Processing Sequences Using RNNs and CNNs.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "kernelspec": {
      "language": "python",
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "pygments_lexer": "ipython3",
      "nbconvert_exporter": "python",
      "version": "3.6.4",
      "file_extension": ".py",
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "name": "python",
      "mimetype": "text/x-python"
    },
    "colab": {
      "name": "Processing Sequences Using RNNs and CNNs.ipynb",
      "provenance": [],
      "include_colab_link": true
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/Shrey-Viradiya/HandsOnMachineLearning/blob/master/Processing%20Sequences%20Using%20RNNs%20and%20CNNs.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7zt2fb59nSJP",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 372
        },
        "outputId": "605ce566-398b-4720-c507-4d89a9aee690"
      },
      "source": [
        "!nvidia-smi"
      ],
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Mon Jun 22 17:40:06 2020       \n",
            "+-----------------------------------------------------------------------------+\n",
            "| NVIDIA-SMI 450.36.06    Driver Version: 418.67       CUDA Version: 10.1     |\n",
            "|-------------------------------+----------------------+----------------------+\n",
            "| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |\n",
            "| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |\n",
            "|                               |                      |               MIG M. |\n",
            "|===============================+======================+======================|\n",
            "|   0  Tesla T4            Off  | 00000000:00:04.0 Off |                    0 |\n",
            "| N/A   74C    P8    12W /  70W |      0MiB / 15079MiB |      0%      Default |\n",
            "|                               |                      |                 ERR! |\n",
            "+-------------------------------+----------------------+----------------------+\n",
            "                                                                               \n",
            "+-----------------------------------------------------------------------------+\n",
            "| Processes:                                                                  |\n",
            "|  GPU   GI   CI        PID   Type   Process name                  GPU Memory |\n",
            "|        ID   ID                                                   Usage      |\n",
            "|=============================================================================|\n",
            "|  No running processes found                                                 |\n",
            "+-----------------------------------------------------------------------------+\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "fV0ZwmQjnOzp",
        "colab_type": "text"
      },
      "source": [
        "# Processing Sequences Using RNNs and CNNs"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "_uuid": "8f2839f25d086af736a60e9eeb907d3b93b6e0e5",
        "_cell_guid": "b1076dfc-b9ad-4769-8c92-a6c4dae69d19",
        "trusted": true,
        "id": "X9U9ye8gnOzr",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 72
        },
        "outputId": "b71ce64d-1a92-4dd9-94a1-db18165f51d3"
      },
      "source": [
        "# This Python 3 environment comes with many helpful analytics libraries installed\n",
        "# It is defined by the kaggle/python Docker image: https://github.com/kaggle/docker-python\n",
        "# For example, here's several helpful packages to load\n",
        "\n",
        "import tensorflow as tf\n",
        "import tensorflow.keras as keras\n",
        "import matplotlib.pyplot as plt\n",
        "import numpy as np\n",
        "import pandas as pd\n",
        "import seaborn as sns\n",
        "import os\n",
        "import matplotlib as mpl\n",
        "\n",
        "# Input data files are available in the read-only \"../input/\" directory\n",
        "# For example, running this (by clicking run or pressing Shift+Enter) will list all files under the input directory\n",
        "\n",
        "import os\n",
        "for dirname, _, filenames in os.walk('/kaggle/input'):\n",
        "    for filename in filenames:\n",
        "        print(os.path.join(dirname, filename))\n",
        "\n",
        "# You can write up to 5GB to the current directory (/kaggle/working/) that gets preserved as output when you create a version using \"Save & Run All\" \n",
        "# You can also write temporary files to /kaggle/temp/, but they won't be saved outside of the current session"
      ],
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/statsmodels/tools/_testing.py:19: FutureWarning: pandas.util.testing is deprecated. Use the functions in the public API at pandas.testing instead.\n",
            "  import pandas.util.testing as tm\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "_qlRnPMVnOzu",
        "colab_type": "text"
      },
      "source": [
        "Suppose you are studying the number of active users per hour on your website, or the daily temperature in your city, or your company’s financial health, measured quarterly using multiple metrics. In all these cases, the data will be a sequence of one or more values per time step. This is called a time series. In the first two examples there is a single value per time step, so these are univariate time series, while in the financial example there are multiple values per time step (e.g., the company’s revenue, debt, and so on), so it is a multivariate time series. A typical task is to predict future values, which is called forecasting. Another common task is to fill in the blanks: to predict (or rather “postdict”) missing values from the past. This is called imputation."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "izcOLvLInOzv",
        "colab_type": "text"
      },
      "source": [
        "## Generating Data"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "Y_6n4A8jnOzw",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def generate_time_series(batch_size, n_steps):\n",
        "    freq1, freq2, offsets1, offsets2 = np.random.rand(4, batch_size, 1)\n",
        "    time = np.linspace(0, 1, n_steps)\n",
        "    series = 0.5 * np.sin((time - offsets1) * (freq1 * 10 + 10))  #   wave 1\n",
        "    series += 0.2 * np.sin((time - offsets2) * (freq2 * 20 + 20)) # + wave 2\n",
        "    series += 0.1 * (np.random.rand(batch_size, n_steps) - 0.5)   # + noise\n",
        "    return series[..., np.newaxis].astype(np.float32)\n",
        "#     return series"
      ],
      "execution_count": 3,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "YgvqMObHnOzy",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "n_steps = 50\n",
        "series = generate_time_series(10000, n_steps + 1)\n",
        "X_train, y_train = series[:7000, :n_steps], series[:7000, -1]\n",
        "X_valid, y_valid = series[7000:9000, :n_steps], series[7000:9000, -1]\n",
        "X_test, y_test = series[9000:, :n_steps], series[9000:, -1]"
      ],
      "execution_count": 4,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "2szjmPZCnOz1",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def plot_learning_curves(loss, val_loss):\n",
        "    plt.plot(np.arange(len(loss)) + 0.5, loss, \"b.-\", label=\"Training loss\")\n",
        "    plt.plot(np.arange(len(val_loss)) + 1, val_loss, \"r.-\", label=\"Validation loss\")\n",
        "    plt.gca().xaxis.set_major_locator(mpl.ticker.MaxNLocator(integer=True))\n",
        "    plt.legend()\n",
        "    plt.xlabel(\"Epochs\")\n",
        "    plt.ylabel(\"Loss\")\n",
        "    plt.grid(True)\n"
      ],
      "execution_count": 5,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "cxDr2SNenOz4",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 296
        },
        "outputId": "9319cd5e-e63b-4533-ed57-59e5d563365c"
      },
      "source": [
        "def plot_series(series, y=None, y_pred=None, x_label=\"$t$\", y_label=\"$x(t)$\"):\n",
        "    plt.plot(series, \".-\")\n",
        "    if y is not None:\n",
        "        plt.plot(n_steps, y, \"bx\", markersize=10)\n",
        "    if y_pred is not None:\n",
        "        plt.plot(n_steps, y_pred, \"ro\")\n",
        "    plt.grid(True)\n",
        "    if x_label:\n",
        "        plt.xlabel(x_label)\n",
        "    if y_label:\n",
        "        plt.ylabel(y_label, rotation=0)\n",
        "    plt.hlines(0, 0, 100, linewidth=1)\n",
        "    plt.axis([0, n_steps + 1, -1, 1])\n",
        "\n",
        "fig, axes = plt.subplots(nrows=1, ncols=3, sharey=True, figsize=(12, 4))\n",
        "for col in range(3):\n",
        "    plt.sca(axes[col])\n",
        "    plot_series(X_valid[col, :, 0], y_valid[col, 0],\n",
        "                y_label=(\"$x(t)$\" if col==0 else None))\n",
        "plt.tight_layout()\n",
        "plt.show()"
      ],
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 864x288 with 3 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ETecABpjnOz7",
        "colab_type": "text"
      },
      "source": [
        "## Computing Some Baselines"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "oEq1FwT4nOz8",
        "colab_type": "text"
      },
      "source": [
        "Before we start using RNNs, it is often a good idea to have a few baseline metrics, or else we may end up thinking our model works great when in fact it is doing worse than basic models. For example, the simplest approach is to predict the last value in each series. This is called naive forecasting, and it is sometimes surprisingly difficult to outperform."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "GgLabpnonOz8",
        "colab_type": "text"
      },
      "source": [
        "\n",
        "Naive predictions (just predict the last observed value):"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "jnppPD_lnOz9",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "99aed43c-fc0e-4253-8d3e-d357514e2de5"
      },
      "source": [
        "y_pred = X_valid[:, -1]\n",
        "np.mean(keras.losses.mean_squared_error(y_valid, y_pred))"
      ],
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0.020416105"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 7
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "nx0lhpyonO0B",
        "colab_type": "text"
      },
      "source": [
        "Another simple approach is to use a fully connected network. Since it expects a flat list of features for each input, we need to add a Flatten layer. Let’s just use a simple Linear Regression model so that each prediction will be a linear combination of the values in the time series:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8pDGWar6Gr4w",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "keras.backend.clear_session()"
      ],
      "execution_count": 8,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "5ZBUo4vqnO0C",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "model = keras.models.Sequential([\n",
        "    keras.layers.Flatten(input_shape = [50,1]),\n",
        "    keras.layers.Dense(1)\n",
        "])"
      ],
      "execution_count": 9,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "5pmF3z10nO0F",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 727
        },
        "outputId": "bbd58090-9d0f-4794-d08f-e3c9169346c6"
      },
      "source": [
        "model.compile(loss=\"mse\", optimizer=\"adam\")\n",
        "history = model.fit(X_train, y_train, epochs=20, validation_data=(X_valid, y_valid))"
      ],
      "execution_count": 10,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/20\n",
            "219/219 [==============================] - 1s 3ms/step - loss: 0.0695 - val_loss: 0.0308\n",
            "Epoch 2/20\n",
            "219/219 [==============================] - 0s 2ms/step - loss: 0.0211 - val_loss: 0.0145\n",
            "Epoch 3/20\n",
            "219/219 [==============================] - 0s 2ms/step - loss: 0.0111 - val_loss: 0.0089\n",
            "Epoch 4/20\n",
            "219/219 [==============================] - 0s 2ms/step - loss: 0.0076 - val_loss: 0.0069\n",
            "Epoch 5/20\n",
            "219/219 [==============================] - 0s 2ms/step - loss: 0.0063 - val_loss: 0.0060\n",
            "Epoch 6/20\n",
            "219/219 [==============================] - 0s 2ms/step - loss: 0.0056 - val_loss: 0.0054\n",
            "Epoch 7/20\n",
            "219/219 [==============================] - 0s 2ms/step - loss: 0.0052 - val_loss: 0.0050\n",
            "Epoch 8/20\n",
            "219/219 [==============================] - 0s 2ms/step - loss: 0.0048 - val_loss: 0.0048\n",
            "Epoch 9/20\n",
            "219/219 [==============================] - 0s 2ms/step - loss: 0.0045 - val_loss: 0.0045\n",
            "Epoch 10/20\n",
            "219/219 [==============================] - 0s 2ms/step - loss: 0.0043 - val_loss: 0.0042\n",
            "Epoch 11/20\n",
            "219/219 [==============================] - 0s 2ms/step - loss: 0.0042 - val_loss: 0.0041\n",
            "Epoch 12/20\n",
            "219/219 [==============================] - 0s 2ms/step - loss: 0.0040 - val_loss: 0.0042\n",
            "Epoch 13/20\n",
            "219/219 [==============================] - 0s 2ms/step - loss: 0.0039 - val_loss: 0.0040\n",
            "Epoch 14/20\n",
            "219/219 [==============================] - 0s 2ms/step - loss: 0.0039 - val_loss: 0.0039\n",
            "Epoch 15/20\n",
            "219/219 [==============================] - 0s 2ms/step - loss: 0.0038 - val_loss: 0.0039\n",
            "Epoch 16/20\n",
            "219/219 [==============================] - 0s 2ms/step - loss: 0.0037 - val_loss: 0.0038\n",
            "Epoch 17/20\n",
            "219/219 [==============================] - 0s 2ms/step - loss: 0.0037 - val_loss: 0.0037\n",
            "Epoch 18/20\n",
            "219/219 [==============================] - 0s 2ms/step - loss: 0.0036 - val_loss: 0.0036\n",
            "Epoch 19/20\n",
            "219/219 [==============================] - 0s 2ms/step - loss: 0.0036 - val_loss: 0.0036\n",
            "Epoch 20/20\n",
            "219/219 [==============================] - 0s 2ms/step - loss: 0.0035 - val_loss: 0.0035\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "_kg5S1a_nO0H",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        },
        "outputId": "91ab7498-6423-4c38-c6b4-bbb1dfafc7b7"
      },
      "source": [
        "model.evaluate(X_test, y_test)"
      ],
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "32/32 [==============================] - 0s 1ms/step - loss: 0.0035\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0.003532787086442113"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 11
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "3VlNgMuYnO0K",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 278
        },
        "outputId": "7c9aedde-81d1-4ce2-b217-057d57f44a24"
      },
      "source": [
        "plot_learning_curves(history.history[\"loss\"], history.history[\"val_loss\"])\n",
        "plt.show()"
      ],
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "89X6uwMqnO0M",
