Deep_learning_final1 (1).ipynb

{
  "cells": [
    {
      "cell_type": "code",
      "execution_count": 4,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "9peqyTYOS5qN",
        "outputId": "581d95c2-1d37-4808-c821-f184cef53e4a"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Looking in indexes: https://pypi.org/simple, https://us-python.pkg.dev/colab-wheels/public/simple/\n",
            "Requirement already satisfied: wandb in /usr/local/lib/python3.9/dist-packages (0.14.0)\n",
            "Requirement already satisfied: appdirs>=1.4.3 in /usr/local/lib/python3.9/dist-packages (from wandb) (1.4.4)\n",
            "Requirement already satisfied: setuptools in /usr/local/lib/python3.9/dist-packages (from wandb) (63.4.3)\n",
            "Requirement already satisfied: requests<3,>=2.0.0 in /usr/local/lib/python3.9/dist-packages (from wandb) (2.27.1)\n",
            "Requirement already satisfied: typing-extensions in /usr/local/lib/python3.9/dist-packages (from wandb) (4.5.0)\n",
            "Requirement already satisfied: docker-pycreds>=0.4.0 in /usr/local/lib/python3.9/dist-packages (from wandb) (0.4.0)\n",
            "Requirement already satisfied: pathtools in /usr/local/lib/python3.9/dist-packages (from wandb) (0.1.2)\n",
            "Requirement already satisfied: psutil>=5.0.0 in /usr/local/lib/python3.9/dist-packages (from wandb) (5.9.4)\n",
            "Requirement already satisfied: setproctitle in /usr/local/lib/python3.9/dist-packages (from wandb) (1.3.2)\n",
            "Requirement already satisfied: PyYAML in /usr/local/lib/python3.9/dist-packages (from wandb) (6.0)\n",
            "Requirement already satisfied: Click!=8.0.0,>=7.0 in /usr/local/lib/python3.9/dist-packages (from wandb) (8.1.3)\n",
            "Requirement already satisfied: sentry-sdk>=1.0.0 in /usr/local/lib/python3.9/dist-packages (from wandb) (1.17.0)\n",
            "Requirement already satisfied: GitPython!=3.1.29,>=1.0.0 in /usr/local/lib/python3.9/dist-packages (from wandb) (3.1.31)\n",
            "Requirement already satisfied: protobuf!=4.21.0,<5,>=3.15.0 in /usr/local/lib/python3.9/dist-packages (from wandb) (3.19.6)\n",
            "Requirement already satisfied: six>=1.4.0 in /usr/local/lib/python3.9/dist-packages (from docker-pycreds>=0.4.0->wandb) (1.15.0)\n",
            "Requirement already satisfied: gitdb<5,>=4.0.1 in /usr/local/lib/python3.9/dist-packages (from GitPython!=3.1.29,>=1.0.0->wandb) (4.0.10)\n",
            "Requirement already satisfied: idna<4,>=2.5 in /usr/local/lib/python3.9/dist-packages (from requests<3,>=2.0.0->wandb) (3.4)\n",
            "Requirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.9/dist-packages (from requests<3,>=2.0.0->wandb) (2022.12.7)\n",
            "Requirement already satisfied: urllib3<1.27,>=1.21.1 in /usr/local/lib/python3.9/dist-packages (from requests<3,>=2.0.0->wandb) (1.26.15)\n",
