neural_nets_with_keras.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "source": "# Setup",
   "metadata": {
    "cell_id": "1f31532d-ba19-4e47-8d47-8ac1f3fabfaf",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 82
   }
  },
  {
   "cell_type": "markdown",
   "source": "First, let's import a few common modules, ensure MatplotLib plots figures inline and prepare a function to save the figures. We also check that Python 3.5 or later is installed (although Python 2.x may work, it is deprecated so we strongly recommend you use Python 3 instead), as well as Scikit-Learn >= 0.20 and TensorFlow >= 2.0.",
   "metadata": {
    "cell_id": "00001-4439061e-f12e-4fea-9f44-e490585ab559",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 97.171875
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00002-17fbe845-9e89-44a9-9bbe-aae7ef17bead",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "8977a559",
    "execution_start": 1659186490674,
    "execution_millis": 4649,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 886
   },
   "source": "# Python >= 3.5 is required\nimport sys\nassert sys.version_info >= (3, 5)\n\n# Scikit-Learn >= 0.20 is required\nimport sklearn\nassert sklearn.__version__ >= \"0.20\"\n\ntry:\n    # %tensorflow_version only exists in Colab.\n    %tensorflow_version 2.x\nexcept Exception:\n    pass\n\n# TensorFlow >= 2.0 is required\nimport tensorflow as tf\nassert tf.__version__ >= \"2.0\"\n\n# Common imports\nimport numpy as np\nimport os\n\n# To make this notebook's output stable across runs\nnp.random.seed(42)\n\n# To plot pretty figures\n%matplotlib inline\nimport matplotlib as mpl\nimport matplotlib.pyplot as plt\nmpl.rc('axes', labelsize=14)\nmpl.rc('xtick', labelsize=12)\nmpl.rc('ytick', labelsize=12)\n\n# Where to save the figures\nPROJECT_ROOT_DIR = \".\"\nCHAPTER_ID = \"ann\"\nIMAGES_PATH = os.path.join(PROJECT_ROOT_DIR, \"images\", CHAPTER_ID)\nos.makedirs(IMAGES_PATH, exist_ok=True)\n\ndef save_fig(fig_id, tight_layout=True, fig_extension=\"png\", resolution=300):\n    path = os.path.join(IMAGES_PATH, fig_id + \".\" + fig_extension)\n    print(\"Saving figure\", fig_id)\n    if tight_layout:\n        plt.tight_layout()\n    plt.savefig(path, format=fig_extension, dpi=resolution)\n",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "markdown",
   "source": "# Perceptrons",
   "metadata": {
    "cell_id": "00003-3001bf5c-97ef-4d29-b49a-aa0af26e4c60",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 82
   }
  },
  {
   "cell_type": "markdown",
   "source": "**Note**: we set `max_iter` and `tol` explicitly to avoid warnings about the fact that their default value will change in future versions of Scikit-Learn.",
   "metadata": {
    "cell_id": "00004-a3e671af-bb05-40b2-aad0-94898cd01615",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 74.78125
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00005-594bbd89-3efa-496c-a318-a17893bffd6d",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "88b4afd5",
    "execution_start": 1659186495324,
    "execution_millis": 166,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 274
   },
   "source": "import numpy as np\nfrom sklearn.datasets import load_iris\nfrom sklearn.linear_model import Perceptron\n\niris = load_iris()\nX = iris.data[:, (2, 3)]  # petal length, petal width\ny = (iris.target == 0).astype(int)\n\nper_clf = Perceptron(max_iter=1000, tol=1e-3, random_state=42)\nper_clf.fit(X, y)\n\ny_pred = per_clf.predict([[2, 0.5]])",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00006-24b81fa9-bb30-4cbb-bb1c-e58e844041a0",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "ba549d12",
    "execution_start": 1659186495500,
    "execution_millis": 380,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 113.1875,
    "deepnote_output_heights": [
     21.1875
    ]
   },
   "source": "y_pred",
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "execution_count": 3,
     "data": {
      "text/plain": "array([1])"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00007-858aa6ab-b9de-45e9-9133-c9310c41c106",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "aee84082",
    "execution_start": 1659186495519,
    "execution_millis": 899,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 924.1875,
    "deepnote_output_heights": [
     null,
     282
    ]
   },
   "source": "a = -per_clf.coef_[0][0] / per_clf.coef_[0][1]\nb = -per_clf.intercept_ / per_clf.coef_[0][1]\n\naxes = [0, 5, 0, 2]\n\nx0, x1 = np.meshgrid(\n        np.linspace(axes[0], axes[1], 500).reshape(-1, 1),\n        np.linspace(axes[2], axes[3], 200).reshape(-1, 1),\n    )\nX_new = np.c_[x0.ravel(), x1.ravel()]\ny_predict = per_clf.predict(X_new)\nzz = y_predict.reshape(x0.shape)\n\nplt.figure(figsize=(10, 4))\nplt.plot(X[y==0, 0], X[y==0, 1], \"bs\", label=\"Not Iris-Setosa\")\nplt.plot(X[y==1, 0], X[y==1, 1], \"yo\", label=\"Iris-Setosa\")\n\nplt.plot([axes[0], axes[1]], [a * axes[0] + b, a * axes[1] + b], \"k-\", linewidth=3)\nfrom matplotlib.colors import ListedColormap\ncustom_cmap = ListedColormap(['#9898ff', '#fafab0'])\n\nplt.contourf(x0, x1, zz, cmap=custom_cmap)\nplt.xlabel(\"Petal length\", fontsize=14)\nplt.ylabel(\"Petal width\", fontsize=14)\nplt.legend(loc=\"lower right\", fontsize=14)\nplt.axis(axes)\n\nsave_fig(\"perceptron_iris_plot\")\nplt.show()",
   "execution_count": null,
   "outputs": [
    {
     "name": "stdout",
     "text": "Saving figure perceptron_iris_plot\n",
     "output_type": "stream"
    },
    {
     "data": {
      "text/plain": "<Figure size 720x288 with 1 Axes>",
      "image/png": "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\n"
     },
     "metadata": {
      "needs_background": "light",
      "image/png": {
       "width": 712,
       "height": 280
      }
     },
     "output_type": "display_data"
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": "# Activation functions",
   "metadata": {
    "cell_id": "00008-8f42edb2-ed97-4ab2-b6d6-f41a3eb185b5",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 82
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00009-92584274-8203-4def-83c6-375b4c4464a2",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "ca572cf",
    "execution_start": 1659186496419,
    "execution_millis": 5,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 202
   },
   "source": "def sigmoid(z):\n    return 1 / (1 + np.exp(-z))\n\ndef relu(z):\n    return np.maximum(0, z)\n\ndef derivative(f, z, eps=0.000001):\n    return (f(z + eps) - f(z - eps))/(2 * eps)",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00010-ef789bf5-6cb2-42c4-983b-95e664e13fdf",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "f820f334",
    "execution_start": 1659186496425,
    "execution_millis": 861,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 906.1875,
    "deepnote_output_heights": [
     null,
     282
    ]
   },
   "source": "z = np.linspace(-5, 5, 200)\n\nplt.figure(figsize=(11,4))\n\nplt.subplot(121)\nplt.plot(z, np.sign(z), \"r-\", linewidth=1, label=\"Step\")\nplt.plot(z, sigmoid(z), \"g--\", linewidth=2, label=\"Sigmoid\")\nplt.plot(z, np.tanh(z), \"b-\", linewidth=2, label=\"Tanh\")\nplt.plot(z, relu(z), \"m-.\", linewidth=2, label=\"ReLU\")\nplt.grid(True)\nplt.legend(loc=\"center right\", fontsize=14)\nplt.title(\"Activation functions\", fontsize=14)\nplt.axis([-5, 5, -1.2, 1.2])\n\nplt.subplot(122)\nplt.plot(z, derivative(np.sign, z), \"r-\", linewidth=1, label=\"Step\")\nplt.plot(0, 0, \"ro\", markersize=5)\nplt.plot(0, 0, \"rx\", markersize=10)\nplt.plot(z, derivative(sigmoid, z), \"g--\", linewidth=2, label=\"Sigmoid\")\nplt.plot(z, derivative(np.tanh, z), \"b-\", linewidth=2, label=\"Tanh\")\nplt.plot(z, derivative(relu, z), \"m-.\", linewidth=2, label=\"ReLU\")\nplt.grid(True)\n#plt.legend(loc=\"center right\", fontsize=14)\nplt.title(\"Derivatives\", fontsize=14)\nplt.axis([-5, 5, -0.2, 1.2])\n\nsave_fig(\"activation_functions_plot\")\nplt.show()",
   "execution_count": null,
   "outputs": [
    {
     "name": "stdout",
     "text": "Saving figure activation_functions_plot\n",
     "output_type": "stream"
    },
    {
     "data": {
      "text/plain": "<Figure size 792x288 with 2 Axes>",
      "image/png": "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\n"
     },
     "metadata": {
      "needs_background": "light",
      "image/png": {
       "width": 784,
       "height": 280
      }
     },
     "output_type": "display_data"
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00011-7d6ac778-a34d-4d24-9ee9-5b302830478a",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "296a473c",
    "execution_start": 1659186497287,
    "execution_millis": 4,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 148
   },
   "source": "def heaviside(z):\n    return (z >= 0).astype(z.dtype)\n\ndef mlp_xor(x1, x2, activation=heaviside):\n    return activation(-activation(x1 + x2 - 1.5) + activation(x1 + x2 - 0.5) - 0.5)",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "scrolled": true,
    "cell_id": "00012-ed9626f0-78fa-416f-a979-e00df0400cb1",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "cf36b917",
    "execution_start": 1659186497302,
    "execution_millis": 335,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 740,
    "deepnote_output_heights": [
     270
    ]
   },
   "source": "x1s = np.linspace(-0.2, 1.2, 100)\nx2s = np.linspace(-0.2, 1.2, 100)\nx1, x2 = np.meshgrid(x1s, x2s)\n\nz1 = mlp_xor(x1, x2, activation=heaviside)\nz2 = mlp_xor(x1, x2, activation=sigmoid)\n\nplt.figure(figsize=(10,4))\n\nplt.subplot(121)\nplt.contourf(x1, x2, z1)\nplt.plot([0, 1], [0, 1], \"gs\", markersize=20)\nplt.plot([0, 1], [1, 0], \"y^\", markersize=20)\nplt.title(\"Activation function: heaviside\", fontsize=14)\nplt.grid(True)\n\nplt.subplot(122)\nplt.contourf(x1, x2, z2)\nplt.plot([0, 1], [0, 1], \"gs\", markersize=20)\nplt.plot([0, 1], [1, 0], \"y^\", markersize=20)\nplt.title(\"Activation function: sigmoid\", fontsize=14)\nplt.grid(True)",
   "execution_count": null,
   "outputs": [
    {
     "data": {
      "text/plain": "<Figure size 720x288 with 2 Axes>",
      "image/png": "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\n"
     },
     "metadata": {
      "needs_background": "light",
      "image/png": {
       "width": 617,
       "height": 268
      }
     },
     "output_type": "display_data"
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": "# Building an Image Classifier",
   "metadata": {
    "cell_id": "00013-531295f8-e09a-4fca-8fda-7dae111fabc3",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 82
   }
  },
  {
   "cell_type": "markdown",
   "source": "First let's import TensorFlow and Keras.",
   "metadata": {
    "cell_id": "00014-337e392d-43ee-4d96-8ec1-b776fbdc752a",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 52.390625
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00015-95402d13-cf9e-486c-a2bc-685dc46fccb1",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "cc0515f2",
    "execution_start": 1659186497645,
    "execution_millis": 4,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 94
   },
   "source": "import tensorflow as tf\nfrom tensorflow import keras",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00016-26ac36f4-6b28-4120-b137-356046704f23",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "de35829e",
    "execution_start": 1659186497657,
    "execution_millis": 8,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 113.1875,
    "deepnote_output_heights": [
     21.1875
    ]
   },
   "source": "tf.__version__",
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "execution_count": 10,
     "data": {
      "text/plain": "'2.9.1'"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00017-ae254d14-f2a0-45b8-8558-c50a8c788c47",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "6007d765",
    "execution_start": 1659186497666,
    "execution_millis": 7,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 113.1875,
    "deepnote_output_heights": [
     21.1875
    ]
   },
   "source": "keras.__version__",
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "execution_count": 11,
     "data": {
      "text/plain": "'2.9.0'"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": "Let's start by loading the fashion MNIST dataset. Keras has a number of functions to load popular datasets in `keras.datasets`. The dataset is already split for you between a training set and a test set, but it can be useful to split the training set further to have a validation set:",
   "metadata": {
    "cell_id": "00018-24064f7d-6fb2-48db-9f4f-71cc8afaa669",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 97.171875
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00019-39c2a6bd-268e-450e-83b1-b63120aefb8b",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "e40172d1",
    "execution_start": 1659186497677,
    "execution_millis": 458,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 94
   },
   "source": "fashion_mnist = keras.datasets.fashion_mnist\n(X_train_full, y_train_full), (X_test, y_test) = fashion_mnist.load_data()",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "markdown",
   "source": "The training set contains 60,000 grayscale images, each 28x28 pixels:",
   "metadata": {
    "cell_id": "00020-d67b7e1f-f635-40fe-ab20-4a5e8657d667",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 52.390625
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00021-3b4c3c7c-932d-4648-b1be-be206cf258f4",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "3b45a169",
    "execution_start": 1659186498139,
    "execution_millis": 580,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 113.1875
   },
   "source": "X_train_full.shape",
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "execution_count": 13,
     "data": {
      "text/plain": "(60000, 28, 28)"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": "Each pixel intensity is represented as a byte (0 to 255):",
   "metadata": {
    "cell_id": "00022-b5fe5968-dfc0-430d-82d1-7e4d29984a22",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 52.390625
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00023-8c1db79f-0581-4ba9-9297-373a6664e3bf",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "32a91791",
    "execution_start": 1659186498194,
    "execution_millis": 527,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 113.1875,
    "deepnote_output_heights": [
     21.1875
    ]
   },
   "source": "X_train_full.dtype",
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "execution_count": 14,
     "data": {
      "text/plain": "dtype('uint8')"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": "Let's split the full training set into a validation set and a (smaller) training set. We also scale the pixel intensities down to the 0-1 range and convert them to floats, by dividing by 255.",
   "metadata": {
    "cell_id": "00024-89067007-c7c4-4e08-80d9-d860e34dedaa",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 74.78125
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00025-bc7b4ebe-e9c3-4234-8a1e-50f8e2b16465",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "364f401f",
    "execution_start": 1659186498195,
    "execution_millis": 112,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 112
   },
   "source": "X_valid, X_train = X_train_full[:5000] / 255., X_train_full[5000:] / 255.\ny_valid, y_train = y_train_full[:5000], y_train_full[5000:]\nX_test = X_test / 255.",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "markdown",
   "source": "You can plot an image using Matplotlib's `imshow()` function, with a `'binary'`\n color map:",
   "metadata": {
    "cell_id": "00026-cf2d12f6-c521-46ec-a918-9b7bb5116036",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 74.78125
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00027-e33843a0-bf1d-44fa-b7ff-4df2d0183333",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "c03cb47f",
    "execution_start": 1659186498317,
    "execution_millis": 109,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 361,
    "deepnote_output_heights": [
     233
    ]
   },
   "source": "plt.imshow(X_train[0], cmap=\"binary\")\nplt.axis('off')\nplt.show()",
   "execution_count": null,
   "outputs": [
    {
     "data": {
      "text/plain": "<Figure size 432x288 with 1 Axes>",
      "image/png": "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\n"
     },
     "metadata": {
      "needs_background": "light",
      "image/png": {
       "width": 231,
       "height": 231
      }
     },
     "output_type": "display_data"
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": "The labels are the class IDs (represented as uint8), from 0 to 9:",
   "metadata": {
    "cell_id": "00028-b8de4f57-a9c6-4597-8a9b-6c7a8a4634a6",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 52.390625
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00029-73553e5e-509f-4f82-acdf-af18ee76b803",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "ed857f47",
    "execution_start": 1659186498427,
    "execution_millis": 296,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 113.1875
   },
   "source": "y_train",
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "execution_count": 17,
     "data": {
      "text/plain": "array([4, 0, 7, ..., 3, 0, 5], dtype=uint8)"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": "Here are the corresponding class names:",
   "metadata": {
    "cell_id": "00030-a1c11a05-ff5a-4ecd-8f10-f3b3d97ff8b5",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 52.390625
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00031-1c4988f4-a5f9-460c-b933-753ccb861013",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "ca5041f8",
    "execution_start": 1659186498427,
    "execution_millis": 1,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 94
   },
   "source": "class_names = [\"T-shirt/top\", \"Trouser\", \"Pullover\", \"Dress\", \"Coat\",\n               \"Sandal\", \"Shirt\", \"Sneaker\", \"Bag\", \"Ankle boot\"]",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "markdown",
   "source": "So the first image in the training set is a coat:",
   "metadata": {
    "cell_id": "00032-4ab6a297-fa96-4c44-8f2f-609f50280a3c",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 52.390625
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00033-82d304e4-2f49-475a-9606-823520636159",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "cfaa7dfd",
    "execution_start": 1659186498487,
    "execution_millis": 238,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 113.1875
   },
   "source": "class_names[y_train[0]]",
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "execution_count": 19,
     "data": {
      "text/plain": "'Coat'"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": "The validation set contains 5,000 images, and the test set contains 10,000 images:",
   "metadata": {
    "cell_id": "00034-8646866b-db84-435e-98ee-d5af8589a450",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 52.390625
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00035-d7434a5d-c50a-45fc-b761-2cc41d61e50f",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "ffc02169",
    "execution_start": 1659186498492,
    "execution_millis": 234,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 113.1875
   },
   "source": "X_valid.shape",
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "execution_count": 20,
     "data": {
      "text/plain": "(5000, 28, 28)"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00036-8142d8f8-4bb0-441a-9aa2-c7f555134582",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "251bcee5",
    "execution_start": 1659186498496,
    "execution_millis": 232,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 113.1875
   },
   "source": "X_test.shape",
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "execution_count": 21,
     "data": {
      "text/plain": "(10000, 28, 28)"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": "Let's take a look at a sample of the images in the dataset:",
   "metadata": {
    "cell_id": "00037-758278ba-b08f-4c17-8501-e7cc439f73a7",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 52.390625
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00038-8ae2e9bc-8726-45aa-8288-9050be9e644a",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "d7bc000d",
    "execution_start": 1659186498509,
    "execution_millis": 1875,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 647.1875,
    "deepnote_output_heights": [
     null,
     293
    ]
   },
   "source": "n_rows = 4\nn_cols = 10\nplt.figure(figsize=(n_cols * 1.2, n_rows * 1.2))\nfor row in range(n_rows):\n    for col in range(n_cols):\n        index = n_cols * row + col\n        plt.subplot(n_rows, n_cols, index + 1)\n        plt.imshow(X_train[index], cmap=\"binary\", interpolation=\"nearest\")\n        plt.axis('off')\n        plt.title(class_names[y_train[index]], fontsize=12)\nplt.subplots_adjust(wspace=0.2, hspace=0.5)\nsave_fig('fashion_mnist_plot', tight_layout=False)\nplt.show()",
   "execution_count": null,
   "outputs": [
    {
     "name": "stdout",
     "text": "Saving figure fashion_mnist_plot\n",
     "output_type": "stream"
    },
    {
     "data": {
      "text/plain": "<Figure size 864x345.6 with 40 Axes>",
      "image/png": "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\n"
     },
     "metadata": {
      "needs_background": "light",
      "image/png": {
       "width": 681,
       "height": 291
      }
     },
     "output_type": "display_data"
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00039-4af898a6-8e81-4952-a20e-9d9ff85b6ee2",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "7ea9606",
    "execution_start": 1659186500433,
    "execution_millis": 1,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 148
   },
   "source": "model = keras.models.Sequential()\nmodel.add(keras.layers.Flatten(input_shape=[28, 28]))\nmodel.add(keras.layers.Dense(300, activation=\"relu\"))\nmodel.add(keras.layers.Dense(100, activation=\"relu\"))\nmodel.add(keras.layers.Dense(10, activation=\"softmax\"))",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00040-e0b1dfe4-835c-4e32-b90c-41d01ab2bb76",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "b86c7ea",
    "execution_start": 1659186500434,
    "execution_millis": 44,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 112
   },
   "source": "keras.backend.clear_session()\nnp.random.seed(42)\ntf.random.set_seed(42)",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00041-2edbf181-56a7-45e2-8f49-35cb1bd8cd12",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "e2bfe76c",
    "execution_start": 1659186500479,
    "execution_millis": 0,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 166
   },
   "source": "model = keras.models.Sequential([\n    keras.layers.Flatten(input_shape=[28, 28]),\n    keras.layers.Dense(300, activation=\"relu\"),\n    keras.layers.Dense(100, activation=\"relu\"),\n    keras.layers.Dense(10, activation=\"softmax\")\n])",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00042-4da9eca5-b97d-4e99-bdd0-cedd067f022b",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "e081c798",
    "execution_start": 1659186500492,
    "execution_millis": 1,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 170.75
   },
   "source": "model.layers",
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "execution_count": 26,
     "data": {
      "text/plain": "[<keras.layers.reshaping.flatten.Flatten at 0x7f3b266895b0>,\n <keras.layers.core.dense.Dense at 0x7f3b266896d0>,\n <keras.layers.core.dense.Dense at 0x7f3b26689700>,\n <keras.layers.core.dense.Dense at 0x7f3b26689880>]"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00043-df26277b-80c7-40e1-8eb2-055cd6f202dd",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "4e6a3b95",
    "execution_start": 1659186500523,
    "execution_millis": 3,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 430.1875
   },
   "source": "model.summary()",
   "execution_count": null,
   "outputs": [
    {
     "name": "stdout",
     "text": "Model: \"sequential\"\n_________________________________________________________________\n Layer (type)                Output Shape              Param #   \n=================================================================\n flatten (Flatten)           (None, 784)               0         \n                                                                 \n dense (Dense)               (None, 300)               235500    \n                                                                 \n dense_1 (Dense)             (None, 100)               30100     \n                                                                 \n dense_2 (Dense)             (None, 10)                1010      \n                                                                 \n=================================================================\nTotal params: 266,610\nTrainable params: 266,610\nNon-trainable params: 0\n_________________________________________________________________\n",
     "output_type": "stream"
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00044-32b8120a-623c-4737-a6fe-a1463ffc21ef",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "7845e35c",
    "execution_start": 1659186500577,
    "execution_millis": 28,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 610,
    "deepnote_output_heights": [
     518
    ]
   },
   "source": "keras.utils.plot_model(model, \"my_fashion_mnist_model.png\", show_shapes=True)",
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "execution_count": 28,
     "data": {
      "image/png": "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\n",
      "text/plain": "<IPython.core.display.Image object>"
     },
     "metadata": {
      "image/png": {
       "width": 420,
       "height": 516
      }
     }
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00045-a7535fce-b479-4d6f-bf34-05315404042e",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "89350fb9",
    "execution_start": 1659186500606,
    "execution_millis": 10,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 131.1875,
    "deepnote_output_heights": [
     21.1875
    ]
   },
   "source": "hidden1 = model.layers[1]\nhidden1.name",
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "execution_count": 29,
     "data": {
      "text/plain": "'dense'"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00046-3c433c77-4ac9-4611-a62a-6934abce0fa3",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "468ed498",
    "execution_start": 1659186500617,
    "execution_millis": 4,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 113.1875,
    "deepnote_output_heights": [
     21.1875
    ]
   },
   "source": "model.get_layer(hidden1.name) is hidden1",
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "execution_count": 30,
     "data": {
      "text/plain": "True"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00047-ce7e6134-37dd-4448-ab8d-6937832617b1",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "7ca43ca4",
    "execution_start": 1659186500625,
    "execution_millis": 3,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 76
   },
   "source": "weights, biases = hidden1.get_weights()",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00048-6115cb12-0da4-4778-8440-70a48f29e247",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "262735a6",
    "execution_start": 1659186500634,
    "execution_millis": 5,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 343.4375,
    "deepnote_output_heights": [
     251.4375
    ]
   },
   "source": "weights",
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "execution_count": 32,
     "data": {
      "text/plain": "array([[ 0.02448617, -0.00877795, -0.02189048, ..., -0.02766046,\n         0.03859074, -0.06889391],\n       [ 0.00476504, -0.03105379, -0.0586676 , ...,  0.00602964,\n        -0.02763776, -0.04165364],\n       [-0.06189284, -0.06901957,  0.07102345, ..., -0.04238207,\n         0.07121518, -0.07331658],\n       ...,\n       [-0.03048757,  0.02155137, -0.05400612, ..., -0.00113463,\n         0.00228987,  0.05581069],\n       [ 0.07061854, -0.06960931,  0.07038955, ..., -0.00384101,\n         0.00034875,  0.02878492],\n       [-0.06022581,  0.01577859, -0.02585464, ..., -0.00527829,\n         0.00272203, -0.06793761]], dtype=float32)"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00049-922bef0a-6852-4d58-8b80-ea5a4d4e074c",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "6a6092a8",
    "execution_start": 1659186500646,
    "execution_millis": 1,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 113.1875,
    "deepnote_output_heights": [
     21.1875
    ]
   },
   "source": "weights.shape",
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "execution_count": 33,
     "data": {
      "text/plain": "(784, 300)"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00050-c78b1f87-ee4e-47c4-a99d-07a843e19cf7",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "abb7e81b",
    "execution_start": 1659186500654,
    "execution_millis": 5,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 439.375
   },
   "source": "biases",
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "execution_count": 34,
     "data": {
      "text/plain": "array([0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n       0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n       0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n       0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n       0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n       0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n       0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n       0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n       0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n       0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n       0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n       0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n       0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n       0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n       0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n       0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n       0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n       0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.], dtype=float32)"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00051-1de2c631-8f9f-4deb-a93c-9f13255e4943",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "12fb90da",
    "execution_start": 1659186500662,
    "execution_millis": 4,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 113.1875
   },
   "source": "biases.shape",
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "execution_count": 35,
     "data": {
      "text/plain": "(300,)"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00052-076a3d50-5d2c-4031-97d6-ec77d9c47f3d",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "11a72e69",
    "execution_start": 1659186500671,
    "execution_millis": 842,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 112
   },
   "source": "model.compile(loss=\"sparse_categorical_crossentropy\",\n              optimizer=\"sgd\",\n              metrics=[\"accuracy\"])",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "markdown",
   "source": "This is equivalent to:",
   "metadata": {
    "cell_id": "00053-3e1de890-af63-4e0f-b400-bc73380e6bf3",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 52.390625
   }
  },
  {
   "cell_type": "markdown",
   "source": "```python\nmodel.compile(loss=keras.losses.sparse_categorical_crossentropy,\n              optimizer=keras.optimizers.SGD(),\n              metrics=[keras.metrics.sparse_categorical_accuracy])\n```",
   "metadata": {
    "cell_id": "00054-9ca3bd49-5587-4804-8f51-d154058853b5",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 117.171875
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00055-701ecd99-a6e0-4bef-88b2-2fd4d369ae24",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "5c55cc88",
    "execution_start": 1659186500686,
    "execution_millis": 135027,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 714
   },
   "source": "history = model.fit(X_train, y_train, epochs=30,\n                    validation_data=(X_valid, y_valid))",
   "execution_count": null,
   "outputs": [
    {
     "name": "stdout",
     "text": "Epoch 1/30\n1719/1719 [==============================] - 5s 3ms/step - loss: 0.7237 - accuracy: 0.7644 - val_loss: 0.5212 - val_accuracy: 0.8228\nEpoch 2/30\n1719/1719 [==============================] - 4s 3ms/step - loss: 0.4843 - accuracy: 0.8317 - val_loss: 0.4357 - val_accuracy: 0.8512\nEpoch 3/30\n1719/1719 [==============================] - 4s 3ms/step - loss: 0.4392 - accuracy: 0.8455 - val_loss: 0.5309 - val_accuracy: 0.7994\nEpoch 4/30\n1719/1719 [==============================] - 4s 3ms/step - loss: 0.4123 - accuracy: 0.8564 - val_loss: 0.3915 - val_accuracy: 0.8660\nEpoch 5/30\n1719/1719 [==============================] - 4s 3ms/step - loss: 0.3937 - accuracy: 0.8618 - val_loss: 0.3753 - val_accuracy: 0.8684\nEpoch 6/30\n1719/1719 [==============================] - 4s 3ms/step - loss: 0.3751 - accuracy: 0.8678 - val_loss: 0.3712 - val_accuracy: 0.8726\nEpoch 7/30\n1719/1719 [==============================] - 4s 3ms/step - loss: 0.3633 - accuracy: 0.8716 - val_loss: 0.3627 - val_accuracy: 0.8728\nEpoch 8/30\n1719/1719 [==============================] - 4s 3ms/step - loss: 0.3518 - accuracy: 0.8748 - val_loss: 0.3863 - val_accuracy: 0.8608\nEpoch 9/30\n1719/1719 [==============================] - 4s 3ms/step - loss: 0.3416 - accuracy: 0.8791 - val_loss: 0.3612 - val_accuracy: 0.8696\nEpoch 10/30\n1719/1719 [==============================] - 5s 3ms/step - loss: 0.3320 - accuracy: 0.8822 - val_loss: 0.3441 - val_accuracy: 0.8764\nEpoch 11/30\n1719/1719 [==============================] - 4s 3ms/step - loss: 0.3242 - accuracy: 0.8836 - val_loss: 0.3430 - val_accuracy: 0.8776\nEpoch 12/30\n1719/1719 [==============================] - 4s 3ms/step - loss: 0.3149 - accuracy: 0.8867 - val_loss: 0.3306 - val_accuracy: 0.8812\nEpoch 13/30\n1719/1719 [==============================] - 4s 3ms/step - loss: 0.3080 - accuracy: 0.8889 - val_loss: 0.3278 - val_accuracy: 0.8866\nEpoch 14/30\n1719/1719 [==============================] - 4s 3ms/step - loss: 0.3021 - accuracy: 0.8919 - val_loss: 0.3401 - val_accuracy: 0.8776\nEpoch 15/30\n1719/1719 [==============================] - 4s 3ms/step - loss: 0.2947 - accuracy: 0.8940 - val_loss: 0.3209 - val_accuracy: 0.8864\nEpoch 16/30\n1719/1719 [==============================] - 4s 3ms/step - loss: 0.2891 - accuracy: 0.8971 - val_loss: 0.3091 - val_accuracy: 0.8906\nEpoch 17/30\n1719/1719 [==============================] - 5s 3ms/step - loss: 0.2839 - accuracy: 0.8975 - val_loss: 0.3536 - val_accuracy: 0.8724\nEpoch 18/30\n1719/1719 [==============================] - 4s 3ms/step - loss: 0.2778 - accuracy: 0.9002 - val_loss: 0.3138 - val_accuracy: 0.8892\nEpoch 19/30\n1719/1719 [==============================] - 5s 3ms/step - loss: 0.2730 - accuracy: 0.9025 - val_loss: 0.3118 - val_accuracy: 0.8900\nEpoch 20/30\n1719/1719 [==============================] - 4s 3ms/step - loss: 0.2674 - accuracy: 0.9034 - val_loss: 0.3265 - val_accuracy: 0.8816\nEpoch 21/30\n1719/1719 [==============================] - 4s 3ms/step - loss: 0.2628 - accuracy: 0.9052 - val_loss: 0.3065 - val_accuracy: 0.8910\nEpoch 22/30\n1719/1719 [==============================] - 4s 3ms/step - loss: 0.2579 - accuracy: 0.9073 - val_loss: 0.2966 - val_accuracy: 0.8972\nEpoch 23/30\n1719/1719 [==============================] - 5s 3ms/step - loss: 0.2540 - accuracy: 0.9083 - val_loss: 0.2983 - val_accuracy: 0.8926\nEpoch 24/30\n1719/1719 [==============================] - 5s 3ms/step - loss: 0.2487 - accuracy: 0.9104 - val_loss: 0.3083 - val_accuracy: 0.8882\nEpoch 25/30\n1719/1719 [==============================] - 5s 3ms/step - loss: 0.2442 - accuracy: 0.9121 - val_loss: 0.2967 - val_accuracy: 0.8952\nEpoch 26/30\n1719/1719 [==============================] - 5s 3ms/step - loss: 0.2405 - accuracy: 0.9134 - val_loss: 0.3067 - val_accuracy: 0.8896\nEpoch 27/30\n1719/1719 [==============================] - 4s 3ms/step - loss: 0.2363 - accuracy: 0.9155 - val_loss: 0.3019 - val_accuracy: 0.8948\nEpoch 28/30\n1719/1719 [==============================] - 4s 3ms/step - loss: 0.2329 - accuracy: 0.9167 - val_loss: 0.2994 - val_accuracy: 0.8934\nEpoch 29/30\n1719/1719 [==============================] - 4s 3ms/step - loss: 0.2288 - accuracy: 0.9182 - val_loss: 0.3057 - val_accuracy: 0.8900\nEpoch 30/30\n1719/1719 [==============================] - 4s 3ms/step - loss: 0.2251 - accuracy: 0.9191 - val_loss: 0.2998 - val_accuracy: 0.8938\n",
     "output_type": "stream"
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00056-f3aad56f-c06d-428d-b354-40bb3de0ac41",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "6ebe03f4",
    "execution_start": 1659186635715,
    "execution_millis": 9,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 113.1875,
    "deepnote_output_heights": [
     21.1875
    ]
   },
   "source": "history.params",
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "execution_count": 38,
     "data": {
      "text/plain": "{'verbose': 1, 'epochs': 30, 'steps': 1719}"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00057-cd7ebb5d-ee1f-408c-9d3c-1131e7516170",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "b46972e2",
    "execution_start": 1659186635727,
    "execution_millis": 7,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 107.1875
   },
   "source": "print(history.epoch)",
   "execution_count": null,
   "outputs": [
    {
     "name": "stdout",
     "text": "[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29]\n",
     "output_type": "stream"
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00058-1b44348b-410b-4126-9f2d-35d146966c56",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "56e797ad",
    "execution_start": 1659186635734,
    "execution_millis": 4,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 113.1875,
    "deepnote_output_heights": [
     21.1875
    ]
   },
   "source": "history.history.keys()",
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "execution_count": 40,
     "data": {
      "text/plain": "dict_keys(['loss', 'accuracy', 'val_loss', 'val_accuracy'])"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00059-4fd2fc13-4119-4038-abf4-b5bbdd24ee8f",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "42f51e9c",
    "execution_start": 1659186635785,
    "execution_millis": 926,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 600.1875,
    "deepnote_output_heights": [
     null,
     354
    ]
   },
   "source": "import pandas as pd\n\npd.DataFrame(history.history).plot(figsize=(8, 5))\nplt.grid(True)\nplt.gca().set_ylim(0, 1)\nsave_fig(\"keras_learning_curves_plot\")\nplt.show()",
   "execution_count": null,
   "outputs": [
    {
     "name": "stdout",
     "text": "Saving figure keras_learning_curves_plot\n",
     "output_type": "stream"
    },
    {
     "data": {
      "text/plain": "<Figure size 576x360 with 1 Axes>",
      "image/png": "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\n"
     },
     "metadata": {
      "needs_background": "light",
      "image/png": {
       "width": 567,
       "height": 352
      }
     },
     "output_type": "display_data"
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00060-25e878f7-85c2-4dc8-a64c-b854023d0449",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "ef7ab4e3",
    "execution_start": 1659186636712,
    "execution_millis": 697,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 159.375,
    "deepnote_output_heights": [
     null,
     21.1875
    ]
   },
   "source": "model.evaluate(X_test, y_test)",
   "execution_count": null,
   "outputs": [
    {
     "name": "stdout",
     "text": "313/313 [==============================] - 1s 2ms/step - loss: 0.3335 - accuracy: 0.8837\n",
     "output_type": "stream"
    },
    {
     "output_type": "execute_result",
     "execution_count": 42,
     "data": {
      "text/plain": "[0.33345213532447815, 0.8837000131607056]"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00061-a2255b7b-43bb-429e-a258-be61e5f4d6e3",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "970fa916",
    "execution_start": 1659186637456,
    "execution_millis": 113,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 252.9375
   },
   "source": "X_new = X_test[:3]\ny_proba = model.predict(X_new)\ny_proba.round(2)",
   "execution_count": null,
   "outputs": [
    {
     "name": "stdout",
     "text": "1/1 [==============================] - 0s 78ms/step\n",
     "output_type": "stream"
    },
    {
     "output_type": "execute_result",
     "execution_count": 43,
     "data": {
      "text/plain": "array([[0.  , 0.  , 0.  , 0.  , 0.  , 0.01, 0.  , 0.03, 0.  , 0.96],\n       [0.  , 0.  , 0.98, 0.  , 0.02, 0.  , 0.  , 0.  , 0.  , 0.  ],\n       [0.  , 1.  , 0.  , 0.  , 0.  , 0.  , 0.  , 0.  , 0.  , 0.  ]],\n      dtype=float32)"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": "**Warning**: `model.predict_classes(X_new)` is deprecated. It is replaced with `np.argmax(model.predict(X_new), axis=-1)`.",
   "metadata": {
    "cell_id": "00062-469c5886-5499-4a73-9c05-9988f972d5d7",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 52.390625
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00063-88e47ac2-dc12-4d79-baac-e0bef8e05e1d",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "b2930ea6",
    "execution_start": 1659186637563,
    "execution_millis": 34,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 195.375
   },
   "source": "#y_pred = model.predict_classes(X_new) # deprecated\ny_pred = np.argmax(model.predict(X_new), axis=-1)\ny_pred",
   "execution_count": null,
   "outputs": [
    {
     "name": "stdout",
     "text": "1/1 [==============================] - 0s 14ms/step\n",
     "output_type": "stream"
    },
    {
     "output_type": "execute_result",
     "execution_count": 44,
     "data": {
      "text/plain": "array([9, 2, 1])"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00064-dc45334d-684c-466b-8960-e331a0eef894",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "6a1358ff",
    "execution_start": 1659186637595,
    "execution_millis": 3,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 113.1875,
    "deepnote_output_heights": [
     21.1875
    ]
   },
   "source": "np.array(class_names)[y_pred]",
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "execution_count": 45,
     "data": {
      "text/plain": "array(['Ankle boot', 'Pullover', 'Trouser'], dtype='<U11')"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00065-9cd17e84-11fa-4ab8-aecc-95786235d2f5",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "640f2365",
    "execution_start": 1659186637596,
    "execution_millis": 0,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 131.1875,
    "deepnote_output_heights": [
     21.1875
    ]
   },
   "source": "y_new = y_test[:3]\ny_new",
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "execution_count": 46,
     "data": {
      "text/plain": "array([9, 2, 1], dtype=uint8)"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00066-dc3c6b03-06e1-4a75-bfc2-ebbdda6a3802",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "2b608b2c",
    "execution_start": 1659186637644,
    "execution_millis": 846,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 432.1875,
    "deepnote_output_heights": [
     null,
     150
    ]
   },
   "source": "plt.figure(figsize=(7.2, 2.4))\nfor index, image in enumerate(X_new):\n    plt.subplot(1, 3, index + 1)\n    plt.imshow(image, cmap=\"binary\", interpolation=\"nearest\")\n    plt.axis('off')\n    plt.title(class_names[y_test[index]], fontsize=12)\nplt.subplots_adjust(wspace=0.2, hspace=0.5)\nsave_fig('fashion_mnist_images_plot', tight_layout=False)\nplt.show()",
   "execution_count": null,
   "outputs": [
    {
     "name": "stdout",
     "text": "Saving figure fashion_mnist_images_plot\n",
     "output_type": "stream"
    },
    {
     "data": {
      "text/plain": "<Figure size 518.4x172.8 with 3 Axes>",
      "image/png": "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\n"
     },
     "metadata": {
      "needs_background": "light",
      "image/png": {
       "width": 416,
       "height": 148
      }
     },
     "output_type": "display_data"
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": "# Regression MLP",
   "metadata": {
    "cell_id": "00067-0e35ddd3-7832-4c54-bfe7-0a2fae973641",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 82
   }
  },
  {
   "cell_type": "markdown",
   "source": "Let's load, split and scale the California housing dataset (the original one, not the modified one as in chapter 2):",
   "metadata": {
    "cell_id": "00068-ed7636cf-4edd-4d36-813b-5a37743128a8",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 52.390625
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00069-dfc18b72-e7c2-4e96-a673-787a426a0085",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "79813033",
    "execution_start": 1659186637927,
    "execution_millis": 0,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 292
   },
   "source": "from sklearn.datasets import fetch_california_housing\nfrom sklearn.model_selection import train_test_split\nfrom sklearn.preprocessing import StandardScaler\n\nhousing = fetch_california_housing()\n\nX_train_full, X_test, y_train_full, y_test = train_test_split(housing.data, housing.target, random_state=42)\nX_train, X_valid, y_train, y_valid = train_test_split(X_train_full, y_train_full, random_state=42)\n\nscaler = StandardScaler()\nX_train = scaler.fit_transform(X_train)\nX_valid = scaler.transform(X_valid)\nX_test = scaler.transform(X_test)",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00070-93199f77-dc9b-489e-9b65-9f467f72a105",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "8905d08a",
    "execution_start": 1659186637928,
    "execution_millis": 2,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 94
   },
   "source": "np.random.seed(42)\ntf.random.set_seed(42)",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00071-1668f9e1-c22b-46ca-b921-7ec158974ed9",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "6355ede6",
    "execution_start": 1659186637993,
    "execution_millis": 20973,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 840
   },
   "source": "model = keras.models.Sequential([\n    keras.layers.Dense(30, activation=\"relu\", input_shape=X_train.shape[1:]),\n    keras.layers.Dense(1)\n])\nmodel.compile(loss=\"mean_squared_error\", optimizer=keras.optimizers.SGD(learning_rate=1e-3))\nhistory = model.fit(X_train, y_train, epochs=20, validation_data=(X_valid, y_valid))\nmse_test = model.evaluate(X_test, y_test)\nX_new = X_test[:3]\ny_pred = model.predict(X_new)",
   "execution_count": null,
   "outputs": [
    {
     "name": "stdout",
     "text": "Epoch 1/20\n363/363 [==============================] - 1s 2ms/step - loss: 1.6419 - val_loss: 0.8560\nEpoch 2/20\n363/363 [==============================] - 1s 1ms/step - loss: 0.7047 - val_loss: 0.6531\nEpoch 3/20\n363/363 [==============================] - 1s 1ms/step - loss: 0.6345 - val_loss: 0.6099\nEpoch 4/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.5977 - val_loss: 0.5658\nEpoch 5/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.5706 - val_loss: 0.5355\nEpoch 6/20\n363/363 [==============================] - 1s 1ms/step - loss: 0.5472 - val_loss: 0.5173\nEpoch 7/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.5288 - val_loss: 0.5081\nEpoch 8/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.5130 - val_loss: 0.4799\nEpoch 9/20\n363/363 [==============================] - 1s 1ms/step - loss: 0.4992 - val_loss: 0.4690\nEpoch 10/20\n363/363 [==============================] - 1s 1ms/step - loss: 0.4875 - val_loss: 0.4656\nEpoch 11/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.4777 - val_loss: 0.4482\nEpoch 12/20\n363/363 [==============================] - 1s 1ms/step - loss: 0.4688 - val_loss: 0.4479\nEpoch 13/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.4615 - val_loss: 0.4296\nEpoch 14/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.4547 - val_loss: 0.4233\nEpoch 15/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.4488 - val_loss: 0.4176\nEpoch 16/20\n363/363 [==============================] - 1s 1ms/step - loss: 0.4435 - val_loss: 0.4123\nEpoch 17/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.4389 - val_loss: 0.4071\nEpoch 18/20\n363/363 [==============================] - 1s 1ms/step - loss: 0.4347 - val_loss: 0.4037\nEpoch 19/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.4306 - val_loss: 0.4000\nEpoch 20/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.4273 - val_loss: 0.3969\n162/162 [==============================] - 0s 1ms/step - loss: 0.4212\n1/1 [==============================] - 0s 37ms/step\n",
     "output_type": "stream"
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00072-e31da946-35c4-4a77-b772-a48d286ac162",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "47027dec",
    "execution_start": 1659186658974,
    "execution_millis": 179,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 404,
    "deepnote_output_heights": [
     258
    ]
   },
   "source": "plt.plot(pd.DataFrame(history.history))\nplt.grid(True)\nplt.gca().set_ylim(0, 1)\nplt.show()",
   "execution_count": null,
   "outputs": [
    {
     "data": {
      "text/plain": "<Figure size 432x288 with 1 Axes>",
      "image/png": "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\n"
     },
     "metadata": {
      "needs_background": "light",
      "image/png": {
       "width": 375,
       "height": 256
      }
     },
     "output_type": "display_data"
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00073-79d03169-e7f8-4926-a5ac-428976761169",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "ba549d12",
    "execution_start": 1659186659152,
    "execution_millis": 12,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 151.5625,
    "deepnote_output_heights": [
     59.5625
    ]
   },
   "source": "y_pred",
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "execution_count": 52,
     "data": {
      "text/plain": "array([[0.38856643],\n       [1.6792021 ],\n       [3.1022794 ]], dtype=float32)"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": "# Functional API",
   "metadata": {
    "cell_id": "00074-ca65f9bf-a4ea-43b0-b6b2-eb6283e4b36f",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 82
   }
  },
  {
   "cell_type": "markdown",
   "source": "Not all neural network models are simply sequential. Some may have complex topologies. Some may have multiple inputs and/or multiple outputs. For example, a Wide & Deep neural network (see [paper](https://ai.google/research/pubs/pub45413)) connects all or part of the inputs directly to the output layer.",
   "metadata": {
    "cell_id": "00075-8a88fee4-6f59-4385-9bf8-b7d09743af5b",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 74.78125
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00076-5b205a20-af75-4d8d-857b-e6a6b1029df3",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "8905d08a",
    "execution_start": 1659186659165,
    "execution_millis": 5,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 94
   },
   "source": "np.random.seed(42)\ntf.random.set_seed(42)",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00077-f55dfea6-6f80-4262-ade4-067a124c880f",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "78219b5b",
    "execution_start": 1659186659174,
    "execution_millis": 27,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 166
   },
   "source": "input_ = keras.layers.Input(shape=X_train.shape[1:])\nhidden1 = keras.layers.Dense(30, activation=\"relu\")(input_)\nhidden2 = keras.layers.Dense(30, activation=\"relu\")(hidden1)\nconcat = keras.layers.concatenate([input_, hidden2])\noutput = keras.layers.Dense(1)(concat)\nmodel = keras.models.Model(inputs=[input_], outputs=[output])",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00078-937f09ce-9a9b-4b9e-83bd-3d63cca7e6d6",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "4e6a3b95",
    "execution_start": 1659186659229,
    "execution_millis": 4,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 490.75
   },
   "source": "model.summary()",
   "execution_count": null,
   "outputs": [
    {
     "name": "stdout",
     "text": "Model: \"model\"\n__________________________________________________________________________________________________\n Layer (type)                   Output Shape         Param #     Connected to                     \n==================================================================================================\n input_1 (InputLayer)           [(None, 8)]          0           []                               \n                                                                                                  \n dense_5 (Dense)                (None, 30)           270         ['input_1[0][0]']                \n                                                                                                  \n dense_6 (Dense)                (None, 30)           930         ['dense_5[0][0]']                \n                                                                                                  \n concatenate (Concatenate)      (None, 38)           0           ['input_1[0][0]',                \n                                                                  'dense_6[0][0]']                \n                                                                                                  \n dense_7 (Dense)                (None, 1)            39          ['concatenate[0][0]']            \n                                                                                                  \n==================================================================================================\nTotal params: 1,239\nTrainable params: 1,239\nNon-trainable params: 0\n__________________________________________________________________________________________________\n",
     "output_type": "stream"
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00079-bef07a32-3140-4e43-98ad-c7c8728b4f6d",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "a3fb3d23",
    "execution_start": 1659186659278,
    "execution_millis": 21922,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 768
   },
   "source": "model.compile(loss=\"mean_squared_error\", optimizer=keras.optimizers.SGD(learning_rate=1e-3))\nhistory = model.fit(X_train, y_train, epochs=20,\n                    validation_data=(X_valid, y_valid))\nmse_test = model.evaluate(X_test, y_test)\ny_pred = model.predict(X_new)",
   "execution_count": null,
   "outputs": [
    {
     "name": "stdout",
     "text": "Epoch 1/20\n363/363 [==============================] - 1s 2ms/step - loss: 1.2611 - val_loss: 3.3940\nEpoch 2/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.6580 - val_loss: 0.9360\nEpoch 3/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.5878 - val_loss: 0.5649\nEpoch 4/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.5582 - val_loss: 0.5712\nEpoch 5/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.5347 - val_loss: 0.5045\nEpoch 6/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.5158 - val_loss: 0.4831\nEpoch 7/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.5002 - val_loss: 0.4639\nEpoch 8/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.4876 - val_loss: 0.4638\nEpoch 9/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.4760 - val_loss: 0.4421\nEpoch 10/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.4659 - val_loss: 0.4313\nEpoch 11/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.4577 - val_loss: 0.4345\nEpoch 12/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.4498 - val_loss: 0.4168\nEpoch 13/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.4428 - val_loss: 0.4230\nEpoch 14/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.4366 - val_loss: 0.4047\nEpoch 15/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.4307 - val_loss: 0.4078\nEpoch 16/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.4257 - val_loss: 0.3938\nEpoch 17/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.4210 - val_loss: 0.3952\nEpoch 18/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.4167 - val_loss: 0.3860\nEpoch 19/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.4121 - val_loss: 0.3827\nEpoch 20/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.4088 - val_loss: 0.4054\n162/162 [==============================] - 0s 1ms/step - loss: 0.4032\n1/1 [==============================] - 0s 43ms/step\n",
     "output_type": "stream"
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": "What if you want to send different subsets of input features through the wide or deep paths? We will send 5 features (features 0 to 4), and 6 through the deep path (features 2 to 7). Note that 3 features will go through both (features 2, 3 and 4).",
   "metadata": {
    "cell_id": "00080-55a6ce40-8314-4d09-b4d8-c0d1688e25f4",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 74.78125
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00081-86f240b4-be86-4017-932c-36e213cf61d2",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "8905d08a",
    "execution_start": 1659186680461,
    "execution_millis": 4,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 94
   },
   "source": "np.random.seed(42)\ntf.random.set_seed(42)",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00082-1a20d521-b65f-4a8e-a149-3bfeca42e741",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "c326d3f2",
    "execution_start": 1659186680469,
    "execution_millis": 43,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 184
   },
   "source": "input_A = keras.layers.Input(shape=[5], name=\"wide_input\")\ninput_B = keras.layers.Input(shape=[6], name=\"deep_input\")\nhidden1 = keras.layers.Dense(30, activation=\"relu\")(input_B)\nhidden2 = keras.layers.Dense(30, activation=\"relu\")(hidden1)\nconcat = keras.layers.concatenate([input_A, hidden2])\noutput = keras.layers.Dense(1, name=\"output\")(concat)\nmodel = keras.models.Model(inputs=[input_A, input_B], outputs=[output])",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00083-dbb7e023-5137-403c-9285-3d9a9793ca7c",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "b9623ae4",
    "execution_start": 1659186680522,
    "execution_millis": 21285,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 876
   },
   "source": "model.compile(loss=\"mse\", optimizer=keras.optimizers.SGD(learning_rate=1e-3))\n\nX_train_A, X_train_B = X_train[:, :5], X_train[:, 2:]\nX_valid_A, X_valid_B = X_valid[:, :5], X_valid[:, 2:]\nX_test_A, X_test_B = X_test[:, :5], X_test[:, 2:]\nX_new_A, X_new_B = X_test_A[:3], X_test_B[:3]\n\nhistory = model.fit((X_train_A, X_train_B), y_train, epochs=20,\n                    validation_data=((X_valid_A, X_valid_B), y_valid))\nmse_test = model.evaluate((X_test_A, X_test_B), y_test)\ny_pred = model.predict((X_new_A, X_new_B))",
   "execution_count": null,
   "outputs": [
    {
     "name": "stdout",
     "text": "Epoch 1/20\n363/363 [==============================] - 1s 2ms/step - loss: 1.8145 - val_loss: 0.8072\nEpoch 2/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.6771 - val_loss: 0.6658\nEpoch 3/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.5979 - val_loss: 0.5687\nEpoch 4/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.5584 - val_loss: 0.5296\nEpoch 5/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.5334 - val_loss: 0.4993\nEpoch 6/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.5120 - val_loss: 0.4811\nEpoch 7/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.4970 - val_loss: 0.4696\nEpoch 8/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.4843 - val_loss: 0.4496\nEpoch 9/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.4730 - val_loss: 0.4404\nEpoch 10/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.4644 - val_loss: 0.4315\nEpoch 11/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.4570 - val_loss: 0.4268\nEpoch 12/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.4510 - val_loss: 0.4166\nEpoch 13/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.4462 - val_loss: 0.4125\nEpoch 14/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.4421 - val_loss: 0.4074\nEpoch 15/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.4385 - val_loss: 0.4044\nEpoch 16/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.4356 - val_loss: 0.4007\nEpoch 17/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.4322 - val_loss: 0.4013\nEpoch 18/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.4305 - val_loss: 0.3987\nEpoch 19/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.4274 - val_loss: 0.3934\nEpoch 20/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.4261 - val_loss: 0.4204\n162/162 [==============================] - 0s 1ms/step - loss: 0.4219\n1/1 [==============================] - 0s 146ms/step\n",
     "output_type": "stream"
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": "Adding an auxiliary output for regularization:",
   "metadata": {
    "cell_id": "00084-d680552b-e8c8-46fe-82b6-f06f4dd0f7f1",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 52.390625
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00085-783f7df8-4a97-48d4-b96c-315d9182bc56",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "8905d08a",
    "execution_start": 1659186701813,
    "execution_millis": 4,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 94
   },
   "source": "np.random.seed(42)\ntf.random.set_seed(42)",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00086-ceff8c74-6ff3-4ce9-ad87-ec2c697216c6",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "1ba57094",
    "execution_start": 1659186701827,
    "execution_millis": 36,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 220
   },
   "source": "input_A = keras.layers.Input(shape=[5], name=\"wide_input\")\ninput_B = keras.layers.Input(shape=[6], name=\"deep_input\")\nhidden1 = keras.layers.Dense(30, activation=\"relu\")(input_B)\nhidden2 = keras.layers.Dense(30, activation=\"relu\")(hidden1)\nconcat = keras.layers.concatenate([input_A, hidden2])\noutput = keras.layers.Dense(1, name=\"main_output\")(concat)\naux_output = keras.layers.Dense(1, name=\"aux_output\")(hidden2)\nmodel = keras.models.Model(inputs=[input_A, input_B],\n                           outputs=[output, aux_output])",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00087-4cd92b94-354c-4acf-b67a-ee630fe4ce96",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "4fe7a41d",
    "execution_start": 1659186701870,
    "execution_millis": 3,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 81
   },
   "source": "model.compile(loss=[\"mse\", \"mse\"], loss_weights=[0.9, 0.1], optimizer=keras.optimizers.SGD(learning_rate=1e-3))",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00088-9abb44e5-474e-46c7-abd3-6725fcdea40d",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "daec0a71",
    "execution_start": 1659186701882,
    "execution_millis": 14293,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 714
   },
   "source": "history = model.fit([X_train_A, X_train_B], [y_train, y_train], epochs=20,\n                    validation_data=([X_valid_A, X_valid_B], [y_valid, y_valid]))",
   "execution_count": null,
   "outputs": [
    {
     "name": "stdout",
     "text": "Epoch 1/20\n363/363 [==============================] - 1s 2ms/step - loss: 2.1365 - main_output_loss: 1.9196 - aux_output_loss: 4.0890 - val_loss: 1.6233 - val_main_output_loss: 0.8468 - val_aux_output_loss: 8.6117\nEpoch 2/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.8905 - main_output_loss: 0.6969 - aux_output_loss: 2.6326 - val_loss: 1.5163 - val_main_output_loss: 0.6836 - val_aux_output_loss: 9.0109\nEpoch 3/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.7429 - main_output_loss: 0.6088 - aux_output_loss: 1.9499 - val_loss: 1.4639 - val_main_output_loss: 0.6229 - val_aux_output_loss: 9.0326\nEpoch 4/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.6771 - main_output_loss: 0.5691 - aux_output_loss: 1.6485 - val_loss: 1.3388 - val_main_output_loss: 0.5481 - val_aux_output_loss: 8.4552\nEpoch 5/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.6381 - main_output_loss: 0.5434 - aux_output_loss: 1.4911 - val_loss: 1.2177 - val_main_output_loss: 0.5194 - val_aux_output_loss: 7.5030\nEpoch 6/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.6079 - main_output_loss: 0.5207 - aux_output_loss: 1.3923 - val_loss: 1.0935 - val_main_output_loss: 0.5106 - val_aux_output_loss: 6.3396\nEpoch 7/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.5853 - main_output_loss: 0.5040 - aux_output_loss: 1.3175 - val_loss: 0.9918 - val_main_output_loss: 0.5115 - val_aux_output_loss: 5.3151\nEpoch 8/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.5666 - main_output_loss: 0.4898 - aux_output_loss: 1.2572 - val_loss: 0.8733 - val_main_output_loss: 0.4733 - val_aux_output_loss: 4.4740\nEpoch 9/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.5504 - main_output_loss: 0.4771 - aux_output_loss: 1.2101 - val_loss: 0.7832 - val_main_output_loss: 0.4555 - val_aux_output_loss: 3.7323\nEpoch 10/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.5373 - main_output_loss: 0.4671 - aux_output_loss: 1.1695 - val_loss: 0.7170 - val_main_output_loss: 0.4604 - val_aux_output_loss: 3.0262\nEpoch 11/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.5266 - main_output_loss: 0.4591 - aux_output_loss: 1.1344 - val_loss: 0.6510 - val_main_output_loss: 0.4293 - val_aux_output_loss: 2.6468\nEpoch 12/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.5173 - main_output_loss: 0.4520 - aux_output_loss: 1.1048 - val_loss: 0.6051 - val_main_output_loss: 0.4310 - val_aux_output_loss: 2.1722\nEpoch 13/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.5095 - main_output_loss: 0.4465 - aux_output_loss: 1.0765 - val_loss: 0.5644 - val_main_output_loss: 0.4161 - val_aux_output_loss: 1.8992\nEpoch 14/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.5027 - main_output_loss: 0.4417 - aux_output_loss: 1.0511 - val_loss: 0.5354 - val_main_output_loss: 0.4119 - val_aux_output_loss: 1.6466\nEpoch 15/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.4967 - main_output_loss: 0.4376 - aux_output_loss: 1.0280 - val_loss: 0.5124 - val_main_output_loss: 0.4047 - val_aux_output_loss: 1.4812\nEpoch 16/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.4916 - main_output_loss: 0.4343 - aux_output_loss: 1.0070 - val_loss: 0.4934 - val_main_output_loss: 0.4034 - val_aux_output_loss: 1.3035\nEpoch 17/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.4867 - main_output_loss: 0.4311 - aux_output_loss: 0.9872 - val_loss: 0.4801 - val_main_output_loss: 0.3984 - val_aux_output_loss: 1.2150\nEpoch 18/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.4829 - main_output_loss: 0.4289 - aux_output_loss: 0.9686 - val_loss: 0.4694 - val_main_output_loss: 0.3962 - val_aux_output_loss: 1.1279\nEpoch 19/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.4785 - main_output_loss: 0.4260 - aux_output_loss: 0.9510 - val_loss: 0.4580 - val_main_output_loss: 0.3936 - val_aux_output_loss: 1.0372\nEpoch 20/20\n363/363 [==============================] - 1s 2ms/step - loss: 0.4756 - main_output_loss: 0.4246 - aux_output_loss: 0.9344 - val_loss: 0.4655 - val_main_output_loss: 0.4048 - val_aux_output_loss: 1.0118\n",
     "output_type": "stream"
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00089-40e94cf1-7132-4880-a184-d04f1d56e24c",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "a53a99ac",
    "execution_start": 1659186716186,
    "execution_millis": 346,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 187.5625
   },
   "source": "total_loss, main_loss, aux_loss = model.evaluate(\n    [X_test_A, X_test_B], [y_test, y_test])\ny_pred_main, y_pred_aux = model.predict([X_new_A, X_new_B])",
   "execution_count": null,
   "outputs": [
    {
     "name": "stdout",
     "text": "162/162 [==============================] - 0s 1ms/step - loss: 0.4668 - main_output_loss: 0.4178 - aux_output_loss: 0.9082\nWARNING:tensorflow:5 out of the last 6 calls to <function Model.make_predict_function.<locals>.predict_function at 0x7f3b26d414c0> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has reduce_retracing=True option that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n1/1 [==============================] - 0s 57ms/step\n",
     "output_type": "stream"
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": "# The subclassing API",
   "metadata": {
    "cell_id": "00090-edc3c866-ec1b-4f20-a0ef-02282ec87986",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 82
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00091-53fa76cf-32fd-477d-8e1b-874cf6bacb28",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "939539cd",
    "execution_start": 1659186716546,
    "execution_millis": 42,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 382
   },
   "source": "class WideAndDeepModel(keras.models.Model):\n    def __init__(self, units=30, activation=\"relu\", **kwargs):\n        super().__init__(**kwargs)\n        self.hidden1 = keras.layers.Dense(units, activation=activation)\n        self.hidden2 = keras.layers.Dense(units, activation=activation)\n        self.main_output = keras.layers.Dense(1)\n        self.aux_output = keras.layers.Dense(1)\n        \n    def call(self, inputs):\n        input_A, input_B = inputs\n        hidden1 = self.hidden1(input_B)\n        hidden2 = self.hidden2(hidden1)\n        concat = keras.layers.concatenate([input_A, hidden2])\n        main_output = self.main_output(concat)\n        aux_output = self.aux_output(hidden2)\n        return main_output, aux_output\n\nmodel = WideAndDeepModel(30, activation=\"relu\")",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00092-fc61ae78-9f1b-4a42-8c63-7e32ff9509ff",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "fd607918",
    "execution_start": 1659186716588,
    "execution_millis": 11167,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 627.3125
   },
   "source": "model.compile(loss=\"mse\", loss_weights=[0.9, 0.1], optimizer=keras.optimizers.SGD(learning_rate=1e-3))\nhistory = model.fit((X_train_A, X_train_B), (y_train, y_train), epochs=10,\n                    validation_data=((X_valid_A, X_valid_B), (y_valid, y_valid)))\ntotal_loss, main_loss, aux_loss = model.evaluate((X_test_A, X_test_B), (y_test, y_test))\ny_pred_main, y_pred_aux = model.predict((X_new_A, X_new_B))",
   "execution_count": null,
   "outputs": [
    {
     "name": "stdout",
     "text": "Epoch 1/10\n363/363 [==============================] - 1s 2ms/step - loss: 2.3298 - output_1_loss: 2.2186 - output_2_loss: 3.3304 - val_loss: 2.1435 - val_output_1_loss: 1.1581 - val_output_2_loss: 11.0117\nEpoch 2/10\n363/363 [==============================] - 1s 2ms/step - loss: 0.9714 - output_1_loss: 0.8543 - output_2_loss: 2.0252 - val_loss: 1.7567 - val_output_1_loss: 0.8205 - val_output_2_loss: 10.1825\nEpoch 3/10\n363/363 [==============================] - 1s 2ms/step - loss: 0.8268 - output_1_loss: 0.7289 - output_2_loss: 1.7082 - val_loss: 1.5664 - val_output_1_loss: 0.7913 - val_output_2_loss: 8.5419\nEpoch 4/10\n363/363 [==============================] - 1s 2ms/step - loss: 0.7636 - output_1_loss: 0.6764 - output_2_loss: 1.5477 - val_loss: 1.3088 - val_output_1_loss: 0.6549 - val_output_2_loss: 7.1933\nEpoch 5/10\n363/363 [==============================] - 1s 2ms/step - loss: 0.7211 - output_1_loss: 0.6402 - output_2_loss: 1.4489 - val_loss: 1.1357 - val_output_1_loss: 0.5964 - val_output_2_loss: 5.9898\nEpoch 6/10\n363/363 [==============================] - 1s 2ms/step - loss: 0.6895 - output_1_loss: 0.6124 - output_2_loss: 1.3833 - val_loss: 1.0036 - val_output_1_loss: 0.5937 - val_output_2_loss: 4.6933\nEpoch 7/10\n363/363 [==============================] - 1s 2ms/step - loss: 0.6632 - output_1_loss: 0.5894 - output_2_loss: 1.3274 - val_loss: 0.8904 - val_output_1_loss: 0.5591 - val_output_2_loss: 3.8714\nEpoch 8/10\n363/363 [==============================] - 1s 2ms/step - loss: 0.6410 - output_1_loss: 0.5701 - output_2_loss: 1.2796 - val_loss: 0.8009 - val_output_1_loss: 0.5243 - val_output_2_loss: 3.2903\nEpoch 9/10\n363/363 [==============================] - 1s 2ms/step - loss: 0.6204 - output_1_loss: 0.5514 - output_2_loss: 1.2416 - val_loss: 0.7357 - val_output_1_loss: 0.5144 - val_output_2_loss: 2.7275\nEpoch 10/10\n363/363 [==============================] - 1s 2ms/step - loss: 0.6024 - output_1_loss: 0.5355 - output_2_loss: 1.2043 - val_loss: 0.6849 - val_output_1_loss: 0.5014 - val_output_2_loss: 2.3370\n162/162 [==============================] - 0s 1ms/step - loss: 0.5841 - output_1_loss: 0.5188 - output_2_loss: 1.1722\nWARNING:tensorflow:6 out of the last 7 calls to <function Model.make_predict_function.<locals>.predict_function at 0x7f3b26bb3160> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has reduce_retracing=True option that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n1/1 [==============================] - 0s 50ms/step\n",
     "output_type": "stream"
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": "# Saving and Restoring",
   "metadata": {
    "cell_id": "00093-51b76e31-46c8-4150-a5ad-7f6f915fe978",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 82
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00094-7a0d1da8-ebaa-40a5-aede-a6896d952f3c",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "8905d08a",
    "execution_start": 1659186727766,
    "execution_millis": 0,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 94
   },
   "source": "np.random.seed(42)\ntf.random.set_seed(42)",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00095-ac8ca286-0d9f-4634-8e18-459b0c23fde9",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "61946d12",
    "execution_start": 1659186727771,
    "execution_millis": 29,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 148
   },
   "source": "model = keras.models.Sequential([\n    keras.layers.Dense(30, activation=\"relu\", input_shape=[8]),\n    keras.layers.Dense(30, activation=\"relu\"),\n    keras.layers.Dense(1)\n])    ",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00096-c0aa2cd6-5ecd-44ef-a410-ddc38be79ade",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "c083ee39",
    "execution_start": 1659186727810,
    "execution_millis": 10833,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 546.9375
   },
   "source": "model.compile(loss=\"mse\", optimizer=keras.optimizers.SGD(learning_rate=1e-3))\nhistory = model.fit(X_train, y_train, epochs=10, validation_data=(X_valid, y_valid))\nmse_test = model.evaluate(X_test, y_test)",
   "execution_count": null,
   "outputs": [
    {
     "name": "stdout",
     "text": "Epoch 1/10\n363/363 [==============================] - 1s 2ms/step - loss: 1.8866 - val_loss: 0.7126\nEpoch 2/10\n363/363 [==============================] - 1s 2ms/step - loss: 0.6577 - val_loss: 0.6880\nEpoch 3/10\n363/363 [==============================] - 1s 1ms/step - loss: 0.5934 - val_loss: 0.5803\nEpoch 4/10\n363/363 [==============================] - 1s 2ms/step - loss: 0.5557 - val_loss: 0.5166\nEpoch 5/10\n363/363 [==============================] - 1s 2ms/step - loss: 0.5272 - val_loss: 0.4895\nEpoch 6/10\n363/363 [==============================] - 1s 2ms/step - loss: 0.5033 - val_loss: 0.4951\nEpoch 7/10\n363/363 [==============================] - 1s 2ms/step - loss: 0.4854 - val_loss: 0.4861\nEpoch 8/10\n363/363 [==============================] - 1s 2ms/step - loss: 0.4709 - val_loss: 0.4554\nEpoch 9/10\n363/363 [==============================] - 1s 2ms/step - loss: 0.4578 - val_loss: 0.4413\nEpoch 10/10\n363/363 [==============================] - 1s 2ms/step - loss: 0.4474 - val_loss: 0.4379\n162/162 [==============================] - 0s 1ms/step - loss: 0.4382\n",
     "output_type": "stream"
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00097-b09d8852-9c4b-4e35-bc48-8d58ce26b8e8",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "cba57df",
    "execution_start": 1659186738556,
    "execution_millis": 100,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 76
   },
   "source": "model.save(\"my_keras_model.h5\")",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00098-ffc0cf25-c3c2-4ccc-93c3-28197f1ba510",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "b7ed004d",
    "execution_start": 1659186738656,
    "execution_millis": 1,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 76
   },
   "source": "model = keras.models.load_model(\"my_keras_model.h5\")",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00099-31005450-daf6-4900-a237-052843a93e73",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "aa544900",
    "execution_start": 1659186738657,
    "execution_millis": 75,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 197.75,
    "deepnote_output_heights": [
     null,
     59.5625
    ]
   },
   "source": "model.predict(X_new)",
   "execution_count": null,
   "outputs": [
    {
     "name": "stdout",
     "text": "1/1 [==============================] - 0s 43ms/step\n",
     "output_type": "stream"
    },
    {
     "output_type": "execute_result",
     "execution_count": 72,
     "data": {
      "text/plain": "array([[0.54002357],\n       [1.650597  ],\n       [3.0098243 ]], dtype=float32)"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00100-42cf3025-0f81-4cac-85df-3828fb291b94",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "3f7f5e19",
    "execution_start": 1659186738730,
    "execution_millis": 1,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 76
   },
   "source": "model.save_weights(\"my_keras_weights.ckpt\")",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00101-eebabd8c-91f7-40ee-9195-d1f6a722116b",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "ba2f3eca",
    "execution_start": 1659186738731,
    "execution_millis": 2,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 113.1875,
    "deepnote_output_heights": [
     21.1875
    ]
   },
   "source": "model.load_weights(\"my_keras_weights.ckpt\")",
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "execution_count": 74,
     "data": {
      "text/plain": "<tensorflow.python.training.tracking.util.CheckpointLoadStatus at 0x7f3b26c9b610>"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": "# Using Callbacks during Training",
   "metadata": {
    "cell_id": "00102-77e74b05-ef4b-4572-8bfd-84facf10c442",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 82
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00103-fcd30b50-b607-48e1-9056-f3db4e57faa7",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "b86c7ea",
    "execution_start": 1659186738777,
    "execution_millis": 0,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 112
   },
   "source": "keras.backend.clear_session()\nnp.random.seed(42)\ntf.random.set_seed(42)",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00104-8ad0fdb7-03f5-4490-abce-5320e6dc3569",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "61946d12",
    "execution_start": 1659186738777,
    "execution_millis": 1,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 148
   },
   "source": "model = keras.models.Sequential([\n    keras.layers.Dense(30, activation=\"relu\", input_shape=[8]),\n    keras.layers.Dense(30, activation=\"relu\"),\n    keras.layers.Dense(1)\n])    ",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00105-163e47b3-fda5-4b97-bbf5-84dcb80871ab",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "b0c0ebcc",
    "execution_start": 1659186738782,
    "execution_millis": 10958,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 618.9375
   },
   "source": "model.compile(loss=\"mse\", optimizer=keras.optimizers.SGD(learning_rate=1e-3))\ncheckpoint_cb = keras.callbacks.ModelCheckpoint(\"my_keras_model.h5\", save_best_only=True)\nhistory = model.fit(X_train, y_train, epochs=10,\n                    validation_data=(X_valid, y_valid),\n                    callbacks=[checkpoint_cb])\nmodel = keras.models.load_model(\"my_keras_model.h5\") # rollback to best model\nmse_test = model.evaluate(X_test, y_test)",
   "execution_count": null,
   "outputs": [
    {
     "name": "stdout",
     "text": "Epoch 1/10\n363/363 [==============================] - 1s 2ms/step - loss: 1.8866 - val_loss: 0.7126\nEpoch 2/10\n363/363 [==============================] - 1s 2ms/step - loss: 0.6577 - val_loss: 0.6880\nEpoch 3/10\n363/363 [==============================] - 1s 2ms/step - loss: 0.5934 - val_loss: 0.5803\nEpoch 4/10\n363/363 [==============================] - 1s 2ms/step - loss: 0.5557 - val_loss: 0.5166\nEpoch 5/10\n363/363 [==============================] - 1s 2ms/step - loss: 0.5272 - val_loss: 0.4895\nEpoch 6/10\n363/363 [==============================] - 1s 2ms/step - loss: 0.5033 - val_loss: 0.4951\nEpoch 7/10\n363/363 [==============================] - 1s 2ms/step - loss: 0.4854 - val_loss: 0.4861\nEpoch 8/10\n363/363 [==============================] - 1s 2ms/step - loss: 0.4709 - val_loss: 0.4554\nEpoch 9/10\n363/363 [==============================] - 1s 2ms/step - loss: 0.4578 - val_loss: 0.4413\nEpoch 10/10\n363/363 [==============================] - 1s 2ms/step - loss: 0.4474 - val_loss: 0.4379\n162/162 [==============================] - 0s 1ms/step - loss: 0.4382\n",
     "output_type": "stream"
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00106-14cab56c-de85-45e2-abae-c2c3307f6712",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "86c885ea",
    "execution_start": 1659186749737,
    "execution_millis": 57081,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 804
   },
   "source": "model.compile(loss=\"mse\", optimizer=keras.optimizers.SGD(learning_rate=1e-3))\nearly_stopping_cb = keras.callbacks.EarlyStopping(patience=10,\n                                                  restore_best_weights=True)\nhistory = model.fit(X_train, y_train, epochs=100,\n                    validation_data=(X_valid, y_valid),\n                    callbacks=[checkpoint_cb, early_stopping_cb])\nmse_test = model.evaluate(X_test, y_test)",
   "execution_count": null,
   "outputs": [
    {
     "name": "stdout",
     "text": "Epoch 1/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.4393 - val_loss: 0.4110\nEpoch 2/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.4315 - val_loss: 0.4266\nEpoch 3/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.4259 - val_loss: 0.3996\nEpoch 4/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.4201 - val_loss: 0.3939\nEpoch 5/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.4154 - val_loss: 0.3889\nEpoch 6/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.4111 - val_loss: 0.3866\nEpoch 7/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.4074 - val_loss: 0.3860\nEpoch 8/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.4040 - val_loss: 0.3793\nEpoch 9/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.4008 - val_loss: 0.3746\nEpoch 10/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3976 - val_loss: 0.3723\nEpoch 11/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3950 - val_loss: 0.3697\nEpoch 12/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3923 - val_loss: 0.3669\nEpoch 13/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3897 - val_loss: 0.3661\nEpoch 14/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3874 - val_loss: 0.3631\nEpoch 15/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3851 - val_loss: 0.3660\nEpoch 16/100\n363/363 [==============================] - 1s 1ms/step - loss: 0.3829 - val_loss: 0.3625\nEpoch 17/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3810 - val_loss: 0.3592\nEpoch 18/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3788 - val_loss: 0.3563\nEpoch 19/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3766 - val_loss: 0.3535\nEpoch 20/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3750 - val_loss: 0.3709\nEpoch 21/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3732 - val_loss: 0.3512\nEpoch 22/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3715 - val_loss: 0.3699\nEpoch 23/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3700 - val_loss: 0.3476\nEpoch 24/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3685 - val_loss: 0.3561\nEpoch 25/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3671 - val_loss: 0.3527\nEpoch 26/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3658 - val_loss: 0.3700\nEpoch 27/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3647 - val_loss: 0.3432\nEpoch 28/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3635 - val_loss: 0.3592\nEpoch 29/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3625 - val_loss: 0.3521\nEpoch 30/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3613 - val_loss: 0.3626\nEpoch 31/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3601 - val_loss: 0.3431\nEpoch 32/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3589 - val_loss: 0.3766\nEpoch 33/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3584 - val_loss: 0.3374\nEpoch 34/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3572 - val_loss: 0.3407\nEpoch 35/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3563 - val_loss: 0.3614\nEpoch 36/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3555 - val_loss: 0.3348\nEpoch 37/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3546 - val_loss: 0.3573\nEpoch 38/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3538 - val_loss: 0.3367\nEpoch 39/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3530 - val_loss: 0.3425\nEpoch 40/100\n363/363 [==============================] - 1s 1ms/step - loss: 0.3523 - val_loss: 0.3368\nEpoch 41/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3515 - val_loss: 0.3514\nEpoch 42/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3511 - val_loss: 0.3426\nEpoch 43/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3500 - val_loss: 0.3677\nEpoch 44/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3496 - val_loss: 0.3563\nEpoch 45/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3490 - val_loss: 0.3336\nEpoch 46/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3481 - val_loss: 0.3456\nEpoch 47/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3478 - val_loss: 0.3433\nEpoch 48/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3471 - val_loss: 0.3658\nEpoch 49/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3466 - val_loss: 0.3286\nEpoch 50/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3460 - val_loss: 0.3268\nEpoch 51/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3454 - val_loss: 0.3438\nEpoch 52/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3449 - val_loss: 0.3263\nEpoch 53/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3444 - val_loss: 0.3910\nEpoch 54/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3439 - val_loss: 0.3275\nEpoch 55/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3435 - val_loss: 0.3560\nEpoch 56/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3430 - val_loss: 0.3237\nEpoch 57/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3423 - val_loss: 0.3242\nEpoch 58/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3419 - val_loss: 0.3764\nEpoch 59/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3417 - val_loss: 0.3289\nEpoch 60/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3410 - val_loss: 0.3502\nEpoch 61/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3404 - val_loss: 0.3456\nEpoch 62/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3402 - val_loss: 0.3445\nEpoch 63/100\n363/363 [==============================] - 1s 1ms/step - loss: 0.3392 - val_loss: 0.3290\nEpoch 64/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3393 - val_loss: 0.3217\nEpoch 65/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3387 - val_loss: 0.3351\nEpoch 66/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3383 - val_loss: 0.3232\nEpoch 67/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3376 - val_loss: 0.3568\nEpoch 68/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3374 - val_loss: 0.3256\nEpoch 69/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3370 - val_loss: 0.3349\nEpoch 70/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3365 - val_loss: 0.3560\nEpoch 71/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3361 - val_loss: 0.3582\nEpoch 72/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3357 - val_loss: 0.3286\nEpoch 73/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3351 - val_loss: 0.3203\nEpoch 74/100\n363/363 [==============================] - 1s 1ms/step - loss: 0.3350 - val_loss: 0.3839\nEpoch 75/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3347 - val_loss: 0.3234\nEpoch 76/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3342 - val_loss: 0.3475\nEpoch 77/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3338 - val_loss: 0.3409\nEpoch 78/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3335 - val_loss: 0.3461\nEpoch 79/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3332 - val_loss: 0.3348\nEpoch 80/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3329 - val_loss: 0.3352\nEpoch 81/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3324 - val_loss: 0.3277\nEpoch 82/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3320 - val_loss: 0.3167\nEpoch 83/100\n363/363 [==============================] - 1s 1ms/step - loss: 0.3317 - val_loss: 0.3280\nEpoch 84/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3312 - val_loss: 0.3636\nEpoch 85/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3310 - val_loss: 0.3175\nEpoch 86/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3308 - val_loss: 0.3156\nEpoch 87/100\n363/363 [==============================] - 1s 1ms/step - loss: 0.3305 - val_loss: 0.3530\nEpoch 88/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3299 - val_loss: 0.3256\nEpoch 89/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3294 - val_loss: 0.3625\nEpoch 90/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3296 - val_loss: 0.3378\nEpoch 91/100\n363/363 [==============================] - 1s 1ms/step - loss: 0.3291 - val_loss: 0.3211\nEpoch 92/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3287 - val_loss: 0.3455\nEpoch 93/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3285 - val_loss: 0.3158\nEpoch 94/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3281 - val_loss: 0.3408\nEpoch 95/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3276 - val_loss: 0.3380\nEpoch 96/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3273 - val_loss: 0.3212\n162/162 [==============================] - 0s 1ms/step - loss: 0.3310\n",
     "output_type": "stream"
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00107-a5f00abc-bcd0-497a-aac8-8e6f92969840",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "d16417ff",
    "execution_start": 1659186806866,
    "execution_millis": 0,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 112
   },
   "source": "class PrintValTrainRatioCallback(keras.callbacks.Callback):\n    def on_epoch_end(self, epoch, logs):\n        print(\"\\nval/train: {:.2f}\".format(logs[\"val_loss\"] / logs[\"loss\"]))",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00108-934f553f-9a78-468c-8df4-7884f89c5702",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "6bbf3d45",
    "execution_start": 1659186806867,
    "execution_millis": 626,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 201.5625
   },
   "source": "val_train_ratio_cb = PrintValTrainRatioCallback()\nhistory = model.fit(X_train, y_train, epochs=1,\n                    validation_data=(X_valid, y_valid),\n                    callbacks=[val_train_ratio_cb])",
   "execution_count": null,
   "outputs": [
    {
     "name": "stdout",
     "text": "352/363 [============================>.] - ETA: 0s - loss: 0.3293\nval/train: 1.08\n363/363 [==============================] - 1s 2ms/step - loss: 0.3302 - val_loss: 0.3560\n",
     "output_type": "stream"
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": "# TensorBoard",
   "metadata": {
    "cell_id": "00109-ce92266e-7719-4e6d-80fb-cd5f39d32add",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 82
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00110-2211dd1c-2136-4602-926b-c1288c75fe53",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "32c569b2",
    "execution_start": 1659186807489,
    "execution_millis": 1,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 76
   },
   "source": "root_logdir = os.path.join(os.curdir, \"my_logs\")",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00111-1f31c930-e076-475e-bbf7-ad4bcd3eedae",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "1f34451f",
    "execution_start": 1659186807490,
    "execution_millis": 3,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 221.1875,
    "deepnote_output_heights": [
     21.1875
    ]
   },
   "source": "def get_run_logdir():\n    import time\n    run_id = time.strftime(\"run_%Y_%m_%d-%H_%M_%S\")\n    return os.path.join(root_logdir, run_id)\n\nrun_logdir = get_run_logdir()\nrun_logdir",
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "execution_count": 82,
     "data": {
      "text/plain": "'./my_logs/run_2022_07_30-13_13_27'"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00112-95c0f883-b7e1-4ccc-8e1d-e1ff40172486",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "b86c7ea",
    "execution_start": 1659186807492,
    "execution_millis": 4,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 112
   },
   "source": "keras.backend.clear_session()\nnp.random.seed(42)\ntf.random.set_seed(42)",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00113-64b98ec5-11ec-4eaf-9b23-b985fb7017f8",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "7cf21963",
    "execution_start": 1659186807541,
    "execution_millis": 1,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 166
   },
   "source": "model = keras.models.Sequential([\n    keras.layers.Dense(30, activation=\"relu\", input_shape=[8]),\n    keras.layers.Dense(30, activation=\"relu\"),\n    keras.layers.Dense(1)\n])    \nmodel.compile(loss=\"mse\", optimizer=keras.optimizers.SGD(learning_rate=1e-3))",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00114-039b8746-2ac8-4cc7-946e-929da37f1596",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "6dfcce10",
    "execution_start": 1659186807542,
    "execution_millis": 19118,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 750
   },
   "source": "tensorboard_cb = keras.callbacks.TensorBoard(run_logdir)\nhistory = model.fit(X_train, y_train, epochs=30,\n                    validation_data=(X_valid, y_valid),\n                    callbacks=[checkpoint_cb, tensorboard_cb])",
   "execution_count": null,
   "outputs": [
    {
     "name": "stdout",
     "text": "Epoch 1/30\n363/363 [==============================] - 1s 2ms/step - loss: 1.8866 - val_loss: 0.7126\nEpoch 2/30\n363/363 [==============================] - 1s 1ms/step - loss: 0.6577 - val_loss: 0.6880\nEpoch 3/30\n363/363 [==============================] - 1s 2ms/step - loss: 0.5934 - val_loss: 0.5803\nEpoch 4/30\n363/363 [==============================] - 1s 2ms/step - loss: 0.5557 - val_loss: 0.5166\nEpoch 5/30\n363/363 [==============================] - 1s 2ms/step - loss: 0.5272 - val_loss: 0.4895\nEpoch 6/30\n363/363 [==============================] - 1s 2ms/step - loss: 0.5033 - val_loss: 0.4951\nEpoch 7/30\n363/363 [==============================] - 1s 2ms/step - loss: 0.4854 - val_loss: 0.4861\nEpoch 8/30\n363/363 [==============================] - 1s 2ms/step - loss: 0.4709 - val_loss: 0.4554\nEpoch 9/30\n363/363 [==============================] - 1s 2ms/step - loss: 0.4578 - val_loss: 0.4413\nEpoch 10/30\n363/363 [==============================] - 1s 2ms/step - loss: 0.4474 - val_loss: 0.4379\nEpoch 11/30\n363/363 [==============================] - 1s 2ms/step - loss: 0.4393 - val_loss: 0.4396\nEpoch 12/30\n363/363 [==============================] - 1s 2ms/step - loss: 0.4318 - val_loss: 0.4507\nEpoch 13/30\n363/363 [==============================] - 1s 2ms/step - loss: 0.4261 - val_loss: 0.3997\nEpoch 14/30\n363/363 [==============================] - 1s 2ms/step - loss: 0.4202 - val_loss: 0.3956\nEpoch 15/30\n363/363 [==============================] - 1s 2ms/step - loss: 0.4155 - val_loss: 0.3916\nEpoch 16/30\n363/363 [==============================] - 1s 2ms/step - loss: 0.4112 - val_loss: 0.3937\nEpoch 17/30\n363/363 [==============================] - 1s 2ms/step - loss: 0.4077 - val_loss: 0.3809\nEpoch 18/30\n363/363 [==============================] - 1s 2ms/step - loss: 0.4040 - val_loss: 0.3793\nEpoch 19/30\n363/363 [==============================] - 1s 1ms/step - loss: 0.4004 - val_loss: 0.3850\nEpoch 20/30\n363/363 [==============================] - 1s 2ms/step - loss: 0.3980 - val_loss: 0.3809\nEpoch 21/30\n363/363 [==============================] - 1s 2ms/step - loss: 0.3949 - val_loss: 0.3701\nEpoch 22/30\n363/363 [==============================] - 1s 2ms/step - loss: 0.3924 - val_loss: 0.3781\nEpoch 23/30\n363/363 [==============================] - 1s 2ms/step - loss: 0.3898 - val_loss: 0.3650\nEpoch 24/30\n363/363 [==============================] - 1s 2ms/step - loss: 0.3874 - val_loss: 0.3655\nEpoch 25/30\n363/363 [==============================] - 1s 2ms/step - loss: 0.3851 - val_loss: 0.3611\nEpoch 26/30\n363/363 [==============================] - 1s 2ms/step - loss: 0.3829 - val_loss: 0.3626\nEpoch 27/30\n363/363 [==============================] - 1s 2ms/step - loss: 0.3809 - val_loss: 0.3564\nEpoch 28/30\n363/363 [==============================] - 1s 2ms/step - loss: 0.3788 - val_loss: 0.3579\nEpoch 29/30\n363/363 [==============================] - 1s 2ms/step - loss: 0.3769 - val_loss: 0.3561\nEpoch 30/30\n363/363 [==============================] - 1s 2ms/step - loss: 0.3750 - val_loss: 0.3548\n",
     "output_type": "stream"
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": "To start the TensorBoard server, one option is to open a terminal, if needed activate the virtualenv where you installed TensorBoard, go to this notebook's directory, then type:\n\n```bash\n$ tensorboard --logdir=./my_logs --port=6006\n```\n\nYou can then open your web browser to [localhost:6006](http://localhost:6006) and use TensorBoard. Once you are done, press Ctrl-C in the terminal window, this will shutdown the TensorBoard server.\n\nAlternatively, you can load TensorBoard's Jupyter extension and run it like this:",
   "metadata": {
    "cell_id": "00115-7d2e8de1-6f78-46dd-adea-67bb9b420447",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 212.34375
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00116-5422c717-8887-4484-a105-368682075434",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "d8e7869",
    "execution_start": 1659186826140,
    "execution_millis": 5539,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 131.1875
   },
   "source": "%load_ext tensorboard\n%tensorboard --logdir=./my_logs --port=6006",
   "execution_count": null,
   "outputs": [
    {
     "data": {
      "text/plain": "Launching TensorBoard..."
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00117-d930eb0f-323c-4025-9cdc-617fe6f6a9eb",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "7150b5ca",
    "execution_start": 1659186831684,
    "execution_millis": 718,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 131.1875
   },
   "source": "run_logdir2 = get_run_logdir()\nrun_logdir2",
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "execution_count": 87,
     "data": {
      "text/plain": "'./my_logs/run_2022_07_30-13_13_51'"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00118-dfb7036f-c92c-4383-938a-dc604816f220",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "b86c7ea",
    "execution_start": 1659186831692,
    "execution_millis": 16,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 112
   },
   "source": "keras.backend.clear_session()\nnp.random.seed(42)\ntf.random.set_seed(42)",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00119-af160032-73f3-423d-ba06-8055befd1385",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "87190c6",
    "execution_start": 1659186831716,
    "execution_millis": 29,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 166
   },
   "source": "model = keras.models.Sequential([\n    keras.layers.Dense(30, activation=\"relu\", input_shape=[8]),\n    keras.layers.Dense(30, activation=\"relu\"),\n    keras.layers.Dense(1)\n])    \nmodel.compile(loss=\"mse\", optimizer=keras.optimizers.SGD(learning_rate=0.05))",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00120-ec4076f9-3891-4174-8a28-3465102e441a",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "36d2bf44",
    "execution_start": 1659186831750,
    "execution_millis": 18023,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 750
   },
   "source": "tensorboard_cb = keras.callbacks.TensorBoard(run_logdir2)\nhistory = model.fit(X_train, y_train, epochs=30,\n                    validation_data=(X_valid, y_valid),\n                    callbacks=[checkpoint_cb, tensorboard_cb])",
   "execution_count": null,
   "outputs": [
    {
     "name": "stdout",
     "text": "Epoch 1/30\n363/363 [==============================] - 1s 2ms/step - loss: 0.5530 - val_loss: 302.8523\nEpoch 2/30\n363/363 [==============================] - 1s 2ms/step - loss: nan - val_loss: nan\nEpoch 3/30\n363/363 [==============================] - 1s 2ms/step - loss: nan - val_loss: nan\nEpoch 4/30\n363/363 [==============================] - 1s 2ms/step - loss: nan - val_loss: nan\nEpoch 5/30\n363/363 [==============================] - 1s 2ms/step - loss: nan - val_loss: nan\nEpoch 6/30\n363/363 [==============================] - 1s 2ms/step - loss: nan - val_loss: nan\nEpoch 7/30\n363/363 [==============================] - 1s 2ms/step - loss: nan - val_loss: nan\nEpoch 8/30\n363/363 [==============================] - 1s 2ms/step - loss: nan - val_loss: nan\nEpoch 9/30\n363/363 [==============================] - 1s 2ms/step - loss: nan - val_loss: nan\nEpoch 10/30\n363/363 [==============================] - 1s 2ms/step - loss: nan - val_loss: nan\nEpoch 11/30\n363/363 [==============================] - 1s 1ms/step - loss: nan - val_loss: nan\nEpoch 12/30\n363/363 [==============================] - 1s 2ms/step - loss: nan - val_loss: nan\nEpoch 13/30\n363/363 [==============================] - 1s 1ms/step - loss: nan - val_loss: nan\nEpoch 14/30\n363/363 [==============================] - 1s 2ms/step - loss: nan - val_loss: nan\nEpoch 15/30\n363/363 [==============================] - 1s 2ms/step - loss: nan - val_loss: nan\nEpoch 16/30\n363/363 [==============================] - 1s 2ms/step - loss: nan - val_loss: nan\nEpoch 17/30\n363/363 [==============================] - 1s 2ms/step - loss: nan - val_loss: nan\nEpoch 18/30\n363/363 [==============================] - 1s 2ms/step - loss: nan - val_loss: nan\nEpoch 19/30\n363/363 [==============================] - 1s 2ms/step - loss: nan - val_loss: nan\nEpoch 20/30\n363/363 [==============================] - 1s 1ms/step - loss: nan - val_loss: nan\nEpoch 21/30\n363/363 [==============================] - 1s 2ms/step - loss: nan - val_loss: nan\nEpoch 22/30\n363/363 [==============================] - 1s 2ms/step - loss: nan - val_loss: nan\nEpoch 23/30\n363/363 [==============================] - 1s 2ms/step - loss: nan - val_loss: nan\nEpoch 24/30\n363/363 [==============================] - 1s 2ms/step - loss: nan - val_loss: nan\nEpoch 25/30\n363/363 [==============================] - 1s 1ms/step - loss: nan - val_loss: nan\nEpoch 26/30\n363/363 [==============================] - 1s 2ms/step - loss: nan - val_loss: nan\nEpoch 27/30\n363/363 [==============================] - 1s 2ms/step - loss: nan - val_loss: nan\nEpoch 28/30\n363/363 [==============================] - 1s 2ms/step - loss: nan - val_loss: nan\nEpoch 29/30\n363/363 [==============================] - 1s 2ms/step - loss: nan - val_loss: nan\nEpoch 30/30\n363/363 [==============================] - 1s 2ms/step - loss: nan - val_loss: nan\n",
     "output_type": "stream"
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": "Notice how TensorBoard now sees two runs, and you can compare the learning curves.",
   "metadata": {
    "cell_id": "00121-175b73fb-2b19-45dd-ae6e-b6797af77bdc",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 52.390625
   }
  },
  {
   "cell_type": "markdown",
   "source": "Check out the other available logging options:",
   "metadata": {
    "cell_id": "00122-e65de519-b658-4545-9f60-d250a182ac59",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 52.390625
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00123-9326cd7c-ca72-43cb-be7c-5f4d48d94981",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "5895030",
    "execution_start": 1659186849642,
    "execution_millis": 137,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 191.9375
   },
   "source": "help(keras.callbacks.TensorBoard.__init__)",
   "execution_count": null,
   "outputs": [
    {
     "name": "stdout",
     "text": "Help on function __init__ in module keras.callbacks:\n\n__init__(self, log_dir='logs', histogram_freq=0, write_graph=True, write_images=False, write_steps_per_second=False, update_freq='epoch', profile_batch=0, embeddings_freq=0, embeddings_metadata=None, **kwargs)\n    Initialize self.  See help(type(self)) for accurate signature.\n\n",
     "output_type": "stream"
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": "# Hyperparameter Tuning",
   "metadata": {
    "cell_id": "00124-40611e03-b778-44ff-bca0-70f1e489cf0c",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 82
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00125-e32e276a-18f0-4a0e-b9fa-2fff40ae42ff",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "b86c7ea",
    "execution_start": 1659186849643,
    "execution_millis": 0,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 112
   },
   "source": "keras.backend.clear_session()\nnp.random.seed(42)\ntf.random.set_seed(42)",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00126-4373267a-bebd-4dce-bcac-dccb9f87e4c2",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "32bf1a74",
    "execution_start": 1659186849644,
    "execution_millis": 2,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 220
   },
   "source": "def build_model(n_hidden=1, n_neurons=30, learning_rate=3e-3, input_shape=[8]):\n    model = keras.models.Sequential()\n    model.add(keras.layers.InputLayer(input_shape=input_shape))\n    for layer in range(n_hidden):\n        model.add(keras.layers.Dense(n_neurons, activation=\"relu\"))\n    model.add(keras.layers.Dense(1))\n    optimizer = keras.optimizers.SGD(learning_rate=learning_rate)\n    model.compile(loss=\"mse\", optimizer=optimizer)\n    return model",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00127-9ad174d3-72db-4944-8244-4b3819f2ee4f",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "69370f20",
    "execution_start": 1659186849649,
    "execution_millis": 2,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 131.375
   },
   "source": "keras_reg = keras.wrappers.scikit_learn.KerasRegressor(build_model)",
   "execution_count": null,
   "outputs": [
    {
     "name": "stderr",
     "text": "<ipython-input-94-651c14c6d32f>:1: DeprecationWarning: KerasRegressor is deprecated, use Sci-Keras (https://github.com/adriangb/scikeras) instead. See https://www.adriangb.com/scikeras/stable/migration.html for help migrating.\n  keras_reg = keras.wrappers.scikit_learn.KerasRegressor(build_model)\n",
     "output_type": "stream"
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00128-8eb28132-0dc2-4e00-87ee-54989bf26cdc",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "d6871ce1",
    "execution_start": 1659186849711,
    "execution_millis": 45862,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 784.1875,
    "deepnote_output_heights": [
     null,
     21.1875
    ]
   },
   "source": "keras_reg.fit(X_train, y_train, epochs=100,\n              validation_data=(X_valid, y_valid),\n              callbacks=[keras.callbacks.EarlyStopping(patience=10)])",
   "execution_count": null,
   "outputs": [
    {
     "name": "stdout",
     "text": "Epoch 1/100\n363/363 [==============================] - 1s 2ms/step - loss: 1.0896 - val_loss: 20.7721\nEpoch 2/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.7606 - val_loss: 5.0266\nEpoch 3/100\n363/363 [==============================] - 1s 1ms/step - loss: 0.5456 - val_loss: 0.5490\nEpoch 4/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.4732 - val_loss: 0.4529\nEpoch 5/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.4503 - val_loss: 0.4188\nEpoch 6/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.4338 - val_loss: 0.4129\nEpoch 7/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.4241 - val_loss: 0.4004\nEpoch 8/100\n363/363 [==============================] - 1s 1ms/step - loss: 0.4168 - val_loss: 0.3944\nEpoch 9/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.4108 - val_loss: 0.3961\nEpoch 10/100\n363/363 [==============================] - 1s 1ms/step - loss: 0.4060 - val_loss: 0.4071\nEpoch 11/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.4021 - val_loss: 0.3855\nEpoch 12/100\n363/363 [==============================] - 1s 1ms/step - loss: 0.3984 - val_loss: 0.4136\nEpoch 13/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3951 - val_loss: 0.3997\nEpoch 14/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3921 - val_loss: 0.3818\nEpoch 15/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3894 - val_loss: 0.3829\nEpoch 16/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3869 - val_loss: 0.3739\nEpoch 17/100\n363/363 [==============================] - 1s 1ms/step - loss: 0.3848 - val_loss: 0.4022\nEpoch 18/100\n363/363 [==============================] - 1s 1ms/step - loss: 0.3829 - val_loss: 0.3873\nEpoch 19/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3807 - val_loss: 0.3768\nEpoch 20/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3791 - val_loss: 0.4191\nEpoch 21/100\n363/363 [==============================] - 1s 1ms/step - loss: 0.3774 - val_loss: 0.3926\nEpoch 22/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3756 - val_loss: 0.4238\nEpoch 23/100\n363/363 [==============================] - 1s 1ms/step - loss: 0.3742 - val_loss: 0.3523\nEpoch 24/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3725 - val_loss: 0.3842\nEpoch 25/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3710 - val_loss: 0.4162\nEpoch 26/100\n363/363 [==============================] - 1s 1ms/step - loss: 0.3700 - val_loss: 0.3980\nEpoch 27/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3691 - val_loss: 0.3474\nEpoch 28/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3677 - val_loss: 0.3920\nEpoch 29/100\n363/363 [==============================] - 1s 1ms/step - loss: 0.3670 - val_loss: 0.3566\nEpoch 30/100\n363/363 [==============================] - 1s 1ms/step - loss: 0.3653 - val_loss: 0.4191\nEpoch 31/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3647 - val_loss: 0.3721\nEpoch 32/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3633 - val_loss: 0.3947\nEpoch 33/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3632 - val_loss: 0.3423\nEpoch 34/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3617 - val_loss: 0.3453\nEpoch 35/100\n363/363 [==============================] - 1s 1ms/step - loss: 0.3610 - val_loss: 0.4068\nEpoch 36/100\n363/363 [==============================] - 1s 1ms/step - loss: 0.3607 - val_loss: 0.3416\nEpoch 37/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3596 - val_loss: 0.3787\nEpoch 38/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3589 - val_loss: 0.3379\nEpoch 39/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3582 - val_loss: 0.3419\nEpoch 40/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3572 - val_loss: 0.3704\nEpoch 41/100\n363/363 [==============================] - 1s 1ms/step - loss: 0.3570 - val_loss: 0.3660\nEpoch 42/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3563 - val_loss: 0.3803\nEpoch 43/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3551 - val_loss: 0.3766\nEpoch 44/100\n363/363 [==============================] - 1s 1ms/step - loss: 0.3548 - val_loss: 0.3814\nEpoch 45/100\n363/363 [==============================] - 1s 1ms/step - loss: 0.3543 - val_loss: 0.3326\nEpoch 46/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3532 - val_loss: 0.3385\nEpoch 47/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3527 - val_loss: 0.3657\nEpoch 48/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3521 - val_loss: 0.3577\nEpoch 49/100\n363/363 [==============================] - 1s 1ms/step - loss: 0.3525 - val_loss: 0.3359\nEpoch 50/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3510 - val_loss: 0.3317\nEpoch 51/100\n363/363 [==============================] - 1s 1ms/step - loss: 0.3504 - val_loss: 0.3564\nEpoch 52/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3502 - val_loss: 0.3524\nEpoch 53/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3496 - val_loss: 0.4589\nEpoch 54/100\n363/363 [==============================] - 1s 1ms/step - loss: 0.3497 - val_loss: 0.3810\nEpoch 55/100\n363/363 [==============================] - 1s 1ms/step - loss: 0.3490 - val_loss: 0.3544\nEpoch 56/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3485 - val_loss: 0.3728\nEpoch 57/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3478 - val_loss: 0.3339\nEpoch 58/100\n363/363 [==============================] - 1s 1ms/step - loss: 0.3469 - val_loss: 0.4001\nEpoch 59/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3474 - val_loss: 0.3264\nEpoch 60/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3465 - val_loss: 0.3271\nEpoch 61/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3452 - val_loss: 0.3350\nEpoch 62/100\n363/363 [==============================] - 1s 1ms/step - loss: 0.3453 - val_loss: 0.3489\nEpoch 63/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3444 - val_loss: 0.3398\nEpoch 64/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3450 - val_loss: 0.3273\nEpoch 65/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3437 - val_loss: 0.3296\nEpoch 66/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3431 - val_loss: 0.3307\nEpoch 67/100\n363/363 [==============================] - 1s 1ms/step - loss: 0.3428 - val_loss: 0.3251\nEpoch 68/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3423 - val_loss: 0.3242\nEpoch 69/100\n363/363 [==============================] - 1s 1ms/step - loss: 0.3419 - val_loss: 0.3253\nEpoch 70/100\n363/363 [==============================] - 1s 1ms/step - loss: 0.3412 - val_loss: 0.3661\nEpoch 71/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3413 - val_loss: 0.3378\nEpoch 72/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3405 - val_loss: 0.3271\nEpoch 73/100\n363/363 [==============================] - 1s 1ms/step - loss: 0.3399 - val_loss: 0.3242\nEpoch 74/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3402 - val_loss: 0.3664\nEpoch 75/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3397 - val_loss: 0.3284\nEpoch 76/100\n363/363 [==============================] - 1s 1ms/step - loss: 0.3395 - val_loss: 0.3242\nEpoch 77/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3383 - val_loss: 0.3375\nEpoch 78/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3384 - val_loss: 0.3365\n",
     "output_type": "stream"
    },
    {
     "output_type": "execute_result",
     "execution_count": 95,
     "data": {
      "text/plain": "<keras.callbacks.History at 0x7f3b26df75b0>"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00129-0ca2a339-c07d-4623-9493-35d6a935c86c",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "a8cd03ed",
    "execution_start": 1659186894806,
    "execution_millis": 768,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 107.1875
   },
   "source": "mse_test = keras_reg.score(X_test, y_test)",
   "execution_count": null,
   "outputs": [
    {
     "name": "stdout",
     "text": "162/162 [==============================] - 0s 1ms/step - loss: 0.3411\n",
     "output_type": "stream"
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00130-3d7188e9-0c58-418a-9b15-67c226694bd2",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "853a8b98",
    "execution_start": 1659186895047,
    "execution_millis": 528,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 107.1875
   },
   "source": "y_pred = keras_reg.predict(X_new)",
   "execution_count": null,
   "outputs": [
    {
     "name": "stdout",
     "text": "1/1 [==============================] - 0s 41ms/step\n",
     "output_type": "stream"
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00131-52f54252-8eb1-44ed-8fa1-b7a93a028622",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "8905d08a",
    "execution_start": 1659186895118,
    "execution_millis": 3,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 94
   },
   "source": "np.random.seed(42)\ntf.random.set_seed(42)",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "markdown",
   "source": "**Warning**: the following cell crashes at the end of training. This seems to be caused by [Keras issue #13586](https://github.com/keras-team/keras/issues/13586), which was triggered by a recent change in Scikit-Learn. [Pull Request #13598](https://github.com/keras-team/keras/pull/13598) seems to fix the issue, so this problem should be resolved soon. In the meantime, I've added `.tolist()` and `.rvs(1000).tolist()` as workarounds.",
   "metadata": {
    "cell_id": "00132-2cfcb821-17aa-4d11-af63-dc6fd7248bc4",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 97.171875
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00133-44731e85-23cd-46a3-874a-c74d89cf12cb",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "ecb7905b",
    "execution_start": 1659186895129,
    "execution_millis": 716981,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 1036.328125,
    "deepnote_output_heights": [
     null,
     93.328125
    ]
   },
   "source": "from scipy.stats import reciprocal\nfrom sklearn.model_selection import RandomizedSearchCV\n\nparam_distribs = {\n    \"n_hidden\": [0, 1, 2, 3],\n    \"n_neurons\": np.arange(1, 100).tolist(),\n    \"learning_rate\": reciprocal(3e-4, 3e-2).rvs(1000).tolist(),\n}\n\nrnd_search_cv = RandomizedSearchCV(keras_reg, param_distribs, n_iter=10, cv=3, verbose=2)\nrnd_search_cv.fit(X_train, y_train, epochs=100,\n                  validation_data=(X_valid, y_valid),\n                  callbacks=[keras.callbacks.EarlyStopping(patience=10)])",
   "execution_count": null,
   "outputs": [
    {
     "name": "stdout",
     "text": "[CV] END learning_rate=0.0020587676114196545, n_hidden=1, n_neurons=49; total time=   5.5s\nEpoch 1/100\n242/242 [==============================] - 1s 2ms/step - loss: 1.7445 - val_loss: 2.5834\nEpoch 2/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.7268 - val_loss: 3.5564\nEpoch 3/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.6419 - val_loss: 1.7895\nEpoch 4/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.6134 - val_loss: 1.7436\nEpoch 5/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.5564 - val_loss: 0.6344\nEpoch 6/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.5303 - val_loss: 0.8713\nEpoch 7/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.5070 - val_loss: 0.5604\nEpoch 8/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.4919 - val_loss: 0.4695\nEpoch 9/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.4759 - val_loss: 0.4942\nEpoch 10/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.4659 - val_loss: 0.4375\nEpoch 11/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.4561 - val_loss: 0.4536\nEpoch 12/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.4481 - val_loss: 0.4276\nEpoch 13/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.4408 - val_loss: 0.4084\nEpoch 14/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.4354 - val_loss: 0.4897\nEpoch 15/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.4300 - val_loss: 0.4018\nEpoch 16/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.4246 - val_loss: 0.5505\nEpoch 17/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.4211 - val_loss: 0.4602\nEpoch 18/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.4171 - val_loss: 0.4347\nEpoch 19/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.4132 - val_loss: 0.3835\nEpoch 20/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.4092 - val_loss: 0.4115\nEpoch 21/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.4065 - val_loss: 0.3817\nEpoch 22/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.4035 - val_loss: 0.3737\nEpoch 23/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.4008 - val_loss: 0.3720\nEpoch 24/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3983 - val_loss: 0.4318\nEpoch 25/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3970 - val_loss: 0.4158\nEpoch 26/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3943 - val_loss: 0.3821\nEpoch 27/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3922 - val_loss: 0.4069\nEpoch 28/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3909 - val_loss: 0.4024\nEpoch 29/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3887 - val_loss: 0.5904\nEpoch 30/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3898 - val_loss: 0.4027\nEpoch 31/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3858 - val_loss: 0.4216\nEpoch 32/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3847 - val_loss: 0.3603\nEpoch 33/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3825 - val_loss: 0.4134\nEpoch 34/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.3819 - val_loss: 0.3633\nEpoch 35/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3805 - val_loss: 0.3542\nEpoch 36/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3786 - val_loss: 0.3568\nEpoch 37/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3783 - val_loss: 0.4216\nEpoch 38/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3770 - val_loss: 0.5522\nEpoch 39/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3792 - val_loss: 0.5648\nEpoch 40/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3763 - val_loss: 0.6416\nEpoch 41/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3785 - val_loss: 0.3847\nEpoch 42/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3729 - val_loss: 0.5255\nEpoch 43/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3744 - val_loss: 0.7023\nEpoch 44/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3741 - val_loss: 0.7508\nEpoch 45/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3771 - val_loss: 0.5608\n121/121 [==============================] - 0s 1ms/step - loss: 0.3745\n[CV] END learning_rate=0.0020587676114196545, n_hidden=1, n_neurons=49; total time=  20.0s\nEpoch 1/100\n242/242 [==============================] - 1s 2ms/step - loss: 1.0682 - val_loss: 6.4183\nEpoch 2/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.7154 - val_loss: 16.7917\nEpoch 3/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.5830 - val_loss: 4.7824\nEpoch 4/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.4475 - val_loss: 8.6078\nEpoch 5/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.4779 - val_loss: 1.8032\nEpoch 6/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.4008 - val_loss: 0.3655\nEpoch 7/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3682 - val_loss: 0.3783\nEpoch 8/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3591 - val_loss: 0.4054\nEpoch 9/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3533 - val_loss: 0.3909\nEpoch 10/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.3488 - val_loss: 0.3907\nEpoch 11/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3445 - val_loss: 0.3554\nEpoch 12/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3395 - val_loss: 0.3610\nEpoch 13/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3352 - val_loss: 0.3665\nEpoch 14/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3334 - val_loss: 0.3635\nEpoch 15/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3294 - val_loss: 0.3581\nEpoch 16/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3272 - val_loss: 0.3566\nEpoch 17/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3240 - val_loss: 0.3580\nEpoch 18/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3224 - val_loss: 0.3481\nEpoch 19/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3187 - val_loss: 0.3540\nEpoch 20/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3166 - val_loss: 0.3415\nEpoch 21/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3148 - val_loss: 0.3362\nEpoch 22/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3122 - val_loss: 0.3555\nEpoch 23/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3099 - val_loss: 0.3351\nEpoch 24/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.3094 - val_loss: 0.3081\nEpoch 25/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3047 - val_loss: 0.3495\nEpoch 26/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3032 - val_loss: 0.3064\nEpoch 27/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.3026 - val_loss: 0.3311\nEpoch 28/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3019 - val_loss: 0.3436\nEpoch 29/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2984 - val_loss: 0.3358\nEpoch 30/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2963 - val_loss: 0.2964\nEpoch 31/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.2953 - val_loss: 0.3071\nEpoch 32/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2934 - val_loss: 0.2941\nEpoch 33/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2909 - val_loss: 0.3172\nEpoch 34/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2889 - val_loss: 0.3220\nEpoch 35/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2878 - val_loss: 0.2902\nEpoch 36/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2865 - val_loss: 0.3640\nEpoch 37/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2859 - val_loss: 0.3461\nEpoch 38/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2855 - val_loss: 0.3547\nEpoch 39/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2838 - val_loss: 0.2935\nEpoch 40/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2820 - val_loss: 0.3328\nEpoch 41/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2808 - val_loss: 0.2871\nEpoch 42/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2796 - val_loss: 0.2931\nEpoch 43/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2789 - val_loss: 0.2863\nEpoch 44/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2766 - val_loss: 0.3160\nEpoch 45/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2767 - val_loss: 0.3112\nEpoch 46/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2755 - val_loss: 0.3287\nEpoch 47/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2736 - val_loss: 0.2909\nEpoch 48/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.2733 - val_loss: 0.2857\nEpoch 49/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2709 - val_loss: 0.2870\nEpoch 50/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2712 - val_loss: 0.3483\nEpoch 51/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2696 - val_loss: 0.3198\nEpoch 52/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.2685 - val_loss: 0.3309\nEpoch 53/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2680 - val_loss: 0.2791\nEpoch 54/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2663 - val_loss: 0.2888\nEpoch 55/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2650 - val_loss: 0.2911\nEpoch 56/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2653 - val_loss: 0.2880\nEpoch 57/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2636 - val_loss: 0.2979\nEpoch 58/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2638 - val_loss: 0.2780\nEpoch 59/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2629 - val_loss: 0.2879\nEpoch 60/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2607 - val_loss: 0.2777\nEpoch 61/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2619 - val_loss: 0.3540\nEpoch 62/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2611 - val_loss: 0.2796\nEpoch 63/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2599 - val_loss: 0.3687\nEpoch 64/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2604 - val_loss: 0.2725\nEpoch 65/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2575 - val_loss: 0.3111\nEpoch 66/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2586 - val_loss: 0.3697\nEpoch 67/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2600 - val_loss: 0.6054\nEpoch 68/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2603 - val_loss: 0.3292\nEpoch 69/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.2587 - val_loss: 0.4400\nEpoch 70/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2582 - val_loss: 0.3052\nEpoch 71/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2576 - val_loss: 0.3678\nEpoch 72/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2560 - val_loss: 0.2706\nEpoch 73/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.2540 - val_loss: 0.3026\nEpoch 74/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2529 - val_loss: 0.2902\nEpoch 75/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2529 - val_loss: 0.2898\nEpoch 76/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2517 - val_loss: 0.2819\nEpoch 77/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2532 - val_loss: 0.2768\nEpoch 78/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2517 - val_loss: 0.2968\nEpoch 79/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2516 - val_loss: 0.2747\nEpoch 80/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2510 - val_loss: 0.3076\nEpoch 81/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2498 - val_loss: 0.2699\nEpoch 82/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2503 - val_loss: 0.3005\nEpoch 83/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2485 - val_loss: 0.2661\nEpoch 84/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2486 - val_loss: 0.3072\nEpoch 85/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2495 - val_loss: 0.2743\nEpoch 86/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2466 - val_loss: 0.2968\nEpoch 87/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2460 - val_loss: 0.2698\nEpoch 88/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2469 - val_loss: 0.2919\nEpoch 89/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2452 - val_loss: 0.3179\nEpoch 90/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.2451 - val_loss: 0.2897\nEpoch 91/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2470 - val_loss: 0.2693\nEpoch 92/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2448 - val_loss: 0.3358\nEpoch 93/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2454 - val_loss: 0.2735\n121/121 [==============================] - 0s 1ms/step - loss: 0.3005\n[CV] END learning_rate=0.005803602934201024, n_hidden=3, n_neurons=74; total time=  45.0s\nEpoch 1/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.8717 - val_loss: 0.7369\nEpoch 2/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.5007 - val_loss: 0.4431\nEpoch 3/100\n242/242 [==============================] - 1s 3ms/step - loss: 0.4272 - val_loss: 0.3919\nEpoch 4/100\n242/242 [==============================] - 1s 4ms/step - loss: 0.3961 - val_loss: 0.3834\nEpoch 5/100\n242/242 [==============================] - 1s 4ms/step - loss: 0.3777 - val_loss: 0.3951\nEpoch 6/100\n242/242 [==============================] - 1s 4ms/step - loss: 0.3661 - val_loss: 0.4649\nEpoch 7/100\n242/242 [==============================] - 1s 4ms/step - loss: 0.3580 - val_loss: 0.6408\nEpoch 8/100\n242/242 [==============================] - 1s 4ms/step - loss: 0.3524 - val_loss: 0.7273\nEpoch 9/100\n242/242 [==============================] - 1s 4ms/step - loss: 0.3462 - val_loss: 0.9104\nEpoch 10/100\n242/242 [==============================] - 1s 4ms/step - loss: 0.3421 - val_loss: 0.6969\nEpoch 11/100\n242/242 [==============================] - 1s 3ms/step - loss: 0.3364 - val_loss: 0.6999\nEpoch 12/100\n242/242 [==============================] - 1s 4ms/step - loss: 0.3327 - val_loss: 0.7835\nEpoch 13/100\n242/242 [==============================] - 1s 4ms/step - loss: 0.3292 - val_loss: 0.8539\nEpoch 14/100\n242/242 [==============================] - 1s 4ms/step - loss: 0.3277 - val_loss: 0.8282\n121/121 [==============================] - 0s 2ms/step - loss: 0.3525\n[CV] END learning_rate=0.005803602934201024, n_hidden=3, n_neurons=74; total time=  21.2s\nEpoch 1/100\n242/242 [==============================] - 1s 3ms/step - loss: 0.9177 - val_loss: 0.9196\nEpoch 2/100\n242/242 [==============================] - 1s 3ms/step - loss: 0.4767 - val_loss: 2.1025\nEpoch 3/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.4329 - val_loss: 3.5511\nEpoch 4/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.4408 - val_loss: 1.5867\nEpoch 5/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3910 - val_loss: 0.4227\nEpoch 6/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3715 - val_loss: 0.3738\nEpoch 7/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3634 - val_loss: 0.3350\nEpoch 8/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3569 - val_loss: 0.3384\nEpoch 9/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3518 - val_loss: 0.3720\nEpoch 10/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3514 - val_loss: 0.3276\nEpoch 11/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3456 - val_loss: 0.3971\nEpoch 12/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.3420 - val_loss: 0.3329\nEpoch 13/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.3373 - val_loss: 0.3227\nEpoch 14/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3347 - val_loss: 0.3675\nEpoch 15/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3324 - val_loss: 0.3209\nEpoch 16/100\n242/242 [==============================] - 1s 3ms/step - loss: 0.3294 - val_loss: 0.3560\nEpoch 17/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3269 - val_loss: 0.3215\nEpoch 18/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3243 - val_loss: 0.3555\nEpoch 19/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3224 - val_loss: 0.3386\nEpoch 20/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3193 - val_loss: 0.3563\nEpoch 21/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3181 - val_loss: 0.3165\nEpoch 22/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3156 - val_loss: 0.3194\nEpoch 23/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3132 - val_loss: 0.3725\nEpoch 24/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3118 - val_loss: 0.3028\nEpoch 25/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3100 - val_loss: 0.3214\nEpoch 26/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3071 - val_loss: 0.3261\nEpoch 27/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3055 - val_loss: 0.3141\nEpoch 28/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3046 - val_loss: 0.3434\nEpoch 29/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3060 - val_loss: 0.4403\nEpoch 30/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3045 - val_loss: 0.3030\nEpoch 31/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3000 - val_loss: 0.3273\nEpoch 32/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2978 - val_loss: 0.3051\nEpoch 33/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2982 - val_loss: 0.3385\nEpoch 34/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.2972 - val_loss: 0.2970\nEpoch 35/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2934 - val_loss: 0.2939\nEpoch 36/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2909 - val_loss: 0.3894\nEpoch 37/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.2907 - val_loss: 0.3670\nEpoch 38/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.2937 - val_loss: 0.4261\nEpoch 39/100\n242/242 [==============================] - 1s 3ms/step - loss: 0.2921 - val_loss: 0.3230\nEpoch 40/100\n242/242 [==============================] - 1s 4ms/step - loss: 0.2913 - val_loss: 0.3934\nEpoch 41/100\n242/242 [==============================] - 1s 4ms/step - loss: 0.2874 - val_loss: 0.3208\nEpoch 42/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2859 - val_loss: 0.3375\nEpoch 43/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2874 - val_loss: 0.2842\nEpoch 44/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2839 - val_loss: 0.3527\nEpoch 45/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2847 - val_loss: 0.3067\nEpoch 46/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2818 - val_loss: 0.3564\nEpoch 47/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2806 - val_loss: 0.2940\nEpoch 48/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2804 - val_loss: 0.2865\nEpoch 49/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2795 - val_loss: 0.3518\nEpoch 50/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2784 - val_loss: 0.2951\nEpoch 51/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2774 - val_loss: 0.3477\nEpoch 52/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.2767 - val_loss: 0.2899\nEpoch 53/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2761 - val_loss: 0.3003\n121/121 [==============================] - 0s 1ms/step - loss: 0.3003\n[CV] END learning_rate=0.005803602934201024, n_hidden=3, n_neurons=74; total time=  41.8s\nEpoch 1/100\n242/242 [==============================] - 1s 3ms/step - loss: 0.9615 - val_loss: 10.9251\nEpoch 2/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.5921 - val_loss: 3.3912\nEpoch 3/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.4539 - val_loss: 0.4039\nEpoch 4/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3914 - val_loss: 0.3692\nEpoch 5/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3713 - val_loss: 0.3555\nEpoch 6/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3601 - val_loss: 0.3875\nEpoch 7/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3523 - val_loss: 0.3633\nEpoch 8/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3454 - val_loss: 0.3991\nEpoch 9/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3411 - val_loss: 0.3797\nEpoch 10/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.3382 - val_loss: 0.3703\nEpoch 11/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3351 - val_loss: 0.3310\nEpoch 12/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.3304 - val_loss: 0.3511\nEpoch 13/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3269 - val_loss: 0.3793\nEpoch 14/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.3261 - val_loss: 0.3316\nEpoch 15/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3217 - val_loss: 0.3488\nEpoch 16/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3194 - val_loss: 0.3472\nEpoch 17/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3172 - val_loss: 0.3264\nEpoch 18/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3157 - val_loss: 0.3422\nEpoch 19/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3123 - val_loss: 0.3366\nEpoch 20/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3108 - val_loss: 0.3223\nEpoch 21/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3085 - val_loss: 0.4004\nEpoch 22/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3087 - val_loss: 0.3110\nEpoch 23/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3050 - val_loss: 0.3629\nEpoch 24/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3068 - val_loss: 0.3638\nEpoch 25/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3030 - val_loss: 0.3312\nEpoch 26/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2990 - val_loss: 0.3076\nEpoch 27/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2981 - val_loss: 0.3395\nEpoch 28/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2973 - val_loss: 0.3367\nEpoch 29/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2943 - val_loss: 0.3433\nEpoch 30/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2922 - val_loss: 0.3107\nEpoch 31/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.2920 - val_loss: 0.3300\nEpoch 32/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2900 - val_loss: 0.2955\nEpoch 33/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2881 - val_loss: 0.3283\nEpoch 34/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2855 - val_loss: 0.3040\nEpoch 35/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.2841 - val_loss: 0.2912\nEpoch 36/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2831 - val_loss: 0.3502\nEpoch 37/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2830 - val_loss: 0.3753\nEpoch 38/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2856 - val_loss: 0.3571\nEpoch 39/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2821 - val_loss: 0.3035\nEpoch 40/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2800 - val_loss: 0.3377\nEpoch 41/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2784 - val_loss: 0.2931\nEpoch 42/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2767 - val_loss: 0.3050\nEpoch 43/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2762 - val_loss: 0.2960\nEpoch 44/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2742 - val_loss: 0.3425\nEpoch 45/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2760 - val_loss: 0.2857\nEpoch 46/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2747 - val_loss: 0.3202\nEpoch 47/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2724 - val_loss: 0.3137\nEpoch 48/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2716 - val_loss: 0.3078\nEpoch 49/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.2700 - val_loss: 0.2757\nEpoch 50/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2701 - val_loss: 0.3335\nEpoch 51/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2686 - val_loss: 0.3330\nEpoch 52/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.2664 - val_loss: 0.3651\nEpoch 53/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2679 - val_loss: 0.2800\nEpoch 54/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2651 - val_loss: 0.3159\nEpoch 55/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2640 - val_loss: 0.2924\nEpoch 56/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.2642 - val_loss: 0.3006\nEpoch 57/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2626 - val_loss: 0.3504\nEpoch 58/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2642 - val_loss: 0.2772\nEpoch 59/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2627 - val_loss: 0.3071\n121/121 [==============================] - 0s 1ms/step - loss: 0.3018\n[CV] END learning_rate=0.0059640580092043885, n_hidden=3, n_neurons=80; total time=  29.0s\nEpoch 1/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.8381 - val_loss: 0.6551\nEpoch 2/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.4570 - val_loss: 0.4129\nEpoch 3/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.4107 - val_loss: 0.6096\nEpoch 4/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3911 - val_loss: 0.6534\nEpoch 5/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3755 - val_loss: 0.6227\nEpoch 6/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3647 - val_loss: 0.8404\nEpoch 7/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3602 - val_loss: 1.0610\nEpoch 8/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3516 - val_loss: 1.1319\nEpoch 9/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3456 - val_loss: 1.2348\nEpoch 10/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.3425 - val_loss: 0.7924\nEpoch 11/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3374 - val_loss: 0.8270\nEpoch 12/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3338 - val_loss: 0.7641\n121/121 [==============================] - 0s 1ms/step - loss: 0.3593\n[CV] END learning_rate=0.0059640580092043885, n_hidden=3, n_neurons=80; total time=  10.8s\nEpoch 1/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.8683 - val_loss: 2.2007\nEpoch 2/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.5072 - val_loss: 3.3028\nEpoch 3/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.4486 - val_loss: 0.9130\nEpoch 4/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.4071 - val_loss: 0.5328\nEpoch 5/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3842 - val_loss: 0.3609\nEpoch 6/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3729 - val_loss: 0.4151\nEpoch 7/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3659 - val_loss: 0.3580\nEpoch 8/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.3596 - val_loss: 0.3516\nEpoch 9/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3528 - val_loss: 0.3983\nEpoch 10/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.3511 - val_loss: 0.3323\nEpoch 11/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3445 - val_loss: 0.4230\nEpoch 12/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3420 - val_loss: 0.3284\nEpoch 13/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3374 - val_loss: 0.3469\nEpoch 14/100\n242/242 [==============================] - 1s 3ms/step - loss: 0.3337 - val_loss: 0.4036\nEpoch 15/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3326 - val_loss: 0.3275\nEpoch 16/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3291 - val_loss: 0.3779\nEpoch 17/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3263 - val_loss: 0.3211\nEpoch 18/100\n242/242 [==============================] - 1s 3ms/step - loss: 0.3227 - val_loss: 0.3306\nEpoch 19/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3216 - val_loss: 0.3842\nEpoch 20/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3192 - val_loss: 0.3373\nEpoch 21/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3178 - val_loss: 0.3721\nEpoch 22/100\n242/242 [==============================] - 1s 3ms/step - loss: 0.3177 - val_loss: 0.3072\nEpoch 23/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3146 - val_loss: 0.3415\nEpoch 24/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3131 - val_loss: 0.3277\nEpoch 25/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3099 - val_loss: 0.3047\nEpoch 26/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3072 - val_loss: 0.3040\nEpoch 27/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3047 - val_loss: 0.4404\nEpoch 28/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.3070 - val_loss: 0.3396\nEpoch 29/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3048 - val_loss: 0.3881\nEpoch 30/100\n242/242 [==============================] - 1s 3ms/step - loss: 0.3033 - val_loss: 0.3153\nEpoch 31/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2999 - val_loss: 0.3274\nEpoch 32/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2977 - val_loss: 0.3988\nEpoch 33/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2985 - val_loss: 0.3137\nEpoch 34/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.2941 - val_loss: 0.3805\nEpoch 35/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3004 - val_loss: 0.3012\nEpoch 36/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2939 - val_loss: 0.3426\nEpoch 37/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2922 - val_loss: 0.2940\nEpoch 38/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2909 - val_loss: 0.3713\nEpoch 39/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2896 - val_loss: 0.2951\nEpoch 40/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2880 - val_loss: 0.3647\nEpoch 41/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2866 - val_loss: 0.3173\nEpoch 42/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2853 - val_loss: 0.3686\nEpoch 43/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2855 - val_loss: 0.2952\nEpoch 44/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2835 - val_loss: 0.3360\nEpoch 45/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2831 - val_loss: 0.3028\nEpoch 46/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2804 - val_loss: 0.3377\nEpoch 47/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2796 - val_loss: 0.2843\nEpoch 48/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2796 - val_loss: 0.2984\nEpoch 49/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2780 - val_loss: 0.3985\nEpoch 50/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2775 - val_loss: 0.3136\nEpoch 51/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.2757 - val_loss: 0.4747\nEpoch 52/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2761 - val_loss: 0.2996\nEpoch 53/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2752 - val_loss: 0.2927\nEpoch 54/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2742 - val_loss: 0.3249\nEpoch 55/100\n242/242 [==============================] - 1s 3ms/step - loss: 0.2729 - val_loss: 0.2812\nEpoch 56/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2729 - val_loss: 0.3926\nEpoch 57/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2720 - val_loss: 0.2858\nEpoch 58/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2714 - val_loss: 0.4391\nEpoch 59/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2711 - val_loss: 0.2852\nEpoch 60/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.2696 - val_loss: 0.3030\nEpoch 61/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2686 - val_loss: 0.2995\nEpoch 62/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2676 - val_loss: 0.3672\nEpoch 63/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2696 - val_loss: 0.3024\nEpoch 64/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2702 - val_loss: 0.3348\nEpoch 65/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2685 - val_loss: 0.3026\n121/121 [==============================] - 0s 1ms/step - loss: 0.2913\n[CV] END learning_rate=0.0059640580092043885, n_hidden=3, n_neurons=80; total time=  32.4s\nEpoch 1/100\n242/242 [==============================] - 1s 2ms/step - loss: 1.2259 - val_loss: 0.5753\nEpoch 2/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.5658 - val_loss: 8.9878\nEpoch 3/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.5484 - val_loss: 11.0986\nEpoch 4/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.5602 - val_loss: 1.1306\nEpoch 5/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.4336 - val_loss: 0.5256\nEpoch 6/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.4072 - val_loss: 0.4498\nEpoch 7/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3939 - val_loss: 0.4056\nEpoch 8/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3835 - val_loss: 0.3999\nEpoch 9/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3766 - val_loss: 0.3957\nEpoch 10/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.3712 - val_loss: 0.3904\nEpoch 11/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3672 - val_loss: 0.3688\nEpoch 12/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3630 - val_loss: 0.3651\nEpoch 13/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3587 - val_loss: 0.3709\nEpoch 14/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3579 - val_loss: 0.3816\nEpoch 15/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3534 - val_loss: 0.3620\nEpoch 16/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3528 - val_loss: 0.3670\nEpoch 17/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3490 - val_loss: 0.3671\nEpoch 18/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3474 - val_loss: 0.3605\nEpoch 19/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3445 - val_loss: 0.3552\nEpoch 20/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3421 - val_loss: 0.3538\nEpoch 21/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3417 - val_loss: 0.3520\nEpoch 22/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3388 - val_loss: 0.3477\nEpoch 23/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3372 - val_loss: 0.3509\nEpoch 24/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3374 - val_loss: 0.3304\nEpoch 25/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3333 - val_loss: 0.3686\nEpoch 26/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3334 - val_loss: 0.3246\nEpoch 27/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3336 - val_loss: 0.3389\nEpoch 28/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3322 - val_loss: 0.3368\nEpoch 29/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3278 - val_loss: 0.3387\nEpoch 30/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3276 - val_loss: 0.3212\nEpoch 31/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3255 - val_loss: 0.3220\nEpoch 32/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3256 - val_loss: 0.3150\nEpoch 33/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3239 - val_loss: 0.3453\nEpoch 34/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3221 - val_loss: 0.3164\nEpoch 35/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3209 - val_loss: 0.3140\nEpoch 36/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3191 - val_loss: 0.3903\nEpoch 37/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3189 - val_loss: 0.3813\nEpoch 38/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3190 - val_loss: 0.3408\nEpoch 39/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3160 - val_loss: 0.3149\nEpoch 40/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3154 - val_loss: 0.3380\nEpoch 41/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3188 - val_loss: 0.3146\nEpoch 42/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3149 - val_loss: 0.3146\nEpoch 43/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3150 - val_loss: 0.3101\nEpoch 44/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3114 - val_loss: 0.3410\nEpoch 45/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3154 - val_loss: 0.3052\nEpoch 46/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3107 - val_loss: 0.3328\nEpoch 47/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3088 - val_loss: 0.3337\nEpoch 48/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3114 - val_loss: 0.3183\nEpoch 49/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3073 - val_loss: 0.3022\nEpoch 50/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3069 - val_loss: 0.3594\nEpoch 51/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3060 - val_loss: 0.3319\nEpoch 52/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3047 - val_loss: 0.3446\nEpoch 53/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3052 - val_loss: 0.3080\nEpoch 54/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3034 - val_loss: 0.3094\nEpoch 55/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.3036 - val_loss: 0.3467\nEpoch 56/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3043 - val_loss: 0.3008\nEpoch 57/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3015 - val_loss: 0.3702\nEpoch 58/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3037 - val_loss: 0.3175\nEpoch 59/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3005 - val_loss: 0.2984\nEpoch 60/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2989 - val_loss: 0.3179\nEpoch 61/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2999 - val_loss: 0.3294\nEpoch 62/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2986 - val_loss: 0.3430\nEpoch 63/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2987 - val_loss: 0.3159\nEpoch 64/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2961 - val_loss: 0.3646\nEpoch 65/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2961 - val_loss: 0.2956\nEpoch 66/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3030 - val_loss: 0.3076\nEpoch 67/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2951 - val_loss: 0.3313\nEpoch 68/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2946 - val_loss: 0.2963\nEpoch 69/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2945 - val_loss: 0.4154\nEpoch 70/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2977 - val_loss: 0.2994\nEpoch 71/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2925 - val_loss: 0.4090\nEpoch 72/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2927 - val_loss: 0.2960\nEpoch 73/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2923 - val_loss: 0.3041\nEpoch 74/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2897 - val_loss: 0.3494\nEpoch 75/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.2916 - val_loss: 0.2979\n121/121 [==============================] - 0s 1ms/step - loss: 0.3269\n[CV] END learning_rate=0.004591455636549438, n_hidden=2, n_neurons=59; total time=  34.4s\nEpoch 1/100\n242/242 [==============================] - 1s 2ms/step - loss: 1.1975 - val_loss: 0.8898\nEpoch 2/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.5319 - val_loss: 0.5270\nEpoch 3/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.4524 - val_loss: 0.4844\nEpoch 4/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.4199 - val_loss: 0.4250\nEpoch 5/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.4023 - val_loss: 0.3735\nEpoch 6/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3907 - val_loss: 0.3859\nEpoch 7/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3833 - val_loss: 0.4576\nEpoch 8/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3754 - val_loss: 0.4926\nEpoch 9/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3691 - val_loss: 0.6246\nEpoch 10/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3660 - val_loss: 0.5262\nEpoch 11/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3611 - val_loss: 0.5952\nEpoch 12/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3570 - val_loss: 0.6355\nEpoch 13/100\n242/242 [==============================] - 1s 3ms/step - loss: 0.3550 - val_loss: 0.7437\nEpoch 14/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3530 - val_loss: 0.7102\nEpoch 15/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3502 - val_loss: 0.6823\n121/121 [==============================] - 0s 1ms/step - loss: 0.3613\n[CV] END learning_rate=0.004591455636549438, n_hidden=2, n_neurons=59; total time=   8.0s\nEpoch 1/100\n242/242 [==============================] - 1s 2ms/step - loss: 1.1315 - val_loss: 2.8528\nEpoch 2/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.6016 - val_loss: 2.3412\nEpoch 3/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.5108 - val_loss: 0.9015\nEpoch 4/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.4594 - val_loss: 0.8313\nEpoch 5/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.4291 - val_loss: 0.5217\nEpoch 6/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.4115 - val_loss: 0.4956\nEpoch 7/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.4016 - val_loss: 0.3745\nEpoch 8/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3917 - val_loss: 0.4012\nEpoch 9/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3848 - val_loss: 0.4170\nEpoch 10/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3814 - val_loss: 0.3841\nEpoch 11/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3756 - val_loss: 0.6120\nEpoch 12/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3743 - val_loss: 0.3577\nEpoch 13/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.3671 - val_loss: 0.3498\nEpoch 14/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3632 - val_loss: 0.5168\nEpoch 15/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3635 - val_loss: 0.4302\nEpoch 16/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3592 - val_loss: 0.5771\nEpoch 17/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3601 - val_loss: 0.5042\nEpoch 18/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.3547 - val_loss: 0.4974\nEpoch 19/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3535 - val_loss: 0.3362\nEpoch 20/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3484 - val_loss: 0.4162\nEpoch 21/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.3485 - val_loss: 0.3327\nEpoch 22/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.3459 - val_loss: 0.3290\nEpoch 23/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3435 - val_loss: 0.3294\nEpoch 24/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3410 - val_loss: 0.5383\nEpoch 25/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3439 - val_loss: 0.7764\nEpoch 26/100\n242/242 [==============================] - 1s 2ms/step - loss: 0.3428 - val_loss: 0.9064\nEpoch 27/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3451 - val_loss: 0.4924\nEpoch 28/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3370 - val_loss: 0.6255\nEpoch 29/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3373 - val_loss: 0.3391\nEpoch 30/100\n242/242 [==============================] - 1s 3ms/step - loss: 0.3327 - val_loss: 0.5513\nEpoch 31/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3329 - val_loss: 0.3596\nEpoch 32/100\n242/242 [==============================] - 0s 2ms/step - loss: 0.3306 - val_loss: 0.6998\n121/121 [==============================] - 0s 1ms/step - loss: 0.3341\n[CV] END learning_rate=0.004591455636549438, n_hidden=2, n_neurons=59; total time=  21.0s\nEpoch 1/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.8571 - val_loss: 6.5088\nEpoch 2/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.5640 - val_loss: 7.3062\nEpoch 3/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.4741 - val_loss: 0.5231\nEpoch 4/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3858 - val_loss: 0.4529\nEpoch 5/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3723 - val_loss: 0.3947\nEpoch 6/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3606 - val_loss: 0.5391\nEpoch 7/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3567 - val_loss: 0.3491\nEpoch 8/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3485 - val_loss: 0.3398\nEpoch 9/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3426 - val_loss: 0.3289\nEpoch 10/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3381 - val_loss: 0.4486\nEpoch 11/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3368 - val_loss: 0.9015\nEpoch 12/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3382 - val_loss: 0.9359\nEpoch 13/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3346 - val_loss: 0.3400\nEpoch 14/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3267 - val_loss: 0.3275\nEpoch 15/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3240 - val_loss: 0.3164\nEpoch 16/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3209 - val_loss: 0.3049\nEpoch 17/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3179 - val_loss: 0.4308\nEpoch 18/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3170 - val_loss: 0.3051\nEpoch 19/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3147 - val_loss: 0.3356\nEpoch 20/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3125 - val_loss: 0.4239\nEpoch 21/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3101 - val_loss: 0.3352\nEpoch 22/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3071 - val_loss: 0.6459\nEpoch 23/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3081 - val_loss: 1.1851\nEpoch 24/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3120 - val_loss: 0.5988\nEpoch 25/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3044 - val_loss: 0.3239\nEpoch 26/100\n363/363 [==============================] - 1s 2ms/step - loss: 0.3009 - val_loss: 0.3598\n",