        "colab_type": "text"
      },
      "source": [
        "## Implementing a Simple RNN"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GxTIjlEdGtDp",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "keras.backend.clear_session()"
      ],
      "execution_count": 13,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "8napkRmTnO0N",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 727
        },
        "outputId": "9bb2f671-2560-4ef6-b7f1-689b5f992b8a"
      },
      "source": [
        "model = keras.models.Sequential([\n",
        "  keras.layers.SimpleRNN(1, input_shape=[None, 1])\n",
        "])\n",
        "\n",
        "model.compile(loss=\"mse\", optimizer=\"adam\")\n",
        "history = model.fit(X_train, y_train, epochs=20, validation_data=(X_valid, y_valid))"
      ],
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/20\n",
            "219/219 [==============================] - 7s 33ms/step - loss: 0.6748 - val_loss: 0.5277\n",
            "Epoch 2/20\n",
            "219/219 [==============================] - 7s 31ms/step - loss: 0.3907 - val_loss: 0.3146\n",
            "Epoch 3/20\n",
            "219/219 [==============================] - 7s 31ms/step - loss: 0.2993 - val_loss: 0.2650\n",
            "Epoch 4/20\n",
            "219/219 [==============================] - 7s 32ms/step - loss: 0.2512 - val_loss: 0.2221\n",
            "Epoch 5/20\n",
            "219/219 [==============================] - 7s 32ms/step - loss: 0.2090 - val_loss: 0.1876\n",
            "Epoch 6/20\n",
            "219/219 [==============================] - 7s 31ms/step - loss: 0.1717 - val_loss: 0.1566\n",
            "Epoch 7/20\n",
            "219/219 [==============================] - 7s 31ms/step - loss: 0.1464 - val_loss: 0.1438\n",
            "Epoch 8/20\n",
            "219/219 [==============================] - 7s 31ms/step - loss: 0.1392 - val_loss: 0.1417\n",
            "Epoch 9/20\n",
            "219/219 [==============================] - 7s 31ms/step - loss: 0.1395 - val_loss: 0.1467\n",
            "Epoch 10/20\n",
            "219/219 [==============================] - 7s 32ms/step - loss: 0.1381 - val_loss: 0.1428\n",
            "Epoch 11/20\n",
            "219/219 [==============================] - 7s 32ms/step - loss: 0.1388 - val_loss: 0.1415\n",
            "Epoch 12/20\n",
            "219/219 [==============================] - 7s 31ms/step - loss: 0.1386 - val_loss: 0.1422\n",
            "Epoch 13/20\n",
            "219/219 [==============================] - 7s 32ms/step - loss: 0.1382 - val_loss: 0.1485\n",
            "Epoch 14/20\n",
            "219/219 [==============================] - 7s 31ms/step - loss: 0.1387 - val_loss: 0.1440\n",
            "Epoch 15/20\n",
            "219/219 [==============================] - 7s 32ms/step - loss: 0.1383 - val_loss: 0.1410\n",
            "Epoch 16/20\n",
            "219/219 [==============================] - 7s 32ms/step - loss: 0.1387 - val_loss: 0.1411\n",
            "Epoch 17/20\n",
            "219/219 [==============================] - 7s 32ms/step - loss: 0.1383 - val_loss: 0.1412\n",
            "Epoch 18/20\n",
            "219/219 [==============================] - 7s 31ms/step - loss: 0.1380 - val_loss: 0.1413\n",
            "Epoch 19/20\n",
            "219/219 [==============================] - 7s 31ms/step - loss: 0.1382 - val_loss: 0.1433\n",
            "Epoch 20/20\n",
            "219/219 [==============================] - 7s 31ms/step - loss: 0.1384 - val_loss: 0.1415\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "TzUU6Wk7nO0T",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        },
        "outputId": "983ecd66-2ad4-4b08-97f2-4416710da6ac"
      },
      "source": [
        "model.evaluate(X_test, y_test)"
      ],
      "execution_count": 15,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "32/32 [==============================] - 0s 4ms/step - loss: 0.1372\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0.13721401989459991"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 15
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "DyoEqn5fnO0W",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 283
        },
        "outputId": "0522bc2b-c4e1-427a-cb5c-e3e7fe61f054"
      },
      "source": [
        "plot_learning_curves(history.history[\"loss\"], history.history[\"val_loss\"])\n",
        "plt.show()"
      ],
      "execution_count": 16,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5TY7QYFsGuE_",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "keras.backend.clear_session()"
      ],
      "execution_count": 17,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "fSy7STAhnO0Y",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "model = keras.models.Sequential([\n",
        "    keras.layers.SimpleRNN(20, return_sequences=True, input_shape=[None, 1]),\n",
        "    keras.layers.SimpleRNN(20, return_sequences=True),\n",
        "    keras.layers.SimpleRNN(1)\n",
        "])"
      ],
      "execution_count": 18,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "E8ZhZmrVnO0b",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 727
        },
        "outputId": "a7fa2da2-1a47-4e83-e91f-84934f0b2daf"
      },
      "source": [
        "model.compile(loss=\"mse\", optimizer=\"adam\")\n",
        "history = model.fit(X_train, y_train, epochs=20, validation_data=(X_valid, y_valid))"
      ],
      "execution_count": 19,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/20\n",
            "219/219 [==============================] - 21s 95ms/step - loss: 0.0343 - val_loss: 0.0057\n",
            "Epoch 2/20\n",
            "219/219 [==============================] - 20s 93ms/step - loss: 0.0043 - val_loss: 0.0039\n",
            "Epoch 3/20\n",
            "219/219 [==============================] - 20s 92ms/step - loss: 0.0035 - val_loss: 0.0034\n",
            "Epoch 4/20\n",
            "219/219 [==============================] - 20s 90ms/step - loss: 0.0033 - val_loss: 0.0034\n",
            "Epoch 5/20\n",
            "219/219 [==============================] - 20s 91ms/step - loss: 0.0032 - val_loss: 0.0033\n",
            "Epoch 6/20\n",
            "219/219 [==============================] - 20s 89ms/step - loss: 0.0031 - val_loss: 0.0033\n",
            "Epoch 7/20\n",
            "219/219 [==============================] - 20s 91ms/step - loss: 0.0031 - val_loss: 0.0032\n",
            "Epoch 8/20\n",
            "219/219 [==============================] - 19s 89ms/step - loss: 0.0031 - val_loss: 0.0033\n",
            "Epoch 9/20\n",
            "219/219 [==============================] - 20s 90ms/step - loss: 0.0030 - val_loss: 0.0031\n",
            "Epoch 10/20\n",
            "219/219 [==============================] - 20s 91ms/step - loss: 0.0030 - val_loss: 0.0032\n",
            "Epoch 11/20\n",
            "219/219 [==============================] - 20s 89ms/step - loss: 0.0030 - val_loss: 0.0031\n",
            "Epoch 12/20\n",
            "219/219 [==============================] - 19s 89ms/step - loss: 0.0030 - val_loss: 0.0030\n",
            "Epoch 13/20\n",
            "219/219 [==============================] - 20s 89ms/step - loss: 0.0029 - val_loss: 0.0031\n",
            "Epoch 14/20\n",
            "219/219 [==============================] - 19s 88ms/step - loss: 0.0030 - val_loss: 0.0030\n",
            "Epoch 15/20\n",
            "219/219 [==============================] - 19s 88ms/step - loss: 0.0030 - val_loss: 0.0033\n",
            "Epoch 16/20\n",
            "219/219 [==============================] - 19s 88ms/step - loss: 0.0029 - val_loss: 0.0031\n",
            "Epoch 17/20\n",
            "219/219 [==============================] - 20s 91ms/step - loss: 0.0029 - val_loss: 0.0031\n",
            "Epoch 18/20\n",
            "219/219 [==============================] - 20s 91ms/step - loss: 0.0029 - val_loss: 0.0030\n",
            "Epoch 19/20\n",
            "219/219 [==============================] - 19s 87ms/step - loss: 0.0029 - val_loss: 0.0032\n",
            "Epoch 20/20\n",
            "219/219 [==============================] - 19s 88ms/step - loss: 0.0029 - val_loss: 0.0032\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "TyGowSwLnO0i",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        },
        "outputId": "e23c1c58-4aeb-4856-eea4-2062922d6cad"
      },
      "source": [
        "model.evaluate(X_test, y_test)"
      ],
      "execution_count": 20,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "32/32 [==============================] - 0s 10ms/step - loss: 0.0032\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0.003191437106579542"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 20
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "EGLWRESZnO0n",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 278
        },
        "outputId": "a339ba4d-03a4-4844-ec42-8223273c9c4b"
      },
      "source": [
        "plot_learning_curves(history.history[\"loss\"], history.history[\"val_loss\"])\n",
        "plt.show()"
      ],
      "execution_count": 21,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PRaKnlJMGvD1",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "keras.backend.clear_session()"
      ],
      "execution_count": 22,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "9XOzRe02nO0q",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "model = keras.models.Sequential([\n",
        "    keras.layers.SimpleRNN(20, return_sequences = True, input_shape=[None, 1]),\n",
        "    keras.layers.SimpleRNN(20),\n",
        "    keras.layers.Dense(1)\n",
        "])"
      ],
      "execution_count": 23,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "A9niQOIfnO0s",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 727
        },
        "outputId": "b13007d8-3dcd-4184-af87-45b18a15ef29"
      },
      "source": [
        "model.compile(loss=\"mse\", optimizer=\"adam\")\n",
        "history = model.fit(X_train, y_train, epochs=20, validation_data=(X_valid, y_valid))"
      ],
      "execution_count": 24,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/20\n",
            "219/219 [==============================] - 13s 61ms/step - loss: 0.0309 - val_loss: 0.0082\n",
            "Epoch 2/20\n",
            "219/219 [==============================] - 13s 60ms/step - loss: 0.0060 - val_loss: 0.0048\n",
            "Epoch 3/20\n",
            "219/219 [==============================] - 13s 61ms/step - loss: 0.0041 - val_loss: 0.0041\n",
            "Epoch 4/20\n",
            "219/219 [==============================] - 14s 62ms/step - loss: 0.0036 - val_loss: 0.0033\n",
            "Epoch 5/20\n",
            "219/219 [==============================] - 13s 61ms/step - loss: 0.0031 - val_loss: 0.0030\n",
            "Epoch 6/20\n",
            "219/219 [==============================] - 13s 61ms/step - loss: 0.0030 - val_loss: 0.0029\n",
            "Epoch 7/20\n",
            "219/219 [==============================] - 13s 61ms/step - loss: 0.0029 - val_loss: 0.0030\n",
            "Epoch 8/20\n",
            "219/219 [==============================] - 13s 61ms/step - loss: 0.0029 - val_loss: 0.0029\n",
            "Epoch 9/20\n",
            "219/219 [==============================] - 13s 60ms/step - loss: 0.0028 - val_loss: 0.0028\n",
            "Epoch 10/20\n",
            "219/219 [==============================] - 13s 60ms/step - loss: 0.0028 - val_loss: 0.0028\n",
            "Epoch 11/20\n",
            "219/219 [==============================] - 13s 61ms/step - loss: 0.0028 - val_loss: 0.0030\n",
            "Epoch 12/20\n",
            "219/219 [==============================] - 13s 60ms/step - loss: 0.0029 - val_loss: 0.0028\n",
            "Epoch 13/20\n",
            "219/219 [==============================] - 13s 60ms/step - loss: 0.0028 - val_loss: 0.0030\n",
            "Epoch 14/20\n",
            "219/219 [==============================] - 13s 61ms/step - loss: 0.0028 - val_loss: 0.0029\n",
            "Epoch 15/20\n",
            "219/219 [==============================] - 13s 61ms/step - loss: 0.0027 - val_loss: 0.0029\n",
            "Epoch 16/20\n",
            "219/219 [==============================] - 13s 61ms/step - loss: 0.0028 - val_loss: 0.0027\n",
            "Epoch 17/20\n",
            "219/219 [==============================] - 13s 61ms/step - loss: 0.0028 - val_loss: 0.0027\n",
            "Epoch 18/20\n",
            "219/219 [==============================] - 13s 61ms/step - loss: 0.0027 - val_loss: 0.0026\n",
            "Epoch 19/20\n",
            "219/219 [==============================] - 14s 62ms/step - loss: 0.0027 - val_loss: 0.0026\n",
            "Epoch 20/20\n",
            "219/219 [==============================] - 14s 63ms/step - loss: 0.0027 - val_loss: 0.0026\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "bj6fcXAfnO0v",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        },
        "outputId": "03b304c8-e07a-4138-b134-1bc1692b774a"
      },
      "source": [
        "model.evaluate(X_test, y_test)"
      ],
      "execution_count": 25,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "32/32 [==============================] - 0s 8ms/step - loss: 0.0025\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0.0025107134133577347"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 25
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "azluoo5AnO0x",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 278
        },
        "outputId": "7e24901b-ebae-4332-e117-fefa119b3f11"
      },
      "source": [
        "plot_learning_curves(history.history[\"loss\"], history.history[\"val_loss\"])\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": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "lYGgsDJ3nO0z",
        "colab_type": "text"
      },
      "source": [
        "## Forecasting Several Steps Ahead"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "-dsEUzIlnO00",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "np.random.seed(42)\n",
        "\n",
        "series = generate_time_series(1, n_steps + 10)\n",
        "X_new, Y_new = series[:, :n_steps], series[:, n_steps:]\n",
        "X = X_new\n",
        "for step_ahead in range(10):\n",
        "    y_pred_one = model.predict(X[:, step_ahead:])[:, np.newaxis, :]\n",
        "    X = np.concatenate([X, y_pred_one], axis=1)\n",
        "\n",
        "Y_pred = X[:, n_steps:]"
      ],
      "execution_count": 27,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "A7kfFK-znO03",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def plot_multiple_forecasts(X, Y, Y_pred):\n",
        "    n_steps = X.shape[1]\n",
        "    ahead = Y.shape[1]\n",
        "    plot_series(X[0, :, 0])\n",
        "    plt.plot(np.arange(n_steps, n_steps + ahead), Y[0, :, 0], \"ro-\", label=\"Actual\")\n",
        "    plt.plot(np.arange(n_steps, n_steps + ahead), Y_pred[0, :, 0], \"bx-\", label=\"Forecast\", markersize=10)\n",
        "    plt.axis([0, n_steps + ahead, -1, 1])\n",
        "    plt.legend()"
      ],
      "execution_count": 28,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "joA3rr4jnO06",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 284
        },
        "outputId": "1f521b1a-67fc-4228-9928-70a2d2a22620"
      },
      "source": [
        "plot_multiple_forecasts(X_new, Y_new, Y_pred)"
      ],
      "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": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "fMMwf_GdnO0-",
        "colab_type": "text"
      },
      "source": [
        "Now let's use this model to predict the next 10 values. We first need to regenerate the sequences with 9 more time steps."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "HLnknMabGyka",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "keras.backend.clear_session()"
      ],
      "execution_count": 30,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "MLq3qZeInO0_",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "series = generate_time_series(10000, n_steps + 10)\n",
        "X_train, Y_train = series[:7000, :n_steps], series[:7000, -10:, 0]\n",
        "X_valid, Y_valid = series[7000:9000, :n_steps], series[7000:9000, -10:, 0]\n",
        "X_test, Y_test = series[9000:, :n_steps], series[9000:, -10:, 0]"
      ],
      "execution_count": 31,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "NzpawjOJnO1B",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "model = keras.models.Sequential([\n",
        "    keras.layers.SimpleRNN(20, return_sequences=True, input_shape=[None, 1]),\n",
        "    keras.layers.SimpleRNN(20),\n",
        "    keras.layers.Dense(10)\n",
        "])"
      ],
      "execution_count": 32,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "AVuSB_xTnO1H",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 727
        },
        "outputId": "108357e9-89fa-4029-a781-04c76a8f5c27"
      },
      "source": [
        "model.compile(loss=\"mse\", optimizer=\"adam\")\n",
        "history = model.fit(X_train, Y_train, epochs=20, validation_data=(X_valid, Y_valid))"
      ],
      "execution_count": 33,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/20\n",
            "219/219 [==============================] - 14s 63ms/step - loss: 0.0504 - val_loss: 0.0280\n",
            "Epoch 2/20\n",
            "219/219 [==============================] - 13s 61ms/step - loss: 0.0237 - val_loss: 0.0180\n",
            "Epoch 3/20\n",
            "219/219 [==============================] - 13s 61ms/step - loss: 0.0168 - val_loss: 0.0138\n",
            "Epoch 4/20\n",
            "219/219 [==============================] - 13s 61ms/step - loss: 0.0133 - val_loss: 0.0110\n",
            "Epoch 5/20\n",
            "219/219 [==============================] - 13s 61ms/step - loss: 0.0124 - val_loss: 0.0107\n",
            "Epoch 6/20\n",
            "219/219 [==============================] - 13s 62ms/step - loss: 0.0109 - val_loss: 0.0124\n",
            "Epoch 7/20\n",
            "219/219 [==============================] - 13s 60ms/step - loss: 0.0106 - val_loss: 0.0120\n",
            "Epoch 8/20\n",
            "219/219 [==============================] - 13s 60ms/step - loss: 0.0103 - val_loss: 0.0105\n",
            "Epoch 9/20\n",
            "219/219 [==============================] - 13s 61ms/step - loss: 0.0098 - val_loss: 0.0088\n",
            "Epoch 10/20\n",
            "219/219 [==============================] - 13s 60ms/step - loss: 0.0094 - val_loss: 0.0096\n",
            "Epoch 11/20\n",
            "219/219 [==============================] - 13s 60ms/step - loss: 0.0095 - val_loss: 0.0091\n",
            "Epoch 12/20\n",
            "219/219 [==============================] - 13s 60ms/step - loss: 0.0096 - val_loss: 0.0092\n",
            "Epoch 13/20\n",
            "219/219 [==============================] - 13s 60ms/step - loss: 0.0091 - val_loss: 0.0084\n",
            "Epoch 14/20\n",
            "219/219 [==============================] - 13s 60ms/step - loss: 0.0090 - val_loss: 0.0085\n",
            "Epoch 15/20\n",
            "219/219 [==============================] - 13s 62ms/step - loss: 0.0089 - val_loss: 0.0085\n",
            "Epoch 16/20\n",
            "219/219 [==============================] - 13s 61ms/step - loss: 0.0089 - val_loss: 0.0077\n",
            "Epoch 17/20\n",
            "219/219 [==============================] - 13s 61ms/step - loss: 0.0087 - val_loss: 0.0079\n",
            "Epoch 18/20\n",
            "219/219 [==============================] - 13s 60ms/step - loss: 0.0087 - val_loss: 0.0082\n",
            "Epoch 19/20\n",
            "219/219 [==============================] - 14s 62ms/step - loss: 0.0087 - val_loss: 0.0095\n",
            "Epoch 20/20\n",
            "219/219 [==============================] - 14s 62ms/step - loss: 0.0090 - val_loss: 0.0086\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "nYW5-Jf5nO1L",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        },
        "outputId": "aa2b9cec-4cb3-4f8f-ced8-467d9230302c"
      },
      "source": [
        "model.evaluate(X_test, Y_test)"
      ],
      "execution_count": 34,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "32/32 [==============================] - 0s 7ms/step - loss: 0.0089\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0.008948233909904957"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 34
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "SuXwIXEdnO1S",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 278
        },
        "outputId": "70775574-320a-4e2d-f058-3a956f1c5f03"
      },
      "source": [
        "plot_learning_curves(history.history[\"loss\"], history.history[\"val_loss\"])\n",
        "plt.show()"
      ],
      "execution_count": 35,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "3zanp65inO1W",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "Y_pred = model.predict(X_new)"
      ],
      "execution_count": 36,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "-hhF78GBnO1Z",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "np.random.seed(42)\n",
        "\n",
        "series = generate_time_series(1, 50 + 10)\n",
        "X_new, Y_new = series[:, :50, :], series[:, -10:, :]\n",
        "Y_pred = model.predict(X_new)[..., np.newaxis]"
      ],
      "execution_count": 37,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "iQDq4qyBnO1d",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 284
        },
        "outputId": "949f572c-a02d-4f40-84ba-a2c8800f08bf"
      },
      "source": [
        "plot_multiple_forecasts(X_new, Y_new, Y_pred)\n",
        "plt.show()"
      ],
      "execution_count": 38,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ic2rKEHUnO1g",
        "colab_type": "text"
      },
      "source": [
        "\n",
        "Now let's create an RNN that predicts the next 10 steps at each time step. That is, instead of just forecasting time steps 50 to 59 based on time steps 0 to 49, it will forecast time steps 1 to 10 at time step 0, then time steps 2 to 11 at time step 1, and so on, and finally it will forecast time steps 50 to 59 at the last time step. Notice that the model is causal: when it makes predictions at any time step, it can only see past time steps."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "f5P-qE3_nO1h",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "Y = np.empty((10000, n_steps, 10))\n",
        "for step_ahead in range(1, 10 + 1):\n",
        "    Y[:, :, step_ahead - 1] = series[:, step_ahead : step_ahead + n_steps, 0]\n",
        "Y_train = Y[:7000]\n",
        "Y_valid = Y[7000:9000]\n",
        "Y_test  = Y[9000:]"
      ],
      "execution_count": 39,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "8QDoF6dVnO1j",
        "colab_type": "text"
      },
      "source": [
        "It may be surprising that the targets will contain values that appear in the inputs (there is a lot of overlap between X_train and Y_train). Isn’t that cheating? Fortunately, not at all: at each time step, the model only knows about past time steps, so it cannot look ahead. It is said to be a causal model."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "e1Mrw4BenO1k",
        "colab_type": "text"
      },
      "source": [
        "To turn the model into a sequence-to-sequence model, we must set return_sequences=True in all recurrent layers (even the last one), and we must apply the output Dense layer at every time step."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "dVqr2Yt1G0sU",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "keras.backend.clear_session()"
      ],
      "execution_count": 40,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "LLTzgxednO1k",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "model = keras.models.Sequential([\n",
        "    keras.layers.SimpleRNN(20, return_sequences = True, input_shape=[None, 1]),\n",
        "    keras.layers.SimpleRNN(20, return_sequences = True),\n",
        "    keras.layers.TimeDistributed(keras.layers.Dense(10))\n",
        "])\n",
        "\n",
        "def last_time_step_mse(Y_true, Y_pred):\n",
        "    return keras.metrics.mean_squared_error(Y_true[:, -1], Y_pred[:, -1])"
      ],
      "execution_count": 41,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "Q7VzKPkhnO1m",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 747
        },
        "outputId": "6fb3af60-d434-49ef-8518-31e0810c2710"
      },
      "source": [
        "model.compile(loss = 'mse', optimizer = 'nadam', metrics = [last_time_step_mse])\n",
        "history = model.fit(X_train, Y_train, validation_data = (X_valid, Y_valid), epochs=20)"
      ],
      "execution_count": 42,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/20\n",
            "219/219 [==============================] - 14s 65ms/step - loss: 0.1045 - last_time_step_mse: 0.0552 - val_loss: 0.0246 - val_last_time_step_mse: 0.0175\n",
            "Epoch 2/20\n",
            "219/219 [==============================] - 14s 65ms/step - loss: 0.0214 - last_time_step_mse: 0.0193 - val_loss: 0.0119 - val_last_time_step_mse: 0.0161\n",
            "Epoch 3/20\n",
            "219/219 [==============================] - 14s 62ms/step - loss: 0.0046 - last_time_step_mse: 0.0058 - val_loss: 0.0026 - val_last_time_step_mse: 0.0042\n",
            "Epoch 4/20\n",
            "219/219 [==============================] - 14s 62ms/step - loss: 0.0024 - last_time_step_mse: 0.0032 - val_loss: 0.0033 - val_last_time_step_mse: 0.0019\n",
            "Epoch 5/20\n",
            "219/219 [==============================] - 14s 62ms/step - loss: 0.0018 - last_time_step_mse: 0.0025 - val_loss: 0.0015 - val_last_time_step_mse: 0.0026\n",
            "Epoch 6/20\n",
            "219/219 [==============================] - 14s 62ms/step - loss: 0.0017 - last_time_step_mse: 0.0024 - val_loss: 0.0039 - val_last_time_step_mse: 0.0023\n",
            "Epoch 7/20\n",
            "219/219 [==============================] - 14s 63ms/step - loss: 0.0016 - last_time_step_mse: 0.0023 - val_loss: 0.0011 - val_last_time_step_mse: 8.0034e-04\n",
            "Epoch 8/20\n",
            "219/219 [==============================] - 14s 62ms/step - loss: 0.0014 - last_time_step_mse: 0.0019 - val_loss: 0.0017 - val_last_time_step_mse: 0.0043\n",
            "Epoch 9/20\n",
            "219/219 [==============================] - 14s 62ms/step - loss: 0.0014 - last_time_step_mse: 0.0021 - val_loss: 0.0016 - val_last_time_step_mse: 7.2654e-04\n",
            "Epoch 10/20\n",
            "219/219 [==============================] - 14s 63ms/step - loss: 0.0012 - last_time_step_mse: 0.0016 - val_loss: 0.0017 - val_last_time_step_mse: 0.0045\n",
            "Epoch 11/20\n",
            "219/219 [==============================] - 14s 62ms/step - loss: 0.0011 - last_time_step_mse: 0.0015 - val_loss: 7.6239e-04 - val_last_time_step_mse: 5.7045e-04\n",
            "Epoch 12/20\n",
            "219/219 [==============================] - 14s 62ms/step - loss: 0.0011 - last_time_step_mse: 0.0016 - val_loss: 0.0014 - val_last_time_step_mse: 0.0041\n",
            "Epoch 13/20\n",
            "219/219 [==============================] - 14s 63ms/step - loss: 9.8918e-04 - last_time_step_mse: 0.0014 - val_loss: 7.1987e-04 - val_last_time_step_mse: 4.4668e-04\n",
            "Epoch 14/20\n",
            "219/219 [==============================] - 14s 63ms/step - loss: 9.9142e-04 - last_time_step_mse: 0.0015 - val_loss: 8.6490e-04 - val_last_time_step_mse: 3.3145e-04\n",
            "Epoch 15/20\n",
            "219/219 [==============================] - 14s 63ms/step - loss: 9.6675e-04 - last_time_step_mse: 0.0015 - val_loss: 9.4717e-04 - val_last_time_step_mse: 0.0025\n",
            "Epoch 16/20\n",
            "219/219 [==============================] - 14s 63ms/step - loss: 9.1589e-04 - last_time_step_mse: 0.0014 - val_loss: 6.4680e-04 - val_last_time_step_mse: 0.0012\n",
            "Epoch 17/20\n",
            "219/219 [==============================] - 14s 64ms/step - loss: 8.4446e-04 - last_time_step_mse: 0.0013 - val_loss: 9.5427e-04 - val_last_time_step_mse: 4.5127e-04\n",
            "Epoch 18/20\n",
            "219/219 [==============================] - 14s 63ms/step - loss: 7.9889e-04 - last_time_step_mse: 0.0012 - val_loss: 7.6217e-04 - val_last_time_step_mse: 0.0020\n",
            "Epoch 19/20\n",
            "219/219 [==============================] - 14s 63ms/step - loss: 7.9068e-04 - last_time_step_mse: 0.0012 - val_loss: 5.2777e-04 - val_last_time_step_mse: 3.2601e-04\n",
            "Epoch 20/20\n",
            "219/219 [==============================] - 14s 64ms/step - loss: 6.6621e-04 - last_time_step_mse: 8.7132e-04 - val_loss: 5.0061e-04 - val_last_time_step_mse: 6.4484e-04\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "sVEX0ogmnO1p",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        },
        "outputId": "b67f3c8c-e2d9-4f6b-fe67-3d8e55b74af6"
      },
      "source": [
        "model.evaluate(X_test, Y_test)"
      ],
      "execution_count": 43,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "32/32 [==============================] - 0s 8ms/step - loss: 4.9944e-04 - last_time_step_mse: 6.3864e-04\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.0004994429764337838, 0.0006386360037140548]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 43
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "x7f3_0uGnO1r",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "np.random.seed(42)\n",
        "\n",
        "series = generate_time_series(1, 50 + 10)\n",
        "X_new, Y_new = series[:, :50, :], series[:, 50:, :]\n",
        "Y_pred = model.predict(X_new)[:, -1][..., np.newaxis]"
      ],
      "execution_count": 44,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "trusted": true,
        "id": "KO3ec0krnO1w",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 284
        },
        "outputId": "6f145216-bf47-454f-8c5e-1e5d7ccbc589"
      },
      "source": [
        "plot_multiple_forecasts(X_new, Y_new, Y_pred)\n",
        "plt.show()"
      ],
      "execution_count": 45,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "LDHiW53_s3Ii",
        "colab_type": "text"
      },
      "source": [
        "## Deep RNN with Batch Norm"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "IPWFABcKG16M",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "keras.backend.clear_session()"
      ],
      "execution_count": 46,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Ht6elCyZs4QK",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 747
        },
        "outputId": "ced0a581-168f-4c9c-fa24-36647fe533f9"
      },
      "source": [
        "tf.random.set_seed(42)\n",
        "np.random.seed(42)\n",
        "\n",
        "model = keras.models.Sequential([\n",
        "                                 keras.layers.SimpleRNN(20, return_sequences=True, input_shape = [None,1]),\n",
        "                                 keras.layers.BatchNormalization(),\n",
        "                                 keras.layers.SimpleRNN(20, return_sequences=True),\n",
        "                                 keras.layers.BatchNormalization(),\n",
        "                                 keras.layers.TimeDistributed(keras.layers.Dense(10))\n",
        "])\n",
        "\n",
        "model.compile(loss = 'mse', optimizer = 'adam', metrics=[last_time_step_mse])\n",
        "history = model.fit(X_train, Y_train, validation_data=(X_valid, Y_valid), epochs=20)"
      ],
      "execution_count": 47,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/20\n",
            "219/219 [==============================] - 14s 65ms/step - loss: 0.2919 - last_time_step_mse: 0.2247 - val_loss: 0.1656 - val_last_time_step_mse: 0.1127\n",
            "Epoch 2/20\n",
            "219/219 [==============================] - 14s 63ms/step - loss: 0.1296 - last_time_step_mse: 0.0746 - val_loss: 0.0532 - val_last_time_step_mse: 0.0767\n",
            "Epoch 3/20\n",
            "219/219 [==============================] - 14s 66ms/step - loss: 0.0192 - last_time_step_mse: 0.0245 - val_loss: 0.0157 - val_last_time_step_mse: 0.0328\n",
            "Epoch 4/20\n",
            "219/219 [==============================] - 15s 66ms/step - loss: 0.0096 - last_time_step_mse: 0.0124 - val_loss: 0.0084 - val_last_time_step_mse: 0.0110\n",
            "Epoch 5/20\n",
            "219/219 [==============================] - 15s 67ms/step - loss: 0.0070 - last_time_step_mse: 0.0093 - val_loss: 0.0063 - val_last_time_step_mse: 0.0078\n",
            "Epoch 6/20\n",
            "219/219 [==============================] - 14s 65ms/step - loss: 0.0054 - last_time_step_mse: 0.0073 - val_loss: 0.0051 - val_last_time_step_mse: 0.0064\n",
            "Epoch 7/20\n",
            "219/219 [==============================] - 14s 65ms/step - loss: 0.0045 - last_time_step_mse: 0.0063 - val_loss: 0.0042 - val_last_time_step_mse: 0.0049\n",
            "Epoch 8/20\n",
            "219/219 [==============================] - 14s 65ms/step - loss: 0.0038 - last_time_step_mse: 0.0057 - val_loss: 0.0036 - val_last_time_step_mse: 0.0053\n",
            "Epoch 9/20\n",
            "219/219 [==============================] - 14s 64ms/step - loss: 0.0033 - last_time_step_mse: 0.0052 - val_loss: 0.0032 - val_last_time_step_mse: 0.0046\n",
            "Epoch 10/20\n",
            "219/219 [==============================] - 14s 66ms/step - loss: 0.0029 - last_time_step_mse: 0.0050 - val_loss: 0.0029 - val_last_time_step_mse: 0.0044\n",
            "Epoch 11/20\n",
            "219/219 [==============================] - 14s 65ms/step - loss: 0.0026 - last_time_step_mse: 0.0049 - val_loss: 0.0026 - val_last_time_step_mse: 0.0044\n",
            "Epoch 12/20\n",
            "219/219 [==============================] - 14s 64ms/step - loss: 0.0024 - last_time_step_mse: 0.0048 - val_loss: 0.0023 - val_last_time_step_mse: 0.0040\n",
            "Epoch 13/20\n",
            "219/219 [==============================] - 14s 65ms/step - loss: 0.0022 - last_time_step_mse: 0.0048 - val_loss: 0.0021 - val_last_time_step_mse: 0.0039\n",
            "Epoch 14/20\n",
            "219/219 [==============================] - 14s 65ms/step - loss: 0.0020 - last_time_step_mse: 0.0047 - val_loss: 0.0019 - val_last_time_step_mse: 0.0043\n",
            "Epoch 15/20\n",
            "219/219 [==============================] - 14s 65ms/step - loss: 0.0018 - last_time_step_mse: 0.0047 - val_loss: 0.0018 - val_last_time_step_mse: 0.0040\n",
            "Epoch 16/20\n",
            "219/219 [==============================] - 14s 64ms/step - loss: 0.0017 - last_time_step_mse: 0.0047 - val_loss: 0.0017 - val_last_time_step_mse: 0.0056\n",
            "Epoch 17/20\n",
            "219/219 [==============================] - 14s 66ms/step - loss: 0.0016 - last_time_step_mse: 0.0047 - val_loss: 0.0015 - val_last_time_step_mse: 0.0054\n",
            "Epoch 18/20\n",
            "219/219 [==============================] - 14s 65ms/step - loss: 0.0014 - last_time_step_mse: 0.0047 - val_loss: 0.0014 - val_last_time_step_mse: 0.0051\n",
            "Epoch 19/20\n",
            "219/219 [==============================] - 14s 65ms/step - loss: 0.0013 - last_time_step_mse: 0.0046 - val_loss: 0.0013 - val_last_time_step_mse: 0.0044\n",
            "Epoch 20/20\n",
            "219/219 [==============================] - 14s 66ms/step - loss: 0.0013 - last_time_step_mse: 0.0046 - val_loss: 0.0014 - val_last_time_step_mse: 0.0049\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "-V-ymj1nuFoH",
        "colab_type": "text"
      },
      "source": [
        "## Deep RNNs with Layer Norm"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "marSWly-G3Os",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "keras.backend.clear_session()"
      ],
      "execution_count": 48,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vQ5RfJXVt3eV",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "class LNSimpleRNNCell(keras.layers.Layer):\n",
        "    def __init__(self, units, activation=\"tanh\", **kwargs):\n",
        "        super().__init__(**kwargs)\n",
        "        self.state_size = units\n",
        "        self.output_size = units\n",
        "        self.simple_rnn_cell = keras.layers.SimpleRNNCell(units,\n",
        "                                                          activation=None)\n",
        "        self.layer_norm = keras.layers.LayerNormalization()\n",
        "        self.activation = keras.activations.get(activation)\n",
        "    def get_initial_state(self, inputs=None, batch_size=None, dtype=None):\n",
        "        if inputs is not None:\n",
        "            batch_size = tf.shape(inputs)[0]\n",
        "            dtype = inputs.dtype\n",
        "        return [tf.zeros([batch_size, self.state_size], dtype=dtype)]\n",
        "    def call(self, inputs, states):\n",
        "        outputs, new_states = self.simple_rnn_cell(inputs, states)\n",
        "        norm_outputs = self.activation(self.layer_norm(outputs))\n",
        "        return norm_outputs, [norm_outputs]"
      ],
      "execution_count": 49,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VUKdp3n1w3y5",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 747
        },
        "outputId": "a4d1b85a-1884-462a-e675-6dc3b777dd64"
      },
      "source": [
        "np.random.seed(42)\n",
        "tf.random.set_seed(42)\n",
        "\n",
        "model = keras.models.Sequential([\n",
        "    keras.layers.RNN(LNSimpleRNNCell(20), return_sequences=True,\n",
        "                     input_shape=[None, 1]),\n",
        "    keras.layers.RNN(LNSimpleRNNCell(20), return_sequences=True),\n",
        "    keras.layers.TimeDistributed(keras.layers.Dense(10))\n",
        "])\n",
        "\n",
        "model.compile(loss=\"mse\", optimizer=\"adam\", metrics=[last_time_step_mse])\n",
        "history = model.fit(X_train, Y_train, epochs=20,\n",
        "                    validation_data=(X_valid, Y_valid))"
      ],
      "execution_count": 50,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/20\n",
            "219/219 [==============================] - 34s 154ms/step - loss: 0.2020 - last_time_step_mse: 0.1286 - val_loss: 0.1442 - val_last_time_step_mse: 0.0764\n",
            "Epoch 2/20\n",
            "219/219 [==============================] - 34s 154ms/step - loss: 0.0987 - last_time_step_mse: 0.0607 - val_loss: 0.0160 - val_last_time_step_mse: 0.0164\n",
            "Epoch 3/20\n",
            "219/219 [==============================] - 34s 155ms/step - loss: 0.0109 - last_time_step_mse: 0.0097 - val_loss: 0.0079 - val_last_time_step_mse: 0.0067\n",
            "Epoch 4/20\n",
            "219/219 [==============================] - 34s 157ms/step - loss: 0.0063 - last_time_step_mse: 0.0061 - val_loss: 0.0051 - val_last_time_step_mse: 0.0056\n",
            "Epoch 5/20\n",
            "219/219 [==============================] - 33s 151ms/step - loss: 0.0046 - last_time_step_mse: 0.0053 - val_loss: 0.0041 - val_last_time_step_mse: 0.0050\n",
            "Epoch 6/20\n",
            "219/219 [==============================] - 33s 150ms/step - loss: 0.0039 - last_time_step_mse: 0.0051 - val_loss: 0.0036 - val_last_time_step_mse: 0.0053\n",
            "Epoch 7/20\n",
            "219/219 [==============================] - 33s 150ms/step - loss: 0.0035 - last_time_step_mse: 0.0049 - val_loss: 0.0033 - val_last_time_step_mse: 0.0051\n",
            "Epoch 8/20\n",
            "219/219 [==============================] - 33s 150ms/step - loss: 0.0032 - last_time_step_mse: 0.0048 - val_loss: 0.0030 - val_last_time_step_mse: 0.0048\n",
            "Epoch 9/20\n",
            "219/219 [==============================] - 33s 149ms/step - loss: 0.0029 - last_time_step_mse: 0.0047 - val_loss: 0.0028 - val_last_time_step_mse: 0.0047\n",
            "Epoch 10/20\n",
            "219/219 [==============================] - 33s 149ms/step - loss: 0.0028 - last_time_step_mse: 0.0047 - val_loss: 0.0027 - val_last_time_step_mse: 0.0048\n",
            "Epoch 11/20\n",
            "219/219 [==============================] - 33s 150ms/step - loss: 0.0026 - last_time_step_mse: 0.0048 - val_loss: 0.0025 - val_last_time_step_mse: 0.0048\n",
            "Epoch 12/20\n",
            "219/219 [==============================] - 33s 150ms/step - loss: 0.0025 - last_time_step_mse: 0.0048 - val_loss: 0.0023 - val_last_time_step_mse: 0.0048\n",
            "Epoch 13/20\n",
            "219/219 [==============================] - 33s 150ms/step - loss: 0.0023 - last_time_step_mse: 0.0049 - val_loss: 0.0022 - val_last_time_step_mse: 0.0050\n",
            "Epoch 14/20\n",
            "219/219 [==============================] - 33s 150ms/step - loss: 0.0021 - last_time_step_mse: 0.0051 - val_loss: 0.0021 - val_last_time_step_mse: 0.0054\n",
            "Epoch 15/20\n",
            "219/219 [==============================] - 33s 149ms/step - loss: 0.0019 - last_time_step_mse: 0.0050 - val_loss: 0.0019 - val_last_time_step_mse: 0.0049\n",
            "Epoch 16/20\n",
            "219/219 [==============================] - 33s 151ms/step - loss: 0.0018 - last_time_step_mse: 0.0048 - val_loss: 0.0016 - val_last_time_step_mse: 0.0045\n",
            "Epoch 17/20\n",
            "219/219 [==============================] - 33s 152ms/step - loss: 0.0014 - last_time_step_mse: 0.0040 - val_loss: 0.0012 - val_last_time_step_mse: 0.0027\n",
            "Epoch 18/20\n",
            "219/219 [==============================] - 33s 151ms/step - loss: 0.0010 - last_time_step_mse: 0.0021 - val_loss: 8.0829e-04 - val_last_time_step_mse: 0.0013\n",
            "Epoch 19/20\n",
            "219/219 [==============================] - 33s 151ms/step - loss: 7.4972e-04 - last_time_step_mse: 0.0011 - val_loss: 6.6110e-04 - val_last_time_step_mse: 8.0009e-04\n",
            "Epoch 20/20\n",
            "219/219 [==============================] - 33s 151ms/step - loss: 6.3238e-04 - last_time_step_mse: 7.5850e-04 - val_loss: 5.7747e-04 - val_last_time_step_mse: 5.8203e-04\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "avN9V5cX4Z01",
        "colab_type": "text"
      },
      "source": [
        "## LSTMs"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-GM92E_fG4A5",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "keras.backend.clear_session()"
      ],
      "execution_count": 51,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ZbIflYLG1M9T",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 747
        },
        "outputId": "fc21acad-66f1-4ca8-e196-d572476cc845"
      },
      "source": [
        "tf.random.set_seed(42)\n",
        "np.random.seed(42)\n",
        "\n",
        "model = keras.models.Sequential([\n",
        "        keras.layers.LSTM(20, return_sequences=True, input_shape = [None,1]),\n",
        "        keras.layers.LSTM(20, return_sequences=True),\n",
        "        keras.layers.TimeDistributed(keras.layers.Dense(10))\n",
        "])\n",
        "\n",
        "model.compile(loss = 'mse', optimizer = 'adam', metrics=[last_time_step_mse])\n",
        "history = model.fit(X_train, Y_train , epochs = 20, validation_data= (X_valid, Y_valid))"
      ],
      "execution_count": 52,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/20\n",
            "219/219 [==============================] - 2s 11ms/step - loss: 0.0947 - last_time_step_mse: 0.0434 - val_loss: 0.0244 - val_last_time_step_mse: 0.0138\n",
            "Epoch 2/20\n",
            "219/219 [==============================] - 2s 8ms/step - loss: 0.0210 - last_time_step_mse: 0.0144 - val_loss: 0.0189 - val_last_time_step_mse: 0.0149\n",
            "Epoch 3/20\n",
            "219/219 [==============================] - 2s 8ms/step - loss: 0.0169 - last_time_step_mse: 0.0144 - val_loss: 0.0136 - val_last_time_step_mse: 0.0131\n",
            "Epoch 4/20\n",
            "219/219 [==============================] - 2s 8ms/step - loss: 0.0069 - last_time_step_mse: 0.0101 - val_loss: 0.0031 - val_last_time_step_mse: 0.0052\n",
            "Epoch 5/20\n",
            "219/219 [==============================] - 2s 8ms/step - loss: 0.0025 - last_time_step_mse: 0.0035 - val_loss: 0.0021 - val_last_time_step_mse: 0.0020\n",
            "Epoch 6/20\n",
            "219/219 [==============================] - 2s 8ms/step - loss: 0.0019 - last_time_step_mse: 0.0016 - val_loss: 0.0017 - val_last_time_step_mse: 0.0013\n",
            "Epoch 7/20\n",
            "219/219 [==============================] - 2s 8ms/step - loss: 0.0015 - last_time_step_mse: 9.2853e-04 - val_loss: 0.0013 - val_last_time_step_mse: 7.7615e-04\n",
            "Epoch 8/20\n",
            "219/219 [==============================] - 2s 8ms/step - loss: 0.0012 - last_time_step_mse: 6.5593e-04 - val_loss: 0.0011 - val_last_time_step_mse: 6.5751e-04\n",
            "Epoch 9/20\n",
            "219/219 [==============================] - 2s 8ms/step - loss: 0.0010 - last_time_step_mse: 4.8798e-04 - val_loss: 9.5683e-04 - val_last_time_step_mse: 3.9740e-04\n",
            "Epoch 10/20\n",
            "219/219 [==============================] - 2s 8ms/step - loss: 9.0413e-04 - last_time_step_mse: 3.6800e-04 - val_loss: 8.4435e-04 - val_last_time_step_mse: 3.1562e-04\n",
            "Epoch 11/20\n",
            "219/219 [==============================] - 2s 8ms/step - loss: 8.1136e-04 - last_time_step_mse: 2.8569e-04 - val_loss: 8.0265e-04 - val_last_time_step_mse: 3.7657e-04\n",
            "Epoch 12/20\n",
            "219/219 [==============================] - 2s 8ms/step - loss: 7.5425e-04 - last_time_step_mse: 2.5164e-04 - val_loss: 7.3873e-04 - val_last_time_step_mse: 2.0055e-04\n",
            "Epoch 13/20\n",
            "219/219 [==============================] - 2s 8ms/step - loss: 6.9964e-04 - last_time_step_mse: 2.1419e-04 - val_loss: 7.2430e-04 - val_last_time_step_mse: 3.8358e-04\n",
