Deep_learning_with_Keras_(modified_by_Jeff_Chen).ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Deep learning with Keras (modified by Jeff Chen)",
      "provenance": [],
      "collapsed_sections": [
        "o0kSHA98GxAg"
      ],
      "toc_visible": true,
      "authorship_tag": "ABX9TyPsEn24G2AbqKHSSLZQzQOx",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/jeffeuxMartin/CodingTipsForShare/blob/main/Deep_learning_with_Keras_(modified_by_Jeff_Chen).ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "1kuKAJFLAg_T"
      },
      "source": [
        "# 基本神經網路架構 - binary classifier\n",
        "modified by Jeff Chen"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FgrdZlC2ufJ9"
      },
      "source": [
        "import numpy as np, pandas as pd\n",
        "\n",
        "import tensorflow as tf\n",
        "import tensorflow.keras as keras\n",
        "from tensorflow.keras import layers\n",
        "from tensorflow.keras.utils import plot_model\n",
        "from tensorflow.keras.utils import to_categorical"
      ],
      "execution_count": 1,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "aOrh7UohF603"
      },
      "source": [
        "## 創造資料"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "3PTmL51QBH0G"
      },
      "source": [
        "首先隨機創造一些假的data:\n",
        "\n",
        "100筆資料，每筆資料是長度為3的tensor\n",
        "類別只有0和1"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "pikxpKHuunE5"
      },
      "source": [
        "# fix random seed\n",
        "np.random.seed(42)"
      ],
      "execution_count": 2,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9u_-JsfLuoKc"
      },
      "source": [
        "# training set\n",
        "x_train = np.random.random((100, 3))\n",
        "y_train = np.random.randint(0, 2, (100,))\n",
        "\n",
        "# validation set\n",
        "x_valid = np.random.random((100, 3))\n",
        "y_valid = np.random.randint(0, 2, (100,))"
      ],
      "execution_count": 3,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "fiwgg9sO7V_U"
      },
      "source": [
        "觀測生成出來的假資料"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "5tdpptdwursT",
        "outputId": "20f3f548-1197-4c27-ea20-5f65ef455ab0"
      },
      "source": [
        "print(x_train[:5])\n",
        "print(y_train[:5])"
      ],
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[[0.37454012 0.95071431 0.73199394]\n",
            " [0.59865848 0.15601864 0.15599452]\n",
            " [0.05808361 0.86617615 0.60111501]\n",
            " [0.70807258 0.02058449 0.96990985]\n",
            " [0.83244264 0.21233911 0.18182497]]\n",
            "[0 1 1 1 0]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "NwluBKs27cdC"
      },
      "source": [
        "## Two common Model definition methods: <br />Sequential Model (Layer-wise) / Functional API (Tensor-wise)\n",
        "\n",
        "關於定義 model 的方式，主要有兩個角度\n",
        "\n",
        "1. **Layer-wise** / Sequential Model: 把 model 視為一層一層的 layers，將這些層排成一個 list 組成整個 model （這部分比較接近 Keras 的思維）\n",
        "1. **Tensor-wise** / Functional API: 把 model 視為一個一個 tensor 互相接起來的 graph（一堆點和線的集合），將這些 tensor 的傳遞告訴 Keras 組成整個 model（這部分思維比較接近 TensorFlow。顧名思義，「Tensor」的「Flow」）\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "HNbEV-uqCe_A"
      },
      "source": [
        "### What is a *layer* exactly?"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4HTuyHdc8rrj"
      },
      "source": [
        "# 定義一個 layer，代表一排 neurons\n",
        "fully_connected_layer = layers.Dense(16)  # 16 個 neuron 組成的一層"
      ],
      "execution_count": 5,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "xDiMCe-Y85JM",
        "outputId": "413b3c1b-cfc2-4559-ce85-25b09800db9b"
      },
      "source": [
        "input_tensor: np.ndarray = np.random.rand(100, 7)\n",
        "print(input_tensor.shape)   # shape = (100, 7)"
      ],
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(100, 7)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "e06GJ_5B8_TE",
        "outputId": "aff5b1dd-7226-4572-c9a8-97f2b2622503"
      },
      "source": [
        "# 一層 fully connected layer 等於會進行一個矩陣運算，\\\n",
        "#     因此出來的 output tensor 的 shape 會是 \\\n",
        "#     (100, num_neurons) 也就是 (100, 16)\n",
        "output_tensor = fully_connected_layer(input_tensor)\n",
        "print(output_tensor.shape)  # shape = (100, 16)\n",
        "\n",
        "# 之所以不用定義 layer 的 input_shape 是因為當張量被傳進 layer 時，layer 自己會知道 shape"
      ],
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(100, 16)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "tcZ34kLC9AOJ"
      },
      "source": [
        "### 接著就處理 model 的建立"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "k81KVNxE8VoD"
      },
      "source": [
        "### Layer-wise (Sequential model / Sequential API)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "89lSAY3u8fXk"
      },
      "source": [
        "#### Method 1: 把 layer 一個一個加到模型裡面"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Yj_3zkHJuxgi"
      },
      "source": [
        "model_from_empty = keras.Sequential()"
      ],
      "execution_count": 8,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OUPwojCe46rU"
      },
      "source": [
        "# 比較不一樣的是，因為整個模型後面在 compile 時需要知道 input_shape 才能 compile，因此在這邊要\n",
        "# 先 specify input_shape，不然 model 不會先被 build 起來\n",
        "model_from_empty.add(layers.Dense(16, input_shape=(3,))) #16個neurons\n",
        "model_from_empty.add(layers.Dense(16)) #16個neurons\n",
        "model_from_empty.add(layers.Dense(2, activation='sigmoid')) #將輸出壓縮至0-1之間\n",
        "\n",
        "#compile\n",
        "model_from_empty.compile(loss='binary_crossentropy', #loss function\n",
        "              optimizer='adam', \n",
        "              metrics=['accuracy'])"
      ],
      "execution_count": 9,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "8aYuRLH_-erq",
        "outputId": "ee8c09a6-ee62-433a-e38b-adddd9d0a331"
      },
      "source": [
        "model_from_empty.summary()"
      ],
      "execution_count": 10,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "dense_1 (Dense)              (None, 16)                64        \n",
            "_________________________________________________________________\n",
            "dense_2 (Dense)              (None, 16)                272       \n",
            "_________________________________________________________________\n",
            "dense_3 (Dense)              (None, 2)                 34        \n",
            "=================================================================\n",
            "Total params: 370\n",
            "Trainable params: 370\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 422
        },
        "id": "8kLSaFLU-dK5",
        "outputId": "8d31d533-af7f-44e1-fec5-2ff5cc227739"
      },
      "source": [
        "plot_model(model_from_empty, show_shapes=True)"
      ],
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAasAAAGVCAYAAABAYd6wAAAABmJLR0QA/wD/AP+gvaeTAAAgAElEQVR4nOzdaVRUZ7Y38H8BBUVBMSkgAVQoVCKixmgiRKO5JtxEWgaBgEMS44oLTQzidBUcgqgoagMLlfYabbo7GmXQi0YlsY0vpu2oK2lxaIwGieBAFFBkLGTa74c0FcsCrIKCGti/teqDz3nOOfucU8W2Tj3n2QIiIjDGGGO6K8tI2xEwxhhjz8PJijHGmM7jZMUYY0zncbJijDGm80yebTh37hySkpK0EQtjjDGGrKwspTalb1Z37txBdnZ2rwTEGFPd+fPncf78eW2HoVfu3r3Lf8/0SGfXS+mbVZv2MhtjTHvCwsIA8GdTHZmZmQgPD+dzpifarld7+DcrxhhjOo+TFWOMMZ3HyYoxxpjO42TFGGNM53GyYowxpvM4WTHWx5w4cQLW1tb46quvtB2KTpo/fz4EAoH8NXv2bKU+p06dQkxMDA4dOgR3d3d53/fee0+pr5+fHyQSCYyNjeHl5YWLFy/2xmF0WWJiIjw9PWFubg4LCwt4enpizZo1qK6ulvc5evQoEhMT0dLSorBuTk6Owrnr37+/xuLiZMVYH8OFFp7Pzs4Oubm5uHHjBvbu3auw7LPPPkNqaipiY2MREhKCX375BVKpFP369cO+fftw/Phxhf4nT55EVlYWpk2bhoKCAowZM6Y3D0Vt//jHPzBv3jzcvn0bDx48wPr165GYmIjQ0FB5n4CAAIhEIkyZMgWPHz+WtwcGBuLu3bv47rvvMHXqVI3GxcmKsT7G398fVVVVmDZtmrZDgUwmg6+vr7bDUGJubo63334bQ4cOhZmZmbx98+bNOHjwIDIzMyGRSBTWSU1NhZGRESIjI1FVVdXbIWuMqakpPvnkE9jb28PS0hJhYWEICgrC3//+d/z666/yfosWLcKoUaMwdepUNDc3AwAEAgGcnZ0xceJEDBkyRKNxcbJijGnN3r17UVZWpu0wVHLz5k2sWbMG69atg0gkUlru6+uL6Oho3Lt3D8uWLdNChJpx+PBhpeNzdnYGANTW1iq0x8XF4dKlS0hJSenxuDhZMdaHnD17FgMHDoRAIMCOHTsAAGlpabCwsIBYLMaRI0fwzjvvwMrKCi4uLjhw4IB83dTUVIhEIjg4OGD+/PlwcnKCSCSCr68vLly4IO8XFRUFU1NTDBgwQN72ySefwMLCAgKBABUVFQCA6OhoLF26FEVFRRAIBPDw8AAAfP3117CyssLGjRt745SoLDU1FUSEgICADvts2LABQ4cOxZ49e3Dq1KlOt0dESEpKwosvvggzMzPY2toiKCgI169fl/dR9doAQEtLC9auXYuBAwfC3NwcI0eOREZGRvcO+j8KCwthY2ODQYMGKbTb2tpi0qRJSElJ6fHby5ysGOtDJkyYgO+//16h7eOPP8bixYshk8kgkUiQkZGBoqIiuLu7Y968eWhqagLwWxKaM2cO6uvrsWjRIhQXF+PixYtobm7GW2+9hTt37gD47Y/6u+++q7CPnTt3Yt26dQptKSkpmDZtGqRSKYgIN2/eBAD5j/atra09cg666vjx4xg2bBjEYnGHfczNzfGXv/wFRkZGmDdvHurq6jrsGxcXh5iYGKxatQplZWX47rvvcOfOHUycOBEPHjwAoPq1AYCVK1diy5YtSE5Oxq+//opp06Zh5syZ+PHHH7t0vE1NTbh37x527NiBU6dOYfv27TA1NVXq99JLL+HevXu4fPlyl/ajKk5WjDE5X19fWFlZwd7eHhEREairq8Pt27cV+piYmMi/DQwfPhxpaWmoqalBenq6RmLw9/dHdXU11qxZo5HtaUJdXR1u3boFqVT63L4+Pj5YvHgxiouLsXLlynb7yGQyJCUlYfr06Zg9ezasra3h7e2NXbt2oaKiArt371Zap7Nr09DQgLS0NAQHByMkJAQ2NjZYvXo1hEJhl6+Lq6srXFxcEBcXhy1btnQ4Z1/bb1NXr17t0n5UxcmKMdautv9FP/2/9/aMHTsWYrFY4faVoSkrKwMRdfqt6mkbNmzAsGHDsHPnTpw9e1ZpeUFBAWprazF27FiF9nHjxsHU1FThtmp7nr02N27cQH19PUaMGCHvY25ujgEDBnT5uty5cwdlZWX48ssv8de//hUvvfRSu78vtp2Ttm+DPYWTFWOs28zMzFBeXq7tMHpMQ0MDACiMDOyMSCRCeno6BAIB5s6dC5lMprC8bbi3paWl0ro2NjaoqalRK762242rV69WeM6ppKQE9fX1am2rjVAohL29Pfz8/HDw4EEUFBQgISFBqZ+5uTmA389RT+FkxRjrlqamJjx+/BguLi7aDqXHtP1BfvYh2M74+PhgyZIlKCwsxPr16xWW2djYAEC7Sakr59Le3h4AkJycDCJSeJ07d06tbbXHw8MDxsbGKCgoUFrW2NgI4Pdz1FM4WTHGuiUvLw9EhPHjx8vbTExMnnv7UJ84ODhAIBCo/fzU+vXr4enpifz8fIX2ESNGwNLSUmnww4ULF9DY2IiXX35Zrf24urpCJBLh0qVLaq33rIcPH2LmzJlK7YWFhWhpaYGrq6vSsrZz4ujo2K19Pw8nK8aYWlpbW1FZWYnm5mZcuXIF0dHRGDhwIObMmSPv4+HhgUePHiEnJwdNTU0oLy9HSUmJ0rbs7OxQWlqK4uJi1NTUoKmpCbm5uTo3dF0sFsPd3R13795Va72224HGxsZK7UuXLsXhw4exb98+VFdX4+rVq1iwYAGcnJwQGRmp9n4+/PBDHDhwAGlpaaiurkZLSwvu3r0rf5A3IiICjo6OnU73ZGFhgZMnT+L06dOorq5GU1MT8vPz8cEHH8DCwgJLlixRWqftnHh7e6sVs7o4WTHWh+zYsQPjxo0DAKxYsQKBgYFIS0tDcnIyAGDkyJH45Zdf8Pnnn2Pp0qUAgLfffhuFhYXybTQ0NMDb2xvm5uaYOHEihg4div/3//6fwu85H3/8Md544w3MmDEDw4YNw/r16+W3iXx8fOTD3BcsWAAHBwcMHz4cU6dOxaNHj3rlPHSFv78/CgoKFH5/+r//+z94eHigqKgI48aNw6effqq03vjx49v9I//ZZ58hISEB8fHx6N+/PyZNmoTBgwcjLy8PFhYWAKDWtUlJScHixYuRmJiIfv36wcnJCdHR0aisrATw2+26srIyHDlypMNjFIlEeO211/DRRx/B2dkZEokEYWFhGDx4MM6fP68wgKPNDz/8AGdnZ4wcOVLVU9k19IyMjAxqp5kxpmWhoaEUGhqq1RgiIyPJzs5OqzGooyt/zyIjI8nZ2VmpvbCwkExMTOiLL77QVHi9qqWlhSZOnEh79+7V2DYrKipIJBLRtm3blJYtWrSI+vXrp9b2OrlemfzNijGmFnUGGegrmUyGb775BoWFhfIBBB4eHoiPj0d8fLzStEO6rqWlBTk5OaipqUFERITGthsXF4fRo0cjKioKwG+zcpSWluLs2bPyh7w1hZMVY4w949GjR/KJbOfOnStvj4mJQVhYGCIiIvRqstq8vDwcOnQIubm5Kj8r9jxJSUm4dOkSTpw4AaFQCAA4cuSIfCLbZ2ef764eSVYfffQRJBIJBAJBt0enaFtrayuSk5O7PTO0IdQQOn/+PF588UUYGRlBIBDA0dERGzZs0HZYCp6tLzRgwIB26xEx9cXGxiI9PR1VVVVwc3NDdna2tkPqEbt27VIY+r1v3z6F5Rs3bkRUVBQ2bdqkpQjVN2XKFOzfv19hvsbuOHLkCJ48eYK8vDzY2trK24OCghTOXds8kJpgorEtPWXPnj148803MWPGjJ7YfK8pLCzEhx9+iH/+858YNWpUt7ZFBlBDaPz48fjpp5/w9ttv45tvvsGNGzfkz4voipCQEISEhMDDwwMVFRW4f/++tkMyGAkJCe0+FNoX+fn5wc/PT9thaE1gYCACAwN7dZ98G7ADly9fxsqVK7FgwQKMHj2629vjGkI9w5COhTHWsR5LVgKBoKc23StGjRqFQ4cOYdasWSpPsaIv9KmG0PMY0rEwxjqmkWRFRNi6dSuGDRsGMzMzWFtbY/ny5Ur9Oqu3ok7dljNnzuCVV16BWCyGlZUVvL29UV1d/dx9aIuh1xDStWNR1z/+8Q8MHz4c1tbWEIlE8Pb2xjfffAPgt99f237/kkql8pkIPvzwQ4jFYlhbW+Po0aMAOn/vbdmyBWKxGBKJBGVlZVi6dCmcnZ1x48aNLsXMWJ+jxjj3Dq1atYoEAgH98Y9/pMrKSqqvr6edO3cSAMrPz5f3W7ZsGZmZmVF2djZVVlZSbGwsGRkZ0Q8//CDfDgD69ttvqaqqisrKymjixIlkYWFBjY2NRERUW1tLVlZWlJiYSDKZjO7fv0/Tp0+n8vJylfbRFa+++iqNGjWqy+sTEd25c4cA0Pbt2+Vtqhwv0W/PfVhYWNC1a9eooaGBCgoKaNy4cSSRSOj27dvyfrNmzSJHR0eF/W7dupUAyM8PEVFISAhJpVKFfseOHSOJRELx8fHPPZb//u//JgBUWVmpk8dCRCSVSsna2vq5x0JElJWVRXFxcfTo0SN6+PAhjR8/XuH5kJCQEDI2NqZ79+4prDdz5kw6evSo/N+qvr8XLVpE27dvp+nTp9NPP/2kUoxEuvGclb7h50b1S48+ZyWTyZCcnIw333wTS5YsgY2NDczNzWFnZ6fQT516K53VbSkuLkZ1dTW8vLwgEong6OiIQ4cOoX///j1S06U3GFINIV04FnWFhobis88+g62tLezs7BAQEICHDx/KZxFfsGABWlpaFOKrrq7GDz/8gKlTpwJQ7/29efNmLFy4EIcOHYKnp2fvHShjeqzbowFv3ryJ+vp6TJkypdN+Xa238mzdFnd3dzg4OGD27NlYtGgR5syZg8GDB3drH7rEkGoI6euxtD0z0vbw63/9139h6NCh+POf/4zY2FgIBAIcPHgQERER8jnfeuu9l52drfe/B2sDnzP91+1k1TaJYdsU9R15ut7K6tWrFZY5OTmpvD9zc3OcPn0aK1euxMaNGxEfH493330X6enpGtuHvjCkGkLaPJbjx49j69atKCgokE/e+TSBQID58+djyZIl+Pbbb/Hmm2/ib3/7G/bv3y/v01vvvfHjx2Px4sUa256hO3fuHFJSUrT+uzVTTdv1ak+3k5VIJAIAPHnypNN+T9dbiY6O7tY+vby88NVXX6G8vBxJSUnYvHkzvLy85NOIaGIfus6Qagj19rF89913+Ne//oXFixfj9u3bCA4OxvTp0/HnP/8ZL7zwArZv347/+Z//UVhnzpw5iI2NxZ49e+Dq6gorKysMGjRIvlyT7+/OuLi44N133+2x7RuilJQUPmd6pKNk1e3frEaMGAEjIyOcOXOm036aqrdSWlqKa9euAfjtD8SmTZswZswYXLt2TWP70AeGVEOot4/lX//6l3xW66tXr6KpqQkff/wx3N3dIRKJ2r1lZGtri/DwcOTk5GDbtm2YN2+ewvK+9N5jTBu6nazs7e0REhKC7Oxs7N27F9XV1bhy5Qp2796t0E+VeiuqKC0txfz583H9+nU0NjYiPz8fJSUlGD9+vMb2oYsMqYZQTx9LR5qamvDgwQOFEgwDBw4EAJw6dQoNDQ0oLCxUGEb/tAULFuDJkyc4duyY0sPdhvzeY0wnqDF0sEM1NTX00UcfUb9+/cjS0pImTJhAa9euJQDk4uJCly9fJiKiJ0+e0IoVK2jgwIFkYmJC9vb2FBISQgUFBbRz504Si8UEgIYMGUJFRUW0e/dusrKyIgA0aNAg+vnnn6m4uJh8fX3J1taWjI2N6YUXXqBVq1ZRc3Pzc/ehjnPnztFrr71GTk5OBIAA0IABA8jX15fOnDmj1ra2b99OAwYMIAAkFospICBA5eMl+m24t1AoJGdnZzIxMSErKysKCgqioqIihf08fPiQ3njjDRKJROTm5kaffvopLV++nACQh4eHfGj4xYsXadCgQWRubk4TJkyg+/fv04kTJ0gikdCGDRs6PI7z58+Tl5cXGRkZyc/Hxo0bdepY/vSnP5FUKpVfs45ehw8flu9rxYoVZGdnRzY2NhQWFkY7duwgACSVShWG0xMRvfTSSxQTE9Pu+ensvZeYmEjm5uYEgFxdXbtUZoKHrquPh67rl86GrguIFCety8zMRHh4uEHMZWco5s+fj6ysLDx8+FDboXSbvh+Lv78/duzYATc3t17fd1hYGAAgKyur1/etr/jvmX7p5Hpl8dyAesKQagjp07E8fVvxypUrEIlEWklUjPV1fSZZXb9+XT5tTmcvVQuTaXp7TDetWLEChYWF+Pnnn/Hhhx9i/fr12g6J9bD58+crfIbbKzFz6tQpxMTEKJWkee+995T6+vn5QSKRwNjYGF5eXrh48WJvHEaXJSYmwtPTE+bm5rCwsICnpyfWrFkjn9IOAI4ePYrExESl/3jm5OQonLv+/ftrLjA17hkyLYiJiSFTU1MCQIMHD6asrCxth9Rl+ngsq1atIiMjI3J1dVWYWkkb+Dcr9XW1rL2dnR3l5ubSjRs3qKGhQWH52rVradq0aVRdXS1vk0ql1K9fPwJAx44dU9pmbm4uBQYGdu0gepm/vz9t27aNysrKqKamhjIzM0koFNJbb72l0C8lJYUmTZqkMO1aa2sr3b17l7777juaOnWqRsvac7JiTE/oQrKqr68nHx8fvdlHV5OVs7Nzu8s2bdpEQ4cOJZlMptAulUpp//79ZGRkRM7OzvT48WOF5fqUrIKDg5WOLywsjABQaWmpQntUVBT5+PhQU1OT0nYWLVqk0WTVZ24DMsa6rzdKsuhq2ZebN29izZo1WLdunXwyhKf5+voiOjoa9+7dw7Jly7QQoWYcPnxY6ficnZ0BALW1tQrtcXFxuHTpUocP8moSJyvGDBgRISkpST5xsK2tLYKCghTmK+xOSRZ9KGGjKampqSAiBAQEdNhnw4YNGDp0KPbs2YNTp051uj1Vro06pZN6sjxSYWEhbGxsFGZtAX57WH7SpElISUnp+RGXanwNY4xpUVduA65du5ZMTU3piy++oMePH9OVK1dozJgx1L9/f7p//768X3dKsuhaCZunafI2oLu7Ow0fPrzddaRSKd26dYuIiL7//nsyMjKiwYMHU21tLRG1fxtQ1WujavkdTZdHamxspLt379L27dvJzMysw2cDY2JilMpBEfFtQMaYimQyGZKSkjB9+nTMnj0b1tbW8Pb2xq5du1BRUaE0y0x36EsJm66qq6vDrVu3IJVKn9vXx8cHixcvRnFxMVauXNlun65cm87K7/REeSRXV1e4uLggLi4OW7ZsQXh4eLv9hgwZAuC3qct6EicrxgxUQUEBamtrMXbsWIX2cePGwdTUtMNppTRB18q+dFdZWRmICGKxWKX+GzZswLBhw7Bz506cPXtWaXl3r82z5Xd6okTNnTt3UFZWhi+//BJ//etf8dJLL7X7W2LbOXnw4EGX9qMqTlaMGajHjx8DACwtLZWW2djYoKampkf3b0glbBoaGgD8dkyqEIlESE9Ph0AgwNy5cyGTyRSWa/raPF2i5unnnEpKSlBfX6/WttoIhULY29vDz88PBw8eREFBARISEpT6mZubA/j9HPUUTlaMGSgbGxsAaPcPX0+XZDGkEjbA73+Q1Zl9xcfHB0uWLEFhYaHSw+SavjZPl6ghIoXXuXPn1NpWezw8PGBsbIyCggKlZY2NjQB+P0c9hZMVYwZqxIgRsLS0xI8//qjQfuHCBTQ2NuLll1+Wt2m6JIshlbABAAcHBwgEAlRVVam13vr16+Hp6Yn8/HyFdnWujSo0VaLm4cOHmDlzplJ7YWEhWlpa4OrqqrSs7Zw4Ojp2a9/Pw8mKMQMlEomwdOlSHD58GPv27UN1dTWuXr2KBQsWwMnJCZGRkfK+3S3JYkglbNojFovh7u4ur4yuqrbbgcbGxkrtql4bVffzvBI1ERERcHR07HS6JwsLC5w8eRKnT5+WV83Oz8/HBx98AAsLCyxZskRpnbZz4u3trVbMalNj6CBjTIu6MnS9tbWVtm7dSkOGDCGhUEi2trYUHBxMN27cUOjXnfIyulLCpj2aHLoeFRVFQqGQ6uvr5W2HDx+Wl6Tp378/LVy4sN1tLl++XGnouirXRp3yO88rjxQcHEwAaO3atZ0ef0BAALm5uZGlpSWZmZmRVCqliIgIunr1arv9/f39ydnZmVpbWxXaNT10nZMVY3pCF6Zbak/bXHq6SJPJqrCwkExMTLpUi0wXtLS00MSJE2nv3r0a22ZFRQWJRCLatm2b0jJ+zooxpnP0qeyLKmQyGb755hsUFhbKBxB4eHggPj4e8fHxStMO6bqWlhbk5OSgpqZGo5Ug4uLiMHr0aERFRQH4bVaO0tJSnD17Fjdv3tTYfgD+zYoxxpQ8evQIb7/9NoYOHYq5c+fK22NiYhAWFoaIiAi1B1toU15eHg4dOoTc3FyVnxV7nqSkJFy6dAknTpyAUCgEABw5cgTOzs6YOHEijh8/rpH9tOFkxRjrstjYWKSnp6Oqqgpubm7Izs7WdkjdtmvXLoWh3/v27VNYvnHjRkRFRWHTpk1ailB9U6ZMwf79+xXmZuyOI0eO4MmTJ8jLy4Otra28PSgoSOHctc35qAkmGtsSY6zPSUhIaPdBUUPn5+cHPz8/bYehNYGBgQgMDOzVffI3K8YYYzqPkxVjjDGdx8mKMcaYzuNkxRhjTOd1OMAiMzOzN+NgjD1H27Q2/NlUXdskrnzO9ENnk+4KiBRrEWdmZnZYZIsxxhjrac+kJQDIUkpWjLGua/vPHn+sGNOoLP7NijHGmM7jZMUYY0zncbJijDGm8zhZMcYY03mcrBhjjOk8TlaMMcZ0HicrxhhjOo+TFWOMMZ3HyYoxxpjO42TFGGNM53GyYowxpvM4WTHGGNN5nKwYY4zpPE5WjDHGdB4nK8YYYzqPkxVjjDGdx8mKMcaYzuNkxRhjTOdxsmKMMabzOFkxxhjTeZysGGOM6TxOVowxxnQeJyvGGGM6j5MVY4wxncfJijHGmM7jZMUYY0zncbJijDGm8zhZMcYY03mcrBhjjOk8TlaMMcZ0HicrxhhjOo+TFWOMMZ3HyYoxxpjOM9F2AIzpq7t37+KDDz5AS0uLvK2yshISiQSTJ09W6Dts2DD87//+by9HyJjh4GTFWBe5uLigpKQERUVFSsvOnDmj8O/XX3+9t8JizCDxbUDGuuH999+HUCh8br+IiIheiIYxw8XJirFumDVrFpqbmzvt4+XlheHDh/dSRIwZJk5WjHWDVCrFyJEjIRAI2l0uFArxwQcf9HJUjBkeTlaMddP7778PY2Pjdpc1NzcjLCyslyNizPBwsmKsm2bMmIHW1laldiMjI4wfPx6DBw/u/aAYMzCcrBjrJicnJ7z22mswMlL8OBkZGeH999/XUlSMGRZOVoxpwHvvvafURkSYPn26FqJhzPBwsmJMA0JDQxV+tzI2Nsabb74JBwcHLUbFmOHgZMWYBtja2uKtt96SJywiwuzZs7UcFWOGg5MVYxoye/Zs+UALoVCIoKAgLUfEmOHgZMWYhgQEBMDMzAwAMG3aNFhaWmo5IsYMBycrxjTEwsJC/m2KbwEyplkCIiJtB9FdYWFhyM7O1nYYjDGmczIyMvDuu+9qO4zuyjKYWdfHjx+PxYsXazsMpmfCw8MRHR0NHx8fjWyvpaUFGRkZmDlzpka2p4uSk5MBgD9veiA8PFzbIWiMwSQrFxcXQ/jfA+tl4eHh8PHx0eh7Jzg4GCKRSGPb0zVZWVkAwJ83PWBIyYp/s2JMwww5UTGmLZysGGOM6TxOVowxxnQeJyvGGGM6j5MVY4wxncfJijENOHHiBKytrfHVV19pOxS9dOrUKcTExODQoUNwd3eHQCCAQCBodzZ7Pz8/SCQSGBsbw8vLCxcvXtRCxKpLTEyEp6cnzM3NYWFhAU9PT6xZswbV1dXyPkePHkViYiJaWlq0GKlu42TFmAYYwLP1WvPZZ58hNTUVsbGxCAkJwS+//AKpVIp+/fph3759OH78uEL/kydPIisrC9OmTUNBQQHGjBmjpchV849//APz5s3D7du38eDBA6xfvx6JiYkIDQ2V9wkICIBIJMKUKVPw+PFjLUaruzhZMaYB/v7+qKqqwrRp07QdCmQyGXx9fbUdhko2b96MgwcPIjMzExKJRGFZamoqjIyMEBkZiaqqKi1F2H2mpqb45JNPYG9vD0tLS4SFhSEoKAh///vf8euvv8r7LVq0CKNGjcLUqVPR3NysxYh1EycrxgzM3r17UVZWpu0wnuvmzZtYs2YN1q1b1+6zab6+voiOjsa9e/ewbNkyLUSoGYcPH1Y6PmdnZwBAbW2tQntcXBwuXbqElJSUXotPX3CyYqybzp49i4EDB0IgEGDHjh0AgLS0NFhYWEAsFuPIkSN45513YGVlBRcXFxw4cEC+bmpqKkQiERwcHDB//nw4OTlBJBLB19cXFy5ckPeLioqCqakpBgwYIG/75JNPYGFhAYFAgIqKCgBAdHQ0li5diqKiIggEAnh4eAAAvv76a1hZWWHjxo29cUpUkpqaCiJCQEBAh302bNiAoUOHYs+ePTh16lSn2yMiJCUl4cUXX4SZmRlsbW0RFBSE69evy/uoel2A36bOWrt2LQYOHAhzc3OMHDkSGRkZ3Tvo/ygsLISNjQ0GDRqk0G5ra4tJkyYhJSWFby0/iwxAaGgohYaGajsMpocAUEZGRre3c+fOHQJA27dvl7etWrWKANC3335LVVVVVFZWRhMnTiQLCwtqbGyU94uMjCQLCwu6du0aNTQ0UEFBAY0bN44kEgndvn1b3m/WrFnk6OiosN+tW7cSACovL5e3hYSEkFQqVeh37NgxkkgkFB8f3+1j1dTnzd3dnYYPH97uMqlUSrdu3SIiou+//56MjIxo8ODBVFtbS0REubm5FDlngkMAACAASURBVBgYqLDO2rVrydTUlL744gt6/PgxXblyhcaMGUP9+/en+/fvy/upel2WLVtGZmZmlJ2dTZWVlRQbG0tGRkb0ww8/dOl4Gxsb6e7du7R9+3YyMzOjL774ot1+MTExBIDy8/O7tJ+naer9rQMy+ZsVYz3M19cXVlZWsLe3R0REBOrq6nD79m2FPiYmJvJvBMOHD0daWhpqamqQnp6ukRj8/f1RXV2NNWvWaGR73VVXV4dbt25BKpU+t6+Pjw8WL16M4uJirFy5st0+MpkMSUlJmD59OmbPng1ra2t4e3tj165dqKiowO7du5XW6ey6NDQ0IC0tDcHBwQgJCYGNjQ1Wr14NoVDY5Wvi6uoKFxcXxMXFYcuWLR3O2zdkyBAAwNWrV7u0H0PFyYqxXmRqagoAaGpq6rTf2LFjIRaLFW5hGZKysjIQEcRisUr9N2zYgGHDhmHnzp04e/as0vKCggLU1tZi7NixCu3jxo2Dqampwi3V9jx7XW7cuIH6+nqMGDFC3sfc3BwDBgzo8jW5c+cOysrK8OWXX+Kvf/0rXnrppXZ/W2w7Jw8ePOjSfgwVJyvGdJSZmRnKy8u1HUaPaGhoAAB5ZeXnEYlESE9Ph0AgwNy5cyGTyRSWtw33bq86s42NDWpqatSKr66uDgCwevVq+TNfAoEAJSUlqK+vV2tbbYRCIezt7eHn54eDBw+ioKAACQkJSv3Mzc0B/H6O2G84WTGmg5qamvD48WO4uLhoO5Qe0fYHWZ2HYH18fLBkyRIUFhZi/fr1CstsbGwAoN2k1JXzaG9vD+C32l1EpPA6d+6cWttqj4eHB4yNjVFQUKC0rLGxEcDv54j9hpMVYzooLy8PRITx48fL20xMTJ57+1BfODg4QCAQqP381Pr16+Hp6Yn8/HyF9hEjRsDS0hI//vijQvuFCxfQ2NiIl19+Wa39uLq6QiQS4dKlS2qt96yHDx+2W4izsLAQLS0tcHV1VVrWdk4cHR27tW9Dw8mKMR3Q2tqKyspKNDc348qVK4iOjsbAgQMxZ84ceR8PDw88evQIOTk5aGpqQnl5OUpKSpS2ZWdnh9LSUhQXF6OmpgZNTU3Izc3VqaHrYrEY7u7uuHv3rlrrtd0ONDY2VmpfunQpDh8+jH379qG6uhpXr17FggUL4OTkhMjISLX38+GHH+LAgQNIS0tDdXU1WlpacPfuXfmDvBEREXB0dOx0uicLCwucPHkSp0+fRnV1NZqampCfn48PPvgAFhYWWLJkidI6befE29tbrZgNnjbHImoKD11nXQUNDO3dvn07DRgwgACQWCymgIAA2rlzJ4nFYgJAQ4YMoaKiItq9ezdZWVkRABo0aBD9/PPPRPTb0HWhUEjOzs5kYmJCVlZWFBQUREVFRQr7efjwIb3xxhskEonIzc2NPv30U1q+fDkBIA8PD/kw94sXL9KgQYPI3NycJkyYQPfv36cTJ06QRCKhDRs2dOtYiTT3eYuKiiKhUEj19fXytsOHD5NUKiUA1L9/f1q4cGG76y5fvlxp6Hpraytt3bqVhgwZQkKhkGxtbSk4OJhu3Lgh76POdXny5AmtWLGCBg4cSCYmJmRvb08hISFUUFBARETBwcEEgNauXdvpcQYEBJCbmxtZWlqSmZkZSaVSioiIoKtXr7bb39/fn5ydnam1tfX5J/E5NPH+1hGZnKxYn6YLH+bIyEiys7PTagzq0NTnrbCwkExMTDp83kjXtbS00MSJE2nv3r0a22ZFRQWJRCLatm2bRranC+9vDeHnrBjTBX1xtm0PDw/Ex8cjPj5eadohXdfS0oKcnBzU1NQgIiJCY9uNi4vD6NGjERUVpbFtGgpOVv/x0UcfQSKRQCAQdPtHVW1rbW1FcnJytyYzfbZUQ9vL1NQUDg4OmDx5MrZu3YrKykoNRs76mpiYGISFhSEiIkKvJqvNy8vDoUOHkJubq/KzYs+TlJSES5cu4cSJExAKhRrZpiHhZPUfe/bsweeff67tMLqtsLAQr7/+OpYsWdLl50EAKJRqsLa2BhGhtbUVZWVlyMzMhJubG1asWAEvLy+lEVhMdbGxsUhPT0dVVRXc3NyQnZ2t7ZB63caNGxEVFYVNmzZpOxSVTZkyBfv371eYq7E7jhw5gidPniAvLw+2trYa2aahMdF2AExzLl++jPj4eCxYsAB1dXUanwhTIBDAxsYGkydPxuTJk+Hv74/w8HD4+/vj559/hrW1tUb31xckJCS0+2BoX+Pn5wc/Pz9th6E1gYGBCAwM1HYYOo2/WT1FIBBoO4RuGTVqFA4dOoRZs2apPDNAd4SGhmLOnDkoKyvDrl27enx/jLG+q88mKyLC1q1bMWzYMJiZmcHa2hrLly9X6tdZmQB1yg2cOXMGr7zyCsRiMaysrODt7S0va92TpQjao8lyEW3PAeXm5srbDPGcMca0q88mqzVr1mDFihWIjIzEgwcPcP/+/XZndF65ciW2bNmC5ORk/Prrr5g2bRpmzpyJH3/8ER9//DEWL14MmUwGiUSCjIwMFBUVwd3dHfPmzZPPNlBXV4eAgACEhobi0aNHKCwsxNChQ+XTqnS2j57QNvKstbW129saPXo0AOCXX36RtxniOWOMaZmWx85rhLrPfdTX15NYLKa33npLof3AgQMKdWRkMhmJxWKKiIhQWNfMzIw+/vhjIvq9No5MJpP32blzJwGgmzdvEhHRv//9bwJAx44dU4pFlX10xauvvkqjRo3q8vptpFIpWVtbd9pHIBCQjY0NEenfOYPhPIfSa/i5Rv1hQO/vzD45wOLmzZuor6/HlClTOu3X1TIBz5YbcHd3h4ODA2bPno1FixZhzpw5GDx4cLf2oSvaBnJYWVkB0M9zpomJSfuStumAMjMztRwJ61O0nS41Qd3/6Z04cYIAKD15/uw3q3/+858EoN3X+PHjiaj9bwmff/45AaCffvpJ3vbvf/+b/vCHP5CJiQkJBAIKDw+n+vp6lfbRFb31zerixYsEgPz8/IhI/85ZR9vhF78M5WUo36z65G9WIpEIAPDkyZNO+2myTICXlxe++uorlJaWYsWKFcjIyMC2bdt6vBRBT/v6668BAO+88w4A/TxnGRkZStvhV8ev0NBQhIaGaj0Ofj3/ZUj6ZLIaMWIEjIyMcObMmU77aapMQGlpKa5duwbgtz/mmzZtwpgxY3Dt2jWN7UMb7t+/j+TkZLi4uGDu3LkA+JwxxnpGn0xW9vb2CAkJQXZ2Nvbu3Yvq6mpcuXIFu3fvVuinSpkAVZSWlmL+/Pm4fv06GhsbkZ+fj5KSEowfP15j+1CHuuUiiAi1tbVobW0FEaG8vBwZGRl47bXXYGxsjJycHPlvVoZ6zhhjWkYGoCujk2pqauijjz6ifv36kaWlJU2YMIHWrl1LAMjFxYUuX75MRJ2XCVC13EBxcTH5+vqSra0tGRsb0wsvvECrVq2i5ubm5+5DHefOnaPXXnuNnJyc5PerBwwYQL6+vnTmzBl5P1XKRRw9epRGjhxJYrGYTE1NycjIiADIR/698sorFB8fTw8fPlRaV5/OGQznnn6v4dGA+sOA3t+ZAiL9v7EZFhYGAMjKytJyJEzfCAQCZGRk4N1339V2KHqDP2/6w4De31l98jYgY4wx/cLJSoddv35dqURHey9N1tNhjDFdxMlKh3l6eqo0PPXgwYPaDpUxlZ06dQoxMTFKNdPee+89pb5+fn6QSCQwNjaGl5cXLl68qIWI1adKTbmmpiYkJCTAw8MDpqamsLGxwYgRI1BcXAwAOHr0KBITE/tkYc72cLJijPWazz77DKmpqYiNjVWomdavXz/s27cPx48fV+h/8uRJZGVlYdq0aSgoKMCYMWO0FLnqVK0pFx4ejr/97W/Yv38/6uvr8dNPP0EqlcqrJgcEBEAkEmHKlCl4/Phxb4WvszhZMaZFMpmsWxWddWUfqti8eTMOHjyIzMxMSCQShWWpqakwMjJCZGSkXlUMftbly5excuVKLFiwQD7Jc3sOHjyInJwcZGVl4dVXX4WJiQmcnJxw5MgRhWnEFi1ahFGjRmHq1Klobm7ujUPQWZysGNOivXv3oqysTO/38Tw3b97EmjVrsG7dOvkMMk/z9fVFdHQ07t27h2XLlmkhQs1Qtabcn/70J4wZMwbe3t7P3WZcXBwuXbqElJQUTYaqdzhZMaYGIkJSUhJefPFFmJmZwdbWFkFBQQoT6EZFRcHU1FSh5Pknn3wCCwsLCAQCVFRUAACio6OxdOlSFBUVQSAQwMPDA6mpqRCJRHBwcMD8+fPh5OQEkUgEX19fXLhwQSP7ADRb00wVqampICIEBAR02GfDhg0YOnQo9uzZg1OnTnW6PVWugzq103qzPlpjYyPOnz/f6Tevp9na2mLSpElISUkxuCmU1KKFh7s0jh9SZF0FNR+aXLt2LZmamtIXX3xBjx8/pitXrtCYMWOof//+dP/+fXm/WbNmkaOjo8K6W7duJQBUXl4ubwsJCSGpVKrQLzIykiwsLOjatWvU0NBABQUFNG7cOJJIJHT79m2N7OPYsWMkkUgoPj5e5WNv05XPm7u7Ow0fPrzdZVKplG7dukVERN9//z0ZGRnR4MGDqba2loiIcnNzKTAwUGEdVa9D26TJ3377LVVVVVFZWRlNnDiRLCwsqLGxUd5v2bJlZGZmRtnZ2VRZWUmxsbFkZGREP/zwg1rH+bSOJpO+desWAaDRo0fT5MmTacCAAWRmZkaenp60Y8cOam1tVVonJiaGgN8n2VaVuu9vHdY3J7JlrCtkMhmSkpIwffp0zJ49G9bW1vD29sauXbtQUVGhNF1Xd5iYmMi/NQwfPhxpaWmoqalBenq6Rrbv7++P6upqrFmzRiPb60xdXR1u3boFqVT63L4+Pj5YvHgxiouL2y2GCnTtOvj6+sLKygr29vaIiIhAXV0dbt++DQBoaGhAWloagoODERISAhsbG6xevRpCoVBj5/tpbQMo7O3tsXHjRhQUFODBgwcICgrCwoUL8eWXXyqtM2TIEADA1atXNR6PvuBkxZiKCgoKUFtbi7Fjxyq0jxs3Dqampgq36TRt7NixEIvFelHj7FllZWUgIojFYpX6b9iwAcOGDcPOnTtx9uxZpeXdvQ7P1k7r7Zpybb9leXl5wdfXF3Z2drC2tsa6detgbW3dbrJtO3cPHjzQeDz6gpMVYypqGz5saWmptMzGxgY1NTU9un8zMzOUl5f36D56QkNDAwB0OuDgaSKRCOnp6RAIBJg7dy5kMpnCck1fh7q6OgDA6tWrFR62Lykp6XToeVc5OTkBgPx3xTampqYYNGgQioqKlNYxNzcH8Pu57Is4WTGmIhsbGwBo94/h48eP4eLi0mP7bmpq6vF99JS2P7TqPNzq4+ODJUuWoLCwEOvXr1dYpunr0Ns15SwtLTFkyBB5CZynNTc3w9raWqm9sbERwO/nsi/iZMWYikaMGAFLS0v8+OOPCu0XLlxAY2MjXn75ZXmbiYmJ/DaTJuTl5YGIMH78+B7bR09xcHCAQCBQ+/mp9evXw9PTE/n5+Qrt6lwHVWijPlp4eDjy8/Pxyy+/yNvq6+tRUlLS7nD2tnPn6OjYazHqGk5WjKlIJBJh6dKlOHz4MPbt24fq6mpcvXoVCxYsgJOTEyIjI+V9PTw88OjRI+Tk5KCpqQnl5eUoKSlR2qadnR1KS0tRXFyMmpoaefJpbW1FZWUlmpubceXKFURHR2PgwIGYM2eORvahbk2z7hCLxXB3d8fdu3fVWq/tdqCxsbFSu6rXQdX9PK8+WkREBBwdHTU23dOSJUswaNAgzJkzB7dv38bDhw+xYsUKyGSydgeWtJ07VZ7LMljaHIuoKTx0nXUV1Bza29raSlu3bqUhQ4aQUCgkW1tbCg4Ophs3bij0e/jwIb3xxhskEonIzc2NPv30U1q+fDkBIA8PD/kQ9IsXL9KgQYPI3NycJkyYQPfv36fIyEgSCoXk7OxMJiYmZGVlRUFBQVRUVKSxfahS06wjXfm8RUVFkVAopPr6ennb4cOHSSqVEgDq378/LVy4sN11ly9frjR0XZXroGrtNKLn10cLDg4mALR27dpOj1PVmnJERHfu3KEZM2aQra0tmZmZ0SuvvEK5ubntbtff35+cnZ3bHdbeGXXf3zosk5MV69N08cMcGRlJdnZ22g6jQ135vBUWFpKJiQl98cUXPRRVz2ppaaGJEyfS3r17e33fFRUVJBKJaNu2bWqvq4vv7y7i56wY00WGNtO2h4cH4uPjER8fL3/OSF+0tLQgJycHNTU1WinHExcXh9GjRyMqKqrX961LOFkxxnpFTEwMwsLCEBERoVeT1ebl5eHQoUPIzc1V+VkxTUlKSsKlS5dw4sQJCIXCXt23ruFkxZgOiY2NRXp6OqqqquDm5obs7Gxth6RRGzduRFRUFDZt2qTtUFQ2ZcoU7N+/X2Eext5w5MgRPHnyBHl5ebC1te3VfesiE20HwBj7XUJCAhISErQdRo/y8/ODn5+ftsPQeYGBgQgMDNR2GDqDv1kxxhjTeZysGGOM6TxOVowxxnQeJyvGGGM6z2AGWJw/fx5hYWHaDoPpoeTkZGRlZWk7DL1x/vx5AODPG+tVBpGsfHx8tB0C01OhoaEa3d79+/eRn5+Pd955R6Pb1SVPT6bLdFtoaChcXV21HYZGCIiItB0EY4YiMzMT4eHh4I8VYxqVxb9ZMcYY03mcrBhjjOk8TlaMMcZ0HicrxhhjOo+TFWOMMZ3HyYoxxpjO42TFGGNM53GyYowxpvM4WTHGGNN5nKwYY4zpPE5WjDHGdB4nK8YYYzqPkxVjjDGdx8mKMcaYzuNkxRhjTOdxsmKMMabzOFkxxhjTeZysGGOM6TxOVowxxnQeJyvGGGM6j5MVY4wxncfJijHGmM7jZMUYY0zncbJijDGm8zhZMcYY03mcrBhjjOk8TlaMMcZ0HicrxhhjOo+TFWOMMZ3HyYoxxpjO42TFGGNM53GyYowxpvNMtB0AY/qqqakJtbW1Cm11dXUAgMrKSoV2gUAAGxubXouNMUPDyYqxLnr06BGcnZ3R0tKitMzOzk7h32+88QZOnz7dW6ExZnD4NiBjXeTo6IjXX38dRkadf4wEAgFmzJjRS1ExZpg4WTHWDe+9995z+xgbG2P69Om9EA1jhouTFWPdEBISAhOTju+mGxsb4+2330a/fv16MSrGDA8nK8a6wcrKCu+8806HCYuIMHv27F6OijHDw8mKsW6aPXt2u4MsAMDU1BR/+MMfejkixgwPJyvGuukPf/gDxGKxUrtQKERwcDAsLCy0EBVjhoWTFWPdJBKJMH36dAiFQoX2pqYmzJo1S0tRMWZYOFkxpgEzZ85EU1OTQpuVlRXeeustLUXEmGHhZMWYBrz55psKDwILhULMmDEDpqamWoyKMcPByYoxDTAxMcGMGTPktwKbmpowc+ZMLUfFmOHgZMWYhsyYMUN+K9DR0RETJkzQckSMGQ5OVoxpiK+vL5ydnQEA77///nOnYWKMqc4gJrI9d+4c7ty5o+0wGMO4ceNw79499OvXD5mZmdoOhzH4+vrCxcVF22F0m4CISNtBdFdYWBiys7O1HQZjjOmcjIwMvPvuu9oOo7uyDOKbFQCEhoYiKytL22EwPSMQCDT+Yc7OzkZoaKjGtqdrwsLCAIA/b3pAIBBoOwSN4ZvqjGmYIScqxrSFkxVjjDGdx8mKMcaYzuNkxRhjTOdxsmKMMabzOFkxxhjTeZysGNOAEydOwNraGl999ZW2Q9F5p06dQkxMDA4dOgR3d3cIBAIIBAK89957Sn39/PwgkUhgbGwMLy8vXLx4UQsRq6+1tRXJycnw9fXtsE9TUxMSEhLg4eEBU1NT2NjYYMSIESguLgYAHD16FImJiR0W9uxrOFkxpgEG8Gx9r/jss8+QmpqK2NhYhISE4JdffoFUKkW/fv2wb98+HD9+XKH/yZMnkZWVhWnTpqGgoABjxozRUuSqKywsxOuvv44lS5agvr6+w37h4eH429/+hv3796O+vh4//fQTpFIpamtrAQABAQEQiUSYMmUKHj9+3Fvh6yxOVoxpgL+/P6qqqjBt2jRthwKZTNbp/+i1ZfPmzTh48CAyMzMhkUgUlqWmpsLIyAiRkZGoqqrSUoTdd/nyZaxcuRILFizA6NGjO+x38OBB5OTkICsrC6+++ipMTEzg5OSEI0eOYMSIEfJ+ixYtwqhRozB16lQ0Nzf3xiHoLE5WjBmYvXv3oqysTNthKLh58ybWrFmDdevWQSQSKS339fVFdHQ07t27h2XLlmkhQs0YNWoUDh06hFmzZsHMzKzDfn/6058wZswYeHt7P3ebcXFxuHTpElJSUjQZqt7hZMVYN509exYDBw6EQCDAjh07AABpaWmwsLCAWCzGkSNH8M4778DKygouLi44cOCAfN3U1FSIRCI4ODhg/vz5cHJygkgkgq+vLy5cuCDvFxUVBVNTUwwYMEDe9sknn8DCwgICgQAVFRUAgOjoaCxduhRFRUUQCATw8PAAAHz99dewsrLCxo0be+OUKElNTQURISAgoMM+GzZswNChQ7Fnzx6cOnWq0+0REZKSkvDiiy/CzMwMtra2CAoKwvXr1+V9VL0GANDS0oK1a9di4MCBMDc3x8iRI5GRkdG9g+5AY2Mjzp8/3+k3r6fZ2tpi0qRJSElJ6dO3mzlZMdZNEyZMwPfff6/Q9vHHH2Px4sWQyWSQSCTIyMhAUVER3N3dMW/ePHndq6ioKMyZMwf19fVYtGgRiouLcfHiRTQ3N+Ott96SVxNITU1Vmr9w586dWLdunUJbSkoKpk2bBqlUCiLCzZs3AUD+I31ra2uPnIPnOX78OIYNGwaxWNxhH3Nzc/zlL3+BkZER5s2bh7q6ug77xsXFISYmBqtWrUJZWRm+++473LlzBxMnTsSDBw8AqH4NAGDlypXYsmULkpOT8euvv2LatGmYOXMmfvzxR82dhP8oLS1FY2Mj/vWvf+GNN96Q/wflxRdfxM6dO9tNSC+99BLu3buHy5cvazwefcHJirEe5uvrCysrK9jb2yMiIgJ1dXW4ffu2Qh8TExP5t4Thw4cjLS0NNTU1SE9P10gM/v7+qK6uxpo1azSyPXXU1dXh1q1bkEqlz+3r4+ODxYsXo7i4GCtXrmy3j0wmQ1JSEqZPn47Zs2fD2toa3t7e2LVrFyoqKrB7926ldTq7Bg0NDUhLS0NwcDBCQkJgY2OD1atXQygUauz8P61tAIW9vT02btyIgoICPHjwAEFBQVi4cCG+/PJLpXWGDBkCALh69arG49EXnKwY60WmpqYAoPC/+vaMHTsWYrFY4baWviorKwMRdfqt6mkbNmzAsGHDsHPnTpw9e1ZpeUFBAWprazF27FiF9nHjxsHU1FTh9ml7nr0GN27cQH19vcLABnNzcwwYMKBHzn/bb1leXl7w9fWFnZ0drK2tsW7dOlhbW7ebbNvOXdu3xr6IkxVjOsrMzAzl5eXaDqPbGhoaAKDTAQdPE4lESE9Ph0AgwNy5cyGTyRSWtw3jtrS0VFrXxsYGNTU1asXXdrtx9erV8me+BAIBSkpKOh163lVOTk4AIP+dsY2pqSkGDRqEoqIipXXMzc0B/H4u+yJOVozpoKamJjx+/NggKry2/aFV5+FWHx8fLFmyBIWFhVi/fr3CMhsbGwBoNyl15ZzZ29sDAJKTk0FECq9z586ptS1VWFpaYsiQIbh27ZrSsubmZlhbWyu1NzY2Avj9XPZFnKwY00F5eXkgIowfP17eZmJi8tzbh7rIwcEBAoFA7een1q9fD09PT+Tn5yu0jxgxApaWlkqDHy5cuIDGxka8/PLLau3H1dUVIpEIly5dUmu97ggPD0d+fj5++eUXeVt9fT1KSkraHc7edu4cHR17LUZdw8mKMR3Q2tqKyspKNDc348qVK4iOjsbAgQMxZ84ceR8PDw88evQIOTk5aGpqQnl5OUpKSpS2ZWdnh9LSUhQXF6OmpgZNTU3Izc3V2tB1sVgMd3d33L17V6312m4HGhsbK7UvXboUhw8fxr59+1BdXY2rV69iwYIFcHJyQmRkpNr7+fDDD3HgwAGkpaWhuroaLS0tuHv3Ln799VcAQEREBBwdHTU23dOSJUswaNAgzJkzB7dv38bDhw+xYsUKyGSydgeWtJ07VZ7LMlhkAEJDQyk0NFTbYTA9BIAyMjK6tY3t27fTgAEDCACJxWIKCAignTt3klgsJgA0ZMgQKioqot27d5OVlRUBoEGDBtHPP/9MRESRkZEkFArJ2dmZTExMyMrKioKCgqioqEhhPw8fPqQ33niDRCIRubm50aeffkrLly8nAOTh4UG3b98mIqKLFy/SoEGDyNzcnCZMmED379+nEydOkEQioQ0bNnTrWIm69nmLiooioVBI9fX18rbDhw+TVColANS/f39auHBhu+suX76cAgMDFdpaW1tp69atNGTIEBIKhWRra0vBwcF048YNeR91rsGTJ09oxYoVNHDgQDIxMSF7e3sKCQmhgoICIiIKDg4mALR27dpOj/PcuXP02muvkZOTEwEgADRgwADy9fWlM2fOKPS9c+cOzZgxg2xtbcnMzIxeeeUVys3NbXe7/v7+5OzsTK2trZ3u/1maeH/riExOVqxP04UPc2RkJNnZ2Wk1BnV05fNWWFhIJiYm9MUXX/RQVD2rpaWFJk6cSHv37u31fVdUVJBIJKJt27apva4uvL81JJNvAzKmAwx9Zm0PDw/Ex8cjPj5e/pyRvmhpaUFOTg5qamoQERHR6/uPi4vD6NGjERUV1ev71iWcrP7jo48+gkQigUAg6NUfWjUpPj4ew4cPh5WVFczMzODh4YH/+Z//6dIfh2fLN7S9TE1N4eDggMmTJ2Pr1q2orKzsgSNhhigmJgZhYWGIiIjQq8lq80Gg+QAAIABJREFU8/LycOjQIeTm5qr8rJimJCUl4dKlSzhx4gSEQmGv7lvXcLL6jz179uDzzz/Xdhjdcvr0aSxcuBDFxcWoqKhAQkICUlJSEBYWpva2ni7fYG1tDSJCa2srysrKkJmZCTc3N6xYsQJeXl49MiVNXxEbG4v09HRUVVXBzc0N2dnZ2g6pR23cuBFRUVHYtGmTtkNR2ZQpU7B//36FeRl7w5EjR/DkyRPk5eXB1ta2V/eti0y0HQDTHEtLS0RGRspHT7377rs4dOgQMjMzcefOHbi6unZr+wKBADY2Npg8eTImT54Mf39/hIeHw9/fHz///HO7z4ewziUkJCAhIUHbYfQqPz8/+Pn5aTsMnRcYGIjAwEBth6Ez+JvVUwQCgbZD6JZjx44pDfPt378/APTIk/ihoaGYM2cOysrKsGvXLo1vnzHG2vTZZEVE2Lp1K4YNGwYzMzNYW1tj+fLlSv06Kx2gTgmCM2fO4JVXXoFYLIaVlRW8vb1RXV393H10171792Bubg43Nzd5mybLRbQ9B5Sbmytv0/dzxhjTQdoej6gJXRlKu2rVKhIIBPTHP/6RKisrqb6+nnbu3EkAKD8/X95v2bJlZGZmRtnZ2VRZWUmxsbFkZGREP/zwg3w7AOjbb7+lqqoqKisro4kTJ5KFhQU1NjYSEVFtbS1ZWVlRYmIiyWQyun//Pk2fPp3Ky8tV2kdX1dXVkUQioaioKIX2Y8eOkUQiofj4+OduQyqVkrW1dYfLq6urCQC5urrK2/TpnMFwhvb2Gn5URH8Y0Pu7bz5nVV9fT2KxmN566y2F9gMHDigkK5lMRmKxmCIiIhTWNTMzo48//piIfv/DK5PJ5H3akt7NmzeJiOjf//43AaBjx44pxaLKPrpq1apVNHToUKquru7yNp6XrIiIBAIB2djYEJH+nTMD+jD3Gk5W+sOA3t+ZfXKAxc2bN1FfX48pU6Z02q+rpQOeLUHg7u4OBwcHzJ49G4sWLcKcOXMwePDgbu3jeQ4fPozMzEycPHkSEomky9t5nrq6OhARrKysAOjnOUtOTkZWVpba6/VV58+fB4AujTJlrKv65G9WbfNstc223BFNlQ4wNzfH6dOnMWHCBGzcuBHu7u6IiIiATCbrkfIEBw8exObNm5GXlyf/A99Tfv75ZwCAp6cnAP09Z4wx3dYnv1mJRCIAwJMnTzrt93TpgOjo6G7t08vLC1999RXKy8uRlJSEzZs3w8vLS/5EvCb2AQDbt2/HN998g9OnT7db70fTvv76awDAO++8A0A/z9nixYuVSsazjrV9o+Jvo7pP30c4P61PfrMaMWIEjIyMcObMmU77aap0QGlpqbx2jb29PTZt2oQxY8bg2rVrGtsHEWHFihW4evUqcnJyeiVR3b9/H8nJyXBxccHcuXMB6Nc5Y4zpjz6ZrOzt7RESEoLs7Gzs3bsX1dXVuHLlilI5aVVKB6iitLQU8+fPx/Xr19HY2Ij8/HyUlJRg/PjxGtvHtWvXsGXLFnz++ecQCoVK0yRt27ZN3lfdchFEhNraWrS2toKIUF5ejoyMDLz22mswNjZGTk6O/DcrfTpnjDE9ot0BHprRldFJNTU19NFHH1G/fv3I0tKSJkyYQGvXriUA5OLiQpcvXyaizksHqFqCoLi4mHx9fcnW1paMjY3phRdeoFWrVlFzc/Nz96Gqq1evyksStPfaunWrvK8q5SKOHj1KI0eOJLFYTKampmRkZEQA5CP/XnnlFYqPj6eHDx8qrasv54zIoEZL9RoeDag/DOj9nSkgItJCjtQovofOukogECAjI4N/s1IDf970hwG9v7P65G1Axhhj+oWTlQ67fv260m9P7b20UWOHsa46deoUYmJilMrQvPfee0p9/fz8IJFIYGxsDC8vL42Vle9pra2tSE5Ohq+vb4d9mpqakJCQAA8PD5iamsLGxgYjRoxAcXExAODo0aNITEw0+FpnquJkpcM8PT1B/7+9Ow+K6sr+AP5t6KabZkcBiUYEGkUUdYgbnTjGsUJNpGQRFVwyEispNBpE1FGMKyBqdJAhkVhGi1TURAENZpGM4ziYsYJWHFEZzILI4hIFVGTfz++P/OjYNmI3NPRrOJ8q//C9++49vNfNoV/f+w7Rc/8dPXrU0KEyppXNmzcjOTkZ69evVytDM2DAABw+fBjffPONWvvTp08jPT0dM2fORH5+Pnx8fAwUufYKCgrwxz/+EdHR0Z2u+wsNDcWnn36KI0eOoK6uDj/++CPc3d1V9ecCAgIgk8kwffp0VFZW9lb4gsXJijEDqq+v7/Svb2MZQxs7duzA0aNHkZaWpvFUleTkZJiYmCAiIsKoCjM+7erVq1i3bh2WLl2KcePGPbPd0aNHkZmZifT0dEyaNAlisRjOzs44efKk2pNZVqxYgbFjx2LGjBloaWnpjR9BsDhZMWZABw8eRFlZmdGP8Tw3btzAxo0bsXXrVtWi/CcplUpERUXhzp07WL16tQEi1I+xY8fi+PHjWLBgAaRS6TPbffTRR/Dx8YG3t/dz+9yyZQuuXLmCpKQkfYZqdDhZMaYDIkJiYiJGjhwJqVQKOzs7BAUFqT2TMDIyEmZmZmqVZZctWwYLCwuIRCJUVFQAAKKiorBq1SoUFhZCJBJBoVAgOTkZMpkMjo6OWLJkCZydnSGTyaBUKnHx4kW9jAHot0yMNpKTk0FECAgIeGab+Ph4DB8+HAcOHMCZM2c67U+b66BLOZreLDnT1NSECxcudPrJ60l2dnaYOnUqkpKS0Acmb3edAebL6x2v+2BdBR3XoWzatInMzMzo0KFDVFlZSdeuXSMfHx8aOHAg3bt3T9VuwYIF5OTkpHbsrl27CICqzAkRUUhICLm7u6u1i4iIIAsLC7p+/To1NDRQfn4+TZgwgaysrKi0tFQvY+hSJuZpXXm/ubm5kZeXV4f73N3dqaioiIiIvv/+ezIxMaFhw4ZRTU0NERFlZWVRYGCg2jHaXgdtytEQ9UyZnkmTJtHYsWM1thcVFREAGjduHL366qs0aNAgkkql5OnpSR9++CG1tbVpHBMTE6NRvkgbur6+BSyNP1kxpqX6+nokJiZi1qxZWLhwIWxsbODt7Y19+/ahoqJC4wko3SEWi1WfGry8vJCSkoLq6mqkpqbqpX9/f39UVVVh48aNeumvM7W1tSgqKoK7u/tz2/r6+mLlypUoLi7GunXrOmzTleugVCphbW0NBwcHhIWFoba2FqWlpQCAhoYGpKSkIDg4GCEhIbC1tcWGDRsgkUj0dr6f1D6BwsHBAdu2bUN+fj7u37+PoKAgLF++HJ999pnGMR4eHgCAvLw8vcdjLDhZMaal/Px81NTUYPz48WrbJ0yYADMzM7XbdPo2fvx4yOXybpWNMZSysjIQEeRyuVbt4+PjMWLECOzduxfnz5/X2N/d6/B0OZqeKtPzLO3fZY0aNQpKpRL29vawsbHB1q1bYWNj02GybT939+/f13s8xoKTFWNaap8+3NFDgm1tbVFdXd2j40ulUpSXl/foGD2hoaEBADqdcPAkmUyG1NRUiEQiLF68GPX19Wr79X0dervkjLOzMwCovldsZ2ZmBhcXFxQWFmocY25uDuD3c9kfcbJiTEu2trYA0OEvw8rKSgwZMqTHxm5ubu7xMXpK+y9aXRa3+vr6Ijo6GgUFBYiLi1Pbp+/r8GRZG3pqDWNOTo5OfWnD0tISHh4eqqoCT2ppaYGNjY3G9qamJgC/n8v+iJMVY1oaPXo0LC0tcenSJbXtFy9eRFNTE1566SXVNrFYrLrNpA/Z2dkgIkyePLnHxugpjo6OEIlEOq+fiouLg6enJ3Jzc9W263IdtGGIkjOhoaHIzc3FzZs3Vdvq6upQUlLS4XT29nPn5OTUazEKDScrxrQkk8mwatUqnDhxAocPH0ZVVRXy8vKwdOlSODs7IyIiQtVWoVDg4cOHyMzMRHNzM8rLy1FSUqLRp729Pe7evYvi4mJUV1erkk9bWxsePXqElpYWXLt2DVFRURg6dCjCw8P1MoauZWK6Qy6Xw83NTVWhW1vttwNNTU01tmt7HbQd53klZ8LCwuDk5KS3xz1FR0fDxcUF4eHhKC0txYMHD7B27VrU19d3OLGk/dxpsy6rzzLkXER94anrrKug49TetrY22rVrF3l4eJBEIiE7OzsKDg6mn3/+Wa3dgwcPaNq0aSSTycjV1ZXeffddWrNmDQEghUKhmoJ++fJlcnFxIXNzc3rllVfo3r17FBERQRKJhAYPHkxisZisra0pKCiICgsL9TaGNmVinqUr77fIyEiSSCRUV1en2nbixAlyd3cnADRw4EBavnx5h8euWbNGY+q6NtdB23I0RM8vORMcHEwAaNOmTZ3+nDk5OfTyyy+Ts7OzqjzPoEGDSKlU0rlz59Ta3rp1i+bNm0d2dnYklUpp4sSJlJWV1WG//v7+NHjw4A6ntXdG19e3gKVxsmL9mhDfzBEREWRvb2/oMJ6pK++3goICEovFdOjQoR6Kqme1trbSlClT6ODBg70+dkVFBclkMtq9e7fOxwrx9d1FvM6KMSHqa0/aVigUiI2NRWxsrGqdkbFobW1FZmYmqqurDVLhYMuWLRg3bhwiIyN7fWwh4WTFGOsVMTExmDNnDsLCwozqYbXZ2dk4fvw4srKytF4rpi+JiYm4cuUKTp06BYlE0qtjCw0nK8YEZP369UhNTcXjx4/h6uqKjIwMQ4ekV9u2bUNkZCS2b99u6FC0Nn36dBw5ckTtOYy94eTJk2hsbER2djbs7Ox6dWwhEhs6AMbY7xISEpCQkGDoMHqUn58f/Pz8DB2G4AUGBiIwMNDQYQgGf7JijDEmeJysGGOMCR4nK8YYY4LHyYoxxpjgcbJijDEmeH1mNmBGRgZEIpGhw2BGKDQ0FKGhoYYOw+jw+431JhERkaGD6K6cnBzcunXL0GEwhpycHCQlJeHYsWOGDoUxAL9VSTbG0jJPSe8TyYoxoUhLS0NoaCj4bcWYXqXzd1aMMcYEj5MVY4wxweNkxRhjTPA4WTHGGBM8TlaMMcYEj5MVY4wxweNkxRhjTPA4WTHGGBM8TlaMMcYEj5MVY4wxweNkxRhjTPA4WTHGGBM8TlaMMcYEj5MVY4wxweNkxRhjTPA4WTHGGBM8TlaMMcYEj5MVY4wxweNkxRhjTPA4WTHGGBM8TlaMMcYEj5MVY4wxweNkxRhjTPA4WTHGGBM8TlaMMcYEj5MVY4wxweNkxRhjTPA4WTHGGBM8TlaMMcYEj5MVY4wxweNkxRhjTPA4WTHGGBM8saEDYMxYlZeX44svvlDbdunSJQDA/v371bZbWVlh3rx5vRYbY32NiIjI0EEwZowaGxvh6OiImpoamJqaAgDa304ikUjVrrm5GYsWLcInn3xiiDAZ6wvS+TYgY10klUoxe/ZsiMViNDc3o7m5GS0tLWhpaVH9v7m5GQAwf/58A0fLmHHjZMVYN8yfPx9NTU2dtrG1tcWf/vSnXoqIsb6JkxVj3TBt2jQ4ODg8c79EIsHChQshFvPXw4x1BycrxrrBxMQECxYsgEQi6XB/c3MzT6xgTA84WTHWTfPmzVN9N/W0F154Ab6+vr0cEWN9Dycrxrpp4sSJcHFx0dhuZmaGRYsWqc0MZIx1DScrxvTgjTfe0LgV2NTUxLcAGdMTTlaM6cGCBQs0bgUqFAp4e3sbKCLG+hZOVozpgaenJ7y8vFS3/CQSCd58800DR8VY38HJijE9+ctf/qJ6kkVLSwvfAmRMjzhZMaYn8+bNQ2trKwDAx8cHrq6uBo6Isb6DkxVjejJ06FBMmjQJALBo0SIDR8NY39InltUnJiYiJyfH0GEwhsbGRohEIpw+fRrfffedocNhDNHR0X1irV+f+GSVk5ODCxcuGDoMZoQyMjJw+/ZtvfU3ZMgQODk5QSaT6a1Poblw4QK/34xERkYGbt26Zegw9KJPfLICgMmTJyM9Pd3QYTAjIxKJsHLlSsydO1dvfd64cQMKhUJv/QnNnDlzAIDfb0agLy1I7xOfrBgTkr6cqBgzFE5WjDHGBI+TFWOMMcHjZMUYY0zwOFkxxhgTPE5WjOnBqVOnYGNjg6+++srQoQjemTNnEBMTg+PHj8PNzQ0ikQgikQhvvPGGRls/Pz9YWVnB1NQUo0aNwuXLlw0Qse7a2tqwZ88eKJXKZ7Zpbm5GQkICFAoFzMzMYGtri9GjR6O4uBgA8OWXX2Lnzp2qp6L0d5ysGNMDIjJ0CEZh8+bNSE5Oxvr16xESEoKbN2/C3d0dAwYMwOHDh/HNN9+otT99+jTS09Mxc+ZM5Ofnw8fHx0CRa6+goAB//OMfER0djbq6ume2Cw0NxaeffoojR46grq4OP/74I9zd3VFTUwMACAgIgEwmw/Tp01FZWdlb4QsWJyvG9MDf3x+PHz/GzJkzDR0K6uvrO/2L3lB27NiBo0ePIi0tDVZWVmr7kpOTYWJigoiICDx+/NhAEXbf1atXsW7dOixduhTjxo17ZrujR48iMzMT6enpmDRpEsRiMZydnXHy5EmMHj1a1W7FihUYO3YsZsyYgZaWlt74EQSLkxVjfczBgwdRVlZm6DDU3LhxAxs3bsTWrVs7fLqHUqlEVFQU7ty5g9WrVxsgQv0YO3Ysjh8/jgULFkAqlT6z3UcffQQfHx+t6p1t2bIFV65cQVJSkj5DNTqcrBjrpvPnz2Po0KEQiUT48MMPAQApKSmwsLCAXC7HyZMn8frrr8Pa2hpDhgzB559/rjo2OTkZMpkMjo6OWLJkCZydnSGTyaBUKnHx4kVVu8jISJiZmWHQoEGqbcuWLYOFhQVEIhEqKioAAFFRUVi1ahUKCwshEolUC5S//fZbWFtbY9u2bb1xSjQkJyeDiBAQEPDMNvHx8Rg+fDgOHDiAM2fOdNofESExMREjR46EVCqFnZ0dgoKC8NNPP6naaHsNAKC1tRWbNm3C0KFDYW5ujjFjxuDYsWPd+6GfoampCRcuXOj0k9eT7OzsMHXqVCQlJfXr282crBjrpldeeQXff/+92rZ33nkHK1euRH19PaysrHDs2DEUFhbCzc0Nb7/9tqqqcGRkJMLDw1FXV4cVK1aguLgYly9fRktLC1577TXVc92Sk5M1Hgm1d+9ebN26VW1bUlISZs6cCXd3dxARbty4AQCqL+nb2tp65Bw8zzfffIMRI0ZALpc/s425uTk++eQTmJiY4O2330Ztbe0z227ZsgUxMTF47733UFZWhu+++w63bt3ClClTcP/+fQDaXwMAWLduHd5//33s2bMHv/76K2bOnIn58+fj0qVL+jsJ/+/u3btoamrCf//7X0ybNk31B8rIkSOxd+/eDhPSH/7wB9y5cwdXr17VezzGgpMVYz1MqVTC2toaDg4OCAsLQ21tLUpLS9XaiMVi1acELy8vpKSkoLq6GqmpqXqJwd/fH1VVVdi4caNe+tNFbW0tioqK4O7u/ty2vr6+WLlyJYqLi7Fu3boO29TX1yMxMRGzZs3CwoULYWNjA29vb+zbtw8VFRXYv3+/xjGdXYOGhgakpKQgODgYISEhsLW1xYYNGyCRSPR2/p/UPoHCwcEB27ZtQ35+Pu7fv4+goCAsX74cn332mcYxHh4eAIC8vDy9x2MsOFkx1ovMzMwAQO2v+o6MHz8ecrlc7baWsSorKwMRdfqp6knx8fEYMWIE9u7di/Pnz2vsz8/PR01NDcaPH6+2fcKECTAzM1O7fdqRp6/Bzz//jLq6OrWJDebm5hg0aFCPnP/277JGjRoFpVIJe3t72NjYYOvWrbCxsekw2bafu/ZPjf0RJyvGBEoqlaK8vNzQYXRbQ0MDAHQ64eBJMpkMqampEIlEWLx4Merr69X2t0/jtrS01DjW1tYW1dXVOsXXfrtxw4YNqjVfIpEIJSUlnU497ypnZ2cAUH3P2M7MzAwuLi4oLCzUOMbc3BzA7+eyP+JkxZgANTc3o7KyEkOGDDF0KN3W/otWl8Wtvr6+iI6ORkFBAeLi4tT22draAkCHSakr58zBwQEAsGfPHhCR2r+eKOpqaWkJDw8PXL9+XWNfS0sLbGxsNLY3NTUB+P1c9kecrBgToOzsbBARJk+erNomFoufe/tQiBwdHSESiXRePxUXFwdPT0/k5uaqbR89ejQsLS01Jj9cvHgRTU1NeOmll3Qa58UXX4RMJsOVK1d0Oq47QkNDkZubi5s3b6q21dXVoaSkpMPp7O3nzsnJqddiFBpOVowJQFtbGx49eoSWlhZcu3YNUVFRGDp0KMLDw1VtFAoFHj58iMzMTDQ3N6O8vBwlJSUafdnb2+Pu3bsoLi5GdXU1mpubkZWVZbCp63K5HG5ubjpXZG6/HWhqaqqxfdWqVThx4gQOHz6Mqqoq5OXlYenSpXB2dkZERITO47z55pv4/PPPkZKSgqqqKrS2tuL27dv49ddfAQBhYWFwcnLS2+OeoqOj4eLigvDwcJSWluLBgwdYu3Yt6uvrO5xY0n7utFmX1WdRHzB79myaPXu2ocNgRggAHTt2rFt9fPDBBzRo0CACQHK5nAICAmjv3r0kl8sJAHl4eFBhYSHt37+frK2tCQC5uLjQL7/8QkREERERJJFIaPDgwSQWi8na2pqCgoKosLBQbZwHDx7QtGnTSCaTkaurK7377ru0Zs0aAkAKhYJKS0uJiOjy5cvk4uJC5ubm9Morr9C9e/fo1KlTZGVlRfHx8d36WYm69n6LjIwkiURCdXV1qm0nTpwgd3d3AkADBw6k5cuXd3jsmjVrKDAwUG1bW1sb7dq1izw8PEgikZCdnR0FBwfTzz//rGqjyzVobGyktWvX0tChQ0ksFpODgwOFhIRQfn4+EREFBwcTANq0aVOnP2dOTg69/PLL5OzsTAAIAA0aNIiUSiWdO3dOre2tW7do3rx5ZGdnR1KplCZOnEhZWVkd9uvv70+DBw+mtra2Tsd/mj5e3wKRxsmK9WtCeDNHRESQvb29QWPQRVfebwUFBSQWi+nQoUM9FFXPam1tpSlTptDBgwd7feyKigqSyWS0e/dunY8VwutbT9L4NiBjAtDXn6ytUCgQGxuL2NhY1TojY9Ha2orMzExUV1cjLCys18ffsmULxo0bh8jIyF4fW0g4WTHGekVMTAzmzJmDsLAwo3pYbXZ2No4fP46srCyt14rpS2JiIq5cuYJTp05BIpH06thCw8nq/7311luwsrKCSCTq1VlB+rRz5054enrC3NwcFhYW8PT0xMaNG1FVVaVzX0/XGmr/Z2ZmBkdHR7z66qvYtWsXHj161AM/Sf+xfv16pKam4vHjx3B1dUVGRoahQ+pR27ZtQ2RkJLZv327oULQ2ffp0HDlyRO25jL3h5MmTaGxsRHZ2Nuzs7Hp1bCHiZPX/Dhw4gI8//tjQYXTLf/7zH7z99tsoLS3F/fv3ERcXh507d2L27Nk69/VkrSEbGxsQEdra2lBWVoa0tDS4urpi7dq1GDVqVI88P62/SEhIQGNjI4gIRUVFXbpWxsbPzw87duwwdBiCFxgYiJiYGI3ZkP0VJ6s+xMzMDMuWLYODgwMsLS0xZ84cBAUF4Z///KdqCm53iEQi2Nra4tVXX0VqairS0tJw//59VS0nxhjrKZysniASiQwdQrecOHFCo1bQ4MGDAaBHvtSePXs2wsPDUVZWhn379um9f8YYa9dvkxURYdeuXRgxYgSkUilsbGywZs0ajXad1bnRpV7OuXPnMHHiRMjlclhbW8Pb21v1XVJP1tIpKCiAra0tXFxcVNv0WduofdFqVlaWapuxnzPGmAAZeO68XnRl3cd7771HIpGI/va3v9GjR4+orq6O9u7dSwAoNzdX1W716tUklUopIyODHj16ROvXrycTExP64YcfVP0AoH/961/0+PFjKisroylTppCFhQU1NTUREVFNTQ1ZW1vTzp07qb6+nu7du0ezZs2i8vJyrcbQVVNTE92+fZs++OADkkqlGmtbvv76a7KysqLY2Njn9uXu7k42NjbP3F9VVUUA6MUXX1RtM6Zzhr6zDqXX8LpG49GHXt/9c1FwXV0dyeVyeu2119S2f/7552rJqr6+nuRyOYWFhakdK5VK6Z133iGi33/x1tfXq9q0J70bN24QEdH//vc/AkBff/21RizajKErJycnAkADBgygv//976oE0BXPS1ZERCKRiGxtbYnI+M5ZH3oz9xpOVsajD72+08S9/1nO8G7cuIG6ujpMnz6903ZdrXPzdL0cNzc3ODo6YuHChVixYgXCw8MxbNiwbo3RmVu3bqGyshK5ubmIiYnB/v37cfbsWTg6Onapv87U1taCiGBtbQ3AOM9ZaGgoQkNDdT6uvzP273iZcemXyar9oZDtpQGe5ck6Nxs2bFDb116TRhvm5uY4e/Ys1q1bh23btiE2NhZz585Famqq3sZ4kkQigYODA/z8/ODq6orhw4cjISEBSUlJXeqvM7/88gsAwNPTE4BxnrOoqCj4+vrqfFx/tWfPHgDAypUrDRwJe56+9EdYv0xW7TPmGhsbO233ZJ2bqKiobo05atQofPXVVygvL0diYiJ27NiBUaNGqR7foo8xOqJQKGBqaor8/Hy99w38NlkDAF5//XUAxnnOfH19MXfu3G7301+kp6cDAJ8zI9CXklW/nA04evRomJiY4Ny5c52201edm7t376oKrTk4OGD79u3w8fHB9evX9TbGgwcPMH/+fI3tBQUFaG1txYsvvtit/jty79497NmzB0OGDMHixYsBGNc5Y4wZj36ZrBwcHBASEoKMjAwcPHgQVVVVuHbtGvbv36/WTps6N9q4e/culixZgp9++glNTU3Izc1FSUkJJk+erLcxLCwscPr0aZw9exZVVVVobm5Gbm4uFi1aBAsLC0RHR6va6lrbiIhQU1ODtrY2EBHKy8tx7NgxvPwIDxbrAAAJTElEQVTyyzA1NUVmZqbqOytjOmeMMSNi4BkeetGV2UnV1dX01ltv0YABA8jS0pJeeeUV2rRpEwGgIUOG0NWrV4mo8zo32tbLKS4uJqVSSXZ2dmRqakovvPACvffee9TS0vLcMXQREBBArq6uZGlpSVKplNzd3SksLIzy8vLU2mlT2+jLL7+kMWPGkFwuJzMzMzIxMSEAqpl/EydOpNjYWHrw4IHGscZ0ztB3Zkv1Gp4NaDz60Os7TUREZLhUqR9z5swB8Pu9dMa0JRKJcOzYMf7+RQf8fjMefej1nd4vbwMyxhgzLpysBOynn37SKNHR0T9DFIRjTJ/OnDmDmJgYjdI0b7zxhkZbPz8/WFlZwdTUFKNGjcLly5cNELH2YmNj4eXlBWtra0ilUigUCvz1r39Ve17nl19+iZ07d/b5Ipzd0S+nrhsLT09P9IG7tIx1avPmzcjNzcWRI0dgZWWFkJAQKBQKVFZW4vDhwwgLC4O/v7+q/enTp/Htt99i3759yMzMNGDk2jl79iyWL1+OsLAwSCQSZGVlYeHChcjLy1M9UzMgIABFRUWYPn06MjMzYWtra+CohYc/WTFmQPX19VAqlUY/Rlft2LEDR48eRVpaGqysrNT2JScnw8TEBBEREUZdgsbS0hIRERGwt7eHlZUV5s6di+DgYHz77be4deuWqt2KFSswduxYzJgxAy0tLQaMWJg4WTFmQAcPHkRZWZnRj9EVN27cwMaNG7F161aN0jYAoFQqERUVhTt37mD16tUGiFA/vv76a40CigMHDgQA1NXVqW3fsmULrly50iNPmzF2nKwY0wERITExESNHjoRUKoWdnR2CgoLUnkkYGRkJMzMztTLoy5Ytg4WFBUQiESoqKgD89pinVatWobCwECKRCAqFAsnJyZDJZHB0dMSSJUvg7OwMmUwGpVKJixcv6mUMQL9lYroqOTkZRISAgIBntomPj8fw4cNx4MABnDlzptP+tLk2upSo6ckyNHfu3IG5uTlcXV3VttvZ2WHq1KlISkrirwCeZsB583rD6z5YV0HHdSibNm0iMzMzOnToEFVWVtK1a9fIx8eHBg4cSPfu3VO1W7BgATk5Oakdu2vXLgKgKnNCRBQSEkLu7u5q7SIiIsjCwoKuX79ODQ0NlJ+fTxMmTCArKysqLS3Vyxi6lIl5mr7eb25ubuTl5dXhPnd3dyoqKiIiou+//55MTExo2LBhVFNTQ0REWVlZFBgYqHaMttdGmxI1RPov3dOutraWrKysKDIyssP9MTExGqWKukrX17eApfEnK8a0VF9fj8TERMyaNQsLFy6EjY0NvL29sW/fPlRUVGg8AaU7xGKx6hOCl5cXUlJSUF1djdTUVL307+/vj6qqKmzcuFEv/emqtrYWRUVFcHd3f25bX19frFy5EsXFxVi3bl2HbbpybZRKJaytreHg4ICwsDDU1taitLQUANDQ0ICUlBQEBwcjJCQEtra22LBhAyQSSbevQUJCApydnREfH9/hfg8PDwBAXl5et8bpazhZMaal/Px81NTUYPz48WrbJ0yYADMzM7XbdPo2fvx4yOXyLpeNEZqysjIQEeRyuVbt4+PjMWLECOzduxfnz5/X2N/da/N0iZqeKN0DACdOnEBaWhr+8Y9/aEwoadd+Tu7fv9/lcfoiTlaMaamyshLAb7O7nmZra4vq6uoeHV8qlaK8vLxHx+gtDQ0NAH77mbQhk8mQmpoKkUiExYsXo76+Xm2/vq/Nk2VonlzTWFJSojEpQltHjx7Fjh07kJ2drarN1hFzc3MAv58j9htOVoxpqX3tS0e/+CorKzFkyJAeG7u5ubnHx+hN7b+QdVkE6+vri+joaBQUFCAuLk5tn76vzZOlbohI7V9OTo5OfQHABx98gMOHD+Ps2bN44YUXOm3b1NQE4PdzxH7DyYoxLY0ePRqWlpa4dOmS2vaLFy+iqakJL730kmqbWCxW3VLSh+zsbBARJk+e3GNj9CZHR0eIRCKd10/FxcXB09MTubm5att1uTba0FcZGiLC2rVrkZeXh8zMzA4/+T2t/Zw4OTl1a+y+hpMVY1qSyWRYtWoVTpw4gcOHD6Oqqgp5eXlYunQpnJ2dERERoWqrUCjw8OFDZGZmorm5GeXl5SgpKdHo097eHnfv3kVxcTGqq6tVyaetrQ2PHj1CS0sLrl27hqioKAwdOhTh4eF6GUPXMjH6JpfL4ebmpqrara3224FPr1vS5dpoO87zytCEhYXBycmp08c9Xb9+He+//z4+/vhjSCQSjUel7d69W+OY9nPi7e2tU8x9HScrxnSwefNmJCQkIDY2FgMHDsTUqVMxbNgwZGdnw8LCQtXunXfewbRp0zBv3jyMGDECcXFxqts6vr6+qicXLF26FI6OjvDy8sKMGTPw8OFDAL99X+Ht7Q1zc3NMmTIFw4cPx7///W+173i6O4ah+fv7Iz8/X+37py+++AIKhQKFhYWYMGEC3n33XY3jJk+erFafrZ021yYlJQV79uwBAIwZMwY3b97Exx9/jFWrVgEA/vznP6OgoAAAkJSUhJUrV2Lnzp0YMGAAnJ2dERUVhUePHgH47XZdWVkZTp48+cyfkbqwVuqHH37A4MGDMWbMGJ2P7dMMN21ef3idFesqCHAdSkREBNnb2xs6jGfS1/utoKCAxGIxHTp0SA9R9b7W1laaMmUKHTx4UG99VlRUkEwmo927d+ulPyG+vruI11kxJkT94enbCoUCsbGxiI2NVXsCuTFobW1FZmYmqqur9Vr1YMuWLRg3bhwiIyP11mdfwcmKMWYwMTExmDNnDsLCwozqYbXZ2dk4fvw4srKytF4r9jyJiYm4cuUKTp06BYlEopc++xJOVowJyPr165GamorHjx/D1dUVGRkZhg6px23btg2RkZHYvn27oUPR2vTp03HkyBG1ZzN2x8mTJ9HY2Ijs7GzY2dnppc++hutZMSYgCQkJSEhIMHQYvc7Pzw9+fn6GDsNgAgMDERgYaOgwBI0/WTHGGBM8TlaMMcYEj5MVY4wxweNkxRhjTPD6zASL27dvIy0tzdBhMCPUlQeT9mftjwPi9xvrTSIi46+dPGfOnH4xxZcxxnR17NgxzJ0719BhdFd6n0hWjDHG+rR0/s6KMcaY4HGyYowxJnicrBhjjAkeJyvGGGOC93/lYn+Zkxha+AAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<IPython.core.display.Image object>"
            ]
          },
          "metadata": {},
          "execution_count": 11
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0r2riz56-rWg"
      },
      "source": [
        "#### Method 2: 把要加的 layers 用成一個 list 一次加到模型裡面\n",
        "Note: 後面要再新加也是可以的！"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GxJr94Ne-rWg"
      },
      "source": [
        "model_from_list = keras.Sequential([\n",
        "    # 跟 method 1 一樣，因為整個模型後面在 compile 時需要知道 input_shape 才能 compile，因此在這邊要\n",
        "    # 先 specify input_shape，不然 model 不會先被 build 起來\n",
        "    layers.Dense(16, input_shape=(3,)),\n",
        "    layers.Dense(16),\n",
        "    layers.Dense(2, activation='sigmoid')\n",
        "])\n",
        "\n",
        "#compile\n",
        "model_from_list.compile(loss='binary_crossentropy', #loss function\n",
        "              optimizer='adam', \n",
        "              metrics=['accuracy'])"
      ],
      "execution_count": 12,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "XE5Jxr81-rWk",
        "outputId": "5bd292ff-7689-4486-9227-128a653a3de0"
      },
      "source": [
        "model_from_list.summary()"
      ],
      "execution_count": 13,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_1\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "dense_4 (Dense)              (None, 16)                64        \n",
            "_________________________________________________________________\n",
            "dense_5 (Dense)              (None, 16)                272       \n",
            "_________________________________________________________________\n",
            "dense_6 (Dense)              (None, 2)                 34        \n",
            "=================================================================\n",
            "Total params: 370\n",
            "Trainable params: 370\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 422
        },
        "id": "7jYmcEAL-rWk",
        "outputId": "ebcd1766-dc1e-4e30-c1c1-109a82e892b1"
      },
      "source": [
        "plot_model(model_from_empty, show_shapes=True)"
      ],
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<IPython.core.display.Image object>"
            ]
          },
          "metadata": {},
          "execution_count": 14
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "2SC1bn-l_Y2x"
      },
      "source": [
        "### Tensor-wise (Funtional API): 給定一個 Tensor，接下來把 tensor 傳進去\n",
        "\n",
        "優點： 比較彈性（讓 Keras 擁有 TensorFlow 與 PyTorch 般的彈性）\n",
        "\n",
        "缺點： 可能比較難懂難理解，至少會用到類似 graph 的概念 （或者能夠有個「骨架」的概念）"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "vAymp9DjDSBv"
      },
      "source": [
        "附上 TensorFlow 1.x 的寫法作為與 **Functional API** 的對比，當然這種寫法就是比較非人類，可能因此 Google 才想把比較噁心的程式碼用 Keras Functional API 重新包裝，推出 TensorFlow 2.0。因為 Keras 的寫法比較直觀，但對電腦來說運作是一樣的。\n",
        "\n",
        "```python\n",
        "# inputs 是一個 tensor，當作一個空殼，裡面之後可以擺真實的數據進去（就是先放個入口，以便建立骨架）\n",
        "# inputs = keras.Input(shape=(3,))    <-- Keras          \n",
        "inputs = tf.placeholder(\"float\", shape=[None, 3])\n",
        "# # nothing, Keras 會自動幫你算       <-- Keras \n",
        "outputs = tf.placeholder(\"float\", shape=[None, 2])\n",
        "\n",
        "# 接下來，就是把 tensor 每層出來的東西「接起來」\n",
        "# dense1 = layers.Dense(16)(inputs)   <-- Keras           \n",
        "dense1 = tf.layers.dense(inputs, 16)\n",
        "# 現在 dense1 是「經過了 inputs 這層被產生出來的 tensor」\n",
        "# 接著就是往下餵\n",
        "# dense2 = layers.Dense(16)(dense1)   <-- Keras           \n",
        "dense2 = tf.layers.dense(dense1, 16)\n",
        "# outputs = layers.Dense(2)(dense2)   <-- Keras           \n",
        "outputs = tf.layers.dense(dense2, 2)\n",
        "```"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "FUbseI6QGn2H"
      },
      "source": [
        "差別是，TensorFlow 還要刻一堆這種東西\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "o0kSHA98GxAg"
      },
      "source": [
        "###### 不要看，你會怕 (X"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "HTGVi3C4Gzmz"
      },
      "source": [
        "```python\n",
        "loss = compute_tf_loss(outputs, answers)\n",
        "\n",
        "optimizer = tf.train.AdamOptimizer()\n",
        "train_op = optimizer.minimize(loss)\n",
        "init_op = tf.global_variables_initializer()\n",
        "\n",
        "def compute_tf_loss(outputs, answers):\n",
        "    # ...\n",
        "    pass\n",
        "\n",
        "def get_batch(x_data, y_data):\n",
        "    # ...\n",
        "    pass\n",
        "\n",
        "with tf.Session() as session:\n",
        "    session.run(init_op)\n",
        "    for epoch in range(200):\n",
        "        x_data, y_data = get_batch(x_data, y_data)\n",
        "        session.run(train_op, feed_dict={\n",
        "            inputs: inputs, outputs: outputs})\n",
        "```"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "KFEkwVnnHAPn"
      },
      "source": [
        "##### 如果可以理解這種精神（不用理解 TensorFlow 的寫法，理解精神就好），就可以重新看 Function API 的作法了"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "91dqkQCB52sW"
      },
      "source": [
        "# inputs 是一個 tensor，當作一個空殼，裡面之後可以擺真實的數據進去（就是先放個入口，以便建立骨架）\n",
        "inputs = keras.Input(shape=(3,))"
      ],
      "execution_count": 15,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3Z63ZLaF_9BU"