     "output_type": "stream"
    },
    {
     "output_type": "execute_result",
     "execution_count": 99,
     "data": {
      "text/plain": "RandomizedSearchCV(cv=3,\n                   estimator=<keras.wrappers.scikit_learn.KerasRegressor object at 0x7f3b264ee3d0>,\n                   param_distributions={'learning_rate': [0.001683454924600351,\n                                                          0.02390836445593178,\n                                                          0.008731907739399206,\n                                                          0.004725396149933917,\n                                                          0.0006154014789262348,\n                                                          0.0006153331256530192,\n                                                          0.0003920021771415983,\n                                                          0.01619845322936229,\n                                                          0.004779156784872302,\n                                                          0.007821074275112298,...\n                                                          0.005021425736625637,\n                                                          0.0005703073595961105,\n                                                          0.001151888789941251,\n                                                          0.001621231156394198,\n                                                          0.0024505367684280487,\n                                                          0.011155092541719619,\n                                                          0.0007524347058135697,\n                                                          0.0032032448128444043,\n                                                          0.004591455636549438,\n                                                          0.0003715541189658278, ...],\n                                        'n_hidden': [0, 1, 2, 3],\n                                        'n_neurons': [1, 2, 3, 4, 5, 6, 7, 8, 9,\n                                                      10, 11, 12, 13, 14, 15,\n                                                      16, 17, 18, 19, 20, 21,\n                                                      22, 23, 24, 25, 26, 27,\n                                                      28, 29, 30, ...]},\n                   verbose=2)",
      "text/html": "<style>#sk-container-id-1 {color: black;background-color: white;}#sk-container-id-1 pre{padding: 0;}#sk-container-id-1 div.sk-toggleable {background-color: white;}#sk-container-id-1 label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.3em;box-sizing: border-box;text-align: center;}#sk-container-id-1 label.sk-toggleable__label-arrow:before {content: \"▸\";float: left;margin-right: 0.25em;color: #696969;}#sk-container-id-1 label.sk-toggleable__label-arrow:hover:before {color: black;}#sk-container-id-1 div.sk-estimator:hover label.sk-toggleable__label-arrow:before {color: black;}#sk-container-id-1 div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}#sk-container-id-1 div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}#sk-container-id-1 input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}#sk-container-id-1 input.sk-toggleable__control:checked~label.sk-toggleable__label-arrow:before {content: \"▾\";}#sk-container-id-1 div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-1 div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-1 input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}#sk-container-id-1 div.sk-estimator {font-family: monospace;background-color: #f0f8ff;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;margin-bottom: 0.5em;}#sk-container-id-1 div.sk-estimator:hover {background-color: #d4ebff;}#sk-container-id-1 div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}#sk-container-id-1 div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-1 div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: 0;}#sk-container-id-1 div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;padding-right: 0.2em;padding-left: 0.2em;position: relative;}#sk-container-id-1 div.sk-item {position: relative;z-index: 1;}#sk-container-id-1 div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;position: relative;}#sk-container-id-1 div.sk-item::before, #sk-container-id-1 div.sk-parallel-item::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: -1;}#sk-container-id-1 div.sk-parallel-item {display: flex;flex-direction: column;z-index: 1;position: relative;background-color: white;}#sk-container-id-1 div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}#sk-container-id-1 div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}#sk-container-id-1 div.sk-parallel-item:only-child::after {width: 0;}#sk-container-id-1 div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0 0.4em 0.5em 0.4em;box-sizing: border-box;padding-bottom: 0.4em;background-color: white;}#sk-container-id-1 div.sk-label label {font-family: monospace;font-weight: bold;display: inline-block;line-height: 1.2em;}#sk-container-id-1 div.sk-label-container {text-align: center;}#sk-container-id-1 div.sk-container {/* jupyter's `normalize.less` sets `[hidden] { display: none; }` but bootstrap.min.css set `[hidden] { display: none !important; }` so we also need the `!important` here to be able to override the default hidden behavior on the sphinx rendered scikit-learn.org. See: https://github.com/scikit-learn/scikit-learn/issues/21755 */display: inline-block !important;position: relative;}#sk-container-id-1 div.sk-text-repr-fallback {display: none;}</style><div id=\"sk-container-id-1\" class=\"sk-top-container\"><div class=\"sk-text-repr-fallback\"><pre>RandomizedSearchCV(cv=3,\n                   estimator=&lt;keras.wrappers.scikit_learn.KerasRegressor object at 0x7f3b264ee3d0&gt;,\n                   param_distributions={&#x27;learning_rate&#x27;: [0.001683454924600351,\n                                                          0.02390836445593178,\n                                                          0.008731907739399206,\n                                                          0.004725396149933917,\n                                                          0.0006154014789262348,\n                                                          0.0006153331256530192,\n                                                          0.0003920021771415983,\n                                                          0.01619845322936229,\n                                                          0.004779156784872302,\n                                                          0.007821074275112298,...\n                                                          0.005021425736625637,\n                                                          0.0005703073595961105,\n                                                          0.001151888789941251,\n                                                          0.001621231156394198,\n                                                          0.0024505367684280487,\n                                                          0.011155092541719619,\n                                                          0.0007524347058135697,\n                                                          0.0032032448128444043,\n                                                          0.004591455636549438,\n                                                          0.0003715541189658278, ...],\n                                        &#x27;n_hidden&#x27;: [0, 1, 2, 3],\n                                        &#x27;n_neurons&#x27;: [1, 2, 3, 4, 5, 6, 7, 8, 9,\n                                                      10, 11, 12, 13, 14, 15,\n                                                      16, 17, 18, 19, 20, 21,\n                                                      22, 23, 24, 25, 26, 27,\n                                                      28, 29, 30, ...]},\n                   verbose=2)</pre><b>In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. <br />On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.</b></div><div class=\"sk-container\" hidden><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-1\" type=\"checkbox\" ><label for=\"sk-estimator-id-1\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">RandomizedSearchCV</label><div class=\"sk-toggleable__content\"><pre>RandomizedSearchCV(cv=3,\n                   estimator=&lt;keras.wrappers.scikit_learn.KerasRegressor object at 0x7f3b264ee3d0&gt;,\n                   param_distributions={&#x27;learning_rate&#x27;: [0.001683454924600351,\n                                                          0.02390836445593178,\n                                                          0.008731907739399206,\n                                                          0.004725396149933917,\n                                                          0.0006154014789262348,\n                                                          0.0006153331256530192,\n                                                          0.0003920021771415983,\n                                                          0.01619845322936229,\n                                                          0.004779156784872302,\n                                                          0.007821074275112298,...\n                                                          0.005021425736625637,\n                                                          0.0005703073595961105,\n                                                          0.001151888789941251,\n                                                          0.001621231156394198,\n                                                          0.0024505367684280487,\n                                                          0.011155092541719619,\n                                                          0.0007524347058135697,\n                                                          0.0032032448128444043,\n                                                          0.004591455636549438,\n                                                          0.0003715541189658278, ...],\n                                        &#x27;n_hidden&#x27;: [0, 1, 2, 3],\n                                        &#x27;n_neurons&#x27;: [1, 2, 3, 4, 5, 6, 7, 8, 9,\n                                                      10, 11, 12, 13, 14, 15,\n                                                      16, 17, 18, 19, 20, 21,\n                                                      22, 23, 24, 25, 26, 27,\n                                                      28, 29, 30, ...]},\n                   verbose=2)</pre></div></div></div><div class=\"sk-parallel\"><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-2\" type=\"checkbox\" ><label for=\"sk-estimator-id-2\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">estimator: KerasRegressor</label><div class=\"sk-toggleable__content\"><pre>&lt;keras.wrappers.scikit_learn.KerasRegressor object at 0x7f3b264ee3d0&gt;</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-3\" type=\"checkbox\" ><label for=\"sk-estimator-id-3\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">KerasRegressor</label><div class=\"sk-toggleable__content\"><pre>&lt;keras.wrappers.scikit_learn.KerasRegressor object at 0x7f3b264ee3d0&gt;</pre></div></div></div></div></div></div></div></div></div></div>"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00134-e5756b93-d404-47ce-b13e-5dd25bb6f54c",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "9a3880fe",
    "execution_start": 1659187611989,
    "execution_millis": 134,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 113.1875,
    "deepnote_output_heights": [
     21.1875
    ]
   },
   "source": "rnd_search_cv.best_params_",
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "execution_count": 100,
     "data": {
      "text/plain": "{'n_neurons': 80, 'n_hidden': 3, 'learning_rate': 0.0059640580092043885}"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00135-87a01c03-249b-4033-9e5c-5327fa45d185",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "65a033a1",
    "execution_start": 1659187612042,
    "execution_millis": 100,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 113.1875,
    "deepnote_output_heights": [
     21.1875
    ]
   },
   "source": "rnd_search_cv.best_score_",
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "execution_count": 101,
     "data": {
      "text/plain": "-0.3174845178922017"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00136-5d899128-71e3-4f98-9dc3-0da481bc1a28",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "13f8bf78",
    "execution_start": 1659187612043,
    "execution_millis": 1519370,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 113.1875,
    "deepnote_output_heights": [
     21.1875
    ]
   },
   "source": "rnd_search_cv.best_estimator_",
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "execution_count": 102,
     "data": {
      "text/plain": "<keras.wrappers.scikit_learn.KerasRegressor at 0x7f3b24947c40>"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00137-4c0cd004-003e-4cfd-bf80-7dfa0827804b",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "36c7ba14",
    "execution_start": 1659187612043,
    "execution_millis": 3733,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 159.375,
    "deepnote_output_heights": [
     null,
     21.1875
    ]
   },
   "source": "rnd_search_cv.score(X_test, y_test)",
   "execution_count": null,
   "outputs": [
    {
     "name": "stdout",
     "text": "162/162 [==============================] - 0s 1ms/step - loss: 0.3066\n",
     "output_type": "stream"
    },
    {
     "output_type": "execute_result",
     "execution_count": 103,
     "data": {
      "text/plain": "-0.3066314160823822"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00138-15d4de2b-0da0-4e00-a519-fd7ed34475a9",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "302b02e6",
    "execution_start": 1659187615782,
    "execution_millis": 5,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 131.1875,
    "deepnote_output_heights": [
     21.1875
    ]
   },
   "source": "model = rnd_search_cv.best_estimator_.model\nmodel",
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "execution_count": 104,
     "data": {
      "text/plain": "<keras.engine.sequential.Sequential at 0x7f3b24947e50>"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "scrolled": true,
    "cell_id": "00139-137c49e3-a706-44fd-902a-9eb9d6178b31",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "ef7ab4e3",
    "execution_start": 1659187615793,
    "execution_millis": 356,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 159.375,
    "deepnote_output_heights": [
     null,
     21.1875
    ]
   },
   "source": "model.evaluate(X_test, y_test)",
   "execution_count": null,
   "outputs": [
    {
     "name": "stdout",
     "text": "162/162 [==============================] - 0s 2ms/step - loss: 0.3066\n",
     "output_type": "stream"
    },
    {
     "output_type": "execute_result",
     "execution_count": 105,
     "data": {
      "text/plain": "0.3066314160823822"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": "# Exercise solutions",
   "metadata": {
    "cell_id": "00140-81a3f10f-0c40-4df3-8943-e21d6f98f7b9",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 82
   }
  },
  {
   "cell_type": "markdown",
   "source": "## 1. to 9.",
   "metadata": {
    "cell_id": "00141-93db5695-9309-446a-bd3a-40457db7cee9",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 70
   }
  },
  {
   "cell_type": "markdown",
   "source": "See appendix A.",
   "metadata": {
    "cell_id": "00142-b2c257c0-516e-48d7-a00b-f2afa10fa269",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 52.390625
   }
  },
  {
   "cell_type": "markdown",
   "source": "1. The [TensorFlow Playground](https://playground.tensorflow.org/) is a handy neural network simulator built by the TensorFlow team. In this exercise, you will train several binary classifiers in just a few clicks, and tweak the model’s architecture and its hyperparameters to gain some intuition on how neural networks work and what their hyperparameters do. Take some time to explore the following:\n\n    - The patterns learned by a neural net. Try training the default neural network by clicking the Run button (top left). Notice how it quickly finds a good solution for the classification task. The neurons in the first hidden layer have learned simple patterns, while the neurons in the second hidden layer have learned to combine the simple patterns of the first hidden layer into more complex patterns. In general, the more layers there are, the more complex the patterns can be. \n    - Activation functions. Try replacing the tanh activation function with a ReLU activation function, and train the network again. Notice that it finds a solution even faster, but this time the boundaries are linear. This is due to the shape of the ReLU function.\n    - The risk of local minima. Modify the network architecture to have just one hidden layer with three neurons. Train it multiple times (to reset the network weights, click the Reset button next to the Play button). Notice that the training time varies a lot, and sometimes it even gets stuck in a local minimum.\n    - What happens when neural nets are too small. Remove one neuron to keep just two. Notice that the neural network is now incapable of finding a good solution, even if you try multiple times. The model has too few parameters and systematically underfits the training set.\n    - What happens when neural nets are large enough. Set the number of neurons to eight, and train the network several times. Notice that it is now consistently fast and never gets stuck. This highlights an important finding in neural network theory: large neural networks almost never get stuck in local minima, and even when they do these local optima are almost as good as the global optimum. However, they can still get stuck on long plateaus for a long time.\n    - The risk of vanishing gradients in deep networks. Select the spiral dataset (the bottom-right dataset under “DATA”), and change the network architecture to have four hidden layers with eight neurons each. Notice that training takes much longer and often gets stuck on plateaus for long periods of time. Also notice that the neurons in the highest layers (on the right) tend to evolve faster than the  neurons in the lowest layers (on the left).  This problem, called the “vanishing gradients” problem, can be alleviated with better weight initialization and other techniques, better optimizers (such as AdaGrad or Adam), or Batch Normalization (discussed in Chapter 11).\n    - Go further. Take an hour or so to play around with other parameters and get a feel for what they do, to build an intuitive understanding about neural networks.\n",
   "metadata": {
    "cell_id": "d341b6800c874e418fd79074dc430e7b",
    "tags": [],
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 683.546875
   }
  },
  {
   "cell_type": "markdown",
   "source": "Visit the [TensorFlow Playground](https://playground.tensorflow.org/) and play around with it, as described in this exercise.\n",
   "metadata": {
    "cell_id": "89fab1eb5a6b4481bf13e1cd726ce02c",
    "tags": [],
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 52.390625
   }
  },
  {
   "cell_type": "markdown",
   "source": "2. Draw an ANN using the original artificial neurons (like the ones in Figure 10-3) that computes $A \\oplus B$ (where $\\oplus$ represents the XOR operation). Hint: $A \\oplus B = (A \\wedge \\neg B) \\vee (\\neg A \\wedge B).$",
   "metadata": {
    "cell_id": "7653f9ce64484bed915272ed9d24d903",
    "tags": [],
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 91.78125
   }
  },
  {
   "cell_type": "markdown",
   "source": "Here is a neural network based on the original artificial neurons that computes $A \\oplus B$ (where $\\oplus$ represents the exclusive OR), using the fact that $A \\oplus B = (A \\wedge \\neg B) \\vee (\\neg A \\wedge B).$ There are other solutions—for example, using the fact that $A \\oplus B = (A \\vee B) \\wedge \\neg(A\n\\wedge B$), or the fact that $A \\oplus B = (A \\vee B) \\wedge (\\neg A \\vee \\wedge B)$, and so on.\n\n![image](/work/images/ann/2.jpg)\n",
   "metadata": {
    "cell_id": "7e4c4ffb52c646e78acaaf2df02f0c2c",
    "tags": [],
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 137.171875
   }
  },
  {
   "cell_type": "markdown",
   "source": "3. Why is it generally preferable to use a Logistic Regression classifier rather than a classical Perceptron (i.e., a single layer of threshold logic units trained using the Perceptron training algorithm)? How can you tweak a Perceptron to make it equivalent to a Logistic Regression classifier?",
   "metadata": {
    "cell_id": "bde2e196047245abb084a39e7d1334f5",
    "tags": [],
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 114.171875
   }
  },
  {
   "cell_type": "markdown",
   "source": "A classical Perceptron will converge only if the dataset is linearly separable, and it won’t be able to estimate class probabilities. In contrast, a Logistic Regression classifier will converge to a good solution even if the dataset is not linearly separable, and it will output class probabilities. If you change the Perceptron’s activation function to the logistic activation function (or the softmax activation function if there are multiple neurons), and if you train it using Gradient Descent (or some other optimization algorithm minimizing the cost function, typically cross entropy), then it becomes equivalent to a Logistic Regression classifier.",
   "metadata": {
    "cell_id": "dbb113b1a59044a380a97767a157e8ff",
    "tags": [],
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 141.953125
   }
  },
  {
   "cell_type": "markdown",
   "source": "4. Why was the logistic activation function a key ingredient in training the first MLPs?\n",
   "metadata": {
    "cell_id": "ea17ab12e4ee49c881d4acf29a688ad7",
    "tags": [],
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 69.390625
   }
  },
  {
   "cell_type": "markdown",
   "source": "The logistic activation function was a key ingredient in training the first MLPs because its derivative is always nonzero, so Gradient Descent can always roll down the slope. When the activation function is a step function, Gradient Descent cannot move, as there is no slope at all.\n\n",
   "metadata": {
    "cell_id": "86c1de97e4b240b097c92b91221cf66f",
    "tags": [],
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 97.171875
   }
  },
  {
   "cell_type": "markdown",
   "source": "5. Name three popular activation functions. Can you draw them?",
   "metadata": {
    "cell_id": "3397ae521fc64700bea0ad960eaa8ad7",
    "tags": [],
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 69.390625
   }
  },
  {
   "cell_type": "markdown",
   "source": "Popular activation functions include the step function, the logistic (sigmoid) function, the hyperbolic tangent (tanh) function, and the Rectified Linear Unit (ReLU) function (see Figure 10-8). See Chapter 11 for other examples, such as ELU and variants of the ReLU function",
   "metadata": {
    "cell_id": "32b1d16bf157458997f588b5c4f05771",
    "tags": [],
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 97.171875
   }
  },
  {
   "cell_type": "markdown",
   "source": "6. Suppose you have an MLP composed of one input layer with 10 passthrough neurons, followed by one hidden layer with 50 artificial neurons, and finally one output layer with 3 artificial neurons. All artificial neurons use the ReLU activation function.\n\n    - What is the shape of the input matrix $\\mathbf{X}$?\n    - What are the shapes of the hidden layer’s weight vector $\\mathbf{W}_h$ and its bias vector $\\mathbf{b}_h$ ?\n    - What are the shapes of the output layer’s weight vector $\\mathbf{W}_o$ and its bias vector $\\mathbf{b}_o$ ?\n    - What is the shape of the network’s output matrix $\\mathbf{Y}$?\n    - Write the equation that computes the network’s output matrix $\\mathbf{Y}$ as a function of $\\mathbf{X}, \\mathbf{W}_h , \\mathbf{b}_h , \\mathbf{W}_o , \\text{and}\\, \\mathbf{b}_o .$",
   "metadata": {
    "cell_id": "2e09338efd0340c1a666b7e203628997",
    "tags": [],
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 229.734375
   }
  },
  {
   "cell_type": "markdown",
   "source": "Considering the MLP described in the question, composed of one input layer with 10 passthrough neurons, followed by one hidden layer with 50 artificial neurons, and finally one output layer with 3 artificial neurons, where all artificial neurons use the ReLU activation function: \n\n- The shape of the input matrix $\\mathbf{X}$ is $m \\times 10$, where $m$ represents the training batch size.\n- The shape of the hidden layer's weight vector $\\mathbf{W}_h$ is $10 \\times 50$, and the length of its bias vector $\\mathbf{b}_h$ is 50.\n- The shape of the output layer’s weight vector $\\mathbf{W}_o$ is $50 \\times 3$, and the length of its bias vector $\\mathbf{b}_o$ is 3.\n- The shape of the network’s output matrix $\\mathbf{Y}$ is $m \\times 3$.\n- $\\mathbf{Y}* = ReLU(ReLU(\\mathbf{X}\\mathbf{W}_h + \\mathbf{b}_h)\\mathbf{W}_o + \\mathbf{b}_o)$. Recall that the ReLU function just sets every negative number in the matrix to zero. Also note that when you are adding a bias vector to a matrix, it is added to every single row in the matrix, which is called broadcasting.\n",
   "metadata": {
    "cell_id": "daa50fd88b8c4654aa044ee058815407",
    "tags": [],
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 293.90625
   }
  },
  {
   "cell_type": "markdown",
   "source": "7. How many neurons do you need in the output layer if you want to classify email into spam or ham? What activation function should you use in the output layer? If instead you want to tackle MNIST, how many neurons do you need in the output layer, and which activation function should you use? What about for getting your network to predict housing prices, as in Chapter 2?",
   "metadata": {
    "cell_id": "c582a09175524a428b17ccac2e51046a",
    "tags": [],
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 114.171875
   }
  },
  {
   "cell_type": "markdown",
   "source": "To classify email into spam or ham, you just need one neuron in the output layer of a neural network—for example, indicating the probability that the email is spam. You would typically use the logistic activation function in the output layer when estimating a probability. If instead you want to tackle MNIST, you need 10 neurons in the output layer, and you must replace the logistic function with the softmax activation function, which can handle multiple classes, outputting one probability per class. If you want your neural network to predict housing prices like in Chapter 2, then you need one output neuron, using no activation function at all in the output layer.",
   "metadata": {
    "cell_id": "c61e467022f74ca1ad45a3e0781b0e72",
    "tags": [],
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 164.34375
   }
  },
  {
   "cell_type": "markdown",
   "source": "8. What is backpropagation and how does it work? What is the difference between backpropagation and reverse-mode autodiff?",
   "metadata": {
    "cell_id": "155ad8f300eb45ceaf067e34192cb9da",
    "tags": [],
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 69.390625