            "Epoch 14/20\n",
            "219/219 [==============================] - 2s 8ms/step - loss: 6.6682e-04 - last_time_step_mse: 2.1336e-04 - val_loss: 6.4747e-04 - val_last_time_step_mse: 2.1752e-04\n",
            "Epoch 15/20\n",
            "219/219 [==============================] - 2s 8ms/step - loss: 6.3871e-04 - last_time_step_mse: 2.1588e-04 - val_loss: 6.1134e-04 - val_last_time_step_mse: 1.7605e-04\n",
            "Epoch 16/20\n",
            "219/219 [==============================] - 2s 9ms/step - loss: 6.0597e-04 - last_time_step_mse: 1.9307e-04 - val_loss: 6.4596e-04 - val_last_time_step_mse: 3.7976e-04\n",
            "Epoch 17/20\n",
            "219/219 [==============================] - 2s 9ms/step - loss: 5.9271e-04 - last_time_step_mse: 2.1497e-04 - val_loss: 5.7992e-04 - val_last_time_step_mse: 2.0948e-04\n",
            "Epoch 18/20\n",
            "219/219 [==============================] - 2s 9ms/step - loss: 5.9724e-04 - last_time_step_mse: 2.8042e-04 - val_loss: 5.5247e-04 - val_last_time_step_mse: 1.6113e-04\n",
            "Epoch 19/20\n",
            "219/219 [==============================] - 2s 9ms/step - loss: 5.4728e-04 - last_time_step_mse: 1.7133e-04 - val_loss: 5.3839e-04 - val_last_time_step_mse: 1.6552e-04\n",
            "Epoch 20/20\n",
            "219/219 [==============================] - 2s 9ms/step - loss: 5.3454e-04 - last_time_step_mse: 1.6842e-04 - val_loss: 5.2608e-04 - val_last_time_step_mse: 1.5948e-04\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "f6RRFs_E54lQ",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        },
        "outputId": "8236f7ef-3709-4df0-ab6a-ebce9b37f7b6"
      },
      "source": [
        "model.evaluate(X_valid, Y_valid)"
      ],
      "execution_count": 53,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "63/63 [==============================] - 0s 4ms/step - loss: 5.2608e-04 - last_time_step_mse: 1.5948e-04\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.0005260783946141601, 0.00015948005602695048]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 53
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "AvbBZToF56K-",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 278
        },
        "outputId": "7db8c3cb-c73b-453d-f187-749eef48d3cd"
      },
      "source": [
        "plot_learning_curves(history.history[\"loss\"], history.history[\"val_loss\"])\n",
        "plt.show()"
      ],
      "execution_count": 54,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "H2mF4xov6Y_v",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 284
        },
        "outputId": "f2496fa7-e87d-4dfa-a499-0c49860ae50a"
      },
      "source": [
        "np.random.seed(42)\n",
        "\n",
        "series = generate_time_series(1, 50 + 10)\n",
        "X_new, Y_new = series[:, :50, :], series[:, 50:, :]\n",
        "Y_pred = model.predict(X_new)[:, -1][..., np.newaxis]\n",
        "\n",
        "\n",
        "plot_multiple_forecasts(X_new, Y_new, Y_pred)\n",
        "plt.show()"
      ],
      "execution_count": 55,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3F3mc2iGG5QN",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "keras.backend.clear_session()"
      ],
      "execution_count": 56,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "RtMaiSPM5oFx",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 727
        },
        "outputId": "56001634-ef60-4730-c457-6afcb9d85b05"
      },
      "source": [
        "model = keras.models.Sequential([\n",
        "    keras.layers.RNN(keras.layers.LSTMCell(20), return_sequences=True,\n",
        "                     input_shape=[None, 1]),\n",
        "    keras.layers.RNN(keras.layers.LSTMCell(20), return_sequences=True),\n",
        "    keras.layers.TimeDistributed(keras.layers.Dense(10))\n",
        "])\n",
        "\n",
        "model.compile(loss = 'mse', optimizer = 'adam', metrics=[last_time_step_mse])\n",
        "history = model.fit(X_train, Y_train , epochs = 20, validation_data= (X_valid, Y_valid))"
      ],
      "execution_count": 57,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/20\n",
            "219/219 [==============================] - 32s 148ms/step - loss: 0.1002 - last_time_step_mse: 0.0519 - val_loss: 0.0299 - val_last_time_step_mse: 0.0322\n",
            "Epoch 2/20\n",
            "219/219 [==============================] - 32s 145ms/step - loss: 0.0226 - last_time_step_mse: 0.0170 - val_loss: 0.0194 - val_last_time_step_mse: 0.0136\n",
            "Epoch 3/20\n",
            "219/219 [==============================] - 32s 147ms/step - loss: 0.0174 - last_time_step_mse: 0.0178 - val_loss: 0.0145 - val_last_time_step_mse: 0.0227\n",
            "Epoch 4/20\n",
            "219/219 [==============================] - 32s 147ms/step - loss: 0.0100 - last_time_step_mse: 0.0224 - val_loss: 0.0059 - val_last_time_step_mse: 0.0166\n",
            "Epoch 5/20\n",
            "219/219 [==============================] - 32s 146ms/step - loss: 0.0040 - last_time_step_mse: 0.0093 - val_loss: 0.0035 - val_last_time_step_mse: 0.0058\n",
            "Epoch 6/20\n",
            "219/219 [==============================] - 32s 147ms/step - loss: 0.0023 - last_time_step_mse: 0.0027 - val_loss: 0.0020 - val_last_time_step_mse: 0.0016\n",
            "Epoch 7/20\n",
            "219/219 [==============================] - 32s 148ms/step - loss: 0.0018 - last_time_step_mse: 0.0011 - val_loss: 0.0017 - val_last_time_step_mse: 8.7267e-04\n",
            "Epoch 8/20\n",
            "219/219 [==============================] - 32s 146ms/step - loss: 0.0015 - last_time_step_mse: 7.3339e-04 - val_loss: 0.0014 - val_last_time_step_mse: 6.7477e-04\n",
            "Epoch 9/20\n",
            "219/219 [==============================] - 32s 147ms/step - loss: 0.0014 - last_time_step_mse: 6.1592e-04 - val_loss: 0.0013 - val_last_time_step_mse: 5.7170e-04\n",
            "Epoch 10/20\n",
            "219/219 [==============================] - 33s 149ms/step - loss: 0.0012 - last_time_step_mse: 5.1126e-04 - val_loss: 0.0011 - val_last_time_step_mse: 4.5625e-04\n",
            "Epoch 11/20\n",
            "219/219 [==============================] - 32s 147ms/step - loss: 0.0010 - last_time_step_mse: 4.5768e-04 - val_loss: 9.6164e-04 - val_last_time_step_mse: 3.8801e-04\n",
            "Epoch 12/20\n",
            "219/219 [==============================] - 32s 146ms/step - loss: 9.1834e-04 - last_time_step_mse: 4.1003e-04 - val_loss: 8.5725e-04 - val_last_time_step_mse: 4.0059e-04\n",
            "Epoch 13/20\n",
            "219/219 [==============================] - 32s 146ms/step - loss: 8.2173e-04 - last_time_step_mse: 3.6174e-04 - val_loss: 7.7657e-04 - val_last_time_step_mse: 3.2621e-04\n",
            "Epoch 14/20\n",
            "219/219 [==============================] - 32s 147ms/step - loss: 7.5350e-04 - last_time_step_mse: 3.2475e-04 - val_loss: 7.1575e-04 - val_last_time_step_mse: 2.9590e-04\n",
            "Epoch 15/20\n",
            "219/219 [==============================] - 32s 147ms/step - loss: 7.0067e-04 - last_time_step_mse: 2.8839e-04 - val_loss: 6.8017e-04 - val_last_time_step_mse: 2.5337e-04\n",
            "Epoch 16/20\n",
            "219/219 [==============================] - 32s 147ms/step - loss: 6.6638e-04 - last_time_step_mse: 2.6208e-04 - val_loss: 6.5872e-04 - val_last_time_step_mse: 3.2716e-04\n",
            "Epoch 17/20\n",
            "219/219 [==============================] - 33s 149ms/step - loss: 6.4175e-04 - last_time_step_mse: 2.4635e-04 - val_loss: 6.5416e-04 - val_last_time_step_mse: 1.9314e-04\n",
            "Epoch 18/20\n",
            "219/219 [==============================] - 33s 149ms/step - loss: 6.1777e-04 - last_time_step_mse: 2.2772e-04 - val_loss: 5.9771e-04 - val_last_time_step_mse: 2.1158e-04\n",
            "Epoch 19/20\n",
            "219/219 [==============================] - 33s 149ms/step - loss: 6.0413e-04 - last_time_step_mse: 2.1975e-04 - val_loss: 6.1062e-04 - val_last_time_step_mse: 1.8122e-04\n",
            "Epoch 20/20\n",
            "219/219 [==============================] - 33s 150ms/step - loss: 5.8213e-04 - last_time_step_mse: 1.9906e-04 - val_loss: 5.8396e-04 - val_last_time_step_mse: 2.1880e-04\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "b-T1xPq-BFOE",
        "colab_type": "text"
      },
      "source": [
        "## GRU CELLS \n",
        "Gated Recurrent Unit (GRU) cell"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "d6c0-9ryG57E",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "keras.backend.clear_session()"
      ],
      "execution_count": 58,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jxRgtlx05sYi",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 727
        },
        "outputId": "65045961-9720-4619-ae30-7f4ffa4242d2"
      },
      "source": [
        "np.random.seed(42)\n",
        "tf.random.set_seed(42)\n",
        "\n",
        "model = keras.models.Sequential([\n",
        "    keras.layers.GRU(20, return_sequences=True, input_shape=[None, 1]),\n",
        "    keras.layers.GRU(20, return_sequences=True),\n",
        "    keras.layers.TimeDistributed(keras.layers.Dense(10))\n",
        "])\n",
        "\n",
        "model.compile(loss=\"mse\", optimizer=\"adam\", metrics=[last_time_step_mse])\n",
        "history = model.fit(X_train, Y_train, epochs=20,\n",
        "                    validation_data=(X_valid, Y_valid))"
      ],
      "execution_count": 59,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/20\n",
            "219/219 [==============================] - 2s 10ms/step - loss: 0.0813 - last_time_step_mse: 0.0442 - val_loss: 0.0198 - val_last_time_step_mse: 0.0204\n",
            "Epoch 2/20\n",
            "219/219 [==============================] - 2s 8ms/step - loss: 0.0170 - last_time_step_mse: 0.0167 - val_loss: 0.0115 - val_last_time_step_mse: 0.0124\n",
            "Epoch 3/20\n",
            "219/219 [==============================] - 2s 8ms/step - loss: 0.0055 - last_time_step_mse: 0.0075 - val_loss: 0.0029 - val_last_time_step_mse: 0.0054\n",
            "Epoch 4/20\n",
            "219/219 [==============================] - 2s 8ms/step - loss: 0.0022 - last_time_step_mse: 0.0060 - val_loss: 0.0015 - val_last_time_step_mse: 0.0057\n",
            "Epoch 5/20\n",
            "219/219 [==============================] - 2s 8ms/step - loss: 0.0011 - last_time_step_mse: 0.0038 - val_loss: 8.5433e-04 - val_last_time_step_mse: 0.0022\n",
            "Epoch 6/20\n",
            "219/219 [==============================] - 2s 8ms/step - loss: 7.5435e-04 - last_time_step_mse: 0.0013 - val_loss: 6.8710e-04 - val_last_time_step_mse: 8.1520e-04\n",
            "Epoch 7/20\n",
            "219/219 [==============================] - 2s 8ms/step - loss: 6.6170e-04 - last_time_step_mse: 6.4229e-04 - val_loss: 6.3709e-04 - val_last_time_step_mse: 4.9356e-04\n",
            "Epoch 8/20\n",
            "219/219 [==============================] - 2s 8ms/step - loss: 6.1016e-04 - last_time_step_mse: 3.9880e-04 - val_loss: 5.8813e-04 - val_last_time_step_mse: 3.2589e-04\n",
            "Epoch 9/20\n",
            "219/219 [==============================] - 2s 8ms/step - loss: 5.8516e-04 - last_time_step_mse: 3.1080e-04 - val_loss: 5.6422e-04 - val_last_time_step_mse: 2.4794e-04\n",
            "Epoch 10/20\n",
            "219/219 [==============================] - 2s 8ms/step - loss: 5.5692e-04 - last_time_step_mse: 2.3462e-04 - val_loss: 5.5339e-04 - val_last_time_step_mse: 2.3388e-04\n",
            "Epoch 11/20\n",
            "219/219 [==============================] - 2s 8ms/step - loss: 5.5218e-04 - last_time_step_mse: 2.3312e-04 - val_loss: 5.2993e-04 - val_last_time_step_mse: 1.8000e-04\n",
            "Epoch 12/20\n",
            "219/219 [==============================] - 2s 8ms/step - loss: 5.2928e-04 - last_time_step_mse: 1.8126e-04 - val_loss: 5.3618e-04 - val_last_time_step_mse: 2.1267e-04\n",
            "Epoch 13/20\n",
            "219/219 [==============================] - 2s 8ms/step - loss: 5.2670e-04 - last_time_step_mse: 1.8847e-04 - val_loss: 5.1733e-04 - val_last_time_step_mse: 1.7074e-04\n",
            "Epoch 14/20\n",
            "219/219 [==============================] - 2s 8ms/step - loss: 5.1147e-04 - last_time_step_mse: 1.6010e-04 - val_loss: 5.0063e-04 - val_last_time_step_mse: 1.3660e-04\n",
            "Epoch 15/20\n",
            "219/219 [==============================] - 2s 8ms/step - loss: 5.0708e-04 - last_time_step_mse: 1.6193e-04 - val_loss: 5.0030e-04 - val_last_time_step_mse: 1.4783e-04\n",
            "Epoch 16/20\n",
            "219/219 [==============================] - 2s 8ms/step - loss: 5.0462e-04 - last_time_step_mse: 1.6771e-04 - val_loss: 5.6080e-04 - val_last_time_step_mse: 3.2689e-04\n",
            "Epoch 17/20\n",
            "219/219 [==============================] - 2s 8ms/step - loss: 4.9501e-04 - last_time_step_mse: 1.5274e-04 - val_loss: 5.1746e-04 - val_last_time_step_mse: 2.2614e-04\n",
            "Epoch 18/20\n",
            "219/219 [==============================] - 2s 8ms/step - loss: 4.9248e-04 - last_time_step_mse: 1.5645e-04 - val_loss: 4.7746e-04 - val_last_time_step_mse: 1.2559e-04\n",
            "Epoch 19/20\n",
            "219/219 [==============================] - 2s 8ms/step - loss: 4.8148e-04 - last_time_step_mse: 1.3698e-04 - val_loss: 4.7183e-04 - val_last_time_step_mse: 1.1652e-04\n",
            "Epoch 20/20\n",
            "219/219 [==============================] - 2s 8ms/step - loss: 4.7956e-04 - last_time_step_mse: 1.4189e-04 - val_loss: 4.6970e-04 - val_last_time_step_mse: 1.2115e-04\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7jthyORbDLvu",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        },
        "outputId": "6b87ac7c-e83a-4aba-b58b-ce9e7ecacbd2"
      },
      "source": [
        "model.evaluate(X_valid, Y_valid)"
      ],
      "execution_count": 60,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "63/63 [==============================] - 0s 4ms/step - loss: 4.6970e-04 - last_time_step_mse: 1.2115e-04\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.0004697019758168608, 0.00012114915443817154]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 60
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6EUjYQWxDRpw",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 278
        },
        "outputId": "afb09d3f-a2b6-42cd-e18a-f3d594966e77"
      },
      "source": [
        "plot_learning_curves(history.history[\"loss\"], history.history[\"val_loss\"])\n",
        "plt.show()"
      ],
      "execution_count": 61,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DgvH_j4fDUPo",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "np.random.seed(43)\n",
        "\n",
        "series = generate_time_series(1, 50 + 10)\n",
        "X_new, Y_new = series[:, :50, :], series[:, 50:, :]\n",
        "Y_pred = model.predict(X_new)[:, -1][..., np.newaxis]"
      ],
      "execution_count": 62,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8BPPYFVdFiNw",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 284
        },
        "outputId": "b26fa5b1-fa84-46ea-bba2-5399029f4dd5"
      },
      "source": [
        "plot_multiple_forecasts(X_new, Y_new, Y_pred)\n",
        "plt.show()"
      ],
      "execution_count": 63,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZYAAAEMCAYAAADnBZCzAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+j8jraAAAgAElEQVR4nOzdd3yUVbrA8d+ZVEJCEpLQIYXeS0JRaRFUsICuiijYVsXdq27x2vay1tVVsa/rumJFjWDEAio2NBEUAiTSewgJJBBKEkJCSJ1z/3hnwqS3mcxk8nw/n/nMvP28aU/eU56jtNYIIYQQ9mJydgGEEEK4FwksQggh7EoCixBCCLuSwCKEEMKuJLAIIYSwKwksQggh7MppgUUp9Y5S6rhSakcd25VS6l9KqVSl1Dal1BibbTcrpfZbXje3XqmFEEI0xJlPLO8BM+rZPhPob3ktAF4HUEp1Bh4FxgPjgEeVUsEOLakQQohGc1pg0VqvAXLr2WU28L42JAFBSqnuwCXAD1rrXK11HvAD9QcoIYQQrcjT2QWoR0/gsM1ypmVdXetrUEotwHjawdfXN7pPnz6OKakLMJvNmEzu22Qm99e2ufP9ufO9Aezbt++k1jqsKce4cmBpMa31YmAxwMCBA/XevXudXCLHSUxMZOrUqc4uhsPI/bVt7nx/7nxvAEqpjKYe48phNgvobbPcy7KurvVCCCFcgCsHlpXATZbeYROAfK31UeA74GKlVLCl0f5iyzohhBAuwGlVYUqppcBUIFQplYnR08sLQGv9X2AVcCmQChQBt1q25Sql/gFsspzqCa11fZ0AhBBCtCKnBRat9fUNbNfAXXVsewd4p6VlKCsrIzMzk+Li4paeyukCAwPZvXu3w6/j6+tLr1698PLycvi1hBBtk1s33jckMzOTgIAAIiIiUEo5uzgtUlBQQEBAgEOvobUmJyeHzMxMIiMjHXotIUTb5cptLA5XXFxMSEhImw8qrUUpRUhIiFs84QkhHKddBxZAgkoTyddLCNGQdh9YhBBC2JcEFhfwxRdfoJRiz5499e738ssvU1RU1OzrvPfee9x9993NPl4IIRpDAktTxMVBRASYTMZ7XJxdTrt06VImTpzI0qVL692vpYFFCCFagwSWxoqLgwULICMDtDbeFyxocXApLCzkl19+4e2332bZsmUAVFRUcN999zFs2DBGjBjBq6++yr/+9S+OHDlCbGwssbGxAPj7+1ee54svvuCWW24B4Msvv2T8+PGMHj2a6dOnc+zYsRaVUQghmqJddzeu4i9/gS1b6t6elAQlJVXXFRXBbbfBm2/WfsyoUfDyy/VedsWKFcyYMYMBAwYQEhJCSkoKGzduJD09nS1btuDp6Ulubi6dO3fmxRdfJCEhgdDQ0HrPOXHiRJKSklBK8dZbb7Fo0SJeeOGFeo8RQgh7kcDSWNWDSkPrG2np0qX8+c9/BmDu3LksXbqUgwcP8oc//AFPT+Pb07lz5yadMzMzk+uuu46jR49SWloqY06EEK1KAotVA08WREQY1V/VhYdDYmKzLpmbm8tPP/3E9u3bUUpRUVGBUoqxY8c26njbrr+2Y0vuuece7r33XmbNmkViYiKPPfZYs8onhBDNIW0sjfXUU+DnV3Wdn5+xvpmWL1/OjTfeSEZGBunp6Rw+fJjIyEhGjhzJG2+8QXl5OWAEIICAgAAKCgoqj+/atSu7d+/GbDbz1VdfVa7Pz8+nZ09jipolS5Y0u3xCCNEcElgaa948WLzYeEJRynhfvNhY30xLly7lqquuqrLu6quv5ujRo/Tp04cRI0YwcuRIPvroIwAWLFjAjBkzKhvvn3nmGS6//HLOP/98unbtWnmOxx57jGuvvZbo6OgG22OEEMLelJHr0f3VNtHX7t27GTx4sJNKZF+tkSvMyhlfN3efTEnur+1y53sDUEqlaK1jmnKMPLEIIYSwKwksQggh7EoCixBCCLuSwCKEEMKuJLAIIYSwK6cGFqXUDKXUXqVUqlLqoVq2v6SU2mJ57VNKnbLZVmGzbaWjyrhoESQk1L9PQoKxnxBCCCcGFqWUB/AaMBMYAlyvlBpiu4/W+q9a61Fa61HAq8BnNpvPWrdprWc5qpxjx8KcOXUHl4QEY3sjB8vX4OHhwahRoypf6enpzS6rPUkmZSFEcznziWUckKq1TtNalwLLgNn17H89UH9eeQeIjYX4+NqDizWoxMcb+zVHhw4d2LJlS+UrIiKiUcdZR+U7igQWIURzOTOw9AQO2yxnWtbVoJQKByKBn2xW+yqlkpVSSUqpKx1XzNqDiz2CSl22bNnChAkTGDFiBFdddRV5eXkATJ06lb/85S/ExMTwyiuvkJKSwpQpU4iOjubKK6/k6NGjAKSmpjJ9+nRGjhzJmDFjOHDgAIWFhUybNo0xY8YwfPhwVqxYAcCZM2e47LLLGDlyJMOGDePjjz+uNUV/cyWlnWTRt3tIychr2RdFCNFmOG3kvVLqGmCG1vp2y/KNwHitdY0pDpVSDwK9tNb32KzrqbXOUkpFYQScaVrrA9WOWwAsAAgLC4uOj4+vct7AwED69esHwIMP+rB9e/1x9tQpxZ49Jrp31xw9qhg0yExQUN1fv+HDzTz7bP3Zj4OCghg6dCgA4eHhfPTRR5x33nk899xzTJw4kSeffJKCggKeffZZLr30UgYOHMhLL71EWVkZM2fOZNmyZYSGhvLJJ5+QkJDAf/7zH2JjY7n33nu54oorKC4uxmw24+3tTVFREZ06dSInJ4cLL7yQLVu2sHLlSlavXs2rr74KGHnGAgMDGTZsGD///DMhISE1ypyamkp+fn699wWQdKSMN7aVogEvEzw41pd+wR4NHlebwsLCKvPPuBu5v7bLne8NIDY2tskj752Z3TgL6G2z3MuyrjZzgbtsV2itsyzvaUqpRGA0cKDaPouBxWCkdKmedmH37t2VaVC8vcGjgb95ISHQowccOqTo0wdCQuo/wNsbAgK8692nQ4cObNu2rXI5Pz+f06dPM3PmTMDID3bttdcSEBCAh4cHN954IwEBAezYsYPdu3dX5horKyurTDyZnZ3NDTfcAFB5f2VlZTz88MOsWbMGk8nE0aNHKSoqYty4cfz973/nySef5PLLL2fSpEmAkTnZ39+/1jQxvr6+jB49ut77+mb7Ud75YQvWsFtuhpKgcKZO7VfvcXVx97QZcn9tlzvfW3M5M7BsAvorpSIxAspc4IbqOymlBgHBwHqbdcFAkda6RCkVClwAtKhfVkNZ8+Fc9dfDD8Prr8Ojj9q/GqwhHTt2BEBrzdChQ1m/3viyWHOF2WY/thUXF8eJEydISUnBy8uLiIgIiouLGTBgAL/99hurVq3i73//O9OmTeORRx5pdvnKKsws+nYPb649SP8u/hzKLaKk3IwGhvcIbPZ5hRBth9PaWLTW5cDdwHfAbiBea71TKfWEUsq2l9dcYJmuWmc3GEhWSm0FEoBntNa7HFle2zaVJ56ou0G/pQIDAwkODmbt2rUAfPDBB0yZMqXGfgMHDuTEiROVgaWsrIydO3cSEBBAr169+OKLLwAoKSmhqKiI/Px8unTpgpeXFwkJCWRY5pY5cuQIfn5+zJ8/n/vvv5/ffvsNqJmivzFW7zpG7HOJvLn2IDedF87Xf5rER3dM4JbzIwDYfPhU/ScQQrgFp070pbVeBayqtu6RasuP1XLcOmC4Qwtno7aGetsGfXs34C9ZsoQ//OEPFBUVERUVxbvvvltjH29vb5YvX86f/vQn8vPzKS0t5d5772Xo0KF88MEH3HnnnTzyyCN4eXnxySefMG/ePK644gqGDx9OTEwMgwYNAmD79u3cf//9mEwmvLy8eP3114FzKfp79OhBQiOiZ8KeY9z+fjIAXh6K2aN64u1pIjo8mOjwYDLzzvLOrwe5bVIk/j4yv5wQbk1r3S5eAwYM0NXt2rWrxrrqfvpJ69BQ470521vL6dOnW+1atX3drnrtFx3+4Fc6/MGvdNRDX+l//7S/yvbNh/J0+INf6f8mpjbrmgkJCc06rq2Q+2u73PnetNYaSNZN/HsrKV0asGlT/U8k1ieXTZtat1yuJCPnDFszT+FhUngo8PI0MSGqam+yUb2DmNgvlDfXHqS4rMJJJRVCtAYJLA144IGGq7liY4392qvnvtuLt4cHb8yP5t6LBxJ3+wSiw4Nr7Hf3hf04WVhCfPLhWs4ihJuIi4OICDCZjPe4uPrXu6F2X9mttUYp5exitBm62rinrYdP8dW2o/zpwn5MH9KV6UO61nEkjI/sTEx4MG/8nMbcsX3w9pT/a0QbFhcHCxcy5dAh6NMHnnrKWL9gAVizVmRkGMu//gpLltRcDy2a3txVtevfbF9fX3Jycmr8sRS101qTk5ODr69v5fLT3+wmpKM3C6b0bfB4pRR3XdiPrFNn+WJzXUOWhGgD4uKMwJCRgdLaCBS33QZ33HEueFgVFRnjE2pbv3Bh65W5FbXrJ5ZevXqRmZnJiRMnnF2UFisuLq78g+9Ivr6+9OrVC4DEvSdISsvl8VlDG93Ta+qAMIb17MRLq/dxvKCY8/qG1lptJoRLW7iwZqAoqT/LRq0OHbJPeVxMuw4sXl5eREZGOrsYdpGYmNjgaHh7qjBrnvlmDxEhflw/rk+jj1NKcemwbiz6bh8vfL8PH6/UOttkhHBZTQ0IHh5QUUunlT6N/91pS9p1VZhovk9/y2TvsQLuv2RQk9tKzJaaRw2UlptJSsuxfwGFcKS6AkJICPj5VV3n52dUm9W23tou42YksIgmW3/gJP/4chf9uvhz6fBuTT7+vL6h+FiCkVlD37CO9i6iEI711FO1B4pXXoHFiyE8HJQy3hcvhv/859x6qz/+0S0b7kECi2iilIw8bnx7IwUl5RzKKeK3Q01P0xIdHsxHd0zg1gsi8PP24IXv95F/tswBpRXCQebNqwwU2jaAzJtnvNLTwWw23q3Bw7q+tBT69oXVq4193JAEFtEkSWk5lFvqsirMza/Gig4P5tErhvLWTTGk55zhzg+SKSmXgZOiDbEEip9/+qlqAGmIl5eRcHDrVmN0tRuSwCKaZEQvI0OxovYR9k11fr9QFl0zgqS0XO5YksxrCftlUjDh/ubOheHDjVTpZe73tC6BRTRJeYXxtDJnbC+79ea6anQvbhjXmzX7T/L8d/uY91aSBBfh3kwmo50mNRW6dXO70fgSWESTbEzPxctD8dgVw+zaRbhncAdAeoqJduT0aSOg5OaCdZDlggVuEVwksIgm2Xgwl+E9A+ng3bwphusyIepcTzEPk2pxFZsQLm/hwpqN924yGl8Ci2i0s6UVbMs8xbhI+//Rjw4P5qPbx9PR24PxkSEyYFK4v7oGWbrBaHwJLKLRNh/Oo6xCMz6ys0POHx3RmdhBXUg9Xij524T7q2uQpRuMxpfAIhpt48FclILoCMc9TYyPCiH7dDGHc8867BpCuIS6Blm6wWh8CSyi0TYezGVI90508vVy2DWsT0NJB6XxXriW5PRcXktItV+PRZtBloAxvsU6yLKNc2pgUUrNUErtVUqlKqUeqmX7LUqpE0qpLZbX7TbbblZK7be8bm7dkrc/peVmfjuUxzgHVYNZ9e/iT+eO3mw8mOvQ6wjRFMnpuVz7xnqe/26vfbvDW0fjv/SSMZ5lwgT7nNfJnBZYlFIewGvATGAIcL1Sakgtu36stR5leb1lObYz8CgwHhgHPKqUktZeB9pxJJ/iMrPD2leslFKMjQiWwCJcyodJGWgNpzZEcfpAcL3d4RMSYNGiJl5g9mzjfeXK5hfShTjziWUckKq1TtNalwLLgNmNPPYS4Aetda7WOg/4AZjhoHIKqPxDPzbCsYEFYFxkCIdyiziaL+0swvm01uw4choFdOiez/EVY/DMrn2m1IQEmDMHxo5t4kUiI42R+CtWtLi8rsCZ87H0BGwnP8/EeAKp7mql1GRgH/BXrfXhOo7tWf1ApdQCYAFAWFgYiYmJ9im5CyosLHTo/X2TUkz3jortyesddg0rz9NGzrD3vv6V83oYP6KOvj9nk/tzXbtyKkg9XszMCE869j8Ng3bw4B+HUZS9hdGjT1Xe2+bNQTz++BAefXQXSp2iqbcbOXIkfT76iF9XrKA8MNAh99JaXH2iry+BpVrrEqXUncAS4MLGHqy1XgwsBhg4cKCeOnWqQwrpChITE3HU/VWYNfckfs/lI3ozdepwh1yj+vWeT/me075dK6/nyPtzBXJ/ruvNt5LoEqB56bZYfL2MgcGXTYI5c0YRHw/+/oloPZV//hM+/xxiY0c170IdO8KHHzIxP/9c1Vgb5cyqsCygt81yL8u6SlrrHK21db7Pt4Doxh4r7GdP9mkKissd3r5i5WFSxEQEs1F6hgkn23L4FL+m5nD7pMjKoAIQG2skJp4zBx59dAhz5hjLsbEtuFh0NPTo4RbtLM4MLJuA/kqpSKWUNzAXqPIVVUp1t1mcBey2fP4OuFgpFWxptL/Ysk44gLV9xdE9wmyNjwrhwIkznCxsxjziQtjJfxJSCezgxQ3jw2tsi42FMWNgzZouXHFFC4MKGHnDrrgCvv0WiotbeDLnclpg0VqXA3djBITdQLzWeqdS6gml1CzLbn9SSu1USm0F/gTcYjk2F/gHRnDaBDxhWSccYOPBXHoFd6BHUIdWu6Y1iEnvMOEs+48V8P2uY9x8fgT+PjVbDRISICUFvL0riIszllts9mw4cwZ++skOJ3Mep45j0Vqv0loP0Fr31Vo/ZVn3iNZ6peXz37TWQ7XWI7XWsVrrPTbHvqO17md5veuse3B3Wms2pee26tMKYCS69PKQwCKc5vXEA3Tw8uDW8yNqbLP2/vrkE5g37xClpXD11XYILhdeCP7+bb46TEbei3qlnTzDycLSVmtfsfLyMBEdHswGCSzCCQ7nFrFi6xFuGN+H4I7eVbZZg4q1TeV3v8uiUycYOdJY36Lg4uMDM2YYgaUNT1ssgUXU61z7SuunsR8X2Zk92afJL3K/GfaEa3tzbRomBbdPiqyyvnpQAfD3L+euu+Dnn+H55+0QXGbNgqNHITm5BSdxLgksol4bD+YSFuBDRIhfwzvb2bjIzmgNm9LlqUW0nh93HyNuwyGmDAije2DVdsVNm2rv/fXXv4KvrxFQ4uON/ZrtssvAw6NNV4dJYBF1SsnI48fdx+gf5o9SqtWvP6p3EN4eJjZKYBGtJCUjjzs/SKHCrFm7/2SNnGAPPFB776+wMLjzTvjwQ2MQ/QMPtKAQnTvDgAFGXpg2OmWxBBZRq5SMPG54M4nTxeVsTM91yhz0vl4ejOodxAaZpli0knUHTlJuNuYCKq9o/BTZixbBeecZDxrPPlv7Po3OIRYXB6mpRlLKNjplsQQWUauktBxKy43GQ7PWTpuDflxkZ3YcOc3Zcpn4Szieh+XJ3KTAy9PU6Cmyx46Fu+6Ciy+Gd96BI0eqbm9SDrGFC42gYquNTVksgUXUyrZ7sXcTfsHsbXxUZyrMmo92lzjlqUm0LxsO5hLS0Zu/XjSAuNsnNHqKbOtI/HXroLwcXnjh3LbaGvzr5QZTFktgEbU6VVSGBmaP7NGkXzB7M1n+g1ybVWHfeTCEqOZwbhFr9p9g/oRw7rmwf5N/5mNjYflyY76u116DkyebEVTALaYslsAiavX++nS6dfLlhTkjnRZUwMjVZFVW3vg6byGa6uNNh1HAnLG9G9y3LrGx8N//QkkJXHVVM4IKuMWUxRJYRA0HThSydv9J5o3vg6eHc39EJkSF4Gkynlo8PZxXJSfcW1mFmfjkw0wd2IWeLUxddMstRqeuX34x2tybnEPMOmVxr17GcmBgm5uyWAKLqOGD9Rl4eSjmjnP+o3d0eDCvzRsDwNVjejr16Um4r5/2HOd4QQnX2+FnPiEBsrONz//+dzMHS86bB4cPw/jxMGRImwoqIIFFVHOmpJxPUzK5bHh3wgJ8nF0cAC4Z2o0BwSZSMk41vLMQzbB04yG6dfIldmBYi85jbVNZvhy6dIERI1o4En/6dNi4EfLzW1Su1iaBRVTx+eYsCkrKufG8CGcXpYpx3TzZe6yAvdkFzi6KcDOZeUX8vO8Ec8b2blHVr21D/UUXwY03QlISvPFGC4LLRRdBRYWRL6YNkcAiKmmteX99OsN6dmJMnyBnF6eKsd08MSn4atuRhncWogniNxmznF/Xgkb72np/3Xqr0fU4I+PcpGBNDi4TJhgN9z/80OyyOYMEFlEpKS2XfccKuem8CKekcKlPoI/i/L6hfLn1CFrLYElhH+UVZj5OPszUAWEtarSvLYfY0KEwbhy8+y5MndrMHGI+PjBlCqxe3eyyOYMEFlHp/fXpBPl5MWtkD2cXpVZXjOxOek4RO7JOO7sowk0k7D3BsdMtb7SvK4fYrbfC9u3GhGCxsc3MITZ9OuzZA5mZLSpja5LAIgD4fmc23+7MZsqAsCpze7uSS4Z2w8tD8aVUhwk7+e/PB+jo7UGQn5dDzj93rpH1+N1mTkW4aBEkdJptLNg8tSxadK5ara4cZI3OTeYAElgEKRl5/E/cb2gN3+7IdtnR7UF+3kzuH8ZXW49gNkt1mGiZD9ank5KRR1FpBTe9s9EhP/dBQcZAyY8+at409mPHwpy/RZEQdFWVdpaxY402mxdfrD0HWZNykzmAUwOLUmqGUmqvUipVKfVQLdvvVUrtUkptU0r9qJQKt9lWoZTaYnm13YkLXMAv+080K6OrM1wxsgdH8ov57ZBrBj/RNuw+epp/fL0bAI1jszr8/vdw6hSsWNH0Y40cZIo5Z98jYdVZI9uxZf3f/gb33We821bDNSuNjJ05LbAopTyA14CZwBDgeqXUkGq7bQZitNYjgOWA7YPdWa31KMtrVqsU2k1Zsxg3NaOrM0wf0hUfTxNfbpXqMNE8h3OLuPmdjfh7e+LjacLDgT/3ixaBUkaar3feqX2fhqqsYmMh/s/rmHPqDRLeTqs85umnjRkrn366arWYs4MKgKfzLs04IFVrnQaglFoGzAZ2WXfQWtt2zksC5rdqCdsBrTU/7jlOeGc/5oztxYSoUJce3e7v48m0wV34evtRHr58iNNTzoi2JaewhJvf2UhxWQWf/OF8CkvKSUrLYUJUiEN+7q1VVjNnGpOAHT4MvW16NdsGgvrE3jOM+EVzuPKebwl7xkh03K+fcU5/f2O4yyWXGGMpnR1UAJSzum4qpa4BZmitb7cs3wiM11rfXcf+/waytdZPWpbLgS1AOfCM1vqLWo5ZACwACAsLi45v6LvXhhUWFuLv79/k43blVLBoUzG3DfNmUi/HNGDag+39bcou57UtJTww1pchIa7Z0aCpmvv9aytc4f6KyzWLNhVzuMDM/WN9GRBsn5+dhu5t8+YgHn10KAUFXtx2Wxrz5x+qXP/440N49NFdjB59iqVLezNoUAGjR9eeYWLn1d9wd+7TgImuXYvp3bsILy8zXl6aLVsCOX3am7lzD3HnnWl2uS+r2NjYFK11TJMO0lo75QVcA7xls3wj8O869p2P8cTiY7Oup+U9CkgH+tZ3vQEDBmh3lpCQ0Kzjbn13o47+x/f6bGm5fQtkZ7b3d7a0XA95+Bv94PKtziuQnTX3+9dWNPb+ftiZrV/9cZ9OTs+16/U3pJ3UU55L0BEPfqW/35lt13M35t5++klrLy+te/TQ2mw2lkNDjXfbfaqv09rY/4kntDYaWMx67pzyKvv99JPWQUHG9o4dax7fUkCybuLfd2fWI2QBtkNde1nWVaGUmg4sBGZprUus67XWWZb3NCARGO3IwrqjAycK+WnPceZPCHfZLsa18fXy4OKh3fhy6xH+9eN+l+3FJprm89+yuP39ZJ7/fp/d5t45nFvEYyt3cN3iJNJPnsHTQ9G5o7cdSts0sbHwv/9rzCwZFQVXXlmzyso6WZi1t9eiRUY2l//5H3jkEQDNIu5j6SdexHtcz5zZxZW9wj77zKgO8/JqYW4yO3FmYNkE9FdKRSqlvIG5QJXeXUqp0cAbGEHluM36YKWUj+VzKHABNm0zonHe/fUg3h4m5o0Pb3hnFzOoWwBnSit4ebX9/ggJ5zGbNc99v6dyubSZvbRSMvJ4fOVOHli+lZmvrGXSogTeW5dh7UyF2ey8abYffxyGD4f0dCgshOPHa+5j29tLa7jmGmN+F9A87/Eg9/MiaE3ssWX8rfhR7rtPV/YK+/vfjd5nc+c6P7g4LbBorcuBu4HvgN1AvNZ6p1LqCaWUtZfXc4A/8Em1bsWDgWSl1FYgAaONRQJLE5wqKuXTlCxmj+rhMlmMm6K4rAIAs5YJwNzBsk2HOXKquHLuHa0hOrxp+epSMvK47o31vLsunfjkTMxas/DSwbx2w2h8vRzb+6sxfv0Vjh6F2bPBbDYCwOOPV90nIQH++U+47jojUHxhaTl+IfhJ/rfiuXP7MZWny+7j+aAnK3uFTZ4MkyYZx3z4oXODizN7haG1XgWsqrbuEZvP0+s4bh0w3LGlc29LNx7mbFkFt02KdHZRmmVi/zBe/nE/Wrt+F2lRvxMFJTzzzW4mRHXm/osH8tGmQ3yaksXGg3lMiApt9HmWbTpUOR7LQ8GskT24Y3IUAN0COzi091dDqncD/vBDuPlmeOwxSE6GZ54xpjN+6y0jceWyZRAQAAUFRpbkez98tMr5NjGWeOYQe+pnRv/4MJs2nXtqueQSo9eYNTeZM3qIOTWwCOcoqzCzZF06F/QLYVC3Ts4uTrNEhwdz5aierNiSxTs3j3XpLtKifk9+vYviMjNPXTWcvmH+REd0pqTMzL8TUrl8RHeiwhruTZZ6vJCvtx1FUft4rOjwYKf9jNQ2tmT+fGO+liuugK++Ml5gtL/ccAP07AkPPwx/+Qu8/jokdLmO2GPLKs/5AJanlz7hxMYa5120CGJijC7OTz8N+/bVDCoJCUawaVbOsiaQQQDt0KrtR8k+XcxtE9vm04rVrFE9MGtj5LRom9bsO8GKLUf449S+9LUJII9cMQQfTxMLP9/RYDbrvDOl3LZkEx28PPjPvDHce/FA4m6f4BL/bNQ3YPHii+Hbb40ExgD33AMHDsCFFxpBJT4enhiatG8AACAASURBVHjC0qBf9C4JPjOqnsDPD556qnJx7FijCm3WLDh4EJYurb0srZHmRQJLO5OSnsvTq3bTI8iXqQO6OLs4LRITHoxJwQZpX2mTissqeHjFDqJCO/LHqX2rbOsS4MtDMwexPi2HT3+r0Vm0Umm5mT/GpXD0VDGLb4pm5vDu3BXbzyWCCtSeTr86X1+YNs0IBNZeXrbHxMZC/Apf5nh/TkLXucZKHx9YvLjKlMXWXmWvvGI8+fzzn0ZbDrT+iHwJLO1ISkYe17+5gezTJRw/XcLmw217qt8AXy+G9ghkw8FcZxdFNFFKRh63vLuRjJwinrxqWK3d3a8f24eY8GAeX7mD57/bW6PnX0p6Ltf8dx1Jabk8e81wosM7t1bxG62udPpw7o/9558biYvryv0FNsGlYikJc98wIsZVV9U4pzW4nDhhZNr/7DPnpHmRwNKOJKXlUFph/AujtfO6XdrT+MjObD58qrKXmHB9KRl53PBmEklpuXgohY9n7WOoTCbF/AnhFJRU8O+EVOYuXs/Tq3azZF06j63cwZzFSWzLzMfTpOjTuWMr30XL1PbHvry8Zu4vW9agsanDZCgrgw0baj13bKwRUDw84O67nZM7TAJLO2IdGKZwn55U46NCKC03sy0z39lFEY304fp0SiyJT6H+f3CyTp3FOpdpWYXmjTVpPLpyJ++ty6DC0gOsrf2TVNcTxAMPwL331j+NcWwsPPBSdyOz5Zo1dV5j+nS47DI4dgwuvbT1e4ZJYGlHvt2RTYCvJ3+e1t9lGjdbamxEMEraWdqEwpJyHly+jc+3HKmz91Z1E6JC8PEyYVLg42nizRtjSPn7dN7//Th8HZyZ2FEaanepfDKpaxrjwEAYNarewJKQYIyb8fQ0ztXa41mku3E7kZKRx8/7TvDQzEH8YUrfhg9oI4L8vBnYNYANB3O5x9mFEbVKycjjvR0lPPhrIicKS/ifqX2ZMiCM5Iy8BseVRIcHE3f7hBpjUCYPCCPujprr24LGdPW1diGu0+TJRuN9aSl4V01RY30i+uQTePNN+Prr1q8Ok8DSTrzy4346d/TmpvPaXvqWhoyP7Ex8ciZlFWa8JI2+S0lJz+W6xUmUmzWKcv5x5TDmTzB+Bsc38imjrjEozhyb4nSTJxvdv5KT4fzzK1dXr2YrKDB6mz3zjPQKa1dSMvJ4LSHVobmuUjLyWLPvBHdOjsLP2/3+lxgfFcLZsgq2Z0k7i6v5ICmjcjS8SUH+2TInl8hNTJpkvNtUh9XWdnPJJcb0yLt21d92Y28SWJwoJSOPeW8l8cL3e1uUSDElI4+vDpTWefzLq/cR0tGbG93waQVgbITRzXSjdDt2OQdOFALGH5q21hbi0sLCYPDgKoGltrYbHx/43e+MLs3nnddA240dtSiwKKU6KKV+tkwzjFKql1LqOstnb6XUGqWU+/2LbCdJaTkUl5kxaygpM5OUdrLJ5zDGpiSxfH8Z896sGZxSMnJZu/8kd05xz6cVgLAAH/qGdZQGfBez/1gB27NOM298H37X38ttOoy4jMmT4ZdfjNz61D1mZu5co0rsm28svcocnM4FWv7E8nvgM621dRDBNGAMgNa6FPgRuK6F13BbwX7nZmzUwIHjZ5p8jrX7TlTOWV9cbuaNnw9QXmGu3P7y6v2E+ntX1mu7q3GRISSn51V2QRXOt2R9Ot6eJv734oFc3tdbgoq9TZ5sRIytW+vdLTbWyEtWPcWLIzUqsCilEpRSF1k+P6mUetWyaR6wwrJ+IvAicI0lxX0U8IVlH1FNcVkFb/9ykFB/b/48rR8XDenKZ5uzeD3xQJPOk3rcqGpQGF3bv991jEteXsN3O7OJS8pg7f6TXDa8u9s+rVhNiOpMQUk5u4+ednZRBHC6uIzPfsti1sgeTplYq12opZ2lNp6ecO21RqLLgoJWKBeNf2J5FFiolJqHMVPjXyyTc0VprdMBtNa/YEzeNVtrPcoys+MOoBVSnrU9r/60nwMnzvD8tSP560UD+e/8aGaP6sGz3+7hw6SMRp1j6+FTrNpxlEuGduXq/l4sv/M8Ft8YjQbu/CCFhV/sAODjTYfdfiKscZFGO4ukd3ENy5MzKSqt4ObzIpxdFPfVuzdERjYYWMCoDjt7Fr78shXKRSMDi9Z6DcY/xfcCcy1VX6FA9WRTA4E9NsdVAKVKqQD7FNc97DySz39/TuN3Y3oydaCRCNLDpHj+2pFMG9SFh1fs4OXV++rtLVZWYeahz7YTFuDDc9eONKoaIjpz8dBufP+XyVw8pGuVfdvSyOTm6B7YgT6d/aSdxQWYzZr316czpk8Qw3sFOrs47m3yZCOwNJAB+vzzoVev1qsOa2xV2HCgO1CqtbY+TJ0FfG32CQXyLTND2vIBiu1QVrdQVmHmgeXbCPbz5pHLh1TZ5uVh4rV5YxjSPYCXV+/n+e/q7i321tqD7D56msdnDaOTr1eVbZ4eJu6c0tclZs1rTeMjO7MpPReztLM41Zr9J0jPKeLm8yOcXRT3N3ky5OTA7t317mYyGU8t330Hua3wUN9gYFFKdQfigNlAoVJqBoDWOg/wUEpZg0sEcKTasSHASa21dF63eHNtGjuPnOYfs4cS5Fez7tnXy4Ppg7sBRoN+SZmZn/dWnRw7I+cML6/exyVDuzJjWLdar2MdsexKc1M42rjIzuQVlbHf0u4knGPJunRC/X2YOay7s4vi/iZPNt7rqQ5btMgYuzJ3rpG78vPPa+6TkGDsZy/1BhallB/wGfC/WuvdwD8w2lusvgcmWj7vAUKVUjuUUtahoLHA1/Wcf4ZSaq9SKlUp9VAt232UUh9btm9QSkXYbPubZf1epdQlDd+q863cmsUL3+9jQlRnZg6v+5du8oAwfD1NKIzg8kFSBqt3HQOMhHv/9/l2vD1MPD5rWL3Xiw4Pdqm5KRzN+lT23Hd73L5NyVWlnzxD4r4T3DC+D96eMkzO4fr2he7d6w0sY8caAyPz86Ffv9aZAKzerkJa6yLgPJvlNbbLwGvAX4HVWutCYFy1U9wA1AgYAJaxL68BFwGZwCal1Eqt9S6b3W4D8rTW/ZRSc4FngeuUUkOAucBQoAewWik1wKbbs8tZf+Akf162Ba1h86FTpGTk1fkHPzo8uDIPUpi/N+/8ms7t7ycze5TRw+bX1BzumBxFt0DfWo9vr46fNmpcV+8+zi+pJ9vNk5or+SApAw+lmDe+j7OL0j4oVbWdRakau1iTWs6ZAzNnQlwcZGdDt26Om6ulRf9SaK1/AxKsAyRtWXqNfaG13lfH4eOAVK11mmXMyzKM6jZbs4Ells/LgWlKKWVZv0xrXaK1PgikUjOouYzScrNlilVjubwRjenWp405Y/uw8u6J/GV6f77aeoR3f00H4IP16fJfeTVJNj3CisvMfLn1SD17C3s7U1JOfPJhZgzrRtdO8k9Pq/H1hawsYwKWiAgjclRjDS5ffWXMEbZ8uWMnAFMNzSftKEqpa4AZWuvbLcs3AuO11nfb7LPDsk+mZfkAMB54DEjSWn9oWf828I3Wenm1aywAFgD4+flFFxUVOfy+alAmQq+4j46DJ6MrygCFNldwbNlCSo/safBwW8HTFhAQfTlKmdAV5Zz6JY7TSZ84ptxtkHePQXSd+xTKwwuUQinFmT2/kP/rUpR3B3z7DKf40PYmf91F4wRNuYXACdeQ8+2/Kdz6rbOL0y5cD7wNdLBZdwa4A6i9A9hU4AeMSqKOwBwgsaHLpGitY5pSLrceNae1XgwsBhg4cKDeu3dva1+fhz7dzsfJh/m/SwcRHd75XJrvF+rvxVEba26xsnIzXr7efLbsDaLD4wFITExk6tSpdr4D19HY+0vJyCMpLYfhPQJJzsjlHR9POg6aiMlSQ+DtaXLJKrK2/v1LycjjmtfXoYGeV/yJuBUfVfkat/X7q49T7y0iAjKqjnvrCHwUHs5H6em1HjJuHGzaFMHChfDkkw1npFS1VK81xJmBJQvobbPcy7Kutn0yLTnHAoGcRh7rVFprnvx6Nx8nH+aeC/uxYLIxB0pL/qDVNTeFOMc2lfrkgWH8fmIkf/wwhfVpRjVZWblRDSlfO/v6evsRrHUf8jVuRYcONWl9QgLssTyw/+c/MG2aY9LoO7Pbxiagv1Iq0tIeMxdYWW2flcDNls/XAD9po+5uJTDX0mssEugPbGylcjfIeLLYwNu/HOSW8yO496IBdjt3e+vp1VJBft7cd8kgPCyPLJ4e7WNMT2srsKTDb0/jplxCnzo6SdSy3tqm8sorxvKddzoujb7TAotlIOXdwHfAbiBea71TKfWEUmqWZbe3gRClVCrGqP+HLMfuBOKBXcC3wF2u0iMsJSOPuYvXs+5ADh5KcfmI7s16lBT2Ex0ezKtzR6OAy4Z3l6BsZ1prNqXnMbxnp3Y1bsolPPUU+PlVXefnZ6y3YdtQf9NNxi5FRY6bo8WpbSxa61XAqmrrHrH5XAxcW8exTwFP1bbNmZLSciirsFYKaDYczCXGMl+IcJ5LR3RnxrZu/LjnOEWl5W6flLM1bcvMJz2niGevHs51Y6WbcauaZ8nxu3DhubaW5547t57ae3+NGWNMPvnKK+eCiz17h8kIJjsbH3kuiEiVgGu5fVIk+WfL+CQ509lFcSsrtx7By0MxY6iMtHeKefMgPd2YJhKgtLTK5tomAIuJgc2bobz8XFdke04AJoHFzgIsebtmDusmVQIuJjq8M6P7BPH2Lwdl3hY7qTBrvtx6hKkDuxDo59XwAcJxBg+G6Gj48MMqq2ubACwmxsh2bE0xZu8JwCSw2Jl14OP/XTpYgooLumNSFIdyi/hhV7ZdzpeSkVdvFmp3t+FgDscLSpg1soeziyIA5s+HlJQGk1LGWEalJCc7phgSWOxs/YEcegV3oHdnv4Z3Fq3ukqHd6N25A2+uPdjic6Vk5HHDm0m88H3dWajd3cotR/Dz9mD64K4N7ywcb+5cI5VxLaPvbfXvDwEBEljaBLNZs+FgjrSruDAPk+L3F0SSkpHHb4daFgi+35VNSbkZszZSyHz2W/tquyktN/PNjmwuHtKVDt41sjoJZ+jWDS66yAgs9WRVMZmMWjMJLG3AvuMF5BWVSWBxcXNietPJ15O31qY1+xzFZRX8sNOoTrN2Jo/bcIib39nI5kN57aKKbM2+E+SfLWP2qJ7OLoqwNX++0Zi/bl29u8XEwNatNdr67UL6XNpR0gGjfcW2Z5hwPR19PLlhfDiL1xzgcG5Rk6sttdb832fbSTtZxN9mDqLcrBnVO4jtWfksXpPGVf9Z5/IpZOxhxdYjBPt5MbF/qLOLImxdeaUxUOXDD+GCC+rcLSYGSkpg504YPdq+RZAnFjtKSsuV9pU24pbzI1DAn5dtbvJTxbu/pvPZ5iz+On0Ad07py12x/bigXyh/mNKXtQ/EMrl/GGYNZm1UF7njtNBnSspZvesYlw7vjpeH/BlxKf7+cNVV8PHH9T6OOLIBX34i7ETaV9qWrFNnAcVvh05x7X/Xcf/yraw/kENxWUW91VjrD+Tw1KrdXDSkK/dc2K/G9o4+nvx5en98LJNcmTUEdXC/brirdx/jbFmF9AZzVT16QF6ekVK/jlT6UVEQFOSYwCJVYXYi7SttS1JaDtqSNtGsYXlyJp8kZ+LpoTCbNVqDl4fizZtimDKwCykZeXy/K5uPNhwiIsSPF+eMxGSqPVVPdHgwH90xge92ZvPl1iP84+tddA/y5cJB7tFzKiUjj3/9uJ+Qjt6MlawSricuDl57zfistTEif8ECY9lmRL5SxlOLBBYXZm1fmRAlv2htwYSoELw9TcYUBJ4mFt8YQ1mFmdcTD5BseVIprdDc/O4megT5ciy/hApLL5unrhpWORC2LtYsywsmR3HLuxu54/0U7praFx8vjzadmdraxbqk3IyHSbH58Kk2ey9ua+FCIxGYraIiY71NYAEjsLzwAhQXGw839iJVYXaSlJZL784d6BUs7SttgXUKAmvSxMkDwpg2uCt/u3Qwvl4mPBR4e5i4YXwfOnh5VAYVk4LDuWcbfZ1Qfx+W3jGBId0D+NdPqTz/Xdse85KUlkNpuRkwOjG4Y/tRm1dfKv24OKNqzGSCiAhiitZQVgbbt9u3CPLEYgfW9hUZJNa22M7dYruu+pw3VSZYa0b+twBfLy4a0pXtWafRnGvQb4v/6U+ICsFkUlSYNd6SC8819elTY/IvwKgWu+UWI0EYQEYGMYsXAHtIToaxY+1XBAksdrD3mLSvuJPqAcceE6xd0C+M/yQcoNgyoHJoj072LHKriQ4PpkegL54eJp6/dmSbDI5u76mnjDYV2+qwDh2MwFJcXGXXPsV7CTXlkJxs379dUhVmB9bqgPHSvuK2WjrBWnR4MHF3TOCm88LxMMHSjYfQ9YyMdlX5Z8s4nHeW343uKUHFVc2bB4sXQ3i40UIfHg5vvmkMWqlGATHmjXZvwJfAYgdJaTnSviIaFB0ezBOzh/HQjMF8t/MYH2867OwiNdlmSxocCSouzppK32w23ufNq3O2yZjA/ezcWbO9vyUksLSQ0b6Sy4RIqQYTjXPbxEgu6BfC41/u4sCJQmcXp0lSMvLwMClG9g5ydlFEU9U226S3NzG/H0lFhZHexV4ksLTQ3mMFnJL2FdEEJpPihWtH4eNl4i/LtlT2smoLUjLyGNw9gI4+0jzb5lSrIlvk8TcSTNOI+f0IoPbxLMaUxT2a3CvJKYFFKdVZKfWDUmq/5b3Gc7VSapRSar1SaqdSaptS6jqbbe8ppQ4qpbZYXqNa9w7OkfYV0RzdAn155ncj2J6Vz4OfbuOrA6Uu3wW5vMLMlsOniO4j1WBtlk0V2dg3bmdO8RL23v8m3brVDCzWKY3hTJMryZz1xPIQ8KPWuj/wo2W5uiLgJq31UGAG8LJSyvb5+36t9SjLa4vji1y7b3ccJbCDJ8dO12wYE6I+M4Z146LBXfh8cxaf7i9z+fEte7ILKCqtIFpG27uF2NuiiL/uM6779lYiQvKrBBZrUImPB8gvaOq5nRVYZgNLLJ+XAFdW30FrvU9rvd/y+QhwHAhrtRI2QnJ6LhsO5pF/ttzl/ygI1zS0ZyAAGihz8YSV1p9vabh3H7Fvzye+65/YtsuLXbvMFKoAErpdz5zZxcTH15zSuLGcVVHaVWt91PI5G6i3Dk8pNQ7wBg7YrH5KKfUIlicerXWNRwal1AJgAUBYWBiJiYl2KPo5b2w91ye8tMzM0tWbKOjrbddrNFZhYaHd78+VuOv9+RdUYFJg1hqTUvicyiAx0TUnDFu1pZhgH8W+zUnsV7XnSauLu37/oO3fW9/JFfz9k3/wfzzN3bzK18cuY6nXbLr+OJZENb15J9VaO+QFrAZ21PKaDZyqtm9ePefpDuwFJlRbpwAfjCeeRxoqz4ABA7Q9mc1mPe2FRB3x4Fc66qGv9MC/r9LJ6bl2vUZTJCQkOO3arcGd7+/z3zJ1+INf6cdW7nB2Uep1/tM/6v+JS2nWse78/Wvz9xYerrPpokM5rkHrh3lca9A6PFxrrTWQrJv4999hTyxa6zpDnVLqmFKqu9b6qFKqO0Y1V237dQK+BhZqrZNszm192ilRSr0L3GfHojdKUlouqccLWTA5isAOXm06saBwritH9+Tlb7ax8WCus4tSp6P5Z8k6dZbbJkY6uyjC3g4dYhdTAHiYJ3idPxJLArGHfm72KZ1VFbYSuBl4xvK+ovoOSilv4HPgfa318mrbrEFJYbTP7HB8kat6/ecDhPr7cO9FA/D1kvm+Rctc0NOTuN2n2ZN9mkHdXC/di7SvuK+ELtcx59i/iGcOsSQSSwJziCe+y59oZhOL0xrvnwEuUkrtB6ZbllFKxSil3rLsMweYDNxSS7fiOKXUdmA7EAo82ZqF35GVz5p9J7j1gggJKsIuxnf3xNOk+Py3LGcXpVYpGXn4epkY0kZznInaJSTAnKJ3ife5iVgSAYglkXifm5hT9K5lHEvTOeWJRWudA0yrZX0ycLvl84fAh3Ucf6FDC9iA//58AH8fT+ZPCHdmMYQb6eStmDrQ6Hr8wIxBeNQxiZizpGTkMbJXkExD7EYquxSv8CX2yHxYuNtIrd+nD7FPzSe+h69lHEtgQFPPLT8lTZSRc4ZV248yb0IfAt1wylnhPFeP6cnxghJ+TT3p7KJUUVRazs4jp4mJkGowd7JpE+e6FNeSWyw21jqOpWOTkyBKXoYmWrwmDU+TidsukEZMYV8XDu5CJ19PPvstk8kDXGfI1tbD+VSYtbSvuJkHHmh4H2Mcy5FjTT23PLE0wfGCYj5JyeTq6J506WTHeTyFAHw8PbhiZA++3ZlNYUm5s4tT6TdLRuMxkspFNJIEliZ499d0yirMLJjc19lFEW7qd2N6UVxm5pvtRxveuZWkZOTRr4s/QX7OGfwr2h4JLI20dv8J3l57kPMiQ4gM7ejs4gg3NaZPEBEhfnzmIr3DzGZNSkaeJJ4UTSKBpRFSMvK49d1NlFaYST6UJznBhMMopfjdmF6sT8shM8+OMy81U9rJQvLPlkn7imgSCSyNkJR2knKzMY1sRYVrJwoUbd9Vo3sC8LfPtjv9n5jkdMvASOkRJppAAksjhPn7AEZyMi9Pk0zqJRzqeEEJJgVr9590etbs73dm4+tl4tSZUqeVQbQ9ElgaYU92IZ4mxT0X9iPu9glSLSAcKiktByPXKpQ6MZV+SnouCXtPUFxmZt7bG5z+9CTaDgksDagwa77cdoQLB3Xh3osHSlARDjchKgQfT+NXU6Gc9oS8dNMhLPHN5eeKEa5FAksD1h04yYmCEq601HsL4WjR4cHE3TGBEb0C8TApBnT1b/UyaK3ZnnUaAA8lVcCiaSSwNOCLzUcI8PHkwkFdnF0U0Y5EhwfzxOxhlFaYWbHlSKtfPyktl73ZBdw+KZJ7Lx4oVcCiSSSlSz2Kyyr4bmc2M4d1kyzGotWN7BXIoG4BfLzpcKsnPP13wn7CAny47+KB8rMvmkyeWOrx4+7jFJaUSzWYcAqlFHPH9mZ7Vj47svJb7bopGXn8mprDgklRElREs0hgqccXW7LoEuAjdcvCaa4a3QsfTxMfbzrcatd8LSGVYD8vbhjfp9WuKdyLBJY6nCoqJXHvca4Y2cPl5sYQ7UegnxeXDu/OF1uyOFta4fDr7cjK56c9x7ltYiQdfaSmXDSPBJY6rNqeTVmF5spRUg0mnOu6sb0pKC7n61ZITPnvn1IJ8PXkpvMjHH4t4b4ksNThiy1ZRIV1ZFhPmYpVONf4yM5Ehnbk402HHHqdfccK+HZnNrecH0EnX5nETjSfUwKLUqqzUuoHpdR+y3ut/RiVUhU2892vtFkfqZTaoJRKVUp9rJSyaz7vrFNn2XgwlytH9UQpqQYTzqWU4rqxvdmUnkfq8QKHXCMlI48/L9uMj6eJ38skdqKFnPXE8hDwo9a6P/CjZbk2Z7XWoyyvWTbrnwVe0lr3A/KA2+xZuJWWcQOzR/Ww52mFaLarx/TC06Qc0oifkpHHDW8msftoAeVmTdrJM3a/hmhfnBVYZgNLLJ+XAFc29kBlPEJcCCxvzvGNsWzjIboH+nKyUBLvCdcQFuDDRUO68ulvWZSU27cRPykth9Jys7GgtaRuES2mtDXbXWteVKlTWusgy2cF5FmXq+1XDmwByoFntNZfKKVCgSTL0wpKqd7AN1rrYbUcvwBYABAWFhYdHx/fYNl+ySrjre1GQPE2wQNjfekX7Pp9+QsLC/H3b/3UH61F7g+2nSjnxZQSxnfz4KJwL7v9XKbmVfDUhmI0jvuZd+fvnzvfG0BsbGyK1jqmKcc4rD+hUmo10K2WTQttF7TWWilVV3QL11pnKaWigJ+UUtuBRo8U01ovBhYDDBw4UE+dOrXBY5588WfACCwVGkqCwpk6tV9jL+k0iYmJNOb+2iq5P+iYnotKWc+G7Aq25mi7pVkZfLqYJzf8yJQBYfxpWn+HpG5x5++fO99bczkssGitp9e1TSl1TCnVXWt9VCnVHThexzmyLO9pSqlEYDTwKRCklPLUWpcDvQC7zOO6PTOf1ONGinyttSTeEy5l48Hcys/WdPr2CAJr9p0A4MEZgxjSQ3pBipZzVhvLSuBmy+ebgRXVd1BKBSulfCyfQ4ELgF3aqLtLAK6p7/jmeHn1PgI7ePHurWMl8Z5wOROiQvC2pNM3Kful01+7/ySh/j4M6hZgl/MJ4azA8gxwkVJqPzDdsoxSKkYp9ZZln8FAslJqK0YgeUZrvcuy7UHgXqVUKhACvN3SAm05fIof9xxnweQoJvUP467YfhJUhEuJDg/mozsm0Du4A6H+3ozpU6NZssnMZs0vqSeZ3D8Uk2SYEHbilJwNWuscYFot65OB2y2f1wHD6zg+DRhnzzK9+MM+Onf05mYZcSxcWHR4MH+ZPoD//WQrGw/mMr6FTy07j5wm90wpkwaE2qmEQsjIewCS03NZs+8Ed06Owl/yIwkXN3N4N/x9PPkkJbPF51qz32hfmdgvrMXnEsJKAgvw0up9hPp7c+N5rTvnhRDN4eftyeUjuvP1tqMUlpS36Fxr9p1gSPdOhAX42Kl0QkhgISkth19Tc/jDlL74ecvTimgbro3pxdmyClZta35iysKSclIy8pg8QJ5WhH2168CitebFH/bRJcCn1WfoE6IlxvQJJiqsI5+kND/Fy/oDOZSbNZP7S/uKsK92HViWrEtn48FcZo3sITPliTZFKcU10b3YlJ7HwWbm9lq7/wQdvDyIjpDej8K+2m1gScnI4/GvjN7LHyZlkJKR5+QSCdE0V4/phUnB8mY+tazZd4Lz+obg4yn/VAn7areB5dOUTKxp0soqzJJ4T7Q5XTv5MmVAGJ+mZFFhblrOv0M5RaTnFDFJqsGEA7TbwLLvmDGvhYdCUreINuvamN5kny5mraXbcGNZq/5RYAAAClVJREFUuxlLw71whHbZDWpvdgHJGXlcF9ObPiF+TIgKkVH2ok2aNrgLQX5eLF6Txs4jpxv9s7x2/wl6BnUgKrRjK5RStDftMrC8seYAHbw8eGjmIII72nXySSFalY+nB+f3DWHV9myS0nLw9jQ1mOOurMLMutQcLh/ZXWZIFQ7R7qrCsk6dZeWWI8wd11uCinALXQN8ATBrKCtvuL1w6+FTFJSUM6m/VIMJx2h3geXttQcBuH1SlJNLIoR9XD6yBx6WBJIepobbC9fsO4FJwQV9peFeOEa7Cix5Z0pZtukQs0b2oGdQB2cXRwi7iA4P5oPfjyPIz4sgP08Gd68//f03O7Lp2smX1BOFrVRC0d60q8Dy/voMikoruHNKX2cXRQi7Or9fKItvjOF4QSnPfbe3zv3eXJvG/uOFZOcXM++tJBm/JRyi3QQWDSxZn860QV0YKBMaCTc0LrIzN58Xznvr0klOz62xfXlKJv9ctRswfh8a0x4jRHO0m8BSUKrJPVPKH6bK04pwXw/MGESPwA48sHwbxWUVgJET77WEVO77ZCvDenTC19Mk47eEQ7Wb7sanijWTuwUwNqKzs4sihMN09PHk2atHMP/tDby0eh8PXDKIx7/cyfvrM5g9qgfPXTOS7Vn5JKXlyPgt4TDtJrCYgbQThaRk5Mkvk3BrE/uHMndsbxb/nMa327PJyC3ijkmR/G3mYEwmRXR4sPwOCIdySlWYUqqzUuoHpdR+y3uNn3KlVKxSaovNq1gpdaVl23tKqYM220Y15rpms5Y6ZdEuXDaiOxrIyC3C06SYMay7zGkvWo2z2lgeAn7UWvcHfrQsV6G1TtBaj9JajwIuBIqA7212ud+6XWu9pTEXlTpl0V5sy8zHGke0ln+oROtyVmCZDSyxfF4CXNnA/tcA32iti5p7wWAf1WCqCyHcxYSoELylkV44ibPaWLpqra1zqmYDXRvYfy7wYrV1TymlHsHyxKO1LqnvBIE+SoKKaDeiw4OJu32CNNILp1BaN20eh0afWKnVQLdaNi0Elmitg2z2zdNa1/qTr5TqDmwDemity2zWZQPewGLggNb6iVqOXQAsAAgLC4uOj49v2U25sMLCQvz9/Z1dDIeR+2vb3Pn+3PneAGJjY1O01jFNOcZhgaXeiyq1F5iqtT5qCRKJWuuBdez7Z2Co1npBHdunAvdprS+v75oDBw7Ue/fWPSK5rUtMTGTq1KnOLobDyP21be58f+58bwBKqSYHFme1sawEbrZ8vhlYUc++1wNLbVdYghHKyPl9JbDDAWUUQgjRDM4KLM8AFyml9gPTLcsopWKUUm9Zd1JKRQC9gZ+rHR+nlNoObAdCgSdbocxCCCEawSmN91rrHGBaLeuTgdttltOBnrXsd6EjyyeEEKL52k2uMCGEEK1DAosQQgi7ksAihBDCriSwCCGEsCsJLEIIIexKAosQQgi7ksAihBDCriSwCCGEsCsJLEIIIexKAosQQgi7ksAihBDCriSwCCGEsCsJLEIIIexKAosQQgi7ksAihBDCriSwCCGEsCsJLEIIIexKAosQQgi7ksAihBDCrpwSWJRS1/5/e3cSYleZhnH8/5AyODVGjWg0SiKKkoVGadSoiB0HoohDEAcURAQ3Nth0i210IQou3DgsRAiOC6cQjREVo0ahexVnsbQSTKfTGKdqbdOKi4j6uDhf4TVUrCHn5tQ59fyguPd859xb70t9xXum+15JH0r6WdIff2e7JZI2SNoo6aae8fmS1pXxpyTN3DWRR0TEWJo6YhkElgL/2NEGkmYA9wHnAAuAyyUtKKvvBO62fQTwDXBNf8ONiIjxaqSw2B6yvWGMzU4ANtreZPsH4EngAkkCFgMry3aPAhf2L9qIiJiIgaYD+B2HAJ/0LG8BTgT2B7ba/rFn/JDR3kDStcC1ZXGbpME+xToVzAa+ajqIPkp+7dbl/LqcG8BRE31B3wqLpFeBg0ZZdYvt1f36vb1sLweWl3jesr3D6zltl/zaLfm1V5dzgyq/ib6mb4XF9pk7+RafAof2LM8tY18DsyQNlKOWkfGIiJgCpvLtxm8CR5Y7wGYClwHP2TbwOnBx2e4qYJccAUVExNiaut34IklbgEXAC5LWlPGDJb0IUI5G/gysAYaAFbY/LG/xd+CvkjZSXXN5cBy/dnnNaUw1ya/dkl97dTk3mER+qg4AIiIi6jGVT4VFREQLpbBEREStpkVh2VFrmLaS9JCk4d7P5UjaT9Irkj4uj/s2GeNkSTpU0uuSPiptf64v413Jb3dJb0h6v+R3WxnvVJsiSTMkvSvp+bLcmfwkbZb0gaT3Rm7F7cr8BJA0S9JKSeslDUlaNNH8Ol9YxmgN01aPAEu2G7sJWGv7SGBtWW6jH4G/2V4AnARcV/5eXclvG7DY9rHAQmCJpJPoXpui66luuhnRtfz+ZHthz+dXujI/Ae4FXrJ9NHAs1d9xYvnZ7vQP1Z1na3qWlwHLmo6rhrzmAYM9yxuAOeX5HGBD0zHWlOdq4Kwu5gfsCbxD1VHiK2CgjP9mzrbth+qzZWupWi89D6hj+W0GZm831on5CewD/JtyY9dk8+v8EQujt4YZtQVMyx1o+/Py/AvgwCaDqYOkecBxwDo6lF85TfQeMAy8AvyLcbYpaol7gBuBn8vyuNswtYSBlyW9XdpGQXfm53zgv8DD5VTmA5L2YoL5TYfCMu242q1o9X3kkvYGngb+Yvvb3nVtz8/2T7YXUu3ZnwAc3XBItZF0HjBs++2mY+mjU20fT3V6/TpJp/WubPn8HACOB+63fRzwPdud9hpPftOhsOyoNUzXfClpDkB5HG44nkmTtBtVUXnM9jNluDP5jbC9laqLxCJKm6Kyqs1z9BTgfEmbqTqSL6Y6Z9+V/LD9aXkcBlZR7Rx0ZX5uAbbYXleWV1IVmgnlNx0Ky6itYRqOqR+eo2pvAy1uc1O+FuFBYMj2XT2rupLfAZJmled7UF0/GqIjbYpsL7M91/Y8qv+112xfQUfyk7SXpD+MPAfOpvp+qU7MT9tfAJ9IGulofAbwERPMb1p88l7SuVTnfWcAD9m+o+GQdoqkJ4DTqdp1fwncCjwLrAAOA/4DXGL7f03FOFmSTgX+CXzAr+fob6a6ztKF/I6h+g6hGVQ7dits3y7pcKo9/P2Ad4ErbW9rLtKdJ+l04Abb53Ulv5LHqrI4ADxu+w5J+9OB+QkgaSHwADAT2ARcTZmrjDO/aVFYIiJi15kOp8IiImIXSmGJiIhapbBEREStUlgiIqJWKSwREVGrFJaIiKhVCktERNQqhSWiYZLmSrq06Tgi6pLCEtG8M6j6MUV0Qj55H9Gg0sJmNbAV+A5YantTs1FF7JwUloiGSXqJqqfW4JgbR7RAToVFNO8oYH3TQUTUJYUlokGSZgP/7/l2xYjWS2GJaNY84LOmg4ioUwpLRLPWA7MlDUo6uelgIuqQi/cREVGrHLFEREStUlgiIqJWKSwREVGrFJaIiKhVCktERNQqhSUiImqVwhIREbX6BTAVlrfnRBIeAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "DHXVsz2pJSnK",
        "colab_type": "text"
      },
      "source": [
        "## USING 1D CONVOLUTIONAL LAYERS TO PROCESS SEQUENCES"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VOCfS3BYJVI7",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "keras.backend.clear_session()"
      ],
      "execution_count": 64,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nCrqVJhdJcdo",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 727
        },
        "outputId": "8a596925-6830-4d39-bda5-1f0fb9db8ff1"
      },
      "source": [
        "np.random.seed(42)\n",
        "tf.random.set_seed(42)\n",
        "\n",
        "model = keras.models.Sequential([\n",
        "    keras.layers.Conv1D(filters=20, kernel_size=4, strides=2, padding=\"valid\",\n",
        "                        input_shape=[None, 1]),\n",
        "    keras.layers.GRU(20, return_sequences=True),\n",
        "    keras.layers.GRU(20, return_sequences=True),\n",
        "    keras.layers.TimeDistributed(keras.layers.Dense(10))\n",
        "])\n",
        "\n",
        "model.compile(loss=\"mse\", optimizer=\"adam\", metrics=[last_time_step_mse])\n",
        "history = model.fit(X_train, Y_train[:, 3::2], epochs=20, validation_data=(X_valid, Y_valid[:, 3::2]))"
      ],
      "execution_count": 65,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/20\n",
            "219/219 [==============================] - 2s 9ms/step - loss: 0.0773 - last_time_step_mse: 0.0337 - val_loss: 0.0205 - val_last_time_step_mse: 0.0123\n",
            "Epoch 2/20\n",
            "219/219 [==============================] - 1s 7ms/step - loss: 0.0140 - last_time_step_mse: 0.0110 - val_loss: 0.0031 - val_last_time_step_mse: 0.0035\n",
            "Epoch 3/20\n",
            "219/219 [==============================] - 1s 7ms/step - loss: 0.0012 - last_time_step_mse: 0.0014 - val_loss: 5.7122e-04 - val_last_time_step_mse: 4.5013e-04\n",
            "Epoch 4/20\n",
            "219/219 [==============================] - 1s 7ms/step - loss: 4.4066e-04 - last_time_step_mse: 3.8253e-04 - val_loss: 3.7105e-04 - val_last_time_step_mse: 4.8130e-04\n",
            "Epoch 5/20\n",
            "219/219 [==============================] - 1s 7ms/step - loss: 3.0242e-04 - last_time_step_mse: 3.2176e-04 - val_loss: 2.7985e-04 - val_last_time_step_mse: 3.9229e-04\n",
            "Epoch 6/20\n",
            "219/219 [==============================] - 1s 7ms/step - loss: 2.3301e-04 - last_time_step_mse: 2.6723e-04 - val_loss: 2.1134e-04 - val_last_time_step_mse: 2.4979e-04\n",
            "Epoch 7/20\n",
            "219/219 [==============================] - 1s 7ms/step - loss: 1.9085e-04 - last_time_step_mse: 2.1648e-04 - val_loss: 1.7515e-04 - val_last_time_step_mse: 1.9404e-04\n",
            "Epoch 8/20\n",
            "219/219 [==============================] - 1s 7ms/step - loss: 1.6577e-04 - last_time_step_mse: 1.8579e-04 - val_loss: 1.5989e-04 - val_last_time_step_mse: 1.8570e-04\n",
            "Epoch 9/20\n",
            "219/219 [==============================] - 1s 7ms/step - loss: 1.4767e-04 - last_time_step_mse: 1.6423e-04 - val_loss: 1.3289e-04 - val_last_time_step_mse: 1.3167e-04\n",
            "Epoch 10/20\n",
            "219/219 [==============================] - 1s 7ms/step - loss: 1.2847e-04 - last_time_step_mse: 1.3466e-04 - val_loss: 1.3467e-04 - val_last_time_step_mse: 1.8794e-04\n",
            "Epoch 11/20\n",
            "219/219 [==============================] - 1s 7ms/step - loss: 1.1610e-04 - last_time_step_mse: 1.2740e-04 - val_loss: 1.0621e-04 - val_last_time_step_mse: 1.0963e-04\n",
            "Epoch 12/20\n",
            "219/219 [==============================] - 1s 7ms/step - loss: 1.0468e-04 - last_time_step_mse: 1.2247e-04 - val_loss: 9.0799e-05 - val_last_time_step_mse: 9.5410e-05\n",
            "Epoch 13/20\n",
            "219/219 [==============================] - 1s 7ms/step - loss: 9.0091e-05 - last_time_step_mse: 1.0744e-04 - val_loss: 8.0276e-05 - val_last_time_step_mse: 8.8627e-05\n",
            "Epoch 14/20\n",
            "219/219 [==============================] - 1s 7ms/step - loss: 7.9333e-05 - last_time_step_mse: 1.0149e-04 - val_loss: 7.2998e-05 - val_last_time_step_mse: 1.0007e-04\n",
            "Epoch 15/20\n",
            "219/219 [==============================] - 2s 7ms/step - loss: 7.0465e-05 - last_time_step_mse: 9.9085e-05 - val_loss: 5.9594e-05 - val_last_time_step_mse: 7.5488e-05\n",
            "Epoch 16/20\n",
            "219/219 [==============================] - 1s 7ms/step - loss: 5.8720e-05 - last_time_step_mse: 8.2031e-05 - val_loss: 5.1932e-05 - val_last_time_step_mse: 6.7779e-05\n",
            "Epoch 17/20\n",
            "219/219 [==============================] - 1s 7ms/step - loss: 5.5295e-05 - last_time_step_mse: 8.6700e-05 - val_loss: 4.5739e-05 - val_last_time_step_mse: 6.3427e-05\n",
            "Epoch 18/20\n",
            "219/219 [==============================] - 1s 7ms/step - loss: 5.3638e-05 - last_time_step_mse: 8.9969e-05 - val_loss: 4.0111e-05 - val_last_time_step_mse: 5.6868e-05\n",