      },
      "source": [
        "# 接下來，就是把 tensor 每層出來的東西「接起來」\n",
        "dense1 = layers.Dense(16)(inputs)\n",
        "# 現在 dense1 是「經過了 inputs 這層被產生出來的 tensor」\n",
        "# 接著就是往下餵\n",
        "\n",
        "dense2 = layers.Dense(16)(dense1)\n",
        "outputs = layers.Dense(2)(dense2)\n",
        "\n",
        "model_from_tensor = keras.Model(inputs=inputs, outputs=outputs)\n",
        "\n",
        "#compile\n",
        "model_from_tensor.compile(loss='binary_crossentropy', #loss function\n",
        "              optimizer='adam', \n",
        "              metrics=['accuracy'])"
      ],
      "execution_count": 16,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "e-FFsOWm55dk",
        "outputId": "dc9dec81-bdb7-491c-d587-624fced0cba3"
      },
      "source": [
        "model_from_tensor.summary()"
      ],
      "execution_count": 17,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"model\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "input_1 (InputLayer)         [(None, 3)]               0         \n",
            "_________________________________________________________________\n",
            "dense_7 (Dense)              (None, 16)                64        \n",
            "_________________________________________________________________\n",
            "dense_8 (Dense)              (None, 16)                272       \n",
            "_________________________________________________________________\n",
            "dense_9 (Dense)              (None, 2)                 34        \n",
            "=================================================================\n",
            "Total params: 370\n",
            "Trainable params: 370\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 422
        },
        "id": "HP_wuAKw5_S3",
        "outputId": "7ca9c5f3-60f3-43da-cee7-e82f0385b39d"
      },
      "source": [
        "plot_model(model_from_tensor, show_shapes=True)"
      ],
      "execution_count": 18,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<IPython.core.display.Image object>"
            ]
          },
          "metadata": {},
          "execution_count": 18
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "w7LHFjecIvWV"
      },
      "source": [
        "### 請注意，三種方式建立出來的 model 架構是 **一模一樣** 的\n",
        "也都可以 summary 跟 plot_model\n",
        "\n",
        "###### **好習慣** ： 在開始跑之前，建議都先 model.summary() 一下確認自己建立的模型是正確的"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "3hLz1UG3JK-t"
      },
      "source": [
        "## 訓練 Model\n",
        "\n",
        "在這邊三個 model 都跑下去，證明它們真的是一樣的"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Uu-dLN5nJCXc",
        "outputId": "0fa895c8-f2fe-42a1-a51f-d0dd034e0c63"
      },
      "source": [
        "# 這邊有一個 Keras 的 bug，就是 functional API 裡面 \\\n",
        "#     binary_crossentropy loss 那邊它會自己幫你轉 one-hot\n",
        "history__from_empty = model_from_empty.fit(x_train,\n",
        "                    to_categorical(y_train),\n",
        "                    batch_size=10,\n",
        "                    epochs=30,\n",
        "                    validation_data=(x_valid, \n",
        "                                     to_categorical(y_valid)))"
      ],
      "execution_count": 19,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/30\n",
            "10/10 [==============================] - 1s 19ms/step - loss: 0.6985 - accuracy: 0.4700 - val_loss: 0.6930 - val_accuracy: 0.4800\n",
            "Epoch 2/30\n",
            "10/10 [==============================] - 0s 6ms/step - loss: 0.6946 - accuracy: 0.5100 - val_loss: 0.6904 - val_accuracy: 0.5200\n",
            "Epoch 3/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6920 - accuracy: 0.5300 - val_loss: 0.6887 - val_accuracy: 0.5700\n",
            "Epoch 4/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6905 - accuracy: 0.5400 - val_loss: 0.6880 - val_accuracy: 0.5500\n",
            "Epoch 5/30\n",
            "10/10 [==============================] - 0s 5ms/step - loss: 0.6907 - accuracy: 0.5400 - val_loss: 0.6873 - val_accuracy: 0.5500\n",
            "Epoch 6/30\n",
            "10/10 [==============================] - 0s 5ms/step - loss: 0.6890 - accuracy: 0.5400 - val_loss: 0.6871 - val_accuracy: 0.5500\n",
            "Epoch 7/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6878 - accuracy: 0.5400 - val_loss: 0.6869 - val_accuracy: 0.5500\n",
            "Epoch 8/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6880 - accuracy: 0.5400 - val_loss: 0.6867 - val_accuracy: 0.5500\n",
            "Epoch 9/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6879 - accuracy: 0.5400 - val_loss: 0.6866 - val_accuracy: 0.5500\n",
            "Epoch 10/30\n",
            "10/10 [==============================] - 0s 5ms/step - loss: 0.6866 - accuracy: 0.5400 - val_loss: 0.6865 - val_accuracy: 0.5500\n",
            "Epoch 11/30\n",
            "10/10 [==============================] - 0s 5ms/step - loss: 0.6870 - accuracy: 0.5400 - val_loss: 0.6864 - val_accuracy: 0.5500\n",
            "Epoch 12/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6860 - accuracy: 0.5400 - val_loss: 0.6863 - val_accuracy: 0.5500\n",
            "Epoch 13/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6866 - accuracy: 0.5400 - val_loss: 0.6863 - val_accuracy: 0.5500\n",
            "Epoch 14/30\n",
            "10/10 [==============================] - 0s 5ms/step - loss: 0.6857 - accuracy: 0.5400 - val_loss: 0.6862 - val_accuracy: 0.5500\n",
            "Epoch 15/30\n",
            "10/10 [==============================] - 0s 6ms/step - loss: 0.6847 - accuracy: 0.5600 - val_loss: 0.6862 - val_accuracy: 0.5600\n",
            "Epoch 16/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6846 - accuracy: 0.5600 - val_loss: 0.6862 - val_accuracy: 0.5600\n",
            "Epoch 17/30\n",
            "10/10 [==============================] - 0s 5ms/step - loss: 0.6842 - accuracy: 0.5600 - val_loss: 0.6862 - val_accuracy: 0.5600\n",
            "Epoch 18/30\n",
            "10/10 [==============================] - 0s 5ms/step - loss: 0.6843 - accuracy: 0.5600 - val_loss: 0.6862 - val_accuracy: 0.5600\n",
            "Epoch 19/30\n",
            "10/10 [==============================] - 0s 6ms/step - loss: 0.6844 - accuracy: 0.5600 - val_loss: 0.6863 - val_accuracy: 0.5600\n",
            "Epoch 20/30\n",
            "10/10 [==============================] - 0s 6ms/step - loss: 0.6845 - accuracy: 0.5600 - val_loss: 0.6862 - val_accuracy: 0.5600\n",
            "Epoch 21/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6837 - accuracy: 0.5900 - val_loss: 0.6862 - val_accuracy: 0.5700\n",
            "Epoch 22/30\n",
            "10/10 [==============================] - 0s 5ms/step - loss: 0.6836 - accuracy: 0.6000 - val_loss: 0.6862 - val_accuracy: 0.5700\n",
            "Epoch 23/30\n",
            "10/10 [==============================] - 0s 5ms/step - loss: 0.6833 - accuracy: 0.5900 - val_loss: 0.6863 - val_accuracy: 0.5700\n",
            "Epoch 24/30\n",
            "10/10 [==============================] - 0s 5ms/step - loss: 0.6831 - accuracy: 0.6000 - val_loss: 0.6863 - val_accuracy: 0.5600\n",
            "Epoch 25/30\n",
            "10/10 [==============================] - 0s 6ms/step - loss: 0.6832 - accuracy: 0.6000 - val_loss: 0.6864 - val_accuracy: 0.5600\n",
            "Epoch 26/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6835 - accuracy: 0.6200 - val_loss: 0.6864 - val_accuracy: 0.5800\n",
            "Epoch 27/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6828 - accuracy: 0.6200 - val_loss: 0.6864 - val_accuracy: 0.5800\n",
            "Epoch 28/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6844 - accuracy: 0.6100 - val_loss: 0.6866 - val_accuracy: 0.5600\n",
            "Epoch 29/30\n",
            "10/10 [==============================] - 0s 6ms/step - loss: 0.6827 - accuracy: 0.6300 - val_loss: 0.6866 - val_accuracy: 0.5600\n",
            "Epoch 30/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6826 - accuracy: 0.6200 - val_loss: 0.6866 - val_accuracy: 0.5900\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "oZFQ4K-AP8bP",
        "outputId": "3e628ac7-fb98-4f9c-8eba-1f42ab1d3fcd"
      },
      "source": [
        "# 這邊有一個 Keras 的 bug，就是 functional API 裡面 \\\n",
        "#     binary_crossentropy loss 那邊它會自己幫你轉 one-hot\n",
        "history__from_list = model_from_list.fit(x_train,\n",
        "                    to_categorical(y_train),\n",
        "                    batch_size=10,\n",
        "                    epochs=30,\n",
        "                    validation_data=(x_valid, \n",
        "                                     to_categorical(y_valid)))"
      ],
      "execution_count": 20,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/30\n",
            "10/10 [==============================] - 0s 16ms/step - loss: 0.7144 - accuracy: 0.5400 - val_loss: 0.7020 - val_accuracy: 0.5500\n",
            "Epoch 2/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6974 - accuracy: 0.5400 - val_loss: 0.6926 - val_accuracy: 0.5500\n",
            "Epoch 3/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6919 - accuracy: 0.5400 - val_loss: 0.6907 - val_accuracy: 0.5500\n",
            "Epoch 4/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6908 - accuracy: 0.5400 - val_loss: 0.6899 - val_accuracy: 0.5500\n",
            "Epoch 5/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6900 - accuracy: 0.5400 - val_loss: 0.6891 - val_accuracy: 0.5500\n",
            "Epoch 6/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6894 - accuracy: 0.5400 - val_loss: 0.6884 - val_accuracy: 0.5500\n",
            "Epoch 7/30\n",
            "10/10 [==============================] - 0s 5ms/step - loss: 0.6887 - accuracy: 0.5400 - val_loss: 0.6880 - val_accuracy: 0.5500\n",
            "Epoch 8/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6881 - accuracy: 0.5400 - val_loss: 0.6875 - val_accuracy: 0.5600\n",
            "Epoch 9/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6876 - accuracy: 0.5400 - val_loss: 0.6872 - val_accuracy: 0.5500\n",
            "Epoch 10/30\n",
            "10/10 [==============================] - 0s 6ms/step - loss: 0.6872 - accuracy: 0.5400 - val_loss: 0.6869 - val_accuracy: 0.5500\n",
            "Epoch 11/30\n",
            "10/10 [==============================] - 0s 5ms/step - loss: 0.6864 - accuracy: 0.5400 - val_loss: 0.6867 - val_accuracy: 0.5500\n",
            "Epoch 12/30\n",
            "10/10 [==============================] - 0s 5ms/step - loss: 0.6863 - accuracy: 0.5400 - val_loss: 0.6865 - val_accuracy: 0.5500\n",
            "Epoch 13/30\n",
            "10/10 [==============================] - 0s 5ms/step - loss: 0.6858 - accuracy: 0.5400 - val_loss: 0.6863 - val_accuracy: 0.5600\n",
            "Epoch 14/30\n",
            "10/10 [==============================] - 0s 5ms/step - loss: 0.6862 - accuracy: 0.5500 - val_loss: 0.6862 - val_accuracy: 0.5600\n",
            "Epoch 15/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6852 - accuracy: 0.5700 - val_loss: 0.6861 - val_accuracy: 0.5500\n",
            "Epoch 16/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6855 - accuracy: 0.5500 - val_loss: 0.6861 - val_accuracy: 0.5600\n",
            "Epoch 17/30\n",
            "10/10 [==============================] - 0s 5ms/step - loss: 0.6849 - accuracy: 0.5500 - val_loss: 0.6860 - val_accuracy: 0.5600\n",
            "Epoch 18/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6854 - accuracy: 0.5700 - val_loss: 0.6859 - val_accuracy: 0.5500\n",
            "Epoch 19/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6848 - accuracy: 0.5700 - val_loss: 0.6859 - val_accuracy: 0.5600\n",
            "Epoch 20/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6842 - accuracy: 0.5600 - val_loss: 0.6858 - val_accuracy: 0.5500\n",
            "Epoch 21/30\n",
            "10/10 [==============================] - 0s 5ms/step - loss: 0.6844 - accuracy: 0.5600 - val_loss: 0.6858 - val_accuracy: 0.5500\n",
            "Epoch 22/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6839 - accuracy: 0.5600 - val_loss: 0.6858 - val_accuracy: 0.5500\n",
            "Epoch 23/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6838 - accuracy: 0.5700 - val_loss: 0.6857 - val_accuracy: 0.5800\n",
            "Epoch 24/30\n",
            "10/10 [==============================] - 0s 5ms/step - loss: 0.6846 - accuracy: 0.5700 - val_loss: 0.6858 - val_accuracy: 0.5600\n",
            "Epoch 25/30\n",
            "10/10 [==============================] - 0s 5ms/step - loss: 0.6842 - accuracy: 0.5900 - val_loss: 0.6857 - val_accuracy: 0.5800\n",
            "Epoch 26/30\n",
            "10/10 [==============================] - 0s 7ms/step - loss: 0.6836 - accuracy: 0.5800 - val_loss: 0.6857 - val_accuracy: 0.5800\n",
            "Epoch 27/30\n",
            "10/10 [==============================] - 0s 5ms/step - loss: 0.6834 - accuracy: 0.5900 - val_loss: 0.6857 - val_accuracy: 0.5800\n",
            "Epoch 28/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6834 - accuracy: 0.5900 - val_loss: 0.6857 - val_accuracy: 0.5800\n",
            "Epoch 29/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6834 - accuracy: 0.5800 - val_loss: 0.6858 - val_accuracy: 0.5800\n",
            "Epoch 30/30\n",
            "10/10 [==============================] - 0s 5ms/step - loss: 0.6834 - accuracy: 0.6200 - val_loss: 0.6857 - val_accuracy: 0.6000\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "284sIB_MP-P6",
        "outputId": "70059ad0-fd68-44ad-a6a2-524844db96db"
      },
      "source": [
        "history__from_tensor = model_from_tensor.fit(x_train,\n",
        "                    y_train,\n",
        "                    batch_size=10,\n",
        "                    epochs=30,\n",
        "                    validation_data=(x_valid, y_valid))"
      ],
      "execution_count": 21,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/30\n",
            "10/10 [==============================] - 1s 18ms/step - loss: 3.7253 - accuracy: 0.5400 - val_loss: 3.1559 - val_accuracy: 0.5700\n",
            "Epoch 2/30\n",
            "10/10 [==============================] - 0s 6ms/step - loss: 2.4988 - accuracy: 0.5400 - val_loss: 1.3870 - val_accuracy: 0.5400\n",
            "Epoch 3/30\n",
            "10/10 [==============================] - 0s 6ms/step - loss: 1.1103 - accuracy: 0.4400 - val_loss: 0.7778 - val_accuracy: 0.4600\n",
            "Epoch 4/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.7435 - accuracy: 0.4800 - val_loss: 0.8521 - val_accuracy: 0.5200\n",
            "Epoch 5/30\n",
            "10/10 [==============================] - 0s 5ms/step - loss: 0.7490 - accuracy: 0.4700 - val_loss: 0.9225 - val_accuracy: 0.5500\n",
            "Epoch 6/30\n",
            "10/10 [==============================] - 0s 5ms/step - loss: 0.7355 - accuracy: 0.4900 - val_loss: 0.7738 - val_accuracy: 0.5200\n",
            "Epoch 7/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.7242 - accuracy: 0.4800 - val_loss: 0.7510 - val_accuracy: 0.5200\n",
            "Epoch 8/30\n",
            "10/10 [==============================] - 0s 6ms/step - loss: 0.7200 - accuracy: 0.5000 - val_loss: 0.7428 - val_accuracy: 0.5500\n",
            "Epoch 9/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.7199 - accuracy: 0.4800 - val_loss: 0.7378 - val_accuracy: 0.5500\n",
            "Epoch 10/30\n",
            "10/10 [==============================] - 0s 6ms/step - loss: 0.7137 - accuracy: 0.4600 - val_loss: 0.7354 - val_accuracy: 0.5400\n",
            "Epoch 11/30\n",
            "10/10 [==============================] - 0s 5ms/step - loss: 0.7112 - accuracy: 0.4600 - val_loss: 0.7322 - val_accuracy: 0.5400\n",
            "Epoch 12/30\n",
            "10/10 [==============================] - 0s 5ms/step - loss: 0.7081 - accuracy: 0.4700 - val_loss: 0.7302 - val_accuracy: 0.5400\n",
            "Epoch 13/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.7084 - accuracy: 0.4900 - val_loss: 0.7293 - val_accuracy: 0.5400\n",
            "Epoch 14/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.7047 - accuracy: 0.4800 - val_loss: 0.7270 - val_accuracy: 0.5600\n",
            "Epoch 15/30\n",
            "10/10 [==============================] - 0s 5ms/step - loss: 0.7044 - accuracy: 0.4900 - val_loss: 0.7234 - val_accuracy: 0.5600\n",
            "Epoch 16/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.7038 - accuracy: 0.4900 - val_loss: 0.7211 - val_accuracy: 0.5600\n",
            "Epoch 17/30\n",
            "10/10 [==============================] - 0s 5ms/step - loss: 0.6998 - accuracy: 0.4900 - val_loss: 0.7202 - val_accuracy: 0.5500\n",
            "Epoch 18/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6993 - accuracy: 0.4900 - val_loss: 0.7190 - val_accuracy: 0.5600\n",
            "Epoch 19/30\n",
            "10/10 [==============================] - 0s 6ms/step - loss: 0.6981 - accuracy: 0.4900 - val_loss: 0.7177 - val_accuracy: 0.5800\n",
            "Epoch 20/30\n",
            "10/10 [==============================] - 0s 6ms/step - loss: 0.6981 - accuracy: 0.4900 - val_loss: 0.7179 - val_accuracy: 0.5800\n",
            "Epoch 21/30\n",
            "10/10 [==============================] - 0s 5ms/step - loss: 0.6977 - accuracy: 0.4900 - val_loss: 0.7143 - val_accuracy: 0.5700\n",
            "Epoch 22/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6948 - accuracy: 0.4800 - val_loss: 0.7134 - val_accuracy: 0.5600\n",
            "Epoch 23/30\n",
            "10/10 [==============================] - 0s 6ms/step - loss: 0.6952 - accuracy: 0.4800 - val_loss: 0.7125 - val_accuracy: 0.5700\n",
            "Epoch 24/30\n",
            "10/10 [==============================] - 0s 6ms/step - loss: 0.6930 - accuracy: 0.4800 - val_loss: 0.7116 - val_accuracy: 0.5700\n",
            "Epoch 25/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6921 - accuracy: 0.4800 - val_loss: 0.7107 - val_accuracy: 0.5500\n",
            "Epoch 26/30\n",
            "10/10 [==============================] - 0s 5ms/step - loss: 0.6926 - accuracy: 0.4800 - val_loss: 0.7092 - val_accuracy: 0.5600\n",
            "Epoch 27/30\n",
            "10/10 [==============================] - 0s 5ms/step - loss: 0.6915 - accuracy: 0.4700 - val_loss: 0.7086 - val_accuracy: 0.5400\n",
            "Epoch 28/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6909 - accuracy: 0.4700 - val_loss: 0.7078 - val_accuracy: 0.5400\n",
            "Epoch 29/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6914 - accuracy: 0.4700 - val_loss: 0.7069 - val_accuracy: 0.5500\n",
            "Epoch 30/30\n",
            "10/10 [==============================] - 0s 5ms/step - loss: 0.6890 - accuracy: 0.4800 - val_loss: 0.7066 - val_accuracy: 0.5700\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "bEcoUWNXJj3X",
        "outputId": "01b17066-929d-4d95-d237-ea0be6250ebc"
      },
      "source": [
        "# 然後你也可以只取一個出來訓練\n",
        "model = model_from_tensor\n",
        "\n",
        "y_train_target = y_train if model == model_from_tensor else to_categorical(y_train)\n",
        "y_valid_target = y_valid if model == model_from_tensor else to_categorical(y_valid)\n",
        "\n",
        "history = model.fit(x_train,\n",
        "                    y_train_target,\n",
        "                    batch_size=10,\n",
        "                    epochs=30,\n",
        "                    validation_data=(x_valid, y_valid_target))"
      ],
      "execution_count": 22,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/30\n",
            "10/10 [==============================] - 0s 8ms/step - loss: 0.6889 - accuracy: 0.4800 - val_loss: 0.7063 - val_accuracy: 0.5500\n",
            "Epoch 2/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6892 - accuracy: 0.4700 - val_loss: 0.7051 - val_accuracy: 0.5500\n",
            "Epoch 3/30\n",
            "10/10 [==============================] - 0s 5ms/step - loss: 0.6879 - accuracy: 0.4700 - val_loss: 0.7050 - val_accuracy: 0.5500\n",
            "Epoch 4/30\n",
            "10/10 [==============================] - 0s 5ms/step - loss: 0.6871 - accuracy: 0.4700 - val_loss: 0.7044 - val_accuracy: 0.5500\n",
            "Epoch 5/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6881 - accuracy: 0.4700 - val_loss: 0.7042 - val_accuracy: 0.5600\n",
            "Epoch 6/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6864 - accuracy: 0.4800 - val_loss: 0.7029 - val_accuracy: 0.5500\n",
            "Epoch 7/30\n",
            "10/10 [==============================] - 0s 5ms/step - loss: 0.6862 - accuracy: 0.4900 - val_loss: 0.7024 - val_accuracy: 0.5600\n",
            "Epoch 8/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6861 - accuracy: 0.4900 - val_loss: 0.7018 - val_accuracy: 0.5500\n",
            "Epoch 9/30\n",
            "10/10 [==============================] - 0s 5ms/step - loss: 0.6877 - accuracy: 0.4900 - val_loss: 0.7015 - val_accuracy: 0.5600\n",
            "Epoch 10/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6892 - accuracy: 0.5000 - val_loss: 0.7026 - val_accuracy: 0.5600\n",
            "Epoch 11/30\n",
            "10/10 [==============================] - 0s 5ms/step - loss: 0.6854 - accuracy: 0.5000 - val_loss: 0.7007 - val_accuracy: 0.5600\n",
            "Epoch 12/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6852 - accuracy: 0.5000 - val_loss: 0.7003 - val_accuracy: 0.5700\n",
            "Epoch 13/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6861 - accuracy: 0.5000 - val_loss: 0.6998 - val_accuracy: 0.5700\n",
            "Epoch 14/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6865 - accuracy: 0.5000 - val_loss: 0.6994 - val_accuracy: 0.5600\n",
            "Epoch 15/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6844 - accuracy: 0.5000 - val_loss: 0.7002 - val_accuracy: 0.5600\n",
            "Epoch 16/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6846 - accuracy: 0.5000 - val_loss: 0.6990 - val_accuracy: 0.5600\n",
            "Epoch 17/30\n",
            "10/10 [==============================] - 0s 5ms/step - loss: 0.6853 - accuracy: 0.5000 - val_loss: 0.6997 - val_accuracy: 0.5600\n",
            "Epoch 18/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6845 - accuracy: 0.5000 - val_loss: 0.6980 - val_accuracy: 0.5400\n",
            "Epoch 19/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6839 - accuracy: 0.5000 - val_loss: 0.6976 - val_accuracy: 0.5400\n",
            "Epoch 20/30\n",
            "10/10 [==============================] - 0s 6ms/step - loss: 0.6839 - accuracy: 0.4900 - val_loss: 0.6975 - val_accuracy: 0.5400\n",
            "Epoch 21/30\n",
            "10/10 [==============================] - 0s 6ms/step - loss: 0.6832 - accuracy: 0.5100 - val_loss: 0.6973 - val_accuracy: 0.5400\n",
            "Epoch 22/30\n",
            "10/10 [==============================] - 0s 5ms/step - loss: 0.6926 - accuracy: 0.5200 - val_loss: 0.7009 - val_accuracy: 0.5400\n",
            "Epoch 23/30\n",
            "10/10 [==============================] - 0s 5ms/step - loss: 0.6847 - accuracy: 0.5100 - val_loss: 0.6967 - val_accuracy: 0.5300\n",
            "Epoch 24/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6832 - accuracy: 0.5200 - val_loss: 0.6965 - val_accuracy: 0.5100\n",
            "Epoch 25/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6840 - accuracy: 0.5200 - val_loss: 0.6961 - val_accuracy: 0.5500\n",
            "Epoch 26/30\n",
            "10/10 [==============================] - 0s 6ms/step - loss: 0.6827 - accuracy: 0.5300 - val_loss: 0.6957 - val_accuracy: 0.5300\n",
            "Epoch 27/30\n",
            "10/10 [==============================] - 0s 6ms/step - loss: 0.6859 - accuracy: 0.5200 - val_loss: 0.6969 - val_accuracy: 0.5200\n",
            "Epoch 28/30\n",
            "10/10 [==============================] - 0s 5ms/step - loss: 0.6846 - accuracy: 0.5200 - val_loss: 0.6962 - val_accuracy: 0.5100\n",
            "Epoch 29/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6820 - accuracy: 0.5100 - val_loss: 0.6956 - val_accuracy: 0.4900\n",
            "Epoch 30/30\n",
            "10/10 [==============================] - 0s 4ms/step - loss: 0.6831 - accuracy: 0.5000 - val_loss: 0.6952 - val_accuracy: 0.5000\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "vTyYGbhBGo2T"
      },
      "source": [
        "## Accuracy and loss evaluation"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "id": "lOq8g2kBRKu7",
        "outputId": "e43571cc-ed49-4b76-e068-72dae06903cc"
      },
      "source": [
        "from matplotlib import pyplot as plt\n",
        "plt.plot(history.history['accuracy'])\n",
        "plt.plot(history.history['val_accuracy'])\n",
        "plt.title('Model Accuracy')\n",
        "plt.ylabel('accuracy')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'val'], loc='upper left')\n",
        "plt.show()"
      ],
      "execution_count": 23,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "id": "4_E4ml6eRKu8",
        "outputId": "2ec6ff64-ee3f-4389-a265-1adc5330a2c8"
      },
      "source": [
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.title('Model Loss')\n",
        "plt.ylabel('loss')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'val'], loc='upper left')\n",
        "plt.show()"
      ],
      "execution_count": 24,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "K_2iVjVhQVYi"
      },
      "source": [
        "history = history__from_empty"
      ],
      "execution_count": 25,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "id": "1qaXWu7pRNCU",
        "outputId": "27e6d5d2-d6fc-4196-c8fc-8f08a4ab6cd2"
      },
      "source": [
        "from matplotlib import pyplot as plt\n",
        "plt.plot(history.history['accuracy'])\n",
        "plt.plot(history.history['val_accuracy'])\n",
        "plt.title('Model Accuracy')\n",
        "plt.ylabel('accuracy')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'val'], loc='upper left')\n",
        "plt.show()"
      ],
      "execution_count": 26,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "id": "uqHh13UkRNCU",
        "outputId": "18ccd942-964b-46e0-a6be-2ba9535a4332"
      },
      "source": [
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.title('Model Loss')\n",
        "plt.ylabel('loss')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'val'], loc='upper left')\n",
        "plt.show()"
      ],
      "execution_count": 27,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4w4ctmN8RCqG"
      },
      "source": [
        "history = history__from_list"
      ],
      "execution_count": 28,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "id": "5bcIGe_KRL-k",
        "outputId": "3e9a8914-8226-4638-c4f8-b0a189fbd415"
      },
      "source": [
        "from matplotlib import pyplot as plt\n",
        "plt.plot(history.history['accuracy'])\n",
        "plt.plot(history.history['val_accuracy'])\n",
        "plt.title('Model Accuracy')\n",
        "plt.ylabel('accuracy')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'val'], loc='upper left')\n",
        "plt.show()"
      ],
      "execution_count": 29,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "id": "Ldbw06RPRL-k",
        "outputId": "23ffc960-4e0d-48bf-dcb1-93e20ff478d6"
      },
      "source": [
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.title('Model Loss')\n",
        "plt.ylabel('loss')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'val'], loc='upper left')\n",
        "plt.show()"
      ],
      "execution_count": 30,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "CGMa8o8_RCkk"
      },
      "source": [
        "history = history__from_tensor"
      ],
      "execution_count": 31,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "F6fE4HvTGZC9",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "outputId": "dca8840d-7705-45b6-9730-086bbd4ed7e1"
      },
      "source": [
        "from matplotlib import pyplot as plt\n",
        "plt.plot(history.history['accuracy'])\n",
        "plt.plot(history.history['val_accuracy'])\n",
        "plt.title('Model Accuracy')\n",
        "plt.ylabel('accuracy')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'val'], loc='upper left')\n",
        "plt.show()"
      ],
      "execution_count": 32,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Sn5VNyOJGkVh",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "outputId": "2cd6f24d-23f8-4c4b-e7d9-e2a63ec8b659"
      },
      "source": [
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.title('Model Loss')\n",
        "plt.ylabel('loss')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'val'], loc='upper left')\n",
        "plt.show()"
      ],
      "execution_count": 33,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    }
  ]
}