   }
  },
  {
   "cell_type": "markdown",
   "source": "Backpropagation is a technique used to train artificial neural networks. It first computes the gradients of the cost function with regard to every model parameter (all the weights and biases), then it performs a Gradient Descent step using these gradients. This backpropagation step is typically performed thousands or millions of times, using many training batches, until the model parameters converge to values that (hopefully) minimize the cost function. To compute the gradients,  backpropagation uses reverse-mode autodiff (although it wasn’t called that when backpropagation was invented, and it has been reinvented several times). Reverse-mode\nautodiff performs a forward pass through a computation graph, computing every node’s value for the current training batch, and then it performs a reverse pass, computing all the gradients at once (see Appendix D for more details). So what’s the difference? Well, backpropagation refers to the whole process of training an artificial neural network using multiple backpropagation steps, each of which computes gradients and uses them to perform a Gradient Descent step. In contrast, reverse-mode autodiff is just a technique to compute gradients efficiently, and it happens to be used by backpropagation.",
   "metadata": {
    "cell_id": "1ad16405d71c4c14876cf29375e49ea6",
    "tags": [],
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 253.90625
   }
  },
  {
   "cell_type": "markdown",
   "source": "9. Can you list all the hyperparameters you can tweak in a basic MLP? If the MLP overfits the training data, how could you tweak these hyperparameters to try to solve the problem?",
   "metadata": {
    "cell_id": "d1649d234fbb438c8343413302ca4eab",
    "tags": [],
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 91.78125
   }
  },
  {
   "cell_type": "markdown",
   "source": "Here is a list of all the hyperparameters you can tweak in a basic MLP: the number of hidden layers, the number of neurons in each hidden layer, and the activation function used in each hidden layer and in the output layer. In general, the ReLU activation function\n(or one of its variants; see Chapter 11) is a good default for the hidden layers. For the output layer, in general you will want the logistic activation function for binary classification, the softmax activation function for multiclass classification, or no activation\nfunction for regression. If the MLP overfits the training data, you can try reducing the number of hidden layers and reducing the number of neurons per hidden layer.",
   "metadata": {
    "cell_id": "1d36519f415647e98333f0ae61c02689",
    "tags": [],
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 164.34375
   }
  },
  {
   "cell_type": "markdown",
   "source": "10. Train a deep MLP on the MNIST dataset (you can load it using `keras.datasets.mnist.load_data()`. See if you can get over 98% precision. Try searching for the optimal learning rate by using the approach presented in this chapter (i.e., by growing the learning rate exponentially, plotting the error, and finding the point where the error shoots up). Try adding all the bells and whistles—save checkpoints, use early stopping, and plot learning curves using TensorBoard.",
   "metadata": {
    "cell_id": "874e6ec50eee4521bb129a728a1a298f",
    "tags": [],
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 136.5625
   }
  },
  {
   "cell_type": "markdown",
   "source": "## 10.",
   "metadata": {
    "cell_id": "00143-7729f964-b5b7-489c-a1dc-81fff0ee3ecc",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 70
   }
  },
  {
   "cell_type": "markdown",
   "source": "*Exercise: Train a deep MLP on the MNIST dataset (you can load it using `keras.datasets.mnist.load_data()`. See if you can get over 98% precision. Try searching for the optimal learning rate by using the approach presented in this chapter (i.e., by growing the learning rate exponentially, plotting the loss, and finding the point where the loss shoots up). Try adding all the bells and whistles—save checkpoints, use early stopping, and plot learning curves using TensorBoard.*",
   "metadata": {
    "cell_id": "00144-ebd60ac2-e4b8-4650-882a-22735a1de643",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 119.5625
   }
  },
  {
   "cell_type": "markdown",
   "source": "Let's load the dataset:",
   "metadata": {
    "cell_id": "00145-316d3c75-98e4-4b2a-9e98-09b3cc1166b1",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 52.390625
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00146-1876e957-2728-45d3-acfe-170cdd54b548",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "487f7870",
    "execution_start": 1659187616150,
    "execution_millis": 284,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 76
   },
   "source": "(X_train_full, y_train_full), (X_test, y_test) = keras.datasets.mnist.load_data()",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "markdown",
   "source": "Just like for the Fashion MNIST dataset, the MNIST training set contains 60,000 grayscale images, each 28x28 pixels:",
   "metadata": {
    "cell_id": "00147-6a45de61-68c1-4a45-a943-35be21f84829",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 52.390625
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00148-597345e9-38f4-4a19-9784-c0d4501f2c38",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "3b45a169",
    "execution_start": 1659187616437,
    "execution_millis": 8,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 113.1875,
    "deepnote_output_heights": [
     21.1875
    ]
   },
   "source": "X_train_full.shape",
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "execution_count": 107,
     "data": {
      "text/plain": "(60000, 28, 28)"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": "Each pixel intensity is also represented as a byte (0 to 255):",
   "metadata": {
    "cell_id": "00149-f6412f9c-7559-4471-9ec2-c087bfdde977",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 52.390625
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00150-9271bf97-6e14-4bc4-8a1a-ddeb04c223a5",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "32a91791",
    "execution_start": 1659187616451,
    "execution_millis": 1522671,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 113.1875
   },
   "source": "X_train_full.dtype",
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "execution_count": 108,
     "data": {
      "text/plain": "dtype('uint8')"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": "Let's split the full training set into a validation set and a (smaller) training set. We also scale the pixel intensities down to the 0-1 range and convert them to floats, by dividing by 255, just like we did for Fashion MNIST:",
   "metadata": {
    "cell_id": "00151-56b7f64f-9630-4133-8322-d6e2122e64b1",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 74.78125
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00152-2df54109-731d-4dcd-b42c-87a5c20d798c",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "364f401f",
    "execution_start": 1659187616458,
    "execution_millis": 141,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 112
   },
   "source": "X_valid, X_train = X_train_full[:5000] / 255., X_train_full[5000:] / 255.\ny_valid, y_train = y_train_full[:5000], y_train_full[5000:]\nX_test = X_test / 255.",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "markdown",
   "source": "Let's plot an image using Matplotlib's `imshow()` function, with a `'binary'`\n color map:",
   "metadata": {
    "cell_id": "00153-1850e05e-5c7c-4fb7-a6d9-807719fb33f8",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 74.78125
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00154-8224bcae-3dc9-413b-add1-8ed0ec6c38d7",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "c03cb47f",
    "execution_start": 1659187616606,
    "execution_millis": 116,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 361,
    "deepnote_output_heights": [
     233
    ]
   },
   "source": "plt.imshow(X_train[0], cmap=\"binary\")\nplt.axis('off')\nplt.show()",
   "execution_count": null,
   "outputs": [
    {
     "data": {
      "text/plain": "<Figure size 432x288 with 1 Axes>",
      "image/png": "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\n"
     },
     "metadata": {
      "needs_background": "light",
      "image/png": {
       "width": 231,
       "height": 231
      }
     },
     "output_type": "display_data"
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": "The labels are the class IDs (represented as uint8), from 0 to 9. Conveniently, the class IDs correspond to the digits represented in the images, so we don't need a `class_names` array:",
   "metadata": {
    "cell_id": "00155-d07aa8bd-5f94-4998-b9d0-1657af110fbb",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 74.78125
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00156-3f911ff8-e16b-40aa-9d4e-e6a9a8f2508e",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "ed857f47",
    "execution_start": 1659187616718,
    "execution_millis": 5,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 113.1875
   },
   "source": "y_train",
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "execution_count": 111,
     "data": {
      "text/plain": "array([7, 3, 4, ..., 5, 6, 8], dtype=uint8)"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": "The validation set contains 5,000 images, and the test set contains 10,000 images:",
   "metadata": {
    "cell_id": "00157-a02fb737-60b1-447b-bd94-038ef4551413",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 52.390625
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00158-54fd8a08-ebf6-497b-97a6-13cd3833584b",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "ffc02169",
    "execution_start": 1659187616719,
    "execution_millis": 1,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 113.1875
   },
   "source": "X_valid.shape",
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "execution_count": 112,
     "data": {
      "text/plain": "(5000, 28, 28)"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00159-50dc13e4-4fbe-4b63-b407-b95e3404f072",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "251bcee5",
    "execution_start": 1659187616720,
    "execution_millis": 6,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 113.1875
   },
   "source": "X_test.shape",
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "execution_count": 113,
     "data": {
      "text/plain": "(10000, 28, 28)"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": "Let's take a look at a sample of the images in the dataset:",
   "metadata": {
    "cell_id": "00160-04f40ae9-8c4a-47ff-8bf4-7dfbd875ebd1",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 52.390625
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00161-448ee193-bb07-4d4a-8067-5d6998d69383",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "5c688ef5",
    "execution_start": 1659187616785,
    "execution_millis": 1285,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 583,
    "deepnote_output_heights": [
     293
    ]
   },
   "source": "n_rows = 4\nn_cols = 10\nplt.figure(figsize=(n_cols * 1.2, n_rows * 1.2))\nfor row in range(n_rows):\n    for col in range(n_cols):\n        index = n_cols * row + col\n        plt.subplot(n_rows, n_cols, index + 1)\n        plt.imshow(X_train[index], cmap=\"binary\", interpolation=\"nearest\")\n        plt.axis('off')\n        plt.title(y_train[index], fontsize=12)\nplt.subplots_adjust(wspace=0.2, hspace=0.5)\nplt.show()",
   "execution_count": null,
   "outputs": [
    {
     "data": {
      "text/plain": "<Figure size 864x345.6 with 40 Axes>",
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAqIAAAEjCAYAAADpBWMTAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8qNh9FAAAACXBIWXMAAAsTAAALEwEAmpwYAABjbUlEQVR4nO3debxN1fvA8c8KmYVMIUOlJIr0/TV8EyXRRFK+GpAmpZTSJDSZmzRLSUSRCpVKs9C3gQqV+VvJVIaSIRTt3x/bs/Y59547n3PWPuc879fL617nnnvuuuvus/faz3rWs4zneSillFJKKZVs+7lugFJKKaWUykw6EFVKKaWUUk7oQFQppZRSSjmhA1GllFJKKeWEDkSVUkoppZQTOhBVSimllFJO6EBUKaWUUko5kfSBqDFme5Z/e40xjye7HWFijJlojFlvjNlqjFlujLnSdZvCwBjTwBizyxgz0XVbXDPGdDHGLDHG7DDG/M8Y08J1m1wxxlxvjJlvjNltjBnnuj2uGWPqGWPeNsb8boz5xRjzhDGmuOt2uaLHR3bGmMrGmGn7zh+rjDEXu26TS8aYI40xHxlj/jDGrDTGdHTdJpdcn0OSPhD1PK+c/ANqADuBV5LdjpAZBtTzPK8C0B4YbIxp7rhNYfAkMM91I1wzxrQBRgA9gPLAKcAPThvl1jpgMDDWdUNC4ilgA3AQ0BRoCfRy2SDH9PjI7kngL6A6cAkwyhhzlNsmubFvgPU6MAOoDFwNTDTGHO60YW45PYe4nprvhP/Lz3HcDqc8z/ve87zd8t99/w512CTnjDFdgC3Ah46bEgb3Avd5nve553n/eJ631vO8ta4b5YrneVM9z5sObHbdlpCoD0zxPG+X53m/ADOBjBxkgB4fWRljyuJfawd6nrfd87y5wBtAV7ctc6YhUBMY6XneXs/zPgI+JXP7AxyfQ1wPRLsDL3i6zyjGmKeMMX8CS4H1wNuOm+SMMaYCcB9ws+u2uGaMKQYcB1TdN4W0Zt+0SWnXbVOh8QjQxRhTxhhTCzgT/0KiFMDhwB7P85ZHPLaQDL5ZicEAjV03wqFHcHgOcTYQNcbUxQ//jnfVhjDxPK8X/rRrC2AqsDv370hrg4DnPM9b47ohIVAdKAFcgH9sNAWaAQMctkmFy2z8QcVWYA0wH5juskEqVMrhHxuR/sC/3mSiZfgzsbcaY0oYY87AH4uUcdssp5yeQ1xGRLsCcz3P+9FhG0Jl3zTBXKA2cK3r9rhgjGkKnA6MdNyUsNi57+Pjnuet9zxvE/AwcJbDNqmQMMbshx+5mAqUBaoAlfBzipUC2A5UyPJYBWCbg7Y453ne38B5wNnAL0BfYAr+ACzjhOEc4nIg2g2NhuakOJmbI9oKqAf8bIz5BbgF6GSM+dplo1zxPO93/BNkZPpKxqeyKKsyUAd4wvO83Z7nbQaeR29UVGA5UNwY0yDisWOA7x21xznP8xZ5ntfS87wDPc9rCxwCfOm6XY44P4c4GYgaY04CaqGr5THGVNtXmqecMaaYMaYtcBGZu0jnGfxBeNN9/54G3gLaumuSc88DvfcdK5WAm/BXfGYkY0xxY0wpoBhQzBhTKlPLFe2LkP8IXLuvXyri594vctowh/T4iOZ53g78aNd9xpiyxph/Ax2ACW5b5o4x5uh9x0UZY8wt+KvFxzlulhNhOIe4ioh2B6Z6npeRUwNZePjT8GuA34EHgT6e573htFWOeJ73p+d5v8g//GmlXZ7nbXTdNocG4ZexWg4sAb4BhjhtkVsD8FMW7gAu3fd5JufMng+0AzYCK4G/8W9WMpUeH9n1Akrj50ZOAq71PC9jI6L4qYHr8fujNdAmonJNJnJ6DjG6YF0ppZRSSrngunyTUkoppZTKUDoQVUoppZRSTuhAVCmllFJKOaEDUaWUUkop5YQORJVSSimllBN51VZL9SX1Js6vp/0RTfsjmvZHdton0bQ/oml/RNP+iKb9ES0t+0MjokoppZRSygkdiCqllFJKKSd0IKqUUkoppZzI2P13lUpV//zzD3379gXgiSeeAOCzzz4D4LjjjnPWLqWUUqqgNCKqlFJKKaWc0IioUiliw4YNAAwcOJBnnnkm6ms//vgjkHkR0auuugqAiRMn8umnnwJw7LHHumySCqH77ruPyZMnAzBjxgwADjnkEJdNSqrFixcD8MgjjwDw7LPP0rNnTwCefvppV81SIbBhwwYWLlwIwOuvvw7A7Nmz+e677wDo0aMHAIceeigAffv2pWTJklGv8dtvv1G5cuVCt0EjokoppZRSygmNiIbAqlWrAP8uFWDIkCEY45fb8jy/bNiRRx4JwODBgzn//PMdtFK5sn79egDuv/9+gKhoaIsWLQA4/vjjk9+wEKhbty4Au3btYsWKFYBGRAHmzp3L6NGjAT9anJUcN3Iu6datW5EiGmG1efNmwD+3rlmzBoCvv/4ayJyI6Pjx4xk4cCCA7QNjDG+//XbM50+cOJEOHToAUL58+eQ0UiXdmDFjABg6dKgdgwjP8+wYZNy4cVFfK126NDfddFPUYxdddBHvvvtuoduiA1FHNm7cCMCwYcN48cUXAdi0aRPgnyTkIBDLli0D/LD4KaecAkCVKlWS1dyE+euvvwBo3bo14F9ARcWKFQFYtGgRBx98cNLbFgZ79uxhyJAhADz55JP28euuuw6Ahx9+GID9998/+Y0LARmIgn/BBfjPf/7jqjnO7NmzB4B77rkH8I+VP/74AyDbuQRgzpw5QPB+W7BgQbYLTjqQY0IGYJng77//BrADg6uvvto+lptRo0YBcMMNN1C/fn0ABg0aBKTXe+p///ufTVGQdJ4lS5bYFIXu3bu7alpSyKBz6NChUf8Hf5AJUK5cOXvekHHJP//8A8Att9zCAQccAMDll18OwLp164rUJp2aV0oppZRSTiQtIvr8888D/t35gQceCPh3IQAnnniinSpKd4MHDwawUyXGGDv9LncgderUoWrVqlHfJ3clP/30k42ISgJ6KpJI6BVXXAFER0LPO+88AO644w4Aatasmetr/frrrwBUr1493s10rl+/flGRUICePXvask0qkKlRYYD+/fsD8MADDwDRU2tZnXLKKXzyySdRj7333nts27YNSK/p2FmzZrluQtLJLEm/fv1yfE7Dhg258cYbox6Ta8zevXtZuXIlANdcc439eqpGRSUa/PLLLwN+xFPOFfK+mT9/fsZEROUcIZHQ/fffnwsvvBDATrk3a9bMPn/KlCkADB8+HICFCxeya9euqNfM6xqdF42IKqWUUkopJwocEX3ppZcA+OabbwAYO3Zsvr5vy5YtwQ8t7v9YiYqVKlWKMmXKAHD00UcDwSg8a2Qw1Ul5BIlWREYtGjVqBPh38VnzPyWnq2XLljZfNJU99NBDQPaFFNdddx0PPvgg4B8Xeenbt6+Ntt91110A9OnTJ44tdePuu+8GsH0BcP311wNBxEPBtGnT7OcXXXSRw5Ykn+SF9u/fP9sxUbZsWW6++WYAOnbsCPgzLQAVKlSwuV2Sn16lShV7Xk4HMsMiOYCZQCJ/UoonFsm1f+aZZzj55JPzfE3JM+7Zsyfz588Hgoha2Mn4QmYfZbHnUUcdxciRIwFo06YN4OcQr169GgiutZIvmW4l8SZNmhT1/5NPPpkXXnghx+d37twZgGrVqgHBeo5IsritsPJ95pGT2qOPPgoEiauFIQeI2LVrlw31ylSKTANMmjQpLaZcJQ1h6dKlQHBRqFq1qh10ysVkwIAB3HnnnVHPk9QFmcaHYPX01Vdfnejmx9V3331nk+CFTAc+8sgj+bogzps3D/BX9P3+++/xb6Qjn3/+OQCPP/64fUzq/cl7b7/9dCJDboTfeustwB9ItW/f3mWTkk4GkZEDgyOOOALwb+SbNGmS4/dmTWM47LDD7IU3Hfz2229RH9Pd3r177XEg9VIjSTrXa6+9BmDT4yKdffbZgF+TeMKECfZ1AbZu3cpRRx0V/4YnyO7du7nyyiuBINgh74dx48Zlq6xRu3Ztew2S31Mq1bz//vtJaXOyyHtCgmD5/bs2aNAA8FPgGjduHPW1oowHQafmlVJKKaWUI/mOiL7yyitAMPKVKfSc7qL//e9/A8HCk9x88MEHNjT8008/AfDxxx8D/nSbJBmn8jS93F1JJE+ioJFT8BLhfOaZZ2yUUyKiU6dOBaJLO6VqPdHhw4ezc+dOAEqUKAHAG2+8AZDv6UGZsv7tt99sdCc/x1rYSXqBRHnPPfdcO7WkkdCAzKrIx/322y+tInr5IYsHPM+jadOmAMycOROIvXDvzz//BPxFGzJ1LecfOb+ksxo1agB+9CvdzJs3jwEDBsT82kknncSbb74J5L4QTaKEY8eOtYvZZMe2VLF7927AT22SSKiMVaSclRwHWckYZ+3atUAwa7Bjxw7Kli2buEYnmaTqSJrgyy+/bMtZxSIpGbfddhsA27dvtyUFJdJe1GuTXtmUUkoppZQT+Y6IfvjhhwB2/1FJ8o1HqY8WLVrYkgmSpyK5lB9//LGNlvbt27fIP8u1hg0b5vg1iU4cccQRNodHkqojox8SGU7VgvZfffWV/bxdu3YAtGrVyj4meUlZc4nBL0YMRJWf6dSpEwD16tWLd1OT7ttvv436/1VXXUWtWrUctSa8JNdN+bMkcn6IjITK7NWCBQsAuPTSSwH/3Cq55nK+TTdy3owkkbETTjgh2c1JGMnllAhVpJNOOgnwr91Z9wZPVxL5HTFihJ1NlFmCnCKhInJBNQQbqqRTNBSw0c/ly5cD/mY5UupLyjfNnj3bHlNyzd2xY4d9DZmx/u9//wtgZzgLSyOiSimllFLKiXxHRA8//PCoj/Em+/7KamopsApBNDAdIqJi9uzZgB+dkMim5JEuW7bM7h2+YcMGIFjhVq1aNd55551kNzdhJKdHfPnllzbXKT+rFWvUqGErDKSyGTNmAPDLL78AQf7vOeec46xNYbZ+/XrXTQgVKa0SSSKhscrPyExErBXW6SDWZh/pkEMuJEol5z7Ja4Qgb0+igwWNhq5YsSIq+gVwwAEH2Gt0GG3evBmAW2+9FfC3qJQC9QcddFCe379+/XpeffXVxDUwRCRSLCUCu3TpYktbycfcNsT4v//7P9q2bQsEK+l79uxZpPFZ+hSOSzFSj/WZZ57JtrOS53l2ACpfk+n43r17Zys9kWpuv/12evToAQQh/tNOOw3wp9wLUgriqquuylZKIhVlXSxywQUXALH3Cc/NP//8o4uaMoRMHUIw+DjmmGMA/wKR9cIqA5LevXtz3333Afmr1Zsu0ikNQdKRIgegQurpFjZt7umnn7bXH1GrVi17jIWR1DuVxc7NmjXjzDPPzPH5kv41btw4wN93/YcffkhoG8NCgmD5rUfdsmVLALub36GHHhr3VA+9YimllFJKKSdCExF96qmngKBUQCRJhJVFLs2bN09ewxIsMuIV63O5C5W7l1SPhgL8/PPP9nPZDUQioxAsJpAyE2vXruWxxx6L+VrpsutF1sLbsQpOx/LZZ58B2GmoNWvW2DIklStXjmMLw+Ovv/7KVlYmt0WA6eq5554DoHHjxnYqVRYPfPrpp9mi6fIeuuqqq5LYyuSbMGGCjZCJcuXKUaxYMUctiq8pU6bYxbyibNmynHjiiUDhI7+SFiRlBCMVdS/xZFu9erU9D2Yt6/bGG2/YnRvlOKlXrx6333474C90grwXN6Wa6dOnA0GJQFl4HovnefZ8ITv65SZyo53C0IioUkoppZRyImkRUVlcMHHixJilNXJbfCB3+5JHmPVuNxVdfPHFAKxatYpNmzYBQcmq7du32+dJLlc6RELF5Zdfnm2LQdGlSxe7H7JEMIYNG5btebJP8llnnZWgVibP77//bsuj5ceOHTvsrIBEBiNLXcl2vJL/lG527NiRbQ/x008/3VFrkk+K0UueeU7RCHlcFumkeyRUyu8899xz2RZB3nTTTWlTBu2nn37KVtqucePGvPfee0V63WeffRaILtMjuYASLQyr+vXrA8ECnHvvvdfukR6LXGNkcfQ111xj95qXiKiUv0oHGzZs4MYbbwSwv6fMmJQsWdJujyxF///44w/KlCmT79cv6FqGrBI2EP3ggw+AYDp99OjRQNF2arj88suL3rCQkCn3yARwGYj279/fhtFlJZqslE/V2qGRateuzR133JHv58eq43bDDTcA+d+JKcz27NkTdfORk0mTJgH+ysZly5bl+Lx0uFHLTaybVlkFnq5++OEHe/6TGrpy8o+8CPzf//0f4Nfllb3oP/roIyCoQiE1oNONDEQjawzLQOrQQw910aSk6dChQ6G/V25YZAFPJEmTat26daFfPxnkPXDPPfcA0KhRI3sNFTLV3rlz55i1ZKUqgOxSJrWKc9qxKhXIoPOYY46x1wVZxCa/1+WXX25TwXr16gX4qV5SdeGyyy4Dct896dprry1SO3VqXimllFJKORHXcNKKFSsAP8wtd+Gx1K1bF4BKlSrZxyRELuVEJEE2MvKTKgnTGzduBIKSS/klCy5ee+01W3pCdoWQfXP79OkTp1amjsg7Mfn8sMMOc9WcuCtTpgxHHHEEQLZI59atW3n55ZcBuPrqq/P1eum+57qcKyCos5pOqSuRZMFFt27dsk03i+OPP94uUJGIRuXKle3UpCzok6m5WDU200Gs3V3kGiM796Wrf//734X+3rfeegsI0sAiSTpcquncuXOuU/OxbNu2DQgWjuZ3wWiYDR48GPBnySQ1RRYhxaqrK4vGf/zxR9544w0gSAGSndlikfNOYWlEVCmllFJKORGXiKgsPpKCpz/88APlypUD/B0ZINjDtGbNmjYJWCKjscj3QZDTkAq7zMyePdvmdUqEU/YDLgjZMUOSh3PLCUx3keVEzjjjDMAvWJwuypYta48V+TsPHDgQ8JPMpUhzfjRt2tTuJZyuIhd2ScQrXUrzCHnfd+vWDfB3IJMC9rJnuuwPfeqpp8Zc/Ce5blKuZejQoYC/e5nkkqYTifhGkh1g0t1dd90VVQIvL5s2bbLlv2SBTyTJqe3atWt8GpgCZCZTygtK+cBU9vrrr9vPJbIpC31z06FDB7v4Tfaczy0iWlQaEVVKKaWUUk7EJSIqRbVli6z27dvbqGBBtwWT/ZFXrVplH5OVj7IXexjJ3VTPnj2pXr06ULhIKPjlM3r27AkUvVBsKpNVflu3brWPpWuOrPy9ZaXil19+ma/vk9WiUppn0KBBMfcdTwe//vorEGyCkM4WLlwIYPNC69ata1e95zc/Wkr8fPHFF4BfnSHyY7qQc+/vv/9uH5PcRpmlS3fr16+3233GKlMlUT6ppDBq1CjWrFmT4+tJhY569erFuaXhNWvWrKj/p0OFGhk/eJ5XoA1OOnfubGe6ZbtXuQ5XqFAhzq2M00BUdnWRKaOilDtYuXIlEFx0IDVqBE6bNg3wp1ZbtWpVqNdYsmQJ4O8jLFO0MtDIxJ1jZDC2atUqO/WYrrsFyeI0GUTKLic5kf2kpR5tKqStFJUs1pIyPRD8/ulKLiQXXHBBgRbobd26lQsuuAAIyjalK5mSjtyVT2ogSnm3PXv2pEWpN/Cny2UB4zfffAPA8uXL7eA71jly8+bNQHB9jUVS5bp06ULjxo3j2uZUkHV3u3QgKRabNm3ioYceAoKUntzOJ8WKFbPXXDnfylS9nFcivfvuu0VKg9GpeaWUUkop5URcbhHlDiwehV9lml9UrFjRFi8PsxYtWgB+BEMKKkvJpSOPPNLuhCMk9WDOnDlMnToVCPaC9TzPRkJlKjpWIn666927t/1cFr/961//ctUcJ3r06GEXnVxxxRWAX8Iq3Us0RZIpRNkcA4JZknRdjHLMMccAQTm7yCnm/v37A9jFSxBEvGQm5eKLL7bTsXIuadSoEZBeC/1yMmPGDCAoZTZw4MCY5YlS0UEHHWSvtTIjsHv3bls+Mb9KlCgBBClvEmWVUnIq9clGB1988YXdaU9KwknUO9Y59NFHH7WpcZKicO655+b4c2655RaNiCqllFJKqdQTmqSZJk2aAME2l+KMM87gxBNPdNGkApG7yvPPP99GNqX0ijEmW8FtiVZs2rTJ5oFFbtUnd7ypEA1OlMgC3hIhyhRSdLhXr15pV5qooCRZXhZjQFCgvKh7HIeVRBceeOABwD8PSI7X2LFjgeiFoLLxhbxnImdVjj/+eCDYSzzdoukyIycl/yK3uJWoX7rsMy+ktJDMtC1evDgqdzovjRo1smWbLrzwwri3Lx3ImpdUJotgH3nkEXsele2kZRGjfIwUef6Q944sGo+lqDOVoRmISq1EWdEpJ5VUWyX99NNP20FmZPK8fC5/3MjBpyTWy2C2X79+nH/++UlrcyrIlMFYrH3UVbQWLVrQvn17181ICjknNGzY0A405BiJrBGYVcOGDbnkkksAuO222wBi1hpNB5KmIekLXbt2teksUr0lkTUQXZo7dy4A69ats3UiZY90GWAMGzYs2/nzwgsvzLWOt4IGDRq4bkKRSfrOvHnz7I2oBMq+++67HL+vZcuWdlpfziO5kZvjwtKpeaWUUkop5YTJo05lUopYTpo0yd6xli1bFoAxY8YAFHi/2CziPW+Xr/7YtGkTEOyOAzB69GjAL80E0TXKZCFSEko0OemPwqpfvz7gR8slmiMLNWS3mCJKqf5IgkTMc2ufRCt0f0hJu6yLQj/44ANbu1hmUiQKmgCh6Y+Q0P6IlrL98eCDDwJw6623An66AxS5fnnK9keCxOwPjYgqpZRSSiknnOaIyg4p999/v414SbHUIkZCnZJo56hRo+xjkZ+r/JHyTYMGDbL5cfvtp/dOKjNJ1FNyvZRS8Sc7B5UvX95xSzKHXtWVUkoppZQTTnNEZYX8yJEj7SrHNm3axPNHaH5GNO2PaNof0TRHNDs9RqJpf0TT/oim/RFN+yNazP4IxWKlBNKDIJr2RzTtj2g6EM1Oj5Fo2h/RtD+iaX9E0/6IpouVlFJKKaVUeOQVEVVKKaWUUiohNCKqlFJKKaWc0IGoUkoppZRyQgeiSimllFLKCR2IKqWUUkopJ3QgqpRSSimlnNCBqFJKKaWUckIHokoppZRSygkdiCqllFJKKSecDESNMZWNMdOMMTuMMauMMRe7aEdYGGMmGmPWG2O2GmOWG2OudN0ml4wx1xtj5htjdhtjxrluj0vGmJLGmOf2vU+2GWMWGGPOdN0uV4wx27P822uMedx1u1zS82k0fc9kZ4w50hjzkTHmD2PMSmNMR9dtcs0Y08UYs2Tf++Z/xpgWrtvkiutzSPFk/rAITwJ/AdWBpsBbxpiFnud976g9rg0DrvA8b7cxpiEwyxjzjed5X7lumCPrgMFAW6C047a4VhxYDbQEfgbOAqYYY5p4nveTy4a54HleOfncGFMO+AV4xV2LQkHPp9H0PRPBGFMceB14GmiD3y9vGmOaeZ633GnjHDHGtAFGAP8BvgQOctsi55yeQ5K+xacxpizwO9BY3gTGmAnAWs/z7khqY0LIGHMEMAu40fO8KY6b45QxZjBQ2/O8y1y3JUyMMYuAez3Pe811W1wyxnQH7gYO9TJ0r2I9n+ZPJr9njDGNgc+B8vI+Mca8B3zhed5Ap41zxBjzX+A5z/Oec90W18JwDnExNX84sCfLndhC4CgHbQkNY8xTxpg/gaXAeuBtx01SIWSMqY7/HsrUaFek7sALmToI3UfPp3nQ90xMBmjsuhEuGGOKAccBVfelKawxxjxhjMnU2Tfn5xAXA9FywNYsj/0BlHfQltDwPK8Xfh+0AKYCu922SIWNMaYE8CIw3vO8pa7b45Ixpi7+FON4121xTM+nudD3DADLgA3ArcaYEsaYM/DfO2XcNsuZ6kAJ4AL8621ToBkwwGGbXHJ+DnExEN0OVMjyWAVgm4O2hIrneXs9z5sL1Aaudd0eFR7GmP2ACfh5PNc7bk4YdAXmep73o+uGOKbn0xzoe8bned7fwHnA2fg51X2BKcAah81yaee+j497nrfe87xNwMP4ucSZyPk5xMVAdDlQ3BjTIOKxY9Bpk0jFgUNdN0KFgzHGAM/h38l32ndhyXTd0Ggo6Pk0Jn3PRPM8b5HneS09zzvQ87y2wCH4i3Qyjud5v+MPwiNTejI5vcf5OSTpA1HP83bgTz3fZ4wpa4z5N9AB/8414xhjqu0rI1HOGFPMGNMWuAj40HXbXDHGFDfGlAKKAcWMMaX2rfzMVKOAI4FzPc/bmdeT050x5iSgFrpaXs+nOdP3TARjzNH7zqNljDG34K8SH+e4WS49D/Ted/2tBNwEzHDcJifCcA5xVdC+F35Zng3AJODaDC414uFPw6/BX7n2INDH87w3nLbKrQH40yd3AJfu+zwj83f25UL2xM9j+iWifuYlblvmVHdgqud5GT/9vI+eTyPoeyamrviLYDcArYE2nudl8jqEQcA8/GjgEuAbYIjTFrnl9ByS9PJNSimllFJKgW7xqZRSSimlHNGBqFJKKaWUckIHokoppZRSygkdiCqllFJKKSd0IKqUUkoppZzIqzZjqi+pN3F+Pe2PaNof0bQ/stM+iab9EU37I5r2RzTtj2hp2R8aEVVKKaWUUk7oQFQppZRSSjmhA1GllFJKKeVEJu/frZRSKs38888//PTTT1GPjRs3jqZNmwJw4oknAnDQQQcluWUqFQwY4O8mvWnTJgB69OjB8ccf77JJaU8jokoppZRSygmNiCbZ/PnzAViyZAkAv/76K8uWLQNg9uzZACxfvpzatWsDcNdddwFw1VVXJbupzvTu3RuAJ598EoCPPvqIVq1aOWyRUqlBIoFvvvkmU6dOBWDWrFkAGJN9werHH38MQMuWLZPSvkSaN28eAPfffz+vvfZatq97nr/guFq1agD2OSeffHKSWqjCauHChfYau2jRIgB2795tP0o0vWTJkk7alywPP/wwAK1atbIzBsmYOdCIqFJKKaWUciLhEVG5C508eTIA9957r40AxnLEEUcA8OGHHwJQvXp1ihdP/cDtjBkzAOjYsSMAe/bsAaKjFNJXxhjWrl0LwPXXXx/1/GuvvTY5DXZI+kQ+vvfee2kfEf3ll18AeOedd4AgYr548WLefvttAPr27QvAWWedxZFHHglA6dKlATjggAMA2Lt3Ly+88AIAO3bsAKBnz56UKFEiGb+GckSOmzvvvBMIojqQ/f0U6bzzzgP8iFCdOnUS3Mr42rlzJwCXXnopAO+++y4Af/75p33O2WefDfhRnW3btgHw8ssvA9ChQwcA1qxZY99HKrP069cP8McnWfOKxbhx4+zzDj/88GQ1LeFkjPXoo4+ycOFCAFavXg1AxYoVbfS3bt26AHz++ecJa0vCRnj//PMPEEyv3nDDDfZr++3nB2LLli0L+IMsOanIIFWmphs3bswHH3wA+IPSVCXTQHv37gWCi0L58uU57rjjop579NFHs337dgAmTpwIwKRJkwC48sorM25Q8d133/H3338DpOXvPn78eHr06AHEHizIYw899BAQTJ8AHHLIIQB28Dlnzhx70hQtW7akSZMm8W+4cuqvv/4C/ONBBqCxjp/c/PHHHwA88cQT3H///fFtYILJQPKTTz4Bgpv0c845h5NOOgkIplKLFStmr0lyDn711VcB/3e/9dZbk9fwJJFz5urVq7n33nuB4DyRmxtuuIG7774bgEqVKgEFP67CbPv27TZt5amnngJg69atOT7/qKOOokKFCklpWzJs3rwZgJtvvhnwr69ZyXkBYMuWLQD2PfXSSy9Rr169uLZJp+aVUkoppZQTRqaDc1Do7aSeeeYZwJ8WjFS8eHF7tyVlEn7++Wd7Nz569GggmIoGPyoK8OmnnwIU5O4kNNtrSYTzrLPOAoLo7siRI230N5bbbrsNgAcffBDw79579epV2GaEpj9yI9FziaZ7nmfv0MqXLx/PH+W0P9atWwdAkyZN+P333/0GxYg8yNSITB3lFp3wPM9+vUqVKoA/pVK/fv38NMn5Fp8TJkwA4Isvvij0D5TZleeff94+JtGwQgjde0bO2SNGjACgf//+UWk9OT1fFj4CDBo0KOpr9erV46233gKwaR85CE1/yLlAopmRv19uZMGopPv0798/2yxCAYSmP+QYl/PEmWeeCcCKFSsK3RhJY7jwwgvz+y2h6Y+c9OrVi1GjRuX5vFq1agH+9Uiuw4UQmv6Q2dWnn34agM8++yzbc+Q9ValSJXbt2gXAhg0bop7Tp08fOzsn0dKKFSvmtxm6xadSSimllAqPhOSI7t2715YMyeqOO+6wkVBRp04dnnjiCSAoI3LjjTcCsH79epvDIEnoqZivUa5cOSD4vSRClVs0NPL7xLRp04oSEVUhIsniclcJweKRe+65xz4mkc2NGzfa51922WUArFq1KtvrVq5cGQiiGfmMhobC3LlzARgzZox9LLdoX+Rzsn5d/n/YYYfFu5lOyIICmTWSj5Ekd6t9+/Z2YeQpp5wS9ZwVK1bYiKhYtWoVP//8M5BnRDR0CrqARHLyJb86XXz77bcANGvWLNvXJLc+8trZsGFDICjcLh+3bNli82iHDx8OQJs2bQoS9Qolef/I4s+cyPika9euQHosUHr99dfp1q0bkPt59PXXXwf8cdj69euBYMGf9N/s2bPtwlkpOTlmzBiOOeaYQrcvIQPRDRs22MU14qijjgL8xTa5kSmAkSNHAtjOSBedOnUq0vfntLJPpZ7I6SFZuCcXEVkBDPCvf/0LCOokvvnmmzEHoEJublKx0oAsxBo8eLCttPHbb78BuZ9AN27caBceCLnpu++++xLR1KTyPC/XAaik/AwbNgwgIxanSfpXrMUWuZGL59KlS+PepmSTAePy5cvp0qVLzOccffTRdrGSLPCKRdJhBg0aZAdr33zzDeCngcixlSpkDHLNNdcAweItSd2JVKpUKfu+kioMsqg6lcl0fLdu3cgpDfPSSy+NuYhN6oc2aNAAgAULFgB+astXX30V9dz27dvnek3KS+r3tFJKKaWUSkkJiYhOnz7dfr7//vsD2MVIsvAiLy+99BLg7wssNRbHjx8PwC233EKxYsXi1dxQkkTiadOmRT2ebtNJmUymTOfNm2cXs+W24CLWFHWpUqUAP+UF/PeZTKG8//77gD+tliokMly2bFm7w1Z+vP/++zYiKtOPUp4ka3pLKoks0RQrEgr+TkFSpziTXHTRRYX6PomMRS6ITTVyvrj66quBoE53pOuuuw7wF7wefPDBOb6W1BuWxW95TV2ngsmTJ9sUttxKM0m/3HrrrXbqOh2MHTsWCGaFIq8ZMgZ78cUXgdipHJFkyl3KnWV9PfDrYEs6VV6z3rFoRFQppZRSSjkR14io7FwRWXBbkuclhym/5Pu6d+9u79Qk6nPeeefZHZjSiSzGmjFjho2MSYF/ifJkXeilUpcsBNiwYQPjxo0D8lc4um7duvYu9pZbbgGCYsPbt2+3pTVkt51UiogWliTZQ7AYJa+FgKlA8q769++f7WtSwF2iYip/ss4ypSJZOxEZCZXZR5kJkOMjt2goYPdYj5zJFLJA6cADDyxSe5PlzTffBOCSSy7JV8k2KeVVrVq1hLYr2SQ6GbnLWM2aNQF45ZVXALJtpJMTWbQ1cOBA+zpS5mn58uWAP7sgUfrC0IioUkoppZRyIq4RUclnWrlyZdxes1GjRtkeGz16dFTUNRVJH3366ad2X/GZM2cC0ftEC7m7bdGiRZJaqJLlrrvuKlC+W+PGje2KxtzIcZUJnnzySRtNPvnkkx23Jn5khXfkilfJ8ZIcwMKUW5LXi3zdPDY3SXlSnkpmH0TTpk2T35gi2LlzJ+eee27UY0cddZSdOczv7KNcgyQqGKlt27ZAUIUh7H0klRP+85//ALlvYHHOOefw3HPPAUFpvFikSklk5Z5jjz0WCPdsy59//mnzoCPJzEl+I6FCouFSDrBBgwY2EhpZqUFW3vfp06fAbU7YXvNCdifIZL/99hvNmzcHgt10su57nBOZXj3jjDMS2ELlUr169eK2d+/ixYvt56lWD7IojDF2IJoO+2JLzVgpURT5O0mJu8L+fQcPHpytj1q1apWt3mi6kTJgsjinXbt2AJx++unO2lQYkyZNsgMBmY4fNGhQgdLfZsyYYadcI/cVF5IGF/YBqJDraKzSTELqYY4dO9Yu8pQFnbKLXyQZoEcORKU/unXrxvXXXw8ENVrDomfPnnz99ddRj3Xo0CHfO49lVaZMGQAuuOAC+5gcf5FkkWxh6NS8UkoppZRyIq4RUSmeGqlHjx7x/BEpadu2bYUu9iqRi3QorptfWacN033KMB5kGuntt9+mevXqQJDOkc5k6jrS5s2bAT8yAP4UpEzdyftpwIABNqIRRhKNiZw2lWLkkbtuFcQVV1wBRG+WIG6++WYb+UhHc+fOzbazzAMPPACEL6KVl8holKRnyI5sebn99tsBePbZZ2NGQgGqV69O48aNi9bIJJNFSrHIYk0pP/Tqq6/axTyffPJJgX6OFHVfsGABrVu3BvwNA8JA0hWlLBMEC9WmTp2akJ8ZeW0uSrpk5oxulFJKKaVUqMQ1Ivrjjz/G8+XSxoEHHmiTeteuXQsE+So1atSwz5PC/U8//bTdylNydYQkkaezWLl+sk2jRDGUT3KizjnnHMC/Q5VjSrZmS0Vbtmyx0QfJ7ZNk+UjvvfdetseeeOKJbI+1bNkSCCJHUrImrGJFMGSRUmEjl3PnzgWC/FMI+iWVF0Fu2LABgDfeeAPwo31ZcwWXL1/O7t27geC88tprrwF+6ZlUyYXM6rDDDsvzOVu3brXlzZ599lnAf3/lZPLkySlTrklImalYZLZISj5u2LCBXbt2JaNZSfXtt98C0dfNvIrVF8by5cu58847s/2sokj4YiXl7+wiO0Xlx5VXXpmttpskU7dp0yajpumFVGRQPqnZ2717dwA2bdoE+CcGuclJRV9++SXgT51/+OGHQOwdpXIjg6vIAWms6hthJlOGkVNfI0eOLNRryeKmFStWZPuaTO0ecMABhXptV3bu3Mndd98NwOOPPw5gB5o5kfOmLPCR/deHDRtmV6HLDV3Xrl3tglIZyKxZswYI+jMMZGDZvn17Oxj773//CwRT0V9++WW+FpLI4q3jjz8+AS1NrFh7pQsZdOc2+DbG2PeApCykWkqY7DxpjKFJkyYAtjpAPD311FP88MMPUY9Vq1bNViwojMwb0SillFJKqVBIWERU9oyuU6dO3F87HXdVilS5cmU7bSQ75she0q+++iqdO3d21jYVDrKjSuSOQuCXF5EoVyqSfa4/+OADO+0oESz5f2Qps8GDBwN+iRV5X8TadzvVFLUU1Y4dO2w9PzmXRL6W7JISWZIlFciisyuvvNJGzyUFRc6VrVq1yrZItkaNGowePRoIor+S5rNgwQLbR/Jx3bp1/P7770AQVZKIj+uI6KGHHmo/l/SVU045xe6+V9gyOrK7kJQ2ygRyrixXrpxNRxg6dCiQewS1WbNmVK1aNeHtK6xevXoBRdsR69dffwWCVJ5BgwYB/s5MWc9LpUqVyldt65xoRFQppZRSSjmRsIioVPbfunVrob5fdsF48MEHs33N9R1pXmTP1eLF/e4tyh2m5O18/vnngJ/PpBHRzCQLMK644gobORRHHXUU4EcIi3Jn6pr8Hn369LF34DK7EsuoUaOAYKFfppN8xttuuy1maSvwNxlJtbJ6H3/8MRBEsJYuXcpll10GBJFNOe9K3jQEO+C89dZbNm9OSAH/1atX28LmV155JQD9+/e3kdMBAwYA0Ldv3/j+UoV0+eWX2xxiWXuQ20LhAw880P6uUsLq/fff56mnnop6XrrPNMYiG8Z4npevxdYSCezevXuoz7PHHHNMkb5/wYIFNm9aNuHJTfv27Yv08zQiqpRSSimlnIhrRDSysKuscpZ8i6x74+bl0ksvBYKSBIDdSzfMKzw3btxoV15efPHFANx4440Feo2///7b5jNlXW0vd/0q9ckqx3feeSfXwsr/+te/APjiiy+AYIV8JCldE6+tQl2R2Y7CzHqkWpSvoCR/sVOnTtm+JucYmUnKKRoKflRMSkGlCvndly5dCkDjxo2pWbMmEETRpSTP1q1b7e8nhc5zK9B+8MEH2+0u5dz9zDPPhLa0VbFixexsgazs//777+3so+Q2nnnmmYD/vpCi7mL48OH2c+mrVH7/yGYNMiOQX1lXf+ekZMmSAFxzzTVAwa/pyRC5yl/y5BctWgT45eq++uorIIjqRl5z5Poxa9asqOfk9XOkvKSM8worrgNR2flDDgoIagDm17Bhw4DgogvQsGFDINgppVixYkVqZyJ9++23NoleksY3btyY65t82rRpQHS9RJkmyFq65tFHH01Mw1XSSO3LyGMitxJFslAt8jlyYpQbllQfgBaUXGxl33BIr0UWJ598MhBdcknKUclUmfztFy9enOvxI1+Tc6vsrpNKstaQ/e677+zCJSGLeIYPH24HDPkluyvJgh2Zjg+r+vXrA0Gg4rfffrPpcFLGKtZCYaklG7mg6cQTTwSia1qnGrmBkGCYDMDioVmzZsycORMIjo8wilzg+NhjjwFBeuDgwYPtjYo8L9bOWvldJBnvmxedmldKKaWUUk7ENSIqEYnGjRvbu1UJfcu+zjfffDOHHHJItu/94IMPAGyRYrm7a9iwod0bOcxT8qJGjRp2lw4prTF06FCGDBkCBHcasSIYsR4rXbo04PcbkG2KJd3s3bs37dMP5O461l1nfsr1GGNo3rw54JeqyUTy3lq1apXbhiTItddeCwTlrDZs2GCn27NOu0ceM5Gfy5T1JZdcAgTnkFQkEcqbbroJ8I97KVckCzolFUoezySVK1fO1/Puv/9+gKidhXr37p2QNiVTrVq1gOD98tZbb3HLLbcAweYfuSlRogTHHnts1GMyA9uuXbtQR0KFRCklPQf8XcPA34yhoBuDCJl9k2tOixYtbCQ0Xrv3aURUKaWUUko5YfLYxqpQe1z9+uuvnH766QDZ8ngaNGhgi62K8ePH87///Q/Ifvfy5JNPZnt+AcRnI9RAvvpj/vz5QFAAdubMmbluUSl3KJLj9NNPP9m71PPPPx8IcsaKyEl/FMTatWuz5Tbtv//+dtGBHFdxkvT+2LZtG6eddhoAX3/9dfCN+bhbjfUcWbAhOVGVKlUqaJsjxbs/IAHHCARJ9dKXJUuWtJFCWdwVJ07fM7KIoGPHjrm/6L5jo3z58oCfKzdx4kSAeC9MctIfMkMm22zWqlXLbnTgWOjPqRDkFcsMysqVK22eqeSNxqkcUWj6Q3LrpbSbXFPLlCljFz6LUqVK2QXScZa0/pDtXocOHWrf87IIb9GiRXa7cImGy6LX4sWL2+uILH7bb7/9bH/JGp2zzjorHu2P2R8JGYhCsM+t7OebdUCak8MPPxzATsfXqVOnKHurh+JNMWfOHHtRkJXSbdu2BfyBpvx+5513HgDLly+3YfA4C0V/5Gbbtm22vp0cA3fffbddwRpnSe+PhQsXZpsCgpwHoueee64dfMtzHnvssWyrPdevXw8UOZk+5QairVu3Bvy95CMrbMSR0/eMLMqaOHFirit15diQXYASuAI69OeQJEuJ/pApeVnlDMEAVHakipOU6I8kCl1/LF++HAjSGMqXLx+1wDzBYvaHTs0rpZRSSiknEhYRFZIsK/uWjh49mjlz5gDR9dkuv/xyINgJQ8oOFFHo7kYc0/6IlvT+WLdunV008sorr9jHy5QpA8Bdd90FBLvDVK5cOdt74Y8//rAlWr7//nsgmI4uV65cUdqfshHR7t27M3bs2ET8KH3PRNP+iBb6/li7dq1NYZFyYO3atbPpTnEuhxj6/kgy7Y9oGhFVSimllFLhkbC95u0P2BfNkfIK9913X6J/pFKhVbNmTbvrhXwsqMgyZqlQViQRZFMAUdS9jpVKV6tXr47aGAGgZcuWod4YRmUWjYgqpZRSSiknEh4RVUqpeJPNM5o0aQIEFSeUUtFOOOEEW5ZHqTBK+GIlxzRROJr2RzTtj2gps1gpifQYiab9EU37I5r2RzTtj2i6WEkppZRSSoVHXhFRpZRSSimlEkIjokoppZRSygkdiCqllFJKKSd0IKqUUkoppZzQgahSSimllHJCB6JKKaWUUsoJHYgqpZRSSikndCCqlFJKKaWc0IGoUkoppZRyQgeiSimllFLKiaQPRI0x1xtj5htjdhtjxiX754eVMaaLMWaJMWaHMeZ/xpgWrtvkgjFme5Z/e40xj7tul0v6nolmjDnSGPORMeYPY8xKY0xH121ySd8zsek5NWCMmWWM2RVxjCxz3SaXtD+iGWPqGWPeNsb8boz5xRjzhDGmeLJ+vouI6DpgMDDWwc8OJWNMG2AE0AMoD5wC/OC0UY54nldO/gE1gJ3AK46b5Zq+Z/bZd3J8HZgBVAauBiYaYw532jCH9D2TnZ5TY7o+4lg5wnVjQkD7I/AUsAE4CGgKtAR6JeuHJ30g6nneVM/zpgObk/2zQ+xe4D7P8z73PO8fz/PWep631nWjQqAT/ptjjuuGuKTvmSgNgZrASM/z9nqe9xHwKdDVbbNCQ98zPj2nKpV/9YEpnuft8jzvF2AmcFSyfrjmiDpmjCkGHAdU3TfNuGZfWLy067aFQHfgBc/zPNcNUaFmgMauGxESGf+e0XNqjoYZYzYZYz41xrRy3ZgQ0P4IPAJ0McaUMcbUAs7EH4wmhQ5E3asOlAAuAFrgh8WbAQMctsk5Y0xd/OmB8a7bokJlGX7E71ZjTAljzBn4x0kZt81yT98zlp5Ts7sdOASoBTwDvGmMOdRtk5zS/og2Gz8CuhVYA8wHpifrh+tA1L2d+z4+7nnees/zNgEPA2c5bFMYdAXmep73o+uGqPDwPO9v4DzgbOAXoC8wBf/kmen0PePTc2oWnud94XneNs/zdnueNx4/nUX7Q/sDY8x++NHPqUBZoApQCT/HOil0IOqY53m/419EI6fSMnZaLUI3NLKjYvA8b5HneS09zzvQ87y2+JGNL123KwT0PYOeU/PJw09pUb5M7o/KQB3giX0D883A8yRxYO6ifFNxY0wpoBhQzBhTKpllAkLqeaC3MaaaMaYScBP+quCMZIw5CX/KJKNX/gp9z0Qzxhy9rw/KGGNuwV/pOc5xs5zS90w2ek7dxxhT0RjTVs4bxphL8KsIJC0HMEy0P6LtmzH4Ebh2X39UxM81X5SsNriIiA7Anzq5A7h03+eZnLsDMAiYBywHlgDfAEOctsit7sBUz/O2uW5ISOh7JlpXYD1+rmhroI3nebvdNsk5fc9E03NqoAR++beNwCagN3Ce53nLnbbKHe2P7M4H2uH3yUrgb/ybt6QwGby4UimllFJKOaQ5okoppZRSygkdiCqllFJKKSd0IKqUUkoppZzQgahSSimllHIirxIwqb6SKd51wbQ/oml/RNP+yE77JJr2RzTtj2jaH9G0P6KlZX9oRFQppZRSSjmhA1GllFJKKeWEDkRDaPHixVSuXJnKlSvTq1cvevXqhed5aM1XpZRSSqUTHYgqpZRSSikn8tpZKdVDcCmVKLxz504ArrvuOp5//vmor/31118AlChRoig/IqX6Iwm0P6LpYqXs9BiJllL98cUXXwAwYcIEZs+eDcCuXbsAOOOMM+zHtm3bAlCyZMmC/oiU6o8k0P6Ipv0RTRcrKaWUUkqp8Eh6RHT79u0MHToUgK5duwJw5JFHxvvHiJS6G5k7dy4ALVq0sI/VqFEDgNWrVwNQvHheFbdylVL9kQQp2R+vvPIK//nPfwCYMmUKABdccEE8Xlojotml5DGSQCnRH/PnzwfgnHPOAWDjxo02x96Y7L/CZZddBsBzzz1X0B+VEv2RRNof0bQ/omlEVCmllFJKhUeRwmuF8dVXX/HQQw8B2Mhoptu+fTsAjz32WLavXXTRRUCRI6EqjQwaNChmVEcp5eeAnn/++YAfCQX4v//7Py6++GIAunTpAmDz8F999VXGjRsHBDmiTz31VDKbrFRGczK6kYU348ePB6B79+4umhEaM2fOBPwpV1G/fn0Arr32WidtSqRnn33W/s7y+51++um5fs+aNWsA+PDDD4HMPGYmTZoEwIoVKxy3JLmeeOIJAHr37g1Aw4YNOfDAA4EgnSWTfPbZZwD8+9//BqB58+a8/vrrANSsWdNZu1zbsWMH4E+zr127FoCKFSsCMGTIEE477bSo5992220A9OjRg/bt2wPwzjvvALBlyxb7vcq3ZMmSqP8nMKUuLrZt2wYEAa9DDjkEgG+//dY+57333gOgVKlSLFy4MMfX6tmzJwCPPvooUKhFbaH11Vdf2c+HDBkCwPTp020qi/ydq1atav9/4403Rn2tqHRqXimllFJKOeF0vnfPnj0uf3wo7NixgwcffDDb45MnTwagQYMGyW5Swrz99tsA3HzzzTYd4aOPPgLgsMMOA6Bjx47UqlULCCJhAFu3bgVg3bp1ALRp0wbIrAjQsmXLgGBGId0NGzYMgAEDBgDBIr6///6b77//HoBrrrkGgHvuuccu7MsUkp7x9ddfc/jhhwNw9tln268fe+yxAJxyyikA9jkSTU43kr4kCzsBjjnmGIBs0dBIVatW5a233gLgl19+AYpcJi+0JKrZrVs35s2bl+fzp06dCvjvxaVLl0Z9rV+/fgDceeedcW5lfLz//vsAjBgxIl/Pzy3d6ZtvvgHg119/BaBOnTpFbJ07kq4i59dHHnnE/u6xFvTJdUf+/nPnzrVRVPnbd+zYsUht0oioUkoppZRyIukRUYlkQJAjesUVVyS7Gc7t3bsX8CMYUnRZGGOoUKGCi2YllOToVKtWzUZEt2zZAgTlVuRjXkaOHAnAAw88EOdWhtd9990H5H7nnk5kVkAi5BMmTACgbt26nHXWWQCMHj0agKOOOsrmkGaiP//8E/AX3gj5XKIccS71FTpjx44F4Msvv7SPSTQ4L5UrV476mG4kCiaLuJYtW2bPtXI+kYL/06dPt59HRsqyRs1kpiKsEdGspP3Nmze3Ub6rr74a8N8TP/74I4DNL+7UqZP93mrVqgFQpkyZpLU33uQYkN8l698z8vOGDRtStmzZmK+zdOlSe+z0798f8GdZ8vteiyXpA9Hy5cvbzyX5NRP98ccfAHzyySf2sf333x/wB1kNGzZ00q5Ekt+pU6dO2QaQctDXr1+fcuXKAfD5558nt4Ehl0fN37Qyc+ZMFi1aBMBLL70E+ANQcfTRRwPB4pJMErm4oCAGDRoE+IucDjrooHg2KRRkUYoxxi40uu666xy2KDy6desGBNOsxhj+7//+z34O0dOy8ljkTa98Lgt3Uo2kaUTeqEQ6/vjjk9mcpJOp+Kx/244dO9oBpWjYsGGOg+6hQ4famxA5nsaMGWNfL7IOen7p1LxSSimllHIi6RFRWbAC2Om1TBS5EEfIHWqvXr2S3Zykuueee+x+z1KySpK/X375ZUqVKgUE00hSrgaCyGm691EskXeyVapUAaB27doum5QwkaXMZB/wSMOHDweChQgff/xxxkzNy7SpRLBatWplF/3JYr6XX37ZPv/mm28GgrI1q1atSquIqES4ZErVGGOjdpk86yauueYa3n33XSD2NGys/8t7ThahyBR2Kpk+fXrMxz/55BP7PnnzzTeB6HQVuQ6n27lV0k7k7yxT6a+99lq25y5ZsoRVq1YBQT9KGpQxJtuxM2HCBCZOnBj1+nL9jvX6WWlEVCmllFJKOeG0fJMUJ8+kxUpyJxYZ8SldujQAV155pZM2JVuZMmXsLlKSwyVR0Lp16/Ldd98BQSmVSFKiRQr+ZwIpohxJSvGccMIJyW5OQklpqkWLFnHmmWcCuS8gOfHEE4HgfZUJsuZ4rV+/3n5NypnddNNN9rG+fftGPT/dZN2hr1q1alELTTLV4MGDAZg2bZr920sULFYh8quuusp+LqW/UtU///zDzz//HPXY4sWLAWjXrh27d++O+posjAR/4SMEsy3pMnMrv5ccC1KOKXKjAnkvTZ8+3W4QkfV8EytvOOvnAI0aNcp32zQiqpRSSimlnEh6RFRG5QArV65M9o93TgoES9QPgjIYmbht5RFHHJHtsR9++AHAltMAqF69OhBsc5kpNm/ezNNPP53t8a5duzpoTeLJtnzz58+321bm5oADDgCCMmCZaPny5TEfz/peadKkCRB9Dk4H//vf/6L+f91119G8eXNHrXFHyvPIDIpEtzzPo2XLlgDMmjXLSduS7ZNPPrG51CK/G4FIiUnJrW7dunVabOkp+b7nnXceEOR+NmrUKGblhKx5oJHvKckfle8bNGiQzQmV9QsFkfSBaGGW9qeTyDqqQuokKt/DDz+c7TEp13PqqacmuzlOTZo0KdtAo2nTppx77rmOWpRYb7zxhv28Xr16OT5PEuNnzpwJ+FPRl112GQCnn346AJdeemliGumYTJtKndCcBl2RC0MB+vTpA0SX0Etl8r6QtAy5cGbiNWbjxo12ClnKe0VOlRZ155tUIzVzY2ndurVN+5GFSRAsApTAkBxfUvM71UkQTKbkc5tWN8bwwgsvAMEUe2S6huxoJ4PPM844o0ht06l5pZRSSinlhNPFSplE7kIik6LBT6zv3Llzjt8nC7lk39w33niDpk2bJqaRIfD+++/HLDicaXf0Ys6cOTbSIx8bNGiQVuV3Iv373/+2n8vfXKIXskALyJauIDttAbaYebpGRG+//XYgiPzGioguWLDARpfluJGNItKFREIlLSO3xVgrV660i2OltNXJJ58MBDuWpbJTTjnFFhePVZZJFq/J++arr75K6V2C8lK6dGn7+8kCLdnJsWLFipQoUSLb98j5RSKiYsyYMdxwww2JbG7CTZw40S5a3LBhA5D7zkodO3a0JbxilUCLlS5WFBoRVUoppZRSTmhENEm+/vprIPuiilatWtnyTWLPnj22ILPsnyw6duwYtYgnXUjR+meffZadO3dm+7psfypREClTkwmyRnruuusuRy1JPMkL7dy5s83zevLJJ3N8vkRQ69evbwu2S3H3dJfbgpzZs2ezfft2IH3LNuXHBx98APjbCstCOOmPOXPmAP7CUcmfSzVSemfZsmUx8/yyfi5R06lTp6btjAH46wzkGhprQWxBxLoepQo5rvv27cumTZuAYK95ye+86qqrGDJkCOCX+gJ/IZNsqRxrzUa8OR2IynT1smXLinywpKpYSb6TJ0/ONgAV//zzT6KblFRz584FgmlYebNkJTVWZTpN6rBWqVLF1hbNFLJSPB3JDcfLL79sp1Dlbx1Z+08GrHI8gL9jFwS7Li1cuBAI9pjOBJs3bwZg1KhR9jHpo0xa6CcDNElt2r59u92VTRb6yfGVDhUXLrnkEns9lcVad955J+AvZJJBtwzOhg4dmtYDUSj6ADSVSQ1QSTPYsGGDvRmRMUfkOUJ2P4ocuEq6k7xvBg0alLD26tS8UkoppZRywmkoSaZK5GM6q1SpEhBEfPbs2QP4EU6ZPpIoqPw/lq1bt9odIgqyc0FYyVR7TpHQrCSCKot1HnroIVvvLZ3Igq0vvvjCPnbOOecAue80lI4uvPDCAj1fIqeyWCeTIqJSOzSy5JfMNki0I3IHnXQ0ZMgQW/dxzZo19nEp+SWLVtKB7JA0YcKEHJ9TpUoVe36NnKKXqHGsXZZUapMpdknFMMbY6Gjt2rWBIDUlstyZTNfPnTuXRx55JOq1NCKqlFJKKaXSTtIjogceeKAtqJwJkVCRtYCuRPauvvrqAr1OkyZN0iISKiTK9+mnnwLR5Xvy4/XXX0/LiKgUH1+9erV9TMpvZF3cpmLLacehdCQzLO+++y7gl2GR82zkvvPppFWrVkCQ9yg5b7FmlCZPnmzPvddffz0QlKqR8j7pTHa7kY+bNm2yOzFpRNQns3Pp4NlnnwWCY/ySSy6xJajyu/ORfG8y1qVoRFQppZRSSjmR9Iho48aN7ZaWssovk/znP/8BgohofkkUTLb1SxdSdFjyVkqWLBm1Ohr8vJWLLroo5vcXZl/bMJMo3vr164HoYsOyX7SKLR32gy4syQOVbT2NMWld5itS//79gaDIdqxyVaVLl7bPk8hXqVKlAGjXrl0ymhlXku8rOX15kain9I0xJq1m1uJBShhl1aVLlyS3JH7k73311VcX+Fop37vffomPVzpZrCRvfBmIrlmzhuOOO85FU5Kua9euALz44osAfP7557k+v0mTJgDcdtttQFADLN3UqVMHgJtvvplhw4ZFfa1Ro0ZccMEFLpqVdFJy6Oeffwb8k0E676QVT3KMSNmaTLF8+XLeeustILh4tG7dmt69e7tsVtJIYCM3F110EX/++ScQ9JGkv5xwwgkJa1uiyF7f8jvlVYpp8ODBQLCrzjXXXJN2N/FFMWfOHGbMmBH1WLdu3YDg2pRKDjzwQCAIZEgaRl6kZNOLL75of+/IMk+JolPzSimllFLKCScR0fPOOw/AlgcYPnw4rVu3BoKp2mLFirloWsJJMXLZ/WXz5s12IdN3330H+HuJy7SaTL2k877AKncSRVe5a9CgARAsCFy7dq3L5iRNrLIqp59+ui0VlykkMhpr0Unk7jgyq9SpU6fkNCwBJJo5ZswYwN9lK6dFR4MHD2bEiBFA8LunewkvCKKAUtpKZmIjUxI+/vhjwD8WJCVMrtEXX3wxkJo7k0WmYBSELHLatGmTHXPE2ms+3jQiqpRSSimlnHASET3ppJMAqF69OuAX7Za7ua+++irqa+lKFh/Vrl3b7pGtYps+fbrN55LcF6VyIlEAiXZs2rQpLfPhZAOMF1980eaC3XHHHUCQU55JZBOIcePG2YUnEglt0qQJL730EgAVKlQAggWSqUjOh7LotVGjRvYaKseCFDM/4ogj7JaPst3rsccem8zmOiG5jQ888ABAVJk/KeElazX++OMP+zVZoxBr++1UcckllwDw3nvvAfDoo4/aveObN28OBBHjd9991+bDyrFjjOGhhx4CoGHDhglvr5OBaIkSJYBghWe7du3sySHdB6AqdyeddJJdzbpr1y7AT1mQE2mmDUS7d++eEXUO40lq08quVK+88grXXnutyybFldRffvTRRwH/oiED7V69ejlrl2uy21q/fv3o16+f49YkltSflsHnxIkT7cAzcjAB/oBU0hCSsfAkLGbNmgXAX3/9BQQD0alTp9rFoKJcuXI2xaWgtb3DSHZLkvPCnDlzOPvsswE4+OCDgWA3w1WrVmWbwu/UqVO+KzLEg07NK6WUUkopJ0xkncIYcv1iCoh3lrH2R7SE9Mfxxx8PBFNt4N+1QdxLaaREfyRRIrLyk94nsruQLE44/fTT7WOFqIkXumNEpmUff/xx+9i8efOApEy5hq4/HHPaH7Jf/NChQ3Osy92hQwe7z3gShOb46NGjBwDjx4/P9jWZlZX61H369ElUmbxQ9IcxxkY9s0bMPc+zi9hkSv/OO+9MVDpTzP7QiKhSSimllHLCSY6oUrmRHVA6dOjguCUqFZ166qkAdO7cGYApU6Ywbdo0ILVL9gjJhRSjRo3KiMUnKjvJEZUSRSogMweyMHjlypWA32cy6yYlmtLdzJkz7SKs2bNnA0FEtGfPnracl6vziEZElVJKKaWUE5ojWjDaH9G0P6Jpf2SnfRJN+yOa9kc07Y9o2h/R0rI/NCKqlFJKKaWc0IGoUkoppZRyIq+peaWUUkoppRJCI6JKKaWUUsoJHYgqpZRSSikndCCqlFJKKaWc0IGoUkoppZRyQgeiSimllFLKCR2IKqWUUkopJ3QgqpRSSimlnNCBqFJKKaWUcsLJQNQYU88Y87Yx5ndjzC/GmCeMMcVdtCUMjDETjTHrjTFbjTHLjTFXum5TGBhjGhhjdhljJrpuSxhof/j0/BHNGLM9y7+9xpjHXbfLJT2nZmeM6WKMWWKM2WGM+Z8xpoXrNrmk59NoLo8PVxHRp4ANwEFAU6Al0MtRW8JgGFDP87wKQHtgsDGmueM2hcGTwDzXjQgR7Q+fnj8ieJ5XTv4BNYCdwCuOm+WanlMjGGPaACOAHkB54BTgB6eNck/Pp/u4Pj5cDUTrA1M8z9vled4vwEzgKEdtcc7zvO89z9st/93371CHTXLOGNMF2AJ86LgpoaD9EUXPHznrhD9In+O6IS7pOTWbe4H7PM/73PO8fzzPW+t53lrXjXJFz6fZOD0+XA1EHwG6GGPKGGNqAWfiX0wyljHmKWPMn8BSYD3wtuMmOWOMqQDcB9zsui1hoP2RzSPo+SMn3YEXPM/zXDfENT2n+owxxYDjgKrGmJXGmDX70llKu26bC3o+jRaG48PVQHQ2fgRjK7AGmA9Md9SWUPA8rxd+SLwFMBXYnft3pLVBwHOe561x3ZCQ0P6IpuePGIwxdfHTFMa7bksY6DnVqg6UAC7A74umQDNggMM2uaTn02jOj4+kD0SNMfvhRy+mAmWBKkAl/PyEjOZ53l7P8+YCtYFrXbfHBWNMU+B0YKTjpoSC9kc0PX/kqisw1/O8H103JCz0nAr4OcMAj3uet97zvE3Aw8BZDtvkhJ5PY3J+fLhYaVoZqAM8sS+HZ7cx5nlgMHCbg/aEUXEyN5+pFVAP+NkYA1AOKGaMaeR53rEO2+VKK7Q/Iun5I2fdgOGuGxFSGXtO9Tzvd2PMGvw8Wfuwq/Y41go9n0YJw/GR9IjovtH2j8C1xpjixpiK+HlNi5LdljAwxlTbVzahnDGmmDGmLXARmZtE/Qz+BaPpvn9PA28Bbd01ySntjwh6/ojNGHMSUAtdLa/n1NieB3rv65tKwE3ADMdtckHPp7E5PT5c5YieD7QDNgIrgb/xf/FM5OFPGa0BfgceBPp4nveG01Y54nnen57n/SL/gO3ALs/zNrpumwvaHzHp+SO77sBUz/O2uW5ICOg5NbtB+KWKlgNLgG+AIU5b5ICeT3Pk9PgwurhSKaWUUkq5oFt8KqWUUkopJ3QgqpRSSimlnNCBqFJKKaWUckIHokoppZRSyom86oim+komE+fX0/6Ipv0RTfsjO+2TaNof0bQ/oml/RNP+iJaW/aERUaWUUkop5YQORJVSSimllBM6EFVKKaWUUk7oQFQppZRSSjmhA1GllFJKAfDTTz9x4YUXcuGFF1KjRg1q1KjBwoULXTdLObZixQpWrFjByJEjqVmzJjVr1qR+/frUr1+fiy66qEivrQNRpZRSSinlRF7lm1SCjR8/ntdeew2AGTNmAOB5HsbErvowcOBArrzySgCqVasGQMmSJZPQ0sKT30U+lixZks8//xyAY445xlm7lFJK+b777jsA2rVrx7p16wD/WgQwefJkPVdnmJUrVwIwevRoACZMmADAr7/+mu25u3btYuPGjQBUrVq1wD9LI6JKKaWUUsoJI3c8OUhI8dRTTz0VgFmzZtnH7r77bgDuueeeeP6o0BaTHT9+PAB33XUXa9asif4huUREI7/26quvAtCxY8f8/lgn/bHffv79TrFixexjZ599NgDTp0+Pc5MKJLTHhyMpV9B+ypQpAAwbNowFCxbk+/suu+wynn/++fw8NRTHSNWqVenWrRsADz30UFwbVECh6A+A3bt3A/DNN98AMHfuXAA+/fRTO+Pyyy+/ZPs+Ofc8+OCDADRs2LCwTYAQ9Udhvf322wB2pi2yz2R88OSTT9KrV6/8vFzK90ecpVR/7N27F4DnnnuOW265BYBt27YBUKVKFQBOPPFEmjdv7jdm3/Exbtw4PvnkEwDq1q2b24+I2R9OBqI5DbIiffzxxwC0atWqSD+qKN8cQ9z6Y968eQCccMIJ9rHDDz8ciD1dvXz5cgAWLFhg+0+e98knn1C+fPn8/Fgn/fHFF18A0SH+du3aATB16lQASpQoketr/PXXXwD06dMHCC4i+++/P8WLFzrDJLTHxx9//AFA/fr1adq0KQAfffRRvr5Xptjq168PQNmyZfP7Y0M9EF28eDHgH0eTJk0CYOvWrUBwfOSXMcYO7PIYkIbiGKlWrRqbNm0CsAPuo48+Om6NKoBQ9AfAHXfcAcCIESMK9f1y3pg3b559jxVCaPqjoMaMGQNAv379ANi8eTMA1atX57777gP88yvApZdeGhVIyEXo+mPXrl0AnHvuuYAf/GnRokVRXza/QtcfscgAtGvXrgBMmjTJ/u1PO+00AEaOHAnEvnEbNWoUl156KUBeYxHdWUkppZRSSoVHKCKirVq1ipqmj/Txxx8XJSoa2ruRLVu2ADBkyBAOO+wwALp06QLAAQcckO35Eh4/7bTT+PrrrwG48MILAT+RPJ+c9sfAgQMBGD58uH1MIps33nhjrt/bo0cPACZOnBj1+OTJk+nUqVNBmhEptMeH3F1OmjTJRso//fRTACpXrpzj961cudI+X2YVWrZsmd8fG+qIaJ06dQCypbIU1T///JPbl0NxjERGRNu2bQvAK6+8AkC5cuXi1LR8CUV/AMycOROAM888E8BGNU844QQb7bziiivs8/v27Qtkn1mYNGmSPfcWQmj6oyBmzJhhI4RyPZYp1Q8//JBDDjmksC8duv4YPHgwEFx/xo4da68nuZEUhQceeMCmDOZz5jFS6Pojq02bNtn+kAXTEKQ9yTgjTjQiqpRSSimlwiMU5ZtatmxpozcSGZUFTaeeeqqNiMqCpiLmjYZCxYoVAf9uKz/kTqxZs2Z89dVXACxbtgzwo6WFuFNLGfPmzbO/c7p79tlnAWxJL4BDDz0UyD0SKjmSktsFfgI5FCgiGiqStyQRjbVr1+br+2rXrg0EEY09e/YkoHXuSCRQIjySu5VpTj/9dAB+/PFHIHh/VKhQoUCvc9BBB8W3YSE2e/ZsIMiXhCA/X8rzFCEaGiqS8/rEE08AUK9ePYA8o6Fy3pDja+XKlXYW4owzzkhEU51q27atnWWVvNBPP/2U4447LmltCMVANJIMMiVl4J577uHee+8FgkFqHukEaeXbb78FsCvSxowZY6dSatWq5axdySADkXnz5rFkyZKor8l0gZws0kXPnj2BYLrsyCOP5Omnn87z+2Qa5cUXX7SPVapUKQEtTJ6dO3cC8NhjjwHR7/uDDz4YwK7kbdCggf2a1NedP38+4Ke/yEUpkqwCTVWjRo0CoEmTJlx++eX5/r7ly5fb519yySUAXHvttfFvYILJ9LsMMHLzxRdf2AWiQgZcRxxxRNzbFjYrVqwAgqo0xhg7AJUbm5NPPtlJ2xLl3XffBYK6l3fddVeuz5cbXamq8P333wN+SlA6DkBvvfVWwF/8KOdMSf+SdMFk0al5pZRSSinlROgiolndc8898a4tGnp//vmnjfRInU1ZrBRpwIABQKESqFOC7O4RayGTTMPFWtiViqQcUVZTp061U82xSNQwcnpWIqGpGOWKJGWnpL5hZBrLG2+8AUSXOpOIqUQ+hgwZkuNrG2Ps+ycVeJ5nF6F1794dgEGDBgH+ghw5P8i0Y4UKFfj777+BoHyaRIhHjBhhpx+lLFyqHytZSaqKzCjdeOON2c6hUraoRo0ayW2cA48//jgQXbv7ww8/BNIvEiokNSm/ZHZg4cKFQJCykNt5JBUtWrQICBYKQ3BuTXYkVGhEVCmllFJKORGKiKjkP+ZFIqPpECGVu3MpKgzBQpUtW7bE3M9VSG7Ysccem8AWuvPSSy8BQY5Pulu+fDnXX389EET1JC9SomA5kWLesrtMpUqVbNQjMm8yFcl7JL8L+iQKlp8IRtWqVbnhhhsK37gkM8bYAvYSyZNFbN27d+emm24C4JFHHgGgVKlStiyV5AfG0qRJk0Q12anrrrsOiD6/Cuk/ibSnu8GDB9trS8mSJQH//JKukVCxYcOGfD93xIgR/Pe//4167KSTTgKCUnqpTjZJyVrusH///nTu3NlFkyyNiCqllFJKKSdCERGdNWtWtiinREljFbpv1apVypVwkn2QZUtL+f0kkhUpcj/5MmXKANCmTRvAzwuVfV7TgRS3b9asGQA///yzzQmV7RsjyQrZvFZAppJ169bZDQ7k7y6rvKVoOQSrOE888UQb2ZDV8vJ9AwcOTJsol+Q+y1aOkRshSFRL9smG4D2WH9u3b7fvwVQpb9W4ceOo/0sUY7/99qN///5AEP2MPIfEcuSRRwJB7mC6ee+997I9VrVqVQA7+5DupATc8OHD2b17N4DdP/yqq65y1i5XZAbkt99+syWd5Hr87bffZtvYIt0qsqxcuTLqo1TdueOOOyhVqpSzdoGDgWhOOyhJiaaspHYoBKWdUm0QCvDyyy8D8NRTTwHBFGxOFwuphSf7YHfs2DHRTUw42fmkSpUqdpcY+di6det8vYYkkEu5iVS2Y8cOAG677bZsX4usB5pVsWLF7L7PMh0tJ810usjKe0NuUuR33rt3r607LOeN+fPn5zvFB2Do0KEpMwAFWLp0aY6LEi+44AI7jSjlu8aOHctPP/0EYAchomLFiowfPx6ARo0aJajFbsniPdlZacuWLWzcuBHAlkPL7T2WDmSxzo4dO2jXrh2Q/r9zbmQHrsgyXlI3s0KFCjYYIGSxZDrYsWMHH3zwQdRjL7zwApD0ndli0ql5pZRSSinlRNL2ms+6Y1JOJNIRp6hnaPZ5lYiFlFKRfj/ppJP4888/gWBhypQpU7jssssAP7IRR6Hoj5NPPtn2Q0HJApzFixcX6vuzcNofkkwfubNLbpFymUrxPM+WtpLIuZSpya3UUz6Eeq95mT6SKHBBSDqLLF45+OCDc92pKkIo3jMF9euvv9pIspRqEsOHD48Zhc+nlOqP7du3A34Kgixik7JWn332GVDkRZ+h6w9Jc5PZgnLlyjFnzhwg+vwgGwLILn9xEpr+kIWcsgOZ/N2LFy9ur8eS9vP333/ToUMHICjjJBHD/fYrUrwuFP2xZcsWm84k59G8dmSTKfzRo0cDwUK3O+64oyhRVN1rXimllFJKhUfSckQjc0Mj8z4hOj80FfM/80OSw2Xf67POOguA22+/3T4mOWD33HOPTaI+4YQTAJg2bRqQHvsiT5482W6ZJjmiv//+O+DfvcvvKFt8/vDDDw5amXilS5cG/CLCcvcppLBwhw4dbHRcInhdunSxEVH5WhEjoaG0evVqIDhGCrq1rxS7P+aYY2zubGQB/HQkZd9OOeUU1q9fH/U1OQcVIRqaciRy069fP7uVpey3Lovb0q0MnhSqF3v27LEbFshWySVKlLCLc6TAebptGX377bcDwWyIbP5RsmTJbFt2ynkU4KGHHgKKHAkNlS1btticYdn0Iha55g4cONAu6Mq6EUS9evXiXvos6YuV7r777rSqB5pf8sfP7SAQDRs2tG+CL7/8EgjC4+nQZ7Vr17ZT67KyU1IymjVrZhcYyEBE9oRON3LjsWzZsnw9X3bZ+uSTT+zFc+jQoQlpmysy7b5kyRIuvPBCgGyD9JzIFON5550HBO8ZWeCWzuRmdvDgwYBfm1bSO6RuYE4LQjOFvGdkICoVA1KpnmxupGJC1kU3u3fvtgNQqUX75Zdf2kVsssd6ug1ERW6r32UQPm3aNA4++GAgmIJOJ9u3b2fXrl0AtG/fPtvX5QZWdm3LrYb3d999F/f2pc+QXymllFJKpZSERURlKl7uwmU6Pt2m3uWuUu5Gs9b6KyxZuHTEEUcAcPbZZ8fldcNGojVZd3tQ2UnSvTHG1gOU6f1UtnfvXhuZkAUGUh81vypWrGij6uk+/Z7V+vXr7bR7ZF1VOYfceuutQHqk9RTF0qVLo/4vUaDt27eHooRNUcmUstQbFqVKlbL7iq9ZswaAc845x05dy85BUt6nTp06SWlvGMiCna1bt3L++ecDcV+8FQpfffWV/fzHH3+M+tqmTZtsaavIuuYyIyWzVK+//nrC2qcRUaWUUkop5UTCI6LyUYpHp1tEVPYElzvNRx99NC6vm3WfcIkQ/etf/4rL66vUIYsqli9fbh+rWbOmq+bE3WWXXWYLsRdWuXLlMi4SKr788suoSCj4m0fIY5keCQV/B52si3iknM+WLVvSIiIqCxazLuobPHhwzMUlEhGV2bwFCxYAmRURlR3sAK677jqHLUksKVcF8N///hfALmBr27atjYTKzoX333+/LegvkWL5v3xfPGlEVCmllFJKOeF0r/nIVZyS35VqJNoge5///fffdhvPwtq2bRvPPfccEETBInM8MpmUeZLoRn63Bk1lslWjrIw++OCD02JrxosuuggoeD7oCSecwOeffx712Nq1a+3Wjddcc018GhhysgpWfu9IM2fOpHr16sluUmhItPPGG28E4Nlnn7XvHyF56elS+kxm56Ragvz9O3XqZNcyyMydVCaBoLC9bHeZCaR0VeQxkU5beuZm/vz5gF9GEfxI+JFHHgnAjBkzAP+cLBV65NiRCi2ybiWekjYQjSw7FGt3pVSdspcSPPLHevrpp+1etgMGDADgxBNPzLY3upQmWr9+vT0gJDw+a9YsezIpU6YMADfffHMif42UITUlZXHGCy+8YKep+/TpA/g3BcOGDQOCC5J44403ktTS+Ni5c2e2HTAee+wxqlat6qhF8SPHfaxdpHIjU4mRPM/L9rdOV1u3bgX8PeYB3n//ffs1eSwTB6EyqJg8eTIPP/wwEL34Qsj154EHHkhe4xyQHbXOOOMMW4M41o525557LoDdjz4TyMK1d955B/DLWtWoUcNlkxKqcuXKNGzYEAh+96uvvhrwB+VSr1yCA1I2EqBbt24A9O7dO2Ht06l5pZRSSinlRNIjoq1atYraZQlSd1oeoE2bNkCQDPzZZ5/x9ddfA9CxY0cAqlSpki0K/MorrwCxo0HGGPu43LWcc845CWh9uElh4Vg7D0lEuU2bNravJCL2wQcf2OLVqb47xogRI2xUp1KlSgB2T+RMFbnAQFSvXt2WG0l3sj+2lNsxxnDooYcCwf7Y6UpmRCZOnGgfkw0hJF1DFt1EKlmypC15JtcimZJOFwcccAAQ9JGINYMAwQLiMWPGJLZhKaBevXr2/JqODjzwQDsb2KRJEyB6x6Tnn38+2/dI+bdBgwYBid0YJLWv0koppZRSKmUl7ZZQFiZFLlBKhyL3ksM5duxYwC8Cm3ULrM2bN/Pqq6/m+zU7d+5s8x4zeVs+yat96aWX7PaWkpC/fft2wM8Lk3y4yAR8Kf6eqlEPifhOmTLFRnzld0oXst2t7IFcGKVKlQLgyCOPTOscLwgioKNGjQKiZ1N69uwJpMcGB7HIgj2ZZYoV9Ywk0a3jjjsO8LfzTMQiizCRBa5Zr6elSpXikksuAaBChQqAv4ApsqSPSn9SEnLChAmAP87ISgrbH3fccXbRZzJKBZqsNceyyPWLuZHp99wWJiVhSr5gqyDylmd/bNiwwa6af+uttwDsVD1A8+bN/Rfa1+81a9bkrLPOAoLdkxK4ijPp/RFvUgNN9qh/7LHHeO+994DYx1oeQtsfcky8++67djXrwoULgWDKJAHi3R+QS5/IDdu5555r/3Zyw7Flyxa7EDDW4EpWQx911FFA7P2T4yQ0x4jc7MoNl0yVDR061C5mTEIqipP+yM++8LIjzuWXX06/fv0APy0qwUJzfIRESvTH4sWLgejzR4J2DkqJ/kiimP2hU/NKKaWUUsqJhEVEhUQ6JEJ69913R5VySjC9G4mm/REtdP2xY8cOAE4++WQAFi1aZKfVkrAQJakR0VjWrVsH+OkYshDQ8Y5JoTlGZG90SUGQun4S/UsSJ/0hqSqy4GjVqlV2ylAW70ld0CTvkhSa4yMkUqI/NCLqjEZElVJKKaVUeCQ8IuqY3o1E0/6IFrr+kLxJiQKWLl2aOXPmANCsWbOivnxenEdEQyh0x4hj2h/RtD+ipUR/yOxCly5dAH8jmgQtZkuJ/kgijYgqpZRSSqnw0IhowWh/RNP+iKb9kZ32STTtj2jaH9G0P6Jpf0RLy/7QiKhSSimllHJCB6JKKaWUUsqJvKbmlVJKKaWUSgiNiCqllFJKKSd0IKqUUkoppZzQgahSSimllHJCB6JKKaWUUsoJHYgqpZRSSikndCCqlFJKKaWc+H+k1iwIe7CztQAAAABJRU5ErkJggg==\n"
     },
     "metadata": {
      "needs_background": "light",
      "image/png": {
       "width": 674,
       "height": 291
      }
     },
     "output_type": "display_data"
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": "Let's build a simple dense network and find the optimal learning rate. We will need a callback to grow the learning rate at each iteration. It will also record the learning rate and the loss at each iteration:",
   "metadata": {
    "cell_id": "00162-658b73da-c3f7-4121-8983-24acee4fabf6",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 74.78125
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00163-2d18400c-72e7-47f9-8bb4-d5089733b527",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "d85ee5c7",
    "execution_start": 1659187618071,
    "execution_millis": 6,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 256
   },
   "source": "K = keras.backend\n\nclass ExponentialLearningRate(keras.callbacks.Callback):\n    def __init__(self, factor):\n        self.factor = factor\n        self.rates = []\n        self.losses = []\n    def on_batch_end(self, batch, logs):\n        self.rates.append(K.get_value(self.model.optimizer.learning_rate))\n        self.losses.append(logs[\"loss\"])\n        K.set_value(self.model.optimizer.learning_rate, self.model.optimizer.learning_rate * self.factor)",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00164-a4a8f0a6-08aa-445e-a756-85ffe0b19cea",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "b86c7ea",
    "execution_start": 1659187618081,
    "execution_millis": 7,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 112
   },
   "source": "keras.backend.clear_session()\nnp.random.seed(42)\ntf.random.set_seed(42)",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00165-75291e58-158a-4354-8582-2a3215d2a8fe",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "e2bfe76c",
    "execution_start": 1659187618096,
    "execution_millis": 46,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 166
   },
   "source": "model = keras.models.Sequential([\n    keras.layers.Flatten(input_shape=[28, 28]),\n    keras.layers.Dense(300, activation=\"relu\"),\n    keras.layers.Dense(100, activation=\"relu\"),\n    keras.layers.Dense(10, activation=\"softmax\")\n])",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "markdown",
   "source": "We will start with a small learning rate of 1e-3, and grow it by 0.5% at each iteration:",
   "metadata": {
    "cell_id": "00166-aede6a80-6189-4e34-9921-e0f88c71ee07",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 52.390625
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00167-e673d1bd-51ed-41b8-8c25-cfb518de89b4",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "df88f995",
    "execution_start": 1659187618142,
    "execution_millis": 0,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 130
   },
   "source": "model.compile(loss=\"sparse_categorical_crossentropy\",\n              optimizer=keras.optimizers.SGD(learning_rate=1e-3),\n              metrics=[\"accuracy\"])\nexpon_lr = ExponentialLearningRate(factor=1.005)",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "markdown",
   "source": "Now let's train the model for just 1 epoch:",
   "metadata": {
    "cell_id": "00168-cdf66aec-7236-40ce-9cef-287924bbf2cb",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 52.390625
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00169-d3ba4328-804d-473d-ae67-239e32ac360b",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "d0b90c6e",
    "execution_start": 1659187618185,
    "execution_millis": 5957,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 147.1875
   },
   "source": "history = model.fit(X_train, y_train, epochs=1,\n                    validation_data=(X_valid, y_valid),\n                    callbacks=[expon_lr])",
   "execution_count": null,
   "outputs": [
    {
     "name": "stdout",
     "text": "1719/1719 [==============================] - 6s 3ms/step - loss: 51261587931141196574883840.0000 - accuracy: 0.5922 - val_loss: 2.3912 - val_accuracy: 0.1126\n",
     "output_type": "stream"
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": "We can now plot the loss as a functionof the learning rate:",
   "metadata": {
    "cell_id": "00170-b23da088-c0b7-4bef-bbc8-eecc7230275f",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 52.390625
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00171-b39e1889-b938-42e7-b9cc-72182f05aad9",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "4ca6abf7",
    "execution_start": 1659187624145,
    "execution_millis": 507,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 527.1875,
    "deepnote_output_heights": [
     21.1875,
     275
    ]
   },
   "source": "plt.plot(expon_lr.rates, expon_lr.losses)\nplt.gca().set_xscale('log')\nplt.hlines(min(expon_lr.losses), min(expon_lr.rates), max(expon_lr.rates))\nplt.axis([min(expon_lr.rates), max(expon_lr.rates), 0, expon_lr.losses[0]])\nplt.grid()\nplt.xlabel(\"Learning rate\")\nplt.ylabel(\"Loss\")",
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "execution_count": 120,
     "data": {
      "text/plain": "Text(0, 0.5, 'Loss')"
     },
     "metadata": {}
    },
    {
     "data": {
      "text/plain": "<Figure size 432x288 with 1 Axes>",
      "image/png": "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\n"
     },
     "metadata": {
      "needs_background": "light",
      "image/png": {
       "width": 393,
       "height": 273
      }
     },
     "output_type": "display_data"
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": "The loss starts shooting back up violently when the learning rate goes over 6e-1, so let's try using half of that, at 3e-1:",
   "metadata": {
    "cell_id": "00172-e719558e-6337-4eda-90ab-a10bca979468",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 52.390625
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00173-cc5d0b56-4ab7-40a5-bd38-fee96d6ab637",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "b86c7ea",
    "execution_start": 1659187624656,
    "execution_millis": 9,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 112
   },
   "source": "keras.backend.clear_session()\nnp.random.seed(42)\ntf.random.set_seed(42)",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00174-492545c8-53d8-4ec5-8f58-93c98a7d7c05",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "e2bfe76c",
    "execution_start": 1659187624674,
    "execution_millis": 25,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 166
   },
   "source": "model = keras.models.Sequential([\n    keras.layers.Flatten(input_shape=[28, 28]),\n    keras.layers.Dense(300, activation=\"relu\"),\n    keras.layers.Dense(100, activation=\"relu\"),\n    keras.layers.Dense(10, activation=\"softmax\")\n])",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00175-342e7e7a-8558-4507-ae31-f9c02be8ac42",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "3a042439",
    "execution_start": 1659187624706,
    "execution_millis": 4,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 112
   },
   "source": "model.compile(loss=\"sparse_categorical_crossentropy\",\n              optimizer=keras.optimizers.SGD(learning_rate=3e-1),\n              metrics=[\"accuracy\"])",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00176-11a7c299-f19e-4e48-9910-ada71de67eb2",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "ffc9457c",
    "execution_start": 1659187624715,
    "execution_millis": 4,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 149.1875,
    "deepnote_output_heights": [
     21.1875
    ]
   },
   "source": "run_index = 1 # increment this at every run\nrun_logdir = os.path.join(os.curdir, \"my_mnist_logs\", \"run_{:03d}\".format(run_index))\nrun_logdir",
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "execution_count": 124,
     "data": {
      "text/plain": "'./my_mnist_logs/run_001'"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00177-e5f22ee8-f332-42b4-af60-0bb785562582",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "3aee099d",
    "execution_start": 1659187624723,
    "execution_millis": 106216,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 804
   },
   "source": "early_stopping_cb = keras.callbacks.EarlyStopping(patience=20)\ncheckpoint_cb = keras.callbacks.ModelCheckpoint(\"my_mnist_model.h5\", save_best_only=True)\ntensorboard_cb = keras.callbacks.TensorBoard(run_logdir)\n\nhistory = model.fit(X_train, y_train, epochs=100,\n                    validation_data=(X_valid, y_valid),\n                    callbacks=[checkpoint_cb, early_stopping_cb, tensorboard_cb])",
   "execution_count": null,
   "outputs": [
    {
     "name": "stdout",
     "text": "Epoch 1/100\n1719/1719 [==============================] - 5s 3ms/step - loss: 0.2360 - accuracy: 0.9263 - val_loss: 0.1022 - val_accuracy: 0.9720\nEpoch 2/100\n1719/1719 [==============================] - 5s 3ms/step - loss: 0.0951 - accuracy: 0.9699 - val_loss: 0.1032 - val_accuracy: 0.9706\nEpoch 3/100\n1719/1719 [==============================] - 5s 3ms/step - loss: 0.0660 - accuracy: 0.9792 - val_loss: 0.0691 - val_accuracy: 0.9790\nEpoch 4/100\n1719/1719 [==============================] - 4s 3ms/step - loss: 0.0487 - accuracy: 0.9835 - val_loss: 0.0778 - val_accuracy: 0.9798\nEpoch 5/100\n1719/1719 [==============================] - 5s 3ms/step - loss: 0.0352 - accuracy: 0.9889 - val_loss: 0.0815 - val_accuracy: 0.9786\nEpoch 6/100\n1719/1719 [==============================] - 5s 3ms/step - loss: 0.0300 - accuracy: 0.9902 - val_loss: 0.0856 - val_accuracy: 0.9796\nEpoch 7/100\n1719/1719 [==============================] - 5s 3ms/step - loss: 0.0215 - accuracy: 0.9931 - val_loss: 0.0842 - val_accuracy: 0.9806\nEpoch 8/100\n1719/1719 [==============================] - 5s 3ms/step - loss: 0.0194 - accuracy: 0.9933 - val_loss: 0.0849 - val_accuracy: 0.9800\nEpoch 9/100\n1719/1719 [==============================] - 5s 3ms/step - loss: 0.0142 - accuracy: 0.9951 - val_loss: 0.0921 - val_accuracy: 0.9820\nEpoch 10/100\n1719/1719 [==============================] - 4s 3ms/step - loss: 0.0129 - accuracy: 0.9955 - val_loss: 0.0790 - val_accuracy: 0.9828\nEpoch 11/100\n1719/1719 [==============================] - 5s 3ms/step - loss: 0.0147 - accuracy: 0.9954 - val_loss: 0.0820 - val_accuracy: 0.9842\nEpoch 12/100\n1719/1719 [==============================] - 4s 3ms/step - loss: 0.0103 - accuracy: 0.9967 - val_loss: 0.0915 - val_accuracy: 0.9802\nEpoch 13/100\n1719/1719 [==============================] - 5s 3ms/step - loss: 0.0119 - accuracy: 0.9965 - val_loss: 0.0971 - val_accuracy: 0.9804\nEpoch 14/100\n1719/1719 [==============================] - 5s 3ms/step - loss: 0.0081 - accuracy: 0.9975 - val_loss: 0.0884 - val_accuracy: 0.9834\nEpoch 15/100\n1719/1719 [==============================] - 5s 3ms/step - loss: 0.0057 - accuracy: 0.9983 - val_loss: 0.1142 - val_accuracy: 0.9792\nEpoch 16/100\n1719/1719 [==============================] - 5s 3ms/step - loss: 0.0029 - accuracy: 0.9991 - val_loss: 0.0910 - val_accuracy: 0.9836\nEpoch 17/100\n1719/1719 [==============================] - 4s 3ms/step - loss: 3.8993e-04 - accuracy: 1.0000 - val_loss: 0.0893 - val_accuracy: 0.9854\nEpoch 18/100\n1719/1719 [==============================] - 5s 3ms/step - loss: 1.2783e-04 - accuracy: 1.0000 - val_loss: 0.0911 - val_accuracy: 0.9846\nEpoch 19/100\n1719/1719 [==============================] - 4s 3ms/step - loss: 9.1791e-05 - accuracy: 1.0000 - val_loss: 0.0924 - val_accuracy: 0.9846\nEpoch 20/100\n1719/1719 [==============================] - 5s 3ms/step - loss: 7.6138e-05 - accuracy: 1.0000 - val_loss: 0.0936 - val_accuracy: 0.9846\nEpoch 21/100\n1719/1719 [==============================] - 4s 3ms/step - loss: 6.5733e-05 - accuracy: 1.0000 - val_loss: 0.0945 - val_accuracy: 0.9848\nEpoch 22/100\n1719/1719 [==============================] - 5s 3ms/step - loss: 5.8258e-05 - accuracy: 1.0000 - val_loss: 0.0953 - val_accuracy: 0.9850\nEpoch 23/100\n1719/1719 [==============================] - 4s 3ms/step - loss: 5.2673e-05 - accuracy: 1.0000 - val_loss: 0.0960 - val_accuracy: 0.9850\n",
     "output_type": "stream"
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00178-5d767fe1-e54d-4c46-8c34-c51992e08940",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "61259277",
    "execution_start": 1659187730939,
    "execution_millis": 1164,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 177.375,
    "deepnote_output_heights": [
     null,
     21.1875
    ]
   },
   "source": "model = keras.models.load_model(\"my_mnist_model.h5\") # rollback to best model\nmodel.evaluate(X_test, y_test)",
   "execution_count": null,
   "outputs": [
    {
     "name": "stdout",
     "text": "313/313 [==============================] - 1s 2ms/step - loss: 0.0833 - accuracy: 0.9744\n",
     "output_type": "stream"
    },
    {
     "output_type": "execute_result",
     "execution_count": 126,
     "data": {
      "text/plain": "[0.08327189832925797, 0.974399983882904]"
     },
     "metadata": {}
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": "We got over 98% accuracy. Finally, let's look at the learning curves using TensorBoard:",
   "metadata": {
    "cell_id": "00179-4b8508b1-4000-49f8-871f-45d78a84b9bc",
    "deepnote_cell_type": "markdown",
    "deepnote_cell_height": 52.390625
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00180-59c7ef7b-717c-4a43-8d4d-a43306ab23d0",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "c522cde0",
    "execution_start": 1659187731813,
    "execution_millis": 6055,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 113.1875
   },
   "source": "%tensorboard --logdir=./my_mnist_logs --port=6006",
   "execution_count": null,
   "outputs": [
    {
     "data": {
      "text/plain": "Launching TensorBoard..."
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "cell_id": "00181-ae2f5e0b-f63d-475d-af69-188ae58ddd8e",
    "deepnote_to_be_reexecuted": false,
    "source_hash": "b623e53d",
    "execution_start": 1659187737873,
    "execution_millis": 2,
    "deepnote_cell_type": "code",
    "deepnote_cell_height": 76
   },
   "source": "",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "markdown",
   "source": "<a style='text-decoration:none;line-height:16px;display:flex;color:#5B5B62;padding:10px;justify-content:end;' href='https://deepnote.com?utm_source=created-in-deepnote-cell&projectId=8b8747e0-d5e2-4759-af6a-42b7fe569db7' target=\"_blank\">\n<img alt='Created in deepnote.com' style='display:inline;max-height:16px;margin:0px;margin-right:7.5px;' src='data:image/svg+xml;base64,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' > </img>\nCreated in <span style='font-weight:600;margin-left:4px;'>Deepnote</span></a>",
   "metadata": {
    "tags": [],
    "created_in_deepnote_cell": true,
    "deepnote_cell_type": "markdown"
   }
  }
 ],
 "nbformat": 4,
 "nbformat_minor": 4,
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.12"
  },
  "nav_menu": {
   "height": "264px",
   "width": "369px"
  },
  "toc": {
   "navigate_menu": true,
   "number_sections": true,
   "sideBar": true,
   "threshold": 6,
   "toc_cell": false,
   "toc_section_display": "block",
   "toc_window_display": false
  },
  "widgets": {
   "application/vnd.jupyter.widget-state+json": {
    "state": {},
    "version_major": 2,
    "version_minor": 0
   }
  },
  "deepnote_notebook_id": "d3118ad9-4c2f-49e2-915b-80cd78cc7dc0",
  "deepnote": {},
  "deepnote_execution_queue": []
 }
}