            "Epoch 19/20\n",
            "219/219 [==============================] - 1s 7ms/step - loss: 3.8034e-05 - last_time_step_mse: 5.4497e-05 - val_loss: 3.4999e-05 - val_last_time_step_mse: 5.0018e-05\n",
            "Epoch 20/20\n",
            "219/219 [==============================] - 1s 7ms/step - loss: 4.0297e-05 - last_time_step_mse: 7.0471e-05 - val_loss: 3.0754e-05 - val_last_time_step_mse: 4.3673e-05\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Sllfo9RQJlf1",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 280
        },
        "outputId": "d3a43d85-e1f7-454d-c2c1-985db841be5c"
      },
      "source": [
        "plot_learning_curves(history.history[\"loss\"], history.history[\"val_loss\"])\n",
        "plt.show()"
      ],
      "execution_count": 68,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "XJ2NKJdNJmkP",
        "colab_type": "text"
      },
      "source": [
        "## WaveNet"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6bp2sm7SJl-Z",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "keras.backend.clear_session()"
      ],
      "execution_count": 69,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "29pMZwg8Ktmj",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 727
        },
        "outputId": "30c0fc73-3f7f-4904-a585-ddac054e56de"
      },
      "source": [
        "np.random.seed(42)\n",
        "tf.random.set_seed(42)\n",
        "\n",
        "model = keras.models.Sequential()\n",
        "model.add(keras.layers.InputLayer(input_shape=[None, 1]))\n",
        "for rate in (1, 2, 4, 8) * 2:\n",
        "    model.add(keras.layers.Conv1D(filters=20, kernel_size=2, padding=\"causal\",\n",
        "                                  activation=\"relu\", dilation_rate=rate))\n",
        "model.add(keras.layers.Conv1D(filters=10, kernel_size=1))\n",
        "model.compile(loss=\"mse\", optimizer=\"adam\", metrics=[last_time_step_mse])\n",
        "history = model.fit(X_train, Y_train, epochs=20,\n",
        "                    validation_data=(X_valid, Y_valid))"
      ],
      "execution_count": 70,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/20\n",
            "219/219 [==============================] - 2s 8ms/step - loss: 0.0865 - last_time_step_mse: 0.0806 - val_loss: 0.0630 - val_last_time_step_mse: 0.0847\n",
            "Epoch 2/20\n",
            "219/219 [==============================] - 1s 5ms/step - loss: 0.0605 - last_time_step_mse: 0.0856 - val_loss: 0.0593 - val_last_time_step_mse: 0.0855\n",
            "Epoch 3/20\n",
            "219/219 [==============================] - 1s 5ms/step - loss: 0.0590 - last_time_step_mse: 0.0854 - val_loss: 0.0588 - val_last_time_step_mse: 0.0866\n",
            "Epoch 4/20\n",
            "219/219 [==============================] - 1s 5ms/step - loss: 0.0586 - last_time_step_mse: 0.0852 - val_loss: 0.0585 - val_last_time_step_mse: 0.0850\n",
            "Epoch 5/20\n",
            "219/219 [==============================] - 1s 5ms/step - loss: 0.0585 - last_time_step_mse: 0.0853 - val_loss: 0.0583 - val_last_time_step_mse: 0.0847\n",
            "Epoch 6/20\n",
            "219/219 [==============================] - 1s 5ms/step - loss: 0.0583 - last_time_step_mse: 0.0854 - val_loss: 0.0583 - val_last_time_step_mse: 0.0851\n",
            "Epoch 7/20\n",
            "219/219 [==============================] - 1s 5ms/step - loss: 0.0582 - last_time_step_mse: 0.0853 - val_loss: 0.0582 - val_last_time_step_mse: 0.0871\n",
            "Epoch 8/20\n",
            "219/219 [==============================] - 1s 5ms/step - loss: 0.0582 - last_time_step_mse: 0.0852 - val_loss: 0.0583 - val_last_time_step_mse: 0.0874\n",
            "Epoch 9/20\n",
            "219/219 [==============================] - 1s 5ms/step - loss: 0.0581 - last_time_step_mse: 0.0852 - val_loss: 0.0581 - val_last_time_step_mse: 0.0860\n",
            "Epoch 10/20\n",
            "219/219 [==============================] - 1s 5ms/step - loss: 0.0580 - last_time_step_mse: 0.0851 - val_loss: 0.0579 - val_last_time_step_mse: 0.0856\n",
            "Epoch 11/20\n",
            "219/219 [==============================] - 1s 5ms/step - loss: 0.0579 - last_time_step_mse: 0.0852 - val_loss: 0.0579 - val_last_time_step_mse: 0.0854\n",
            "Epoch 12/20\n",
            "219/219 [==============================] - 1s 5ms/step - loss: 0.0580 - last_time_step_mse: 0.0852 - val_loss: 0.0578 - val_last_time_step_mse: 0.0848\n",
            "Epoch 13/20\n",
            "219/219 [==============================] - 1s 5ms/step - loss: 0.0579 - last_time_step_mse: 0.0851 - val_loss: 0.0581 - val_last_time_step_mse: 0.0841\n",
            "Epoch 14/20\n",
            "219/219 [==============================] - 1s 5ms/step - loss: 0.0578 - last_time_step_mse: 0.0851 - val_loss: 0.0577 - val_last_time_step_mse: 0.0848\n",
            "Epoch 15/20\n",
            "219/219 [==============================] - 1s 5ms/step - loss: 0.0578 - last_time_step_mse: 0.0851 - val_loss: 0.0578 - val_last_time_step_mse: 0.0836\n",
            "Epoch 16/20\n",
            "219/219 [==============================] - 1s 6ms/step - loss: 0.0577 - last_time_step_mse: 0.0850 - val_loss: 0.0579 - val_last_time_step_mse: 0.0862\n",
            "Epoch 17/20\n",
            "219/219 [==============================] - 1s 6ms/step - loss: 0.0577 - last_time_step_mse: 0.0850 - val_loss: 0.0578 - val_last_time_step_mse: 0.0833\n",
            "Epoch 18/20\n",
            "219/219 [==============================] - 1s 6ms/step - loss: 0.0577 - last_time_step_mse: 0.0849 - val_loss: 0.0576 - val_last_time_step_mse: 0.0848\n",
            "Epoch 19/20\n",
            "219/219 [==============================] - 1s 6ms/step - loss: 0.0577 - last_time_step_mse: 0.0849 - val_loss: 0.0577 - val_last_time_step_mse: 0.0858\n",
            "Epoch 20/20\n",
            "219/219 [==============================] - 1s 5ms/step - loss: 0.0577 - last_time_step_mse: 0.0849 - val_loss: 0.0576 - val_last_time_step_mse: 0.0830\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nydVOnl3LK6K",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        },
        "outputId": "1155f9f7-0dfd-485f-bde9-8e789104ebfb"
      },
      "source": [
        "model.evaluate(X_test, Y_test)"
      ],
      "execution_count": 71,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "32/32 [==============================] - 0s 5ms/step - loss: 0.0579 - last_time_step_mse: 0.0834\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.057884469628334045, 0.0834352970123291]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 71
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5eJKpq_KLVM8",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 278
        },
        "outputId": "2fc59804-821b-467d-c689-7e0cb2b5c902"
      },
      "source": [
        "plot_learning_curves(history.history[\"loss\"], history.history[\"val_loss\"])\n",
        "plt.show()"
      ],
      "execution_count": 72,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "A9Ch16QYMIEy",
        "colab_type": "text"
      },
      "source": [
        "\n",
        "Here is the original WaveNet defined in the paper: it uses Gated Activation Units instead of ReLU and parametrized skip connections, plus it pads with zeros on the left to avoid getting shorter and shorter sequences:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4ctQrLWWLXxk",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "class GatedActivationUnit(keras.layers.Layer):\n",
        "    def __init__(self, activation=\"tanh\", **kwargs):\n",
        "        super().__init__(**kwargs)\n",
        "        self.activation = keras.activations.get(activation)\n",
        "    def call(self, inputs):\n",
        "        n_filters = inputs.shape[-1] // 2\n",
        "        linear_output = self.activation(inputs[..., :n_filters])\n",
        "        gate = keras.activations.sigmoid(inputs[..., n_filters:])\n",
        "        return self.activation(linear_output) * gate"
      ],
      "execution_count": 73,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XLYGyWhTMV3-",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def wavenet_residual_block(inputs, n_filters, dilation_rate):\n",
        "    z = keras.layers.Conv1D(2 * n_filters, kernel_size=2, padding=\"causal\",\n",
        "                            dilation_rate=dilation_rate)(inputs)\n",
        "    z = GatedActivationUnit()(z)\n",
        "    z = keras.layers.Conv1D(n_filters, kernel_size=1)(z)\n",
        "    return keras.layers.Add()([z, inputs]), z"
      ],
      "execution_count": 74,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BHyLjKUGMaaa",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "keras.backend.clear_session()\n",
        "np.random.seed(42)\n",
        "tf.random.set_seed(42)\n",
        "\n",
        "n_layers_per_block = 3 # 10 in the paper\n",
        "n_blocks = 1 # 3 in the paper\n",
        "n_filters = 32 # 128 in the paper\n",
        "n_outputs = 10 # 256 in the paper\n",
        "\n",
        "inputs = keras.layers.Input(shape=[None, 1])\n",
        "z = keras.layers.Conv1D(n_filters, kernel_size=2, padding=\"causal\")(inputs)\n",
        "skip_to_last = []\n",
        "for dilation_rate in [2**i for i in range(n_layers_per_block)] * n_blocks:\n",
        "    z, skip = wavenet_residual_block(z, n_filters, dilation_rate)\n",
        "    skip_to_last.append(skip)\n",
        "z = keras.activations.relu(keras.layers.Add()(skip_to_last))\n",
        "z = keras.layers.Conv1D(n_filters, kernel_size=1, activation=\"relu\")(z)\n",
        "Y_proba = keras.layers.Conv1D(n_outputs, kernel_size=1, activation=\"softmax\")(z)\n",
        "\n",
        "model = keras.models.Model(inputs=[inputs], outputs=[Y_proba])"
      ],
      "execution_count": 79,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jpceW7EiMw0K",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 727
        },
        "outputId": "19d4b2c8-4537-4f13-fb05-66c750b60fba"
      },
      "source": [
        "model.compile(loss=\"mse\", optimizer=\"adam\", metrics=[last_time_step_mse])\n",
        "history = model.fit(X_train, Y_train, epochs=20,\n",
        "                    validation_data=(X_valid, Y_valid))"
      ],
      "execution_count": 80,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/20\n",
            "219/219 [==============================] - 2s 8ms/step - loss: 0.1572 - last_time_step_mse: 0.0516 - val_loss: 0.1561 - val_last_time_step_mse: 0.0540\n",
            "Epoch 2/20\n",
            "219/219 [==============================] - 1s 6ms/step - loss: 0.1557 - last_time_step_mse: 0.0538 - val_loss: 0.1555 - val_last_time_step_mse: 0.0539\n",
            "Epoch 3/20\n",
            "219/219 [==============================] - 1s 6ms/step - loss: 0.1554 - last_time_step_mse: 0.0534 - val_loss: 0.1553 - val_last_time_step_mse: 0.0537\n",
            "Epoch 4/20\n",
            "219/219 [==============================] - 1s 6ms/step - loss: 0.1552 - last_time_step_mse: 0.0532 - val_loss: 0.1551 - val_last_time_step_mse: 0.0534\n",
            "Epoch 5/20\n",
            "219/219 [==============================] - 1s 6ms/step - loss: 0.1551 - last_time_step_mse: 0.0529 - val_loss: 0.1551 - val_last_time_step_mse: 0.0528\n",
            "Epoch 6/20\n",
            "219/219 [==============================] - 1s 6ms/step - loss: 0.1551 - last_time_step_mse: 0.0528 - val_loss: 0.1551 - val_last_time_step_mse: 0.0535\n",
            "Epoch 7/20\n",
            "219/219 [==============================] - 1s 6ms/step - loss: 0.1551 - last_time_step_mse: 0.0528 - val_loss: 0.1550 - val_last_time_step_mse: 0.0534\n",
            "Epoch 8/20\n",
            "219/219 [==============================] - 1s 6ms/step - loss: 0.1551 - last_time_step_mse: 0.0528 - val_loss: 0.1550 - val_last_time_step_mse: 0.0530\n",
            "Epoch 9/20\n",
            "219/219 [==============================] - 1s 6ms/step - loss: 0.1550 - last_time_step_mse: 0.0528 - val_loss: 0.1550 - val_last_time_step_mse: 0.0527\n",
            "Epoch 10/20\n",
            "219/219 [==============================] - 1s 6ms/step - loss: 0.1550 - last_time_step_mse: 0.0528 - val_loss: 0.1551 - val_last_time_step_mse: 0.0529\n",
            "Epoch 11/20\n",
            "219/219 [==============================] - 1s 6ms/step - loss: 0.1550 - last_time_step_mse: 0.0528 - val_loss: 0.1550 - val_last_time_step_mse: 0.0529\n",
            "Epoch 12/20\n",
            "219/219 [==============================] - 1s 6ms/step - loss: 0.1550 - last_time_step_mse: 0.0528 - val_loss: 0.1550 - val_last_time_step_mse: 0.0523\n",
            "Epoch 13/20\n",
            "219/219 [==============================] - 1s 6ms/step - loss: 0.1550 - last_time_step_mse: 0.0528 - val_loss: 0.1551 - val_last_time_step_mse: 0.0523\n",
            "Epoch 14/20\n",
            "219/219 [==============================] - 1s 6ms/step - loss: 0.1550 - last_time_step_mse: 0.0528 - val_loss: 0.1550 - val_last_time_step_mse: 0.0525\n",
            "Epoch 15/20\n",
            "219/219 [==============================] - 1s 6ms/step - loss: 0.1550 - last_time_step_mse: 0.0527 - val_loss: 0.1550 - val_last_time_step_mse: 0.0534\n",
            "Epoch 16/20\n",
            "219/219 [==============================] - 1s 6ms/step - loss: 0.1550 - last_time_step_mse: 0.0528 - val_loss: 0.1550 - val_last_time_step_mse: 0.0529\n",
            "Epoch 17/20\n",
            "219/219 [==============================] - 1s 6ms/step - loss: 0.1550 - last_time_step_mse: 0.0527 - val_loss: 0.1550 - val_last_time_step_mse: 0.0532\n",
            "Epoch 18/20\n",
            "219/219 [==============================] - 1s 6ms/step - loss: 0.1550 - last_time_step_mse: 0.0527 - val_loss: 0.1550 - val_last_time_step_mse: 0.0527\n",
            "Epoch 19/20\n",
            "219/219 [==============================] - 1s 6ms/step - loss: 0.1550 - last_time_step_mse: 0.0527 - val_loss: 0.1550 - val_last_time_step_mse: 0.0526\n",
            "Epoch 20/20\n",
            "219/219 [==============================] - 1s 6ms/step - loss: 0.1550 - last_time_step_mse: 0.0527 - val_loss: 0.1550 - val_last_time_step_mse: 0.0530\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ltQOi5iIMzKi",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        },
        "outputId": "c705ff33-0f35-42c7-b4fd-cf3e44a198b9"
      },
      "source": [
        "model.evaluate(X_test, Y_test)"
      ],
      "execution_count": 81,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "32/32 [==============================] - 0s 3ms/step - loss: 0.1550 - last_time_step_mse: 0.0529\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.15499015152454376, 0.05293254181742668]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 81
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "m5EADzXnM1FK",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 278
        },
        "outputId": "03e77646-69ca-40c7-925d-ec1986f13ef4"
      },
      "source": [
        "plot_learning_curves(history.history[\"loss\"], history.history[\"val_loss\"])\n",
        "plt.show()"
      ],
      "execution_count": 82,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    }
  ]
}