            "Requirement already satisfied: charset-normalizer~=2.0.0 in /usr/local/lib/python3.9/dist-packages (from requests<3,>=2.0.0->wandb) (2.0.12)\n",
            "Requirement already satisfied: smmap<6,>=3.0.1 in /usr/local/lib/python3.9/dist-packages (from gitdb<5,>=4.0.1->GitPython!=3.1.29,>=1.0.0->wandb) (5.0.0)\n"
          ]
        }
      ],
      "source": [
        "pip install wandb\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 5,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "F2WhcCOQTT8E",
        "outputId": "9036536d-1668-4e2f-8e0c-d908cadaf4ba"
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAGEAAABwCAYAAAAOjB6dAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/bCgiHAAAACXBIWXMAAAsTAAALEwEAmpwYAAAVXUlEQVR4nO2dbYxc11nHf8/MnffdmdkX2+u34CR2auyosWmFkUjSSigVREgtEqpaBIKAqCIViYoXEcEH4BOFD3yqkKigKpUQtBVFQVGlAikVTQUmTRqS1CmO17t24n2f3dl5f7kzhw8zz/GZu7Pr3dn1ZuzsXxrdmTv33nPn/M/zcp7nOXfEGMMB3luE3usbOMABCUOBAxKGAAckDAEOSBgCHJAwBNgVCSLysyLyfyJyTUSe26uber9BBp0niEgYuAo8BbwLvAx82hhzZe9u7/2B3UjCTwLXjDHXjTEN4B+Bj+/Nbb2/4O3i3OPAO87nd4FLW50gIns6PQ+FQoRCIcLhMIlEAs/zGBkZIRqN0m63abfbtFotfN/HGINKfSgUQkSIRqNEo1FarRb1ep1ms8na2hr1ep27FElYMcYcCu7cDQnbgoh8BvjMLq+BiOB5nu34UChEMpkkmUwyPj7OI488wtjYGI8//jhTU1O2UwuFAsvLy/i+j+/7AJawo0ePcuLECQqFAtevX2d5eZnnn3+eGzdu0Gg0LHlKaJDMAXCj387dkHALOOl8PtHd1wNjzBeBL8JgkuB5HvF4nHQ6zQc/+EGy2SxHjx5lbGyMSCRCLBbD8zxSqRTRaJTDhw+TSCSYnJwkkUhYaWi327bzRAS4LRGRSATf9zl8+DDZbJZKpUKtVqNWq1EsFllYWCCfz/PGG29QLBYtQXuF3ZDwMnBGRB6k0/mfAn5pT+7KQTgcJhaLkU6n+cAHPsDU1BRnz55lamqKcDiM53m0223q9TqAlZZMJsP4+LglSjsesKQUi0UKhQIiwvj4OO12mwceeIBQKES5XKZer7O0tMS1a9eYn59ndnaWer2+pwTALkgwxvgi8lvAt4Aw8CVjzA/36sbi8TjxeJxTp05x6dIlMpkMp06dIpVK4Xke9Xrddqwxhmaz2XN+tVplYWGBSCRCNBolFArRbreBDgmtVotms2n1f6vVQkTwfR8RsZ9TqRQnT54km80CkM/nmZmZYWVlhXw+z8rKyq5/68Au6kCN7UAdTUxMkMlk+MhHPsKzzz5LNBqlWCxa3R5UM7r1fZ9Wq0W1WqVer+N5HrFYDMDq82azSbvdtkY9Go0yMjJCKBTC8zxEhHA43GP42+22VVEvvfQS165d4+233+bKlSs7sRGvGGM+HNx51w3zIBARJicnOXXqFEeOHLEjVQlotVo9HR98r9dQSQker1sRsXZBz/V9n1AohDHGkqDfKSGTk5M0Gg1WVlbwPM8OhkExlCSEQiEee+wxnnrqKdLpNNVqlXa7Tblcxvd9wuFwT8e5HQvYkRyPx3u8GlVZ7giPxWJWVSkxrVbLXscl0/M8PM/j/PnznD17lmq1yuuvv069XqfRaAzsNQ0dCdo5qVSK8fFx68erm6gdHgp15pmup+OSEolEekZou90mHA5jjLHGOxKJEIlEeiQiCL2eiNh24/G4tRfJZBIRodls3h8khEIhUqkUiUSCTCZDJpOh3W7bH+iqDu1Uz+v8BO1A3bqE6Xsd4drxsViMaDQKYDvQJU3PUQlSo60SMTU1xYULF1heXubNN9+0HtpOMVQkANal1Fez2bSG1PWGADuClRz31U9VaQdHIhHC4bCVBGCDvXDVW3CSpu2mUikOHTpEs9kkHA4P/JuHioRQKEQsFiORSJBKpRgZGaFer/foa9X3boerCtPO7we3E9XzCYfDVkUpQa5E6H51aWu1mlV14XCYiYkJTp8+DWAlchAMFQkA0WiUeDxOMpkklUoRCoWsL6+dFo1G7SRN9bSOaFeHA331vCs5iuB5LjGAnTe0Wi08zyMcDpPJZHjggQcoFAr3lySMjIyQzWbtLFcNIWBHu2tM1f1UD6pWq+H7Po1Gw9oSd5IGbBj1cFvFJJNJIpEI6XSaVCplj2+327bzo9Foz2DYDQEwZCR4nsfk5CRHjx4lnU5bAvRHqkekI1kNZb1etzGd5eVlqtUq+XyeQqFgCVFXtd1u29iP7lMDH41GOXbsGKOjozzyyCOkUinrGAB2YGi8qlQqEYvF7KAY+Hfvvuv2DjoXWF9f59atW1y7ds12jm5FhGKxSL1et69Go2FJWF1dtYG3crnc0/kqNfo5OGlTvZ5IJAiHw9TrdUZGRhgfH7dEuTZJ7+m+IqFer/PWW29x7do1ZmZm+OY3v8mJEyc4f/48qVSKI0eOICJ873vfY3p6mnw+z+rqqu1owHawSotrtLUTVYWogdaJXKvVolAo0Gw27Qh/4okneOaZZ0in04yMjOB5nm0jlUrZyKtK6SAYKhