Model_creation.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "152267b8",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "152267b8",
    "outputId": "5ca605e3-2892-48d9-c86f-4db86a3348ea"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[33mWARNING: Skipping opencv-python-headless as it is not installed.\u001b[0m\n"
     ]
    }
   ],
   "source": [
    "!pip uninstall opencv-python-headless==4.5.3.56 --yes\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "7d4aa13b",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "7d4aa13b",
    "outputId": "a52cf9b1-9224-4d37-fb2b-9be0b179a9f9"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[K     |████████████████████████████████| 1.8 MB 5.5 MB/s \n",
      "\u001b[K     |████████████████████████████████| 144 kB 24.0 MB/s \n",
      "\u001b[K     |████████████████████████████████| 181 kB 36.1 MB/s \n",
      "\u001b[K     |████████████████████████████████| 63 kB 1.1 MB/s \n",
      "\u001b[?25h  Building wheel for pathtools (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
      "Requirement already satisfied: opencv-python==4.1.2.30 in /usr/local/lib/python3.7/dist-packages (4.1.2.30)\n",
      "Requirement already satisfied: numpy>=1.14.5 in /usr/local/lib/python3.7/dist-packages (from opencv-python==4.1.2.30) (1.21.5)\n",
      "Requirement already satisfied: imageio in /usr/local/lib/python3.7/dist-packages (2.4.1)\n",
      "Requirement already satisfied: pillow in /usr/local/lib/python3.7/dist-packages (from imageio) (7.1.2)\n",
      "Requirement already satisfied: numpy in /usr/local/lib/python3.7/dist-packages (from imageio) (1.21.5)\n",
      "Collecting onnxruntime\n",
      "  Downloading onnxruntime-1.11.0-cp37-cp37m-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (5.2 MB)\n",
      "\u001b[K     |████████████████████████████████| 5.2 MB 5.4 MB/s \n",
      "\u001b[?25hRequirement already satisfied: protobuf in /usr/local/lib/python3.7/dist-packages (from onnxruntime) (3.17.3)\n",
      "Requirement already satisfied: numpy>=1.21.0 in /usr/local/lib/python3.7/dist-packages (from onnxruntime) (1.21.5)\n",
      "Requirement already satisfied: flatbuffers in /usr/local/lib/python3.7/dist-packages (from onnxruntime) (2.0)\n",
      "Requirement already satisfied: six>=1.9 in /usr/local/lib/python3.7/dist-packages (from protobuf->onnxruntime) (1.15.0)\n",
      "Installing collected packages: onnxruntime\n",
      "Successfully installed onnxruntime-1.11.0\n",
      "Collecting keras2onnx\n",
      "  Downloading keras2onnx-1.7.0-py3-none-any.whl (96 kB)\n",
      "\u001b[K     |████████████████████████████████| 96 kB 3.0 MB/s \n",
      "\u001b[?25hRequirement already satisfied: protobuf in /usr/local/lib/python3.7/dist-packages (from keras2onnx) (3.17.3)\n",
      "Collecting onnx\n",
      "  Downloading onnx-1.11.0-cp37-cp37m-manylinux_2_12_x86_64.manylinux2010_x86_64.whl (12.8 MB)\n",
      "\u001b[K     |████████████████████████████████| 12.8 MB 32.9 MB/s \n",
      "\u001b[?25hCollecting fire\n",
      "  Downloading fire-0.4.0.tar.gz (87 kB)\n",
      "\u001b[K     |████████████████████████████████| 87 kB 6.5 MB/s \n",
      "\u001b[?25hRequirement already satisfied: requests in /usr/local/lib/python3.7/dist-packages (from keras2onnx) (2.23.0)\n",
      "Requirement already satisfied: numpy in /usr/local/lib/python3.7/dist-packages (from keras2onnx) (1.21.5)\n",
      "Collecting onnxconverter-common>=1.7.0\n",
      "  Downloading onnxconverter_common-1.9.0-py2.py3-none-any.whl (78 kB)\n",
      "\u001b[K     |████████████████████████████████| 78 kB 7.1 MB/s \n",
      "\u001b[?25hRequirement already satisfied: six in /usr/local/lib/python3.7/dist-packages (from fire->keras2onnx) (1.15.0)\n",
      "Requirement already satisfied: termcolor in /usr/local/lib/python3.7/dist-packages (from fire->keras2onnx) (1.1.0)\n",
      "Requirement already satisfied: typing-extensions>=3.6.2.1 in /usr/local/lib/python3.7/dist-packages (from onnx->keras2onnx) (4.1.1)\n",
      "Requirement already satisfied: chardet<4,>=3.0.2 in /usr/local/lib/python3.7/dist-packages (from requests->keras2onnx) (3.0.4)\n",
      "Requirement already satisfied: urllib3!=1.25.0,!=1.25.1,<1.26,>=1.21.1 in /usr/local/lib/python3.7/dist-packages (from requests->keras2onnx) (1.24.3)\n",
      "Requirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.7/dist-packages (from requests->keras2onnx) (2021.10.8)\n",
      "Requirement already satisfied: idna<3,>=2.5 in /usr/local/lib/python3.7/dist-packages (from requests->keras2onnx) (2.10)\n",
      "Building wheels for collected packages: fire\n",
      "  Building wheel for fire (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
      "  Created wheel for fire: filename=fire-0.4.0-py2.py3-none-any.whl size=115942 sha256=80910d4ca9168e63380150ae17fe096123e8039cf6e7c4b76597ae1f2e4ccac8\n",
      "  Stored in directory: /root/.cache/pip/wheels/8a/67/fb/2e8a12fa16661b9d5af1f654bd199366799740a85c64981226\n",
      "Successfully built fire\n",
      "Installing collected packages: onnx, onnxconverter-common, fire, keras2onnx\n",
      "Successfully installed fire-0.4.0 keras2onnx-1.7.0 onnx-1.11.0 onnxconverter-common-1.9.0\n",
      "Collecting tf2onnx\n",
      "  Downloading tf2onnx-1.9.3-py3-none-any.whl (435 kB)\n",
      "\u001b[K     |████████████████████████████████| 435 kB 5.3 MB/s \n",
      "\u001b[?25hRequirement already satisfied: numpy>=1.14.1 in /usr/local/lib/python3.7/dist-packages (from tf2onnx) (1.21.5)\n",
      "Requirement already satisfied: requests in /usr/local/lib/python3.7/dist-packages (from tf2onnx) (2.23.0)\n",
      "Requirement already satisfied: onnx>=1.4.1 in /usr/local/lib/python3.7/dist-packages (from tf2onnx) (1.11.0)\n",
      "Requirement already satisfied: six in /usr/local/lib/python3.7/dist-packages (from tf2onnx) (1.15.0)\n",
      "Collecting flatbuffers~=1.12\n",
      "  Downloading flatbuffers-1.12-py2.py3-none-any.whl (15 kB)\n",
      "Requirement already satisfied: typing-extensions>=3.6.2.1 in /usr/local/lib/python3.7/dist-packages (from onnx>=1.4.1->tf2onnx) (4.1.1)\n",
      "Requirement already satisfied: protobuf>=3.12.2 in /usr/local/lib/python3.7/dist-packages (from onnx>=1.4.1->tf2onnx) (3.17.3)\n",
      "Requirement already satisfied: urllib3!=1.25.0,!=1.25.1,<1.26,>=1.21.1 in /usr/local/lib/python3.7/dist-packages (from requests->tf2onnx) (1.24.3)\n",
      "Requirement already satisfied: chardet<4,>=3.0.2 in /usr/local/lib/python3.7/dist-packages (from requests->tf2onnx) (3.0.4)\n",
      "Requirement already satisfied: idna<3,>=2.5 in /usr/local/lib/python3.7/dist-packages (from requests->tf2onnx) (2.10)\n",
      "Requirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.7/dist-packages (from requests->tf2onnx) (2021.10.8)\n",
      "Installing collected packages: flatbuffers, tf2onnx\n",
      "  Attempting uninstall: flatbuffers\n",
      "    Found existing installation: flatbuffers 2.0\n",
      "    Uninstalling flatbuffers-2.0:\n",
      "      Successfully uninstalled flatbuffers-2.0\n",
      "\u001b[31mERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.\n",
      "tensorflow 2.8.0 requires tf-estimator-nightly==2.8.0.dev2021122109, which is not installed.\u001b[0m\n",
      "Successfully installed flatbuffers-1.12 tf2onnx-1.9.3\n",
      "Requirement already satisfied: pyyaml in /usr/local/lib/python3.7/dist-packages (3.13)\n",
      "Requirement already satisfied: h5py in /usr/local/lib/python3.7/dist-packages (3.1.0)\n",
      "Requirement already satisfied: numpy>=1.14.5 in /usr/local/lib/python3.7/dist-packages (from h5py) (1.21.5)\n",
      "Requirement already satisfied: cached-property in /usr/local/lib/python3.7/dist-packages (from h5py) (1.5.2)\n"
     ]
    }
   ],
   "source": [
    "# WandB – Install the W&B library\n",
    "!pip install wandb -q\n",
    "!pip install opencv-python==4.1.2.30\n",
    "!pip install imageio\n",
    "!pip install onnxruntime\n",
    "!pip install keras2onnx\n",
    "!pip install tf2onnx\n",
    "!pip install pyyaml h5py\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "7b9d8701",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 86
    },
    "id": "7b9d8701",
    "outputId": "f0274a90-fad1-4c54-88be-bc2eb79837d3"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-04-17 13:50:09.389060: W tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libcudart.so.11.0'; dlerror: libcudart.so.11.0: cannot open shared object file: No such file or directory\n",
      "2022-04-17 13:50:09.389136: I tensorflow/stream_executor/cuda/cudart_stub.cc:29] Ignore above cudart dlerror if you do not have a GPU set up on your machine.\n",
      "Failed to detect the name of this notebook, you can set it manually with the WANDB_NOTEBOOK_NAME environment variable to enable code saving.\n",
      "\u001b[34m\u001b[1mwandb\u001b[0m: Currently logged in as: \u001b[33mmichaelazarzar\u001b[0m (use `wandb login --relogin` to force relogin)\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "import tensorflow as tf\n",
    "\n",
    "from tensorflow import keras\n",
    "from tensorflow.keras import layers\n",
    "from tensorflow.keras.models import Sequential\n",
    "from tensorflow.keras.layers import Conv2D, MaxPooling2D, Dense, Flatten\n",
    "from tensorflow.keras.utils import to_categorical\n",
    "\n",
    "import wandb\n",
    "from wandb.keras import WandbCallback\n",
    "wandb.login()\n",
    "\n",
    "import tf2onnx\n",
    "import onnx\n",
    "\n",
    "import imageio, glob, cv2\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "cf000808",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "cf000808",
    "outputId": "1eeeb052-7276-477b-95eb-bfec28057c01"
   },
   "outputs": [],
   "source": [
    "(train_images, train_labels),(test_images, test_labels) = keras.datasets.fashion_mnist.load_data()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "805c6ca9",
   "metadata": {
    "id": "805c6ca9"
   },
   "source": [
    "# Plot head of dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "e0a4d162",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 92
    },
    "id": "e0a4d162",
    "outputId": "9ecaf027-db35-4133-b39f-ca7ad9bd431d",
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAWAAAABLCAYAAABOfV0NAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/YYfK9AAAACXBIWXMAAAsTAAALEwEAmpwYAAAxpUlEQVR4nO1dW29bV3pdJA/Jw8PDq0hJlmTJlmM7Tu0YadJ0MtMO0AFaFIMiQTEPBYo+FH2d1/6MvvapT+3jpGjReWiBQdsg7SBoBrk5TuqMr7IsyaJ4vx7e2Qd1ffrOEWVbtmRRHi7AkEVSJM8+e6/9Xdb3bd9oNMIUU0wxxRQvH/6T/gJTTDHFFL+pmBLwFFNMMcUJYUrAU0wxxRQnhCkBTzHFFFOcEKYEPMUUU0xxQpgS8BRTTDHFCcE4zIt9Pt9vhGZtNBr5nvW1LzomPp9P/vn9fvl/IBBAOBzGYDBAu93GaDSCz7f/a/l8PgSDQfj9foxGI4xGI/R6PQyHQ1BiyP+/oOSwMBqNss94TUc6T4LBIKLRKEKhEOLxOEajEbrdrowTr/n/P9s1lrxmwzDg8/mQz+dRq9VedCyIZx6T//9uR/Khfr8fhmEgHA4jHo8D2LvHw+GQn+X6ndfLOcSf/X4fw+EQnU5n37x5XrzM9eN5L9f6MU1z3/XqNeRdT8PhEMPhEN1uF/1+X9bMEawd4IC5cigCnuLoEAgEEAgEcOXKFbz11ltYXl7GG2+8gXA4DMuyYJom0uk0+v0+arUa/H4/LMvaN2lINgBk4vDn+vo68vk8bt68iXv37mF9fR0bGxvP+5UfvtgVHw68ztFohHfffRc//elPsbCwgGvXrsHn86Hb7QLYHcfhcIh6vY7RaIRwOIxAICCE2+l0MBqNEIlE4Pf78bd/+7f4xS9+gQcPHmBzc/NFv+ZLHRNidXUV77zzDt566y385Cc/wWg0QrPZRKfTQa1WQyAQwMzMDEajEcrlMjqdDvL5PAaDAWZnZ2GaJgaDAfr9Pm7fvo1CoYD19XUUCgV8++23WFtbO4nL2ge9gY77HdjdWEOhEGZnZ7GysoJEIoH5+XnMzc3h2rVrGAwGKBQKGAwGQs7hcBiGYSAajcLn86HVamEwGLjWUa/Xw927d7G9vY179+5hY2MDjUYDjUbjeS9n7FyZEvBLRjgcRjAYhGmaiEQiWFpawoULF3Du3Dn81m/9FsLhMCKRCEKhEGzbxnA4RK/XQyAQgG3bMkkAuCzewWAgFgytGNu2kcvl0Gq10Ov10O120Ww20W634TjOCY7C4RCNRrG0tITZ2VnE43Gx9oHdRTkcDhGJRABAvAFawN1uF4PBQKyhVCqFmZkZPH78+CQv6YUQDoflOrLZXaOKBBwKhWAYBubm5jAajWCapnhQ/X4fZ86cQSQSQbfbRbfbRaFQQK/XQyqVwnA4hGmaJ3x1e9D32GuFGoaBYDCIcDgM0zSRTCaRzWaRSCQwNzeHbDaLbDYra2IwGGA0GsHv98M0TSFgv9+PYDCIfr/vmjsAZI3UajU4jgPDMOD3+9Hr9WRjf1HLeErALxlvv/023nzzTVy5cgWXL19GKpVCJpNBKBSCaZqum9psNl1udqfTkV0cAAaDgbhMg8FAPiMQCMDv9+PcuXM4e/YsXn/9dTiOg4cPH+Lhw4f4z//8T/zzP//ziVz/s0JP7Gg0iuXlZUSjUbFWms0mgF0y0gu00Wjsc719Ph9mZmYQDoeRTCaxsLCABw8enMh1HQWCwSASiQT8fj8KhQJ8Ph/6/b64zYPBANVqFcDu9QcCAUSjUQyHQ/GuOp0OBoMBLMtCOp1GsVgUgpk0jCO5Cxcu4PLly4jH40ilUjBNE5Zlodvtolqtot1uo9lsiicJ7IZbuJ76/b7MlWaziX6/LyRtWRaCwSAMw0Amk8H8/DwCgQA6nQ7a7TZu3LiB//mf/4HjODIPnxdTAn5JoFu8sLCAK1eu4Pr167h+/ToMwxCXkJZst9sVi4U7MhcWAFkkjN/xpyZnn88Hy7JgGAZmZmbg8/mQSCSQyWRw//592LYtVtCkwzAM2LaNUCiEfr8vFgihx6jb7bq8AD7PsTNNE7ZtIxgMnsi1HAUCgQBCoRAAoNVqwTAM1/wAgF6v5wpX8TUkWT5HS9Lv98umNclg6I4baSKRQDqdlmuq1+uoVCoSYrFtG7ZtIxAIyHtwnDqdDvr9PtrtNvr9PhzHkTGgJxAKhZBOpxGPx2WTKxaLSCQSADAl4NMAn8+HK1eu4OzZs/jjP/5j/OhHP0IwGESv10Ov1xMXUSfLdGy33+8DgCSaCL1gdMyUO32v1xMS4s5+6dIl/PjHP0YikcBnn32Gf/3Xf534hceEpGEYQizhcBjALoEAe2NEC4ePa1ICdq3pmZkZCVmcRgSDQdi2DcMw0Gg0EIlEkEwmMRgM0Gq14Pf75fo6nQ4CgYD8Tu8oFAqJSz4ajVAsFrG+vo56vX6Sl/ZUXL58Gaurq8hms5KQLRQKsm44DwaDAer1OhzHQbvdhm3buHTpEgzDEC+q1+uJ8TIYDJDP5+E4DiKRCGzblph6t9tFuVxGJBKBZVl47bXXkEwm8eWXX+IXv/jFC4UhpgT8EuDz+ZDNZnHhwgWsrq7i/PnzEtDnxCEJknS1NXtQZpqP6d2dj9MK1KQei8UQj8dx/vx59Ho9FIvFscqKSYPP5xPLjddMC1YnZrRShBuYd3MJhUKIRCJC0KcRVEEAu5tyKBRCIBCQzYc/AbgsXf07Nykd76zX6xPlEXmTbn6/H+l0GisrK5InYViAJMq/Y+6E/2j9c97oUB/XS6fTQbPZlIQdE9r0rPx+v4Sx0uk0tre3JX78vEbM6Z2Fpwg+nw9vv/02fvzjH2NlZQWO47isU4YLtFXGUIRXTqbJiIus0+nIc8De5NKvAXatxGq1CsuycPnyZdy8eROxWAztdhvtdvtlDcczg2RpWZY8xjidDrlostULVv/e6/XQarUQDAYRi8XEhT+NaLfbKBaLCIfDrg2WYQi/3y9E6iUH5hWoHiABV6tVPH78+IVd6qOC1zCIRqOIRCJIp9NIp9Not9uoVquuTQeAhB78fj9isRh6vZ4oQ/L5PKLRqKgfms2mWLeO42B2dhbz8/Pw+/0SxggGg/KPCW8m/paWlvDee+8hl8vh9u3bz0XCkxdxfwXh8/mwtLSEq1evIpVKSaxXhxqo66RKwpuRBfZCDoyDGYYhciuvZpEERRKmXKvdbiMYDCKbzSKZTMI0zYmNh1LLGQwGXdfo1bYCbjmeFyQmkpNpmvu8htOEfr+PVqslG68eF/5kvNK7KXW7XdlsSVyj0Qjtdlvc7UkE5Zk0VHw+H9rttktexn8MQ3AtMczXaDTQarVkY2dYi2MSi8WQyWQAQOLBJHjOF36eYRhIJBI4d+4cMpnMc3uSUwv4JSEYDCISieyL8XI33dnZwRdffIFcLodf//rXWF1dxfvvvy8TxO/3I5lMAtgTjFP7StdKJxCYtWWci9lwEncwGIRlWchkMvD7/RIOmSREo1GcOXMGqVRK3Eq6mVwQ2vID9ltODFuQlMLh8KlPwgGQ7P3m5qZYesAuORuGgVQq5Yr90kMol8syn/x+P8rlMorFIlqt1pEUYRwVvJtrLBZDNpuVNQTAZVToxCLvOZUe/Js7d+4gmUzi/PnziEajoh7iuuB7UEHBMdKGgA79mKaJs2fPiiX+PJgS8EsCSY8ukn48EomgXq/j1q1buHPnDj7++GN8//vfxwcffCDZa5/PJzt/q9USK2gwGCCZTMIwDCHacDiMUCgk5Evrj1VznHBUBEyqJph6V9u2AcBl/ZJAn+b26Ti6rpA6zRYw0ev1UKlURCdOizYcDsucoScFQDYhWnx+vx/NZlMs30kiYC8ikQhisZjL0KD1ThIlOZKAuR64Nra3t9Htdl2ESuLV/xhXDoVCQrqMs3PeDIdDhMNhpNNpmZ/PgykBvwToMINXuUCCePToEf7t3/4NOzs72NjYQLFYBACXUPzhw4coFAr4+c9/jlKphHa7Dcuy8Gd/9mc4d+4c4vE4fD4ffvazn+HWrVtYXl5GJpPB5cuXsbKyIkTM900kEjh//jwA4MGDBxO3+CKRCBYXF5FMJsX6pcID2FM8+Hw+yerTLfdCh2EOes1pAYmy0+mgUqmgXq/LfAEgRMX4eTAYRDqdlmuvVqsoFAro9/vI5XKoVqtiOWvjYJLACjfTNOU7etcT8yrcfLjpmqYpoZVOpyMyTBav0BKuVqtoNpviTVBbHA6HxRI2DENIOBaLIZ1OY319fWoBTzp4A8eVUwJALpfDJ598glarJRlp/p1hGGi329ja2sK9e/fw93//91JSPDMzg+9973uYm5sTt/Pf//3f8fOf/xx/8Ad/gKtXr2JxcRGJRAKtVkuSMMCui7+4uIhKpTKRagiGSFgRSALWGX7tJgKQyicNHQvXuunTDGbmG42GaFl5TclkEouLi+L1AJCkY7fbRb1eR7lcRrvdRrlcRrPZxHA43JdzmBT4fD7Yto1MJoNwOCwEzPWkezjoUAQT0aFQSMam0+lga2sLkUjEVcg0HA4lAclNnnFnnfRmzgXYNRBmZmaQSCSmBAyMrxXXz+l4ILWU1Asel/WniRdwKxS0DOj8+fP40z/9U9y5cwe//OUvXX9vmiZyuRz+8R//ETs7O3jjjTdw8eJF9Ho9BINBfPHFF1hbW0Mmk0EgEMDOzg7i8Tjy+Ty+/fZbvPvuu/vGiP0R5ufnsbm5ObEEvLi4iFQqBWCv8ZC+z3RJy+WyaJ1DoZAsKh0zDgaDYgFOarLpWdBut5HP55FIJBCPx2XDpodDbWytVnPphC3LQqPRQK1WQ7ValQ253W5LUcKkbkwMl2mZpmEYcr20ZLVEj2TM5CsLkiqVChqNhryuVquh3+8jEom4EuCWZSEajYqULRqNIplMimqIxRsMazxPMcsrQ8DanQf2ly/SXeVuyQoXBty9gv2jgu5LoL8bvw8n1Llz5/D+++/j448/xieffOL6e9M00Wg08E//9E9wHAd//ud/jnQ6Dcdx4DgOvvjiC1SrVRHn12o1xONxFAoFFAoFcU+92mLTNDE3N4dkMjmRBByNRjE/Py8EDMAVg2N4ptvtolQqYTgcIh6PwzRNsXS5QGm9UFyvK+lOGyhDY5ih0+mgXq8LUQyHQ+TzeQDAnTt3EI1G8bu/+7sIhUJoNBqoVquoVqtSSus4jqsb2iTCNE3E43F0Oh10Op2xPVFYdAK4u58xRpxKpTAajWSzZi8IEjDXI3MolmXBtm3UajWJCSeTSRm7Xq8Hx3FcFauvNAEznsMsbygUQj6fFxdKC6u9YLCez0WjUVy4cAHValU6RR0HwuEwotGo7LZa+6uJmfpU3UIPAIrFIj777DP87//+r+y29+7dw/b2NtrtNnq9nkwIrWSgZlFXB437bslkUnSRk4ZQKIRMJiN9DHRpMS1hx3GQz+fxL//yLxiNRvjLv/xLWVje0my/3+8qPT2t0F4V45WWZYncimPD12opFceE1rJ3g5o0Ag6FQtJgCNgrw+b1NBoNsd5p6er7rlUy7KnCcAy1vYRWPEQiEekZwc9l8x9d1EJOymazqNVqQu7Piokm4HFWYzQahWVZuHjxIuLxOL7++mvp9PWkReWtlY/FYnjjjTeQy+Xw9ddfH9v3tywL8XjcJfwfDocu6QsA6XWriw6A3djwhx9+iIcPH4ql/s0334jlR0uQbme73UYikZCGK4xxjQMnTiwWO5brf1FYliUd0HRHKybdKMXb2NjAP/zDP2AwGOD999/H2bNnXZub9jZYkHHaCZhFFCRg9vYAducSwzSafHUfCFaKcVzp0k/auOhEor6PJMl+vy9FE/Rux3mcwF5ZOhs40VoOBAIyBoFAQCxf9hDh49QPawkj1Unz8/MwDAOVSuXVIWAvGMdhRjQej6PRaCAWi+HRo0euTLAX3MHY/Wlubk7c1OMEu5x5S1914QAJORqNyuQgut0utre3xUqPRqP44Q9/iHQ6LZOS0qJf/vKXyOVyeO+997CysoJKpYJ2u43l5WWxELRlQJ2o9zMnBZz0lOLp8lEmJnd2drCzs4Ner7evjNtrBfG5SUw0HQY6kajzCyQPyuy63S4qlYoUZfA1JDESkJ6Hk6SEIbnFYjEEAgFXLxSdQ2FOhb2gddJV3++D5jhjvrSQdfESQwucg9y09EZmGAZisRhardah19FEE7B3MgQCASwsLGB+fh6vvfYa4vE44vG4xGieRMB08c+cOSN9d3Wd+HGA7gljs97ndDyTLQa1+wzsdru6d+8ecrkc+v0+UqkU/uIv/gKXLl3C7OwsgsEg6vW6SIm+/fZb/NVf/RV++MMfYmNjA6VSCYuLi5Il1yWbhmEgHo9PbGMardnVfS2YoW61WlhbW8Pa2hq63a6rl61eeF43U0uJTiN05j4UCrnCCKwU45jlcjlXjwfq0TkGWpkziQQcj8elWKjb7boUCTocYBiGHFjAa/Naqjr/oRGNRiVEAbjnCMNdJHquI91BLhgMYmZm5tUiYK+UhG3lkskkEomEyHBM04Tf78e1a9eQyWREfM0FyO72JMKZmRksLS3JgJbL5WO1/pgo0QTsdY/oGmn3ZnNzEx9++CG2trYwHA6RSCSwurqKdDqNTz/9FLdv30YikYBlWXjjjTfEuo9Go7h586ZkeQOBADKZjJRk0lXT7tak9kVgq05qfL3k0O128ejRI2xubkqNPoma482SUm8Z9yRa/M8KxvV16TE9AK0b73Q6Uvmm1SC09HQcmPNikggYgCTDGB4hkWpPiA3mvT1TdL8QPVY6jKUtaM05jCeT7PlaPVYE+06MO7HmaZhYAqb1QkttYWEB6XQaS0tLrqN6eDLA5cuXEYlEpEaeO32tVkOlUpFkGG9op9NBqVSS436OA9QvJpNJaZ/oBdvi8dwzy7IwGo3w1Vdf4ac//SlM00Qmk8HS0hI++OADtFot/N3f/R12dnbg8+02Gv+bv/kbXLlyBfF4HOl0Gj/72c9QLBbxe7/3e3j99dexsLAg5ETLCdizGiY1CUcL3efzyQLTBNFsNvHll19ibW1NWg5ygbC5PTPcBC2n02wB61aKJAsd52cZerPZxNbWljQqB9wVmcybUF+tk9STAK6fdDrtik9rAmbDdd1UXRcb0RDjJsXnA4GAFKyQoEmsDNnQi2AjIF3ez/fp9XpyAsnzlCRPJAHrAebuMzc3h3Q6LRavjvMEg0HJlnISsVFNNBqV99HCbe+ud1ygFTeO5L2xyuFwKH1LuWBs28b58+cl7OI4Dn77t38bxWIRgUAAiURCeqOurq4iEAjg/v37aDabCAQCrr4H+vP0+E4i+QJ7vWuBvTiu/q69Xk9kdrQIueEmEol9kiRgr8vccaleXhbGNeDRY+VtxuNVkZC4tXcwzss4abBXitbwczNmqCAUCklPbY2DfvdKQ7VFq5vu6LHTZ8aRzIG9UFc8Hn8uQ2biCFiTEd2EWCyGN998E7OzswAgBMsYK118TTD8x3LBZrOJarUquz0n7HFbQpFIBPF43NW7gJNHT4TBYADHcXDmzBn85Cc/kQRLKpXCm2++iWQyidXVVQwGA5w/fx6dTgemaSIcDmN1dRWWZeGP/uiPUKlU8Pnnn+Obb75BMpnEysqKnJzLyaKLQXR9+6SB2W5aeoCblNvtNu7cuYP19XUJNa2vr8O2bVy8eFE2X735sSuWt7n9acK48AGw16SJJcfaa+BJIrzfDFHQatQx9kkBVUSpVMrl/gOQMBPPUHQcxzVH+Bp97zXJakUIPYFoNCqFPfRKqR3XhT8swCDC4TCWl5dRKBQO7U1PHAFr0LRPpVKuLkQcfBIIb44+Tlpj3C6m5TnPW8XyNHAD8bq8OiPP78FFEI1GcfHiRVfsmw0/+B4MwTAJwe8eiUQwHA5x7do1tNttXLp0CXNzcy5pm45zeb/rpEHLyLw/tTWkjxBvtVrS2QvAvgWhj346rfAWmRC6lNgb09WaX60D5rFUkxZ+IOjJErwubs7NZhP1en1sP2uvRc+x0Zu4hjdxyznG4g+vjM/7PZ/HkHnpBHwQAWjznrBtG9///vflUD12fgJ2SYjdilhiqs/B4o7OiaoXHWVX1PWxWYfe6Y4K42LAXkkUN5BOp4P5+Xl88MEH8jy/+2g0klr12dlZVyiBpxxHo1FkMhn89V//tSQeKVrXn3dQrwQdmphEMMPNayCJ6GNoKpWKhCS8onxgN3F32nXAvV5PjtvxutUcH84bAFIlpvtFkHzr9TpardZYF34SoM9M5IbL058zmQweP36M27dvy7VzDpM8Cf28NxFJQ87bX5tzpFqtSlJ7dnZW/lZ/Brno0Nd3NMP0YvCSMgPkqVRKpFl0lzgoWptH94o12eNif/r9uSiZMebR3TqTelSgm+QtnaR1wqPi+b28E4k6X/6NN27M9+Pzw+FQsrF8XB9iycmjQWE/x3CSweuii60XmSYar+ZZYxKJ5jDgZq1VAYTWv3rnCH9y82IobtKq3zR0FzvAnQugNrhWq8E0TSko0teq17y2pPX7HDQ/+JpOpyNNj3QzHk3Az3ui9EsnYH2x2orTkyAWi+Hdd99FOp2WI6HZvUkH3ll+2ev15HgRCtGpi9RWprb0mA22bRvz8/MIBoMu1/UowBiWtxKO48B2gPfv30c2m5WWkaVSaZ8URrtHXneLr2k2mxIf1xI8klU2m0Umk5ESZl6raZqYnZ1FuVxGPp+fOILyLrp+v498Po9isei6X7SAy+WyS/P8qoHyMiZaOc+1Ntab39Cly47jiNU76aEYWqa6ak9LTRuNBh49eoRsNot0Ou2y/AmGLMkLXOecO9qb4mN6A6tUKtja2sLq6qqML8MQNAjpSZ+qJJw3q81qsGQyiVQqJRIkvci09codUCcQvP/4OeNAC5hNPo5DjkaVhn5vLYup1WrI5XKIRCLyuL4ebyKBGHdN2rLWFnGj0UC9Xodt2/ssIz0O2vKeRGiLhU1kvBYwu32d5hDD06Ar4XivKa3Snp1WuIxbEzouPIn3XBtMXvDa2f9i3P32znXt/enr1WMzLjynDz/Q8WPtperCjcOsoWMnYK/0yXuB+vdUKoXvfe97SCaTOHfunDTb8CactECfx06zrJbuCrO8nIwHWba2bePatWvY3NzEvXv3jtQF9/l8+5rx0BLmGVzfffcdPvroI/z+7/8+fud3fkfCAN5wyTji1GDSTk8GXvOjR4/w3XffIRAIYHl5ed/fsgpvEg/mBOAiGHoqm5ub2N7edllwnU4Ht27dQrPZxJ/8yZ/Ic6+aJcwQBMMt/EnQkHEcR1Qvul0j1wPj4fSIJglaocPYrE6iA7vj0Gq1UCqVpCcv7zWvE9gjaxb0AOM3JGp/SapcT7VaDY8fP4bjOC4e0jJYyuVM03SF/J6GYyXgp018bcmyGQ0tXxKW7nKkEw3eOAxJx0v2tKA50NrC1FKvRqNxbBawjg+RIJldbTabciqrV5/I13vJ90k7rH4936/VaqFQKLikM3ri0X09roKUowTvJxvX63EYDoeoVqvS/8A7FsBkal2fB3SbuTk96bpoKdK48EqzJk1+Bhwcy+ZzwP5EO6GNF75u3PXxvb1elBeUnY2LuevvpD3JZ80nvRABj/siOqzwtBAANb6ZTAZXr14VzZ/P55N+prwJJDGWzjIWqIPguqmMHljubvxdN582DAOvvfaaEOVRIxKJuHpBcDfnEfH1el2ORuGmoRONGk9zbcYlZAzDQLFYxJ07d/D222+7Xq9lXM9iZZ8UvN+t1+shn8+jVCq5Fl6/38fjx4/lNeP6HTzJZT0tINmytSbVPzrRVK/X0Wg0JMxQKBSkhJ/rh7kUKkYmBcydsARZ84lW8ujDZnmvdak9K/6o8NAbkFaMANiXkCR/+Hy7x9eXSqV9B5d6pa/0JIfD4T7j4CC8tBiwNuk5ACSnZDKJTCYjna90pYnX3GfyTMdf9IB63QoOojf2BUDc2kgksq/N3FGBu6L+vsBeNc24JMiTLJknQZMUQxIkHSYp9fvokNDLqAo8LHQcW1/bcDgU19sby+NpBRxrbf3yb8dpxU8r9P3T91Arhph4arVaMi/0utIkNynQTYO83p8mTa1f5vXo3705Eb2Zj2tE5J0XOtmmlSOEJmSGHHU/5qde52EHRutIn5UoGBuJRqOIxWKwbRupVAqWZSGRSIiGjgsIgCSlLMtyTRjvibZM1DHpoDv7axLSRKy7Kvl8PlmwZ86cQblcPuyQHAifzyfXzQw0P5fWi2maWFpaEsufG4S2ZjimXs/iaWPPBiJe6Zn2EihBi8ViqFarR3btRwF98oO2VngYZa1W2+c+0jV/EjG9CqXIBNeCbdsuzbfW/fb7fezs7EjfD3qL3h4RkwK9brTum+uWqohisYhqtSrXyX4nmhC9iUgAQursm8zn6C1w7fHkZPIX+89oPbVOvHEdtdvt4yNg70Dxp/6/93lNvslkEslkEtlsVsp0dYiAA6DjKd4dW+9atJrHie71wHq/mzfzSbfnqF0xLQfyui8sDbVte1+hxrjxfFarjdelx23czq6/IzfBSbKE6D4Ce8UzevwO8h6eNE6TGvN8XnC+8l5rD1Lrx1utFhqNBizLcpGGLj6YJHhDBLxv+hpbrZaEALR1qy1gTcbaK9LE7L12/j6uUnZcwlJ7WrSAnxWHJmAdI2HttLZimRXUySceHsidhM8DcJEjB1e/hyYTv3/vBACdmAMglq8eNFqdzP7Seg6Hw0LMlKG1221p/HyU0MlDjXq9jq2tLfT7fTn1wZsQ0BPqWRaITiqQZHw+n1iMzMzqsA29h1gsNnF9gbvdLqrVKkajkasJE8+yG9fJTlfGAW5pHgCJe05q/4vDIBQKwbZtORswHA5jcXFRekizif9wOMTm5ib8fj8WFxcRjUblRGQtZ5s0MHzGo7j6/T4SiQQSiYTE+0ulkpCi3nw5v0namsg5Joyda9menk/0EoBdfikWi3j06BGGw92yf20164o4bVk/Dc9lAfPi2AwmkUjIyQqc4CTTYDAoVi/VDvzilIoB7iowHTfVoQS67rSYfT6fXDxlOborPnXCJHmWLfNI6m63Ky7GcVmA3rgsJwjLQAFIm8xxf6t/PstneT/H59s7MUM3n9fvyfsyaRawDtMQ9CjYItBLwF4L2Ess45QmpxVcgzoZlUgkAEAsXhoq9XodtVoNgUBAxlMneyd1PLyqJYYGeOIzi0nGKUG8hoYuvCB3aAPJ60FrI5LWb7VaFckZjSOdS/Fq/p+GQxNwMBjEysoKotGoHIzJ3YNEokMCwO5OwgFiYQH/hpNBEzDjtDrJpg/mAyBkQouY/RB0EFwPKDWD/Dtv8o46w6O2BLgRAG4yaDQayOVySKfTyGazrs5d4yxmvhdxkKwG2O9mUytZLpdRLpddXgI/c9yxSSeNer2OBw8e4OzZs1hYWJAN2DAMLC4uolqtPtWSpbvK8dS9b0874vE4Ll26hF6vh0wmg3Q6jevXr+PBgwfY2NiQRvWGYSCXy7k80nfeeQdnzpxBsVhEo9FAqVRCoVA46UsC4I4Bk/xoJPBxHi0PQPJK3IwI3nMtaQX2vGfOCz6viVo3M7IsCzMzMxgMBsjn85iZmUE8HpfTk3lsWK1Wc5Hzs+C5CDiTyUi1GqUeHAjd94C7g3aL9QUz1MAvS2LWDWh0zMZbb83PZbyLv+tFqd1y3gBvEu44LYBxYQEAcpQ4e13QAn5aXOpJn6N/amuA0itmwiORiCToAHf8cJLAJBwTlMDe+XkMaz3N2vDeX22tnHbwUFXmQlKplCSSWZJNsmGFIMltaWlJvNZyuezyMiYBJFvGcen50IsFIAlEhjW9/MB7r4uzgP1HEvE1mrdIwvwuNJAajQbi8bjLiBkMBpKcO+w6OtSKM00TV69exXvvvYdkMilBcHaXYtJEkw4ngLZGA4GA9HXQAWsGsYG93WqcDlAnzHhiKkmYvSC05cvvRuLVCS/GldvtNur1+pFbRtxMxslcuLvOzc0hFovtG68nQROu9z315AF2M+SZTAaGsXtqq9/vl8Yl2nWatGRMpVLB3bt3EYvFXPczGAxiaWkJxWLxwMmuY+B6UXpjwqcZXAsk4nA4jFKphFKptK81K+cee6gsLCwIkdCCnBT4/X7Mzs5icXFRTq2glxaLxWTjbbVaME0Ty8vLSKfTwgFazwuMl2fqRCStYVq9upn9cDhEPB7H0tKSxM51r2W2mzVNE/F4HMvLy2IkPAsORcDBYBCLi4u4ePEiEokEtra20Gw2JQZLstOLZRyhMDZL8uOioLuowwfjLEHtMtMVkAvyVMQRulsajzTXDT663e6xtOTzujMEF0Q4HJZ2lTrB+SzwkrBebPxsYHfjTCaT8Pv9UoSgNZJMgE5aYspxHGxvb6NSqchjnNyxWAyJROJAC3ic7vNVBI0YJlAbjYY0nNdyUZ2EDIfDIgNttVoAMFEWsM/nk+O1eNgCN1pq9plcDoVCyGazsCxr7Drj++kYsJdfvN6RTqwNh0M5SR1wF3EBbu/RNE3MzMwc6pzJQxEwD0F8/Pgx+v0+bNtGPB7H/Py8y7XjT3YqY7tI/tRErS1gSrJCoRASiYRMFsaXtRQtm83CNE2J67KaLJlMIp1Ou1wPDi4lblRHsM9Ep9ORGNJRL9SDiIDfjxaJPjTzaW71uESa93EdR+ZJ0LFYDKVSSZry8G9YjHLQuXUnhXq9jrW1NayursqGzUXRbDYP7F7H3ADbVY67p68CIevQHHX0465Xu+L8R1mfN18zKSB38DtSQcUwp+M42NrakmIuAOKJe3vzct3pWK/OyzBBPa60m15qOp3Gzs4O8vk8Ll++jHfeeQf5fB5ra2sIBAJiwLG/8rMacociYLZPLBQKCAaDWF5ehm3bIkcjuEhoqlMmwzgkH9MEzAGybRuRSASLi4uu0yQYciBhrKyswLZtrK+vS0C+1+shEokgm826LLpxLjp7L1C6RkXHUU9E7Q7p9+ak0Kcm02p5ke8wjpxnZmZw4cIFWJaFWq22r1EIPYpJU0G0Wi08fvwY5XJZkkm0cNjwZNxE1wTshTckcZrhTSSTTMZZ/Dq5q0N/k9oDRLeVJC/w0FqeAbezs4PZ2VnJEbDARieYCZ0M1xsOQw4kfH62JmEWWGxtbSGfzyMcDuPKlSsIh8PY2dkRg7Lb7Yoy41lxKAIeDnebnfzqV78SF1BbcHRrotEowuGwHNVMIgXcBRDexUC1A38Oh0MUi0V0u10Ui0U4joPHjx+LZtcwDAl+0xL+wz/8Q4kFRyIRGVgdC+Zj3DE7nQ42NzeRz+ePvEeA1jFriYplWchms66jhnjTD7MgvDFgPsafjBEmk0n0+31sb29jfn7e5ZYddyLyeUEdcL1el02DY0MrLpFIoNlsotFouKxh3RPAu/FNIuE8D3SRkQ4pjXPDtZRLyzrZbHySuqH1+33cvXsXOzs7uHv3LkzTlNzOzMwMYrEYbty4gXK5LOoPHWLQiinKxwDss/o1IetNifOl0+mg1WrBtm1YloWHDx+iXq/jv/7rv9DpdFAoFLCxsSHcQm21PrnnaTgUAY9Gu8d8f/75567HqaNdXV1FJpPB3NycnNLLpueWZSEcDotMzCtW1nHLwWAg5zwVi0XUajXcuHEDhUIBX3zxBUqlkivGRfh8u8dYZ7NZEWw7joNms+lqwsJSQh10397eRrFYPPIknHYLdZbWsixkMhkp/uD30ARxkBtzkFJCP6+Jx7IsJJNJ5PN5FAoFNJtNWbCEnpCTAp5X1mg00Ol04PPtlY+SgJPJJBqNxr7ewMDeoYzjPI9XAdTD8zq9mlQNr56W46BPe5gUDIdD3L9/3/UYeWNmZgaJRAK5XA7lchmj0QipVErEACRSb8k/vVxd7EWvQdcN6LXHdrbpdBqJRALhcBiNRgOffPIJ/vu//1u8sHFN4J8VR6I7YrZ1Z2dHNIWmaeLRo0cSs+Gg6A5VwP7jiLhDM9hdrVbF8m00Gmg2mwfqdUejEW7evInBYOCq36ZrohtqaEucJynU6/UjT8IdtCB0T2NqkEn+3mTaOKWD9zM0+Hod315YWJBqHu+YT2oSjtCxTh3PpAVcqVSkex6we13ceKPRqCR9uem+Klawt0BBV2ZpeJUxOg/B5yZdmsc1X6vV5Fw4StJM03RZ/bwuNlHXNQc6SQ+4E5T6cc0xnHvxeFxOjikUCmOb8xwWR0bA/X4fW1tbB77mea2OwxLip59+ik8//fTABNXLBEmQN1N/J8Z+GXv1kq73fcY9rm+89/2ZaBwMBnIkUrlcHks8Ojwyidah3iDoLpKAU6nUvoIMLr56vS6bnO4ZMKnXeVhoJQ/1qONOS+ZrNTHrENRpKErRGn4AEtaMRCKwLGtfAcZgMJA8E+cK21vq9/TmaOiNegk4FAohmUzizJkzaLfbco7ji+KlKe9fNgFOks6TN1Qf6pfL5XDz5k2YpomLFy8CwIEWqJdcteVKa1eHHUjY9DjK5TJKpRK2t7fx8OFDXLlyxfXdgMkMQRDNZhNra2tIpVJSEQfsbmLZbFbORiO48Y1rO8lQ1GkgnafBMAxXUc24clrAPR7jwjTHUQF6HNCeoG3bSKfTsG3bRaL0AhhmoWSRnRIZygL2agz4Oz1zPV6cL81mE+FwGPPz8ygWi0d2TZNV+vQKQmdZmaH2+/24f/8+/uM//gPxeBw/+MEPxAo9KGQB7FdzeJNtJCadjAiFQmg0GtjY2MDt27dx8+ZNXL9+3RWCAZ7/VNeXgVKphBs3buDcuXNYXFwUS9g0TaysrEglmIaWoXkfdxxnopJOzws2krJt+0DyBfakVoxXMgnn8/kkfjnJG5L3ukajEdLpNF577TWk02nXRjscDl3ql3A4LGXEVCwAkGpKXYxBL0uv11AoJNauZVm4dOkStre3j+zapgR8zOCN1ZapjlcxSUB3igoNr9ULPLn/g/4/35/iepIOm5V7u4VR5jdpYnxer+M42NzcRCwWcy009oRwHGdfRdxBZOQtCT/NYOUpy//1HPOC80pfO8NUur/tJGJc/sM0TVFh6dASwRoCwzBkzvP6+bgu2NKSPh0L9sbZ6XEcFaYEfIxg7JKZe10MAkDido1GQ5Jyg8FAdtwnqRzG/c5FxF2eUjz2nWCVFCVd3AyYXY7FYhNlBXPMyuUybty4IROfMb5IJIKrV69KwlVDLyRvwnWcguY0gqdqh0IhZDIZiV+OCzPQjSZZ6QMQ6vX6kR5Gexzwbiq2bWN+fl4Ilnp+bjDBYBCzs7PodruimqIagk10qHvn5sVYr27RyVBEv9+X1rtHaahMCfgYMRrtHpNeLpexsLAgZMAF0Wg0cPfuXXz00UciFWPLu3EWMHHQ49o6BIBEIgHLsvDdd9/h4cOHyOVy8ry2GthM5DAVPMcJHcsGICoY9jjQR0+NKyTQSbjZ2VnJiuuqqkmNdx8G7XYb+XxeiqLGacIJTSQkE6qNqIc/TTAMQ5rLc6PVzXY4B1j0xTCVtwmXfj3j4QzJULkFwFU0xoS19jaeNPZPvI6jGIwpxmM0GmF7ext37tzB8vKyy0Kp1WqSFPvwww/3JdCOAnxPSpNisRhisZiraZJhGHAcBw8ePMD29vbEWIZaJsTjh+bm5lCv1+V7j+sBC+x6ALlcDltbW1hZWRGy8fl2DxFg743Tjkqlghs3bqDb7eL69etjKz/5O48oYlUq+3LXajXk8/lnPkb9pOBdH9FoVKz+QCAgFZ66P3gul0Ov10O9Xkc4HMbCwsK+1gcE+8HQA9U9bGhRa9lbMBjc19b0eUh4SsDHiNFohGazKZUx3Kk5SQaDgQjhXwbYxMQb8wIgMq1Jgpbm6YZJlO+xys+yLOnoxeYyAFyaX93OULfiPM2gDI3NlbxWmRfchPR9P62nRFcqFaytrUnvl2AwiHQ67Qo1MTTBe87ub+wrwtCgbdsiC2U+ZDQaSX8UblY8f65SqRyZkTQl4GMELeBQKCTN4vv9vkvu8jJBEmLzFn4Pdklji72ThjdRSfAgxmAwiLm5OVFCdDodXLx4EX6/H3fv3gUAV1abRxANBgPYti3FQacdWgdsGIbrDLiDErbMSfBYLh65PukWMMHrunnzJu7du4e33noLP/jBD5DNZvH6668jFAohEonAcRzk83kJt3Gu0PsEIEU6PCGHjcB4RiOPOWMLylu3buHrr7/G2travpj5866bKQEfI+j6sXMXK3O0u/MyXX5a3K1WC9VqVYiYyTnHcSaCgIHxE5oNWPx+P5LJJIBdS+jRo0cS7yP5VKtVFAoFVCoV2LYtOux6vS6l6acdLNUvlUqoVCqu0v2DusQ1m01Uq1WZexyPSbeAvfOB4YZ8Po+NjQ3Xyci2bUu3Q2Cv/wetYxoaND46nY406GL8l31mms2mKIg2NjaQy+XkKLGjwJSAjxGj0Qj1eh2lUgkbGxu4f/++JEFyuZyUWwPuSfKiHdEIb7llvV6H4zi4c+cOPv/8c7EWvv32W/z617/G1tbWxMSAx6FQKOCjjz5CJpPBxsYGms0mfvWrXyGXy+Gzzz5DrVYTK/Crr75CLpeTht3szPfll1/i1q1bKJVKJ305L4ydnR18/PHHqFQquHjxIlqtFr766itsbGyMtWh7vR5u3bolFmGv18M333yD9fX1iSdgwE3C3EBu3bqF+/fvizfHo9Ki0ShmZ2elKRiTduNkm17PoVKpwHEcrK2tSXycfUloOB0VpgR8zNCZVcZZn3YS7bhJclTfhb0xHMeRGCnDD5O+CJkgiUQiaDabqNfrYgHq4grKq2jV04JhDPlVqYTjOYe00hzHQavVOtCToUfmOA6AvfjvpEvQDgKvR2827BHd7XYRjUZdagZ9CoaGbsBOWShDM4VCQfoFHwd8h3E5fT5fHsDDY/kmk4OV0WiUfdYX/4aMCXCIcZmOyXj8hozLdEzGY+y4HIqAp5hiiimmODqcfi3OFFNMMcUpxZSAp5hiiilOCFMCnmKKKaY4IUwJeIopppjihDAl4CmmmGKKE8KUgKeYYoopTghTAp5iiimmOCFMCXiKKaaY4oQwJeAppphiihPC/wFK5qeAcRrJSAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 5 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# summarize loaded dataset\n",
    "# plot first few images\n",
    "for i in range(5):\n",
    "    plt.subplot(1,5,i+1)\n",
    "    plt.xticks([])\n",
    "    plt.yticks([])\n",
    "    plt.imshow(test_images[i], cmap=plt.get_cmap('gray'))\n",
    "# show the figure\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7ca32cec",
   "metadata": {
    "id": "7ca32cec"
   },
   "source": [
    "# Save images for API test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "52bc158f",
   "metadata": {
    "id": "52bc158f"
   },
   "outputs": [],
   "source": [
    "images_to_save = test_images[:5] \n",
    "labels_to_save = test_labels[:5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "e1cafe38",
   "metadata": {
    "id": "e1cafe38"
   },
   "outputs": [],
   "source": [
    "for i in range(len(images_to_save)):\n",
    "    plt.imsave('images/image_' + str(i) + '.png', images_to_save[i], cmap = 'gray')\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3a52e15c",
   "metadata": {
    "id": "3a52e15c"
   },
   "source": [
    "# Remove saved images from test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "6345af99",
   "metadata": {
    "id": "6345af99"
   },
   "outputs": [],
   "source": [
    "test_images = test_images[5:]\n",
    "test_labels = test_labels[5:]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "000e020c",
   "metadata": {
    "id": "000e020c"
   },
   "source": [
    "# Prepare data for model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "2d3de2c3",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "2d3de2c3",
    "outputId": "f5453328-d445-4384-f473-1d8a2a19f423"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0 1 2 3 4 5 6 7 8 9]\n",
      "(28, 28, 1)\n"
     ]
    }
   ],
   "source": [
    "# Reshape and normalize images.\n",
    "train_images_re = np.expand_dims(train_images, axis=3) \n",
    "test_images_re = np.expand_dims(test_images, axis=3) \n",
    "\n",
    "class_names = ['T-shirt/top', 'Trouser', 'Pullover', 'Dress', 'Coat', \n",
    "               'Sandal', 'Shirt', 'Sneaker', 'Bag', 'Ankle boot']\n",
    "\n",
    "classes = np.unique(train_labels)\n",
    "input_shape = train_images_re[0].shape\n",
    "\n",
    "print(classes)\n",
    "print(input_shape)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2d487c09",
   "metadata": {
    "id": "2d487c09"
   },
   "source": [
    "# Images and Lables"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "191a4976",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 494
    },
    "id": "191a4976",
    "outputId": "58ec5aea-8e43-480a-ca3e-f19428e844e3"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 15 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(10,10))\n",
    "for i in range(15):\n",
    "    plt.subplot(3,5,i+1)\n",
    "    plt.xticks([])\n",
    "    plt.yticks([])\n",
    "    plt.imshow(train_images_re[i], cmap=plt.cm.binary)\n",
    "    plt.xlabel(class_names[train_labels[i]])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "91b37d43",
   "metadata": {
    "id": "91b37d43"
   },
   "source": [
    "# Set W&B"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "5328e3e1",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 86
    },
    "id": "5328e3e1",
    "outputId": "09ccfafc-3880-4fcc-cf48-2541ef7b79b6"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\u001b[34m\u001b[1mwandb\u001b[0m: Currently logged in as: \u001b[33mmichaelazarzar\u001b[0m (use `wandb login --relogin` to force relogin)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "Tracking run with wandb version 0.12.14"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "Run data is saved locally in <code>/content/wandb/run-20220417_134037-1m8r27pg</code>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "Syncing run <strong><a href=\"https://wandb.ai/michaelazarzar/AICourseFashioMnist-AviAndMichael/runs/1m8r27pg\" target=\"_blank\">vague-darkness-8</a></strong> to <a href=\"https://wandb.ai/michaelazarzar/AICourseFashioMnist-AviAndMichael\" target=\"_blank\">Weights & Biases</a> (<a href=\"https://wandb.me/run\" target=\"_blank\">docs</a>)<br/>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Initilize a new wandb run\n",
    "wandb.init(project=\"AICourseFashioMnist-AviAndMichael\")\n",
    "\n",
    "# Default values for hyper-parameters\n",
    "config = wandb.config # Config is a variable that holds and saves hyperparameters and inputs\n",
    "config.learning_rate = 0.05\n",
    "config.epochs = 10\n",
    "config.img_width=28\n",
    "config.img_height=28\n",
    "config.num_classes = 10\n",
    "config.batch_size = 256\n",
    "config.validation_split = 0.2\n",
    "config.weight_decay = 0.0005\n",
    "config.activation = 'relu'\n",
    "config.optimizer = 'Adam'\n",
    "config.seed = 10"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "39c16244",
   "metadata": {
    "id": "39c16244"
   },
   "source": [
    "# CreateModels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "d55db394",
   "metadata": {
    "id": "d55db394"
   },
   "outputs": [],
   "source": [
    "\n",
    "def createModel(input_shape, num_classes):\n",
    "    model = Sequential([keras.Input(shape=input_shape),\n",
    "        layers.Conv2D(32, kernel_size=(3, 3), activation=\"relu\"),\n",
    "        layers.Conv2D(64, kernel_size=(3, 3), activation=\"relu\"),\n",
    "        layers.MaxPooling2D(pool_size=(2, 2)),\n",
    "        layers.Dropout(0.5),\n",
    "        layers.Conv2D(128, kernel_size=(3, 3), activation=\"relu\"),\n",
    "        layers.MaxPooling2D(pool_size=(2, 2)),\n",
    "        layers.Dropout(0.5),\n",
    "        layers.Flatten(),\n",
    "        layers.Dense(64, activation=\"relu\"),\n",
    "        layers.Dense(num_classes, activation=\"softmax\")])\n",
    "    return model\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "90006670",
   "metadata": {
    "id": "90006670"
   },
   "outputs": [],
   "source": [
    "model = createModel(input_shape, len(classes))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "90f4b51a",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "90f4b51a",
    "outputId": "30d6a274-92f3-442e-c29d-44261aa1c172"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_1\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " conv2d_3 (Conv2D)           (None, 26, 26, 32)        320       \n",
      "                                                                 \n",
      " conv2d_4 (Conv2D)           (None, 24, 24, 64)        18496     \n",
      "                                                                 \n",
      " max_pooling2d_2 (MaxPooling  (None, 12, 12, 64)       0         \n",
      " 2D)                                                             \n",
      "                                                                 \n",
      " dropout_2 (Dropout)         (None, 12, 12, 64)        0         \n",
      "                                                                 \n",
      " conv2d_5 (Conv2D)           (None, 10, 10, 128)       73856     \n",
      "                                                                 \n",
      " max_pooling2d_3 (MaxPooling  (None, 5, 5, 128)        0         \n",
      " 2D)                                                             \n",
      "                                                                 \n",
      " dropout_3 (Dropout)         (None, 5, 5, 128)         0         \n",
      "                                                                 \n",
      " flatten_1 (Flatten)         (None, 3200)              0         \n",
      "                                                                 \n",
      " dense_2 (Dense)             (None, 64)                204864    \n",
      "                                                                 \n",
      " dense_3 (Dense)             (None, 10)                650       \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 298,186\n",
      "Trainable params: 298,186\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "f2c706da",
   "metadata": {
    "id": "f2c706da"
   },
   "outputs": [],
   "source": [
    "model.compile(optimizer='adam',\n",
    "              loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True),\n",
    "              metrics=['accuracy'])\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "56f484f5",
   "metadata": {
    "id": "56f484f5"
   },
   "source": [
    "# Train model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "fd4444ca",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 514
    },
    "id": "fd4444ca",
    "outputId": "f20f5015-1932-40fd-9089-aed1ed933d45"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<iframe src=\"https://wandb.ai/michaelazarzar/AICourseFashioMnist-AviAndMichael/runs/1m8r27pg?jupyter=true\" style=\"border:none;width:100%;height:420px;\"></iframe>"
      ],
      "text/plain": [
       "<wandb.jupyter.IFrame at 0x7fd32bdac250>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.7/dist-packages/tensorflow/python/util/dispatch.py:1082: UserWarning: \"`sparse_categorical_crossentropy` received `from_logits=True`, but the `output` argument was produced by a sigmoid or softmax activation and thus does not represent logits. Was this intended?\"\n",
      "  return dispatch_target(*args, **kwargs)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:Your input ran out of data; interrupting training. Make sure that your dataset or generator can generate at least `steps_per_epoch * epochs` batches (in this case, 18750.0 batches). You may need to use the repeat() function when building your dataset.\n"
     ]
    }
   ],
   "source": [
    "%%wandb\n",
    "history = model.fit(\n",
    "    train_images_re,                 \n",
    "    train_labels, \n",
    "    epochs=config.epochs, \n",
    "    steps_per_epoch=len(train_images_re) / 32,\n",
    "    shuffle=True,\n",
    "    validation_split = 0.2,\n",
    "    verbose = 0,\n",
    "    callbacks=[\n",
    "        WandbCallback(),\n",
    "        tf.keras.callbacks.EarlyStopping(patience=10, restore_best_weights=True)\n",
    "    ])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "9a6143d9",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "9a6143d9",
    "outputId": "8bd488e7-7ed5-438d-f298-4e61c991dfb8"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "  8/313 [..............................] - ETA: 2s - loss: 0.2507 - accuracy: 0.9297  "
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.7/dist-packages/tensorflow/python/util/dispatch.py:1082: UserWarning: \"`sparse_categorical_crossentropy` received `from_logits=True`, but the `output` argument was produced by a sigmoid or softmax activation and thus does not represent logits. Was this intended?\"\n",
      "  return dispatch_target(*args, **kwargs)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "313/313 [==============================] - 2s 7ms/step - loss: 0.2933 - accuracy: 0.8966\n"
     ]
    }
   ],
   "source": [
    "test_loss, test_acc = model.evaluate(test_images_re,  test_labels, verbose=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ZJQNCjvyt5Ix",
   "metadata": {
    "id": "ZJQNCjvyt5Ix"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "id": "ee75fa23",
   "metadata": {
    "id": "ee75fa23"
   },
   "source": [
    "# PLOT history"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "b70fa44f",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 295
    },
    "id": "b70fa44f",
    "outputId": "65a9cad7-73e3-4c0c-b6cc-74bd5e7cd139"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(history.history['loss'])\n",
    "plt.plot(history.history['val_loss'])\n",
    "plt.title(\"Loss\")\n",
    "plt.ylabel(\"loss\")\n",
    "plt.xlabel(\"epoch\")\n",
    "plt.legend(['train', 'test'], loc='upper left')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "927a9657",
   "metadata": {
    "id": "927a9657"
   },
   "source": [
    "# Save model to .onnx format"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "92a016c3",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "92a016c3",
    "outputId": "d9d22768-0129-4192-d8a7-e34c429abfa7"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From /usr/local/lib/python3.7/dist-packages/tf2onnx/tf_loader.py:706: extract_sub_graph (from tensorflow.python.framework.graph_util_impl) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Use `tf.compat.v1.graph_util.extract_sub_graph`\n"
     ]
    }
   ],
   "source": [
    "\n",
    "onnx_model, _ = tf2onnx.convert.from_keras(model,\n",
    "                                           #input_sig,\n",
    "                                           opset=13)\n",
    "\n",
    "onnx.save(onnx_model, 'models/model.onnx')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "78bc8b87",
   "metadata": {
    "id": "78bc8b87"
   },
   "outputs": [],
   "source": [
    "model.save('models/single_run_model.h5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "84912dc4",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "84912dc4",
    "outputId": "09b12c51-7498-4928-e698-088685a568d6"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_1\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " conv2d_3 (Conv2D)           (None, 26, 26, 32)        320       \n",
      "                                                                 \n",
      " conv2d_4 (Conv2D)           (None, 24, 24, 64)        18496     \n",
      "                                                                 \n",
      " max_pooling2d_2 (MaxPooling  (None, 12, 12, 64)       0         \n",
      " 2D)                                                             \n",
      "                                                                 \n",
      " dropout_2 (Dropout)         (None, 12, 12, 64)        0         \n",
      "                                                                 \n",
      " conv2d_5 (Conv2D)           (None, 10, 10, 128)       73856     \n",
      "                                                                 \n",
      " max_pooling2d_3 (MaxPooling  (None, 5, 5, 128)        0         \n",
      " 2D)                                                             \n",
      "                                                                 \n",
      " dropout_3 (Dropout)         (None, 5, 5, 128)         0         \n",
      "                                                                 \n",
      " flatten_1 (Flatten)         (None, 3200)              0         \n",
      "                                                                 \n",
      " dense_2 (Dense)             (None, 64)                204864    \n",
      "                                                                 \n",
      " dense_3 (Dense)             (None, 10)                650       \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 298,186\n",
      "Trainable params: 298,186\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model1 = keras.models.load_model('models/single_run_model.h5')\n",
    "model1.summary()"
   ]
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "name": "Model_creation.ipynb",
   "provenance": []
  },
  "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": "3.9.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}