jGyC7IcX4XZZBfmkm3LIbMwB6stD6tRQH4ZA5ZPJcAwqGezEWuzkWCHREs4CcF1HJ+GRDYmsV4r6tVrjTmmMUQmVjjewZsukorC67q9Vp6etwVbOb9eqzsguDrtKes9NWdsHy55bVmV3mvVeX96089ORMTTeMWg2D/++IO9e/diNps57bTT2L9/P126dOGZZ56hQ4cO5T4uLy+PvLy84u8zM8ueBtEb+fvcy7VdOmQxW6gfXp/64fXp3KhfmdvYDTuZOYfJOLyNQ0d2kGHdxaHj+8jITudQ7hEOFR7nmC2PTKMQK3YyzSZsJhN5JhN5JiDIAAr/+XLYnHUIspzyFNzGYhhE2+1E2eyOf+12ogkg1hJGfEg09ULqEhsWT72IBsRGJxMV3RhTZIKjLCYkptpTQVanZt9dTg6v9w9tQ8OYUP4za43bxmWU957qnFSnyu+1qo718aafnYiIt3NboP/7778BmDx5Ms8//zzJyck899xzDBgwgK1bt1K3bt0yHzd16lSmTJniyqaKC7m7dMhsMhMTFktMWCwtGvU55fZGQR5Zx/ZitaaSenAnL/3wOyEWKwHmbMzmPMzmPFokBFJg5JNtyyPblk+OUUC2YSPHsFGIQQFQYIICk4kCTBQW/f+fr8oKNIwTviCQkt8HYBBiNwgzIMxkJswUQJgpkFBLIGHmYEIDggmzhBIWGEZ0cAzRofWIDosjOiKBsPD6mMLqOsJ5aB0IjnLJrDHeMvNQWeH16flb+GXCQLePbynvPVXV91pVx/p4y89ORMQXOD3QT5gwgaeeeqrCbTZt2oTd7lhm/cEHH+TSSy8F4N1336VRo0Z89tln3HzzzWU+duLEiYwfP774+8zMTJKSkpzUeimLq2tgXfVJRVWfV1nbmwKDiajbjIi6zWjYdACpprOrPle/YYCtAGx5UJgPhbnF/zcKcjBh/2eRAcOxbVn/ApgDwRLAwWw7ezILaFgvivjoSEc5yz/3ERAKAUHVP2m1oKKfg7fMPFRReHX3RSqU/56qynutqp+gecvPTkTEFzg90N99992MHj26wm2aNWtGWppjWGK7du2Kbw8ODqZZs2akpqaW+9jg4GCCg4Od0lY5NV+tga3OQl6V2b5a4c1kcoTsgCA46aVd1UKVku089E873T3vS/lOdV49fQanoouR8CBLheHV28vpip5n/1Zx/DJhYKVe357+sxMR8SVuGxSbmZlJfHw8r776avGg2IKCAho1asRjjz3GTTfdVOn9+MqgWE/jq6tZVvV5ect58JZ2FqlKez1xwPjJFyOXnNaQL9fsq9FKyp6ophf1nvizExHxBl4xKDYqKopbbrmFSZMmkZSURJMmTXjmmWcAuPzyy93VLDmBr9bAVvV5ect58JZ2FqlKez2th7usmvkv1+xjztg+ZOfbfSa8OmNgq6f97EREfJFb56F/5plnCAgI4JprriEnJ4devXqxePFi6tSp485myT98tQa2qs/LW86Dt7SziLe190TlXYxk59tdMh2lq3jbRaKIiL+q/WkqKhAYGMizzz7LgQMHyMzMZMGCBbRv396dTZITFNXAWv6ZacVXamCr+ry85Tx4SzuLeFt7T1R0MXIib7kYqQp/eZ4iIt7ObTX0zqIa+tpXXg2st68AWdXaXm+pBfaWdhbx1tfX7JWpVZ/RyAv5y/MUEfE0Vcm4CvRSLb46+414Bm95fXnbxVN1efPz9PQLQxGR8ijQS63yttlUxLvo9SXO4i0XhiIiZalKxnVrDb14p4oGyonnSrPmsGxHBmnWHHc3pUJ6fYkzlDdDj6e//kVEqsOts9yId/Lm2Un8lTf1VOr1Jc6gGXpExJ+oh16qzJtnJ/FH3tZTqdeXOINm6BERf6IeeqmWkT0a079VnNcOlPMn3thTqdeX1FTRheHJM/TotSQivkiBXqpNK0B6B28tYdHrS2pKF4Yi4i9UciPiBq4coKoSFvFnidGh9GleT693EfFp6qEXcTF3DFBVT6WIiIjvUg+9H/OWaQx9iTsHqKqn0vfoPSwiIqAeer/lTdMY+hJvHKDqibT6p97DIiLyL/XQ+yFvm8bQl2gqvZqbvTKVftMWM+qtFfSbtpjZK1Pd3SSX03tYREROpEDvh7QSZ+U5u6RBA1RrRkHWQe9hERE5kUpu/JC3TmNYWc4qx6itkgZPG6DqTeUrKlly8PX3sIiIVI166P2QL/cSO6sco7Z7gj1lgKq3la+oZMnBl9/DIiJSdeqh91Oe1kvsDOWF8P6t4qr8/PyhJ9iZ58tVylr9876hrdmZkVV8v7/wxfewiIhUjwK9H/O1lThPFcKrUlriDyUN3nrRcmKQXbv3KE/N2+y3M7342ntYRESqRyU34jMqKseoammJP5Q0VLV8xZPmPE+MDiU5Nqw4zIP/DpB1J096TYiI+DP10ItL1eYAzLLKMZ4c3gGgWqUlnlzS4IzzWN75Kmt/njjnubd+wuCpqvqa8sTXhIiIv1KgF5dxRQAoK4Qv25FR7eDniSUNzjyPlblo8dRae38oi3KVqr6mPPU1ISLir1RyIy7hyvnDT55BxpdmRqmN83iqGXc8dc5zfyiLcoXqvKY89TUhIuKv1EMvLuHO8oiqlJZ4OnecR0/uCffksihvUZ3XlCe/JkRE/JECvbiEuwOAJwQ/Z9S9u+M8evoFkSeWRXmT6rymPP01ISLib0yGYRin3sxzZWZmEh0djdVqJSoqyt3NkQrMXplaKgD4yyA6Z9a9u+s8pllzfLIn3JtWyq0t1X1N+eprQkTEE1Ql4yrQi0v5YwBIs+bQb9riUj2gv0wYWO1z4I/nsTZoppZ/6TUlIuJZqpJxVXIjLuWP5RG1Uffuj+fR2TRTS0l6TYmIeC/NciNSy3xplp1T8aaFhjRTi4iI+AoFeh/lTcHK1/nL9IpVXY3X3fzpQktERHybauh9kOqCPZMv1yjXxjgBV/DngdoiIuLZVEPvx1QX7Ll8uUbZnesM1IQnTGcqIiJSUwr0PsZbg5V4N3evM1ATvnyhJSIi/kE19D5GdcHiDv4yTkBERMQTqYfex2gFR3EXla+IiIi4hwK9D1KwEnc5sXxFK7CKiIi4hgK9j1JdsLiTZloSERFxHdXQi4hTlTfTktZEEBERqR0K9CLiVJ6yAqsWVxMREX+hkhsRcSpPmMJSJT8iIuJP1EMvIk7l7iksVfIjIiL+Rj30IuJ07pxpSYuriYiIv1GgF6+iqRC9h7tmWvKEkh8RERFXUsmNeI3ZK1PpN20xo95aQb9pi5m9MtXdTRIP5O6SHxEREVczGYZhnHozz5WZmUl0dDRWq5WoqCh3N0dqSZo1h37TFpfqdf1lwkAFNSlTmjVHi6uJiIjXqkrGVcmNeAXVRUtVaXE1ERHxFyq5Ea9QVBd9ItVFi4iIiCjQi5dQXbSIiIhI2VRyI17DnVMhioiIiHgqBXrxKqqLFhERESlJJTciIiIiIl5MgV5ERERExIsp0IuIiIiIeDEFehERERERL+bWQL9161YuuugiYmNjiYqK4vTTT2fJkiXubJKIiIiIiFdxa6AfNmwYhYWFLF68mNWrV9O5c2eGDRvG/v373dksERERERGv4bZAn5GRwbZt25gwYQKdOnWiZcuWTJs2jezsbNavX++uZomIiIiIeBW3Bfp69erRunVrPvjgA7KysigsLOSNN94gPj6ebt26uatZIiIiIiJexW0LS5lMJhYuXMjFF19MZGQkZrOZ+Ph45s+fT506dcp9XF5eHnl5ecXfZ2ZmuqK5IiIiIiIeyek99BMmTMBkMlX4tXnzZgzDYNy4ccTHx/Pzzz/z+++/c/HFF3PBBReQlpZW7v6nTp1KdHR08VdSUpKzn4KIiIiIiNcwGYZhOHOHBw8e5NChQxVu06xZM37++WcGDx7MkSNHiIqKKr6vZcuW3HjjjUyYMKHMx5bVQ5+UlITVai2xHxERERERb5WZmUl0dHSlMq7TS27i4uKIi4s75XbZ2dkAmM0lPyQwm83Y7fZyHxccHExwcHDNGiki4qXSrDnszMiiaWw4idGh7m6OiIh4ALfV0Pfp04c6depw3XXX8cgjjxAaGspbb73Fzp07Of/8893VLBERjzV7ZSoT56zDboDZBFOHd2Rkj8bubpaIiLiZ22a5iY2NZf78+Rw/fpyzzjqL7t2788svv/DVV1/RuXNndzVLRMQjpVlzisM8gN2AB+asJ82a496GiYiI27mthx6ge/fufP/99+5sgoiIV9iZkVUc5ovYDIOUjGyV3oiI+Dm3rhQrIiKV0zQ2HLOp5G0Wk4nk2DD3NEhERDyGAr2IiBdIjA5l6vCOWEyOVG8xmXhyeAf1zouIiHtLbkREpPJG9mhM/1ZxpGRkkxwbpjAvIiKAAr2IiFdJjA5VkBcRkRJUciMiIiIi4sUU6EVEREREvJgCvYiIiIiIF1OgFxERERHxYgr0IiIiIiJeTIFeRERERMSLKdCLiIiIiHgxBXoRERERES+mQC8iIiIi4sUU6EWqIc2aw7IdGaRZc9zdFBEREfFzAe5ugIi3mb0ylYlz1mE3wGyCqcM7MrJHY3c3S0RERPyUeuhFqiDNmlMc5gHsBjwwZ7166kVERMRtFOhFqmBnRlZxmC9iMwxSMrLd0yARERHxewr0IlXQNDYcs6nkbRaTieTYMPc0SERERPyeAr1IFSRGhzJ1eEcsJkeqt5hMPDm8A4nRoW5umYiIiPgrDYoVqaKRPRrTv1UcKRnZJMeGKcyLiIiIWynQi1RDYnSogryIiIh4BJXciIiIiIh4MQV6EREREREvpkAvIiIiIuLFFOhFRERERLyYAr2IiIiIiBdToBcRERER8WIK9CIiIiIiXkyBXkRERETEiynQi4iIiIh4MQV6EREREREvpkAvIiIiIuLFFOhFRERERLyYAr2IiIiIiBdToBcRERER8WIK9CIiIiIiXkyBXkRERETEiynQi4iIiIh4MQV6EREREREvpkAvIiIiIuLFFOhFRERERLyYAr2IiIiIiBdToBcRERER8WIK9CIiIiIiXkyBXtwmzZrDsh0ZpFlz3N0UEREREa8V4O4GiH+avTKViXPWYTfAbIKpwzsyskdjdzdLRERExOuoh15cLs2aUxzmAewGPDBnvXrqRURERKpBgV5cbmdGVnGYL2IzDFIyst3TIBEREREvpkAvLtc0NhyzqeRtFpOJ5Ngw9zRIRERExIsp0IvLJUaHMnV4RywmR6q3mEw8ObwDidGhbm6ZiIiIiPfRoFhxi5E9GtO/VRwpGdkkx4YpzIuIiIhUkwK9uE1idKiCvIiIiEgNqeRGRERERMSLKdCLiIiIiHixWgv0TzzxBH379iUsLIyYmJgyt0lNTeX8888nLCyM+Ph47r33XgoLC2urSSIiIiIiPqfWaujz8/O5/PLL6dOnD//9739L3W+z2Tj//PNJSEhg2bJlpKWlce211xIYGMiTTz5ZW80SEREREfEpJsMwjFNvVn3vvfced955J0ePHi1x+7x58xg2bBj79u2jfv36ALz++uvcf//9HDx4kKCgoErtPzMzk+joaKxWK1FRUc5uvoiIiIiIy1Ul47qthn758uV07NixOMwDDBkyhMzMTDZs2FDu4/Ly8sjMzCzxJSIiIiLir9wW6Pfv318izAPF3+/fv7/cx02dOpXo6Ojir6SkpFptp4iIiIiIJ6tSoJ8wYQImk6nCr82bN9dWWwGYOHEiVqu1+Gv37t21ejwREREREU9WpUGxd999N6NHj65wm2bNmlVqXwkJCfz+++8lbjtw4EDxfeUJDg4mODi4UscQEREREfF1VQr0cXFxxMXFOeXAffr04YknniA9PZ34+HgAFixYQFRUFO3atXPKMUREREREfF2tTVuZmprK4cOHSU1NxWaz8eeffwLQokULIiIiGDx4MO3ateOaa67h6aefZv/+/Tz00EOMGzdOPfAiIv/f3v3HRF3/cQB/AsKBP4CIHwdT6QDDTUkUgx1tAoOvkK4gnSUxg0aghE36YZlbMduaRi62HJu2JtbWtNj8seyHQ+RwGpIiTERkQoSBHi4NOAXC4PX9w3ETEeQO7z538Hxst+nn3m94fZ5734cXn7vPByIionGy2G0rMzIy8M0334zYXl5ejtjYWABAa2srcnJyoNPpMGPGDKSnp2PHjh2YNm38v2fwtpVERERENNmY0uNa/D70lsaGnoiIiIgmG7u4Dz0REREREU0cG3oiIiIiIjtmsYtirWXoE0P8i7FERERENFkM9bbj+XS83Tf0BoMBAPgXY4mIiIho0jEYDPDw8BhzjN1fFDs4OIhr165h1qxZcHBwsPr37+7uxpw5c/DXX3/xolwTMTvzMTvzMbuJYX7mY3bmY3YTw/zMp2R2IgKDwYCAgAA4Oo79KXm7P0Pv6OiI2bNnK10G3N3d+SIxE7MzH7MzH7ObGOZnPmZnPmY3MczPfEpl96gz80N4USwRERERkR1jQ09EREREZMfY0E+QSqVCfn4+VCqV0qXYHWZnPmZnPmY3MczPfMzOfMxuYpif+ewlO7u/KJaIiIiIaCrjGXoiIiIiIjvGhp6IiIiIyI6xoSciIiIismNs6ImIiIiI7BgbehP8+eefyMzMhEajgZubG4KDg5Gfn4/+/v4x5/X19SE3NxdPPvkkZs6cidWrV6Ojo8NKVduWTz/9FNHR0Zg+fTo8PT3HNScjIwMODg7DHklJSZYt1AaZk52I4OOPP4a/vz/c3NyQkJCAK1euWLZQG3Tr1i2kpaXB3d0dnp6eyMzMxO3bt8ecExsbO2LdbdiwwUoVK6uoqAhPPfUUXF1dERUVhd9//33M8SUlJZg/fz5cXV0RFhaGn3/+2UqV2h5Tstu3b9+INebq6mrFam3HyZMn8cILLyAgIAAODg44fPjwI+fodDosWbIEKpUKISEh2Ldvn8XrtEWmZqfT6UasOwcHB+j1eusUbEO2b9+OZ599FrNmzYKvry9SUlLQ2Nj4yHm2eMxjQ2+Cy5cvY3BwEHv27EF9fT0KCwuxe/dubN26dcx5b7/9Nn788UeUlJSgoqIC165dw6pVq6xUtW3p7+/HmjVrkJOTY9K8pKQkXL9+3fjYv3+/hSq0XeZkV1BQgC+//BK7d+9GVVUVZsyYgcTERPT19VmwUtuTlpaG+vp6lJaW4ujRozh58iSys7MfOS8rK2vYuisoKLBCtcr6/vvv8c477yA/Px/nz5/HokWLkJiYiBs3bjx0/G+//YbU1FRkZmaipqYGKSkpSElJwcWLF61cufJMzQ6499cn719jra2tVqzYdty5cweLFi1CUVHRuMa3tLRg5cqViIuLQ21tLfLy8vDGG2/g2LFjFq7U9pia3ZDGxsZha8/X19dCFdquiooK5Obm4syZMygtLcXdu3exfPly3LlzZ9Q5NnvME5qQgoIC0Wg0oz7f2dkpzs7OUlJSYtzW0NAgAKSystIaJdqk4uJi8fDwGNfY9PR0SU5Otmg99mS82Q0ODoparZbPP//cuK2zs1NUKpXs37/fghXalkuXLgkAOXv2rHHbL7/8Ig4ODtLe3j7qvJiYGNm0aZMVKrQtkZGRkpuba/z/wMCABAQEyPbt2x86/uWXX5aVK1cO2xYVFSXr16+3aJ22yNTsTDkOTiUA5NChQ2OOef/992XBggXDtr3yyiuSmJhowcps33iyKy8vFwDyzz//WKUme3Ljxg0BIBUVFaOOsdVjHs/QT1BXVxe8vLxGfb66uhp3795FQkKCcdv8+fMxd+5cVFZWWqPESUGn08HX1xehoaHIycnBzZs3lS7J5rW0tECv1w9bex4eHoiKippSa6+yshKenp5YunSpcVtCQgIcHR1RVVU15tzvvvsO3t7eWLhwIT788EP09PRYulxF9ff3o7q6etiacXR0REJCwqhrprKycth4AEhMTJxSawwwLzsAuH37NgIDAzFnzhwkJyejvr7eGuXaPa67iQsPD4e/vz/+97//4fTp00qXYxO6uroAYMy+zlbX3jRFv7uda2pqwq5du7Bz585Rx+j1eri4uIz4zLOfn9+U/LyaOZKSkrBq1SpoNBo0Nzdj69ateP7551FZWQknJyely7NZQ+vLz89v2Paptvb0ev2It5KnTZsGLy+vMXN49dVXERgYiICAAFy4cAEffPABGhsbcfDgQUuXrJi///4bAwMDD10zly9ffugcvV4/5dcYYF52oaGh2Lt3L5555hl0dXVh586diI6ORn19PWbPnm2Nsu3WaOuuu7sbvb29cHNzU6gy2+fv74/du3dj6dKl+Pfff/H1118jNjYWVVVVWLJkidLlKWZwcBB5eXl47rnnsHDhwlHH2eoxj2foAWzZsuWhF4jc/3jwgNze3o6kpCSsWbMGWVlZClVuG8zJzxRr167Fiy++iLCwMKSkpODo0aM4e/YsdDrd49sJhVg6u8nM0tllZ2cjMTERYWFhSEtLw7fffotDhw6hubn5Me4FTWVarRavvfYawsPDERMTg4MHD8LHxwd79uxRujSaxEJDQ7F+/XpEREQgOjoae/fuRXR0NAoLC5UuTVG5ubm4ePEiDhw4oHQpZuEZegDvvvsuMjIyxhwTFBRk/Pe1a9cQFxeH6OhofPXVV2POU6vV6O/vR2dn57Cz9B0dHVCr1RMp22aYmt9EBQUFwdvbG01NTYiPj39sX1cJlsxuaH11dHTA39/fuL2jowPh4eFmfU1bMt7s1Gr1iIsS//vvP9y6dcuk12BUVBSAe+/MBQcHm1yvPfD29oaTk9OIu3CNdbxSq9UmjZ+szMnuQc7Ozli8eDGamposUeKkMtq6c3d359l5M0RGRuLUqVNKl6GYjRs3Gm+Y8Kh3x2z1mMeGHoCPjw98fHzGNba9vR1xcXGIiIhAcXExHB3HfpMjIiICzs7OKCsrw+rVqwHcu7L86tWr0Gq1E67dFpiS3+PQ1taGmzdvDmtS7ZUls9NoNFCr1SgrKzM28N3d3aiqqjL5LkO2aLzZabVadHZ2orq6GhEREQCAEydOYHBw0Nikj0dtbS0ATIp1NxoXFxdERESgrKwMKSkpAO69DV1WVoaNGzc+dI5Wq0VZWRny8vKM20pLSyfN8W28zMnuQQMDA6irq8OKFSssWOnkoNVqR9wqcCquu8eltrZ2Uh/bRiMieOutt3Do0CHodDpoNJpHzrHZY56il+Tamba2NgkJCZH4+Hhpa2uT69evGx/3jwkNDZWqqirjtg0bNsjcuXPlxIkTcu7cOdFqtaLVapXYBcW1trZKTU2NbNu2TWbOnCk1NTVSU1MjBoPBOCY0NFQOHjwoIiIGg0Hee+89qayslJaWFjl+/LgsWbJE5s2bJ319fUrthiJMzU5EZMeOHeLp6SlHjhyRCxcuSHJysmg0Gunt7VViFxSTlJQkixcvlqqqKjl16pTMmzdPUlNTjc8/+LptamqSTz75RM6dOyctLS1y5MgRCQoKkmXLlim1C1Zz4MABUalUsm/fPrl06ZJkZ2eLp6en6PV6ERFZt26dbNmyxTj+9OnTMm3aNNm5c6c0NDRIfn6+ODs7S11dnVK7oBhTs9u2bZscO3ZMmpubpbq6WtauXSuurq5SX1+v1C4oxmAwGI9pAOSLL76QmpoaaW1tFRGRLVu2yLp164zj//jjD5k+fbps3rxZGhoapKioSJycnOTXX39VahcUY2p2hYWFcvjwYbly5YrU1dXJpk2bxNHRUY4fP67ULigmJydHPDw8RKfTDevpenp6jGPs5ZjHht4ExcXFAuChjyEtLS0CQMrLy43bent75c0335QnnnhCpk+fLi+99NKwXwKmkvT09Ifmd39eAKS4uFhERHp6emT58uXi4+Mjzs7OEhgYKFlZWcYfkFOJqdmJ3Lt15UcffSR+fn6iUqkkPj5eGhsbrV+8wm7evCmpqakyc+ZMcXd3l9dff33YL0IPvm6vXr0qy5YtEy8vL1GpVBISEiKbN2+Wrq4uhfbAunbt2iVz584VFxcXiYyMlDNnzhifi4mJkfT09GHjf/jhB3n66afFxcVFFixYID/99JOVK7YdpmSXl5dnHOvn5ycrVqyQ8+fPK1C18oZupfjgYyiv9PR0iYmJGTEnPDxcXFxcJCgoaNixbyoxNbvPPvtMgoODxdXVVby8vCQ2NlZOnDihTPEKG62nu38t2csxz0FExJLvABARERERkeXwLjdERERERHaMDT0RERERkR1jQ09EREREZMfY0BMRERER2TE29EREREREdowNPRERERGRHWNDT0RERERkx9jQExERERHZMTb0RERERER2jA09EREREZEdY0NPRERERGTH2NATEREREdmx/wPtq5A8iyz+MgAAAABJRU5ErkJggg==",
            "text/plain": [
              "<Figure size 900x600 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "plot_preds(x, y, f, new_model, 'After Training: Keras')"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ng-BY_eGS0bn"
      },
      "source": [
        "请参阅[基本训练循环](basic_training_loops.ipynb)和 [Keras 指南](https://tensorflow.google.cn/guide/keras)，了解更多详细信息。"
      ]
    }
  ],
  "metadata": {
    "colab": {
      "collapsed_sections": [],
      "name": "basics.ipynb",
      "toc_visible": true
    },
    "kernelspec": {
      "display_name": "xxx",
      "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": "3.12.2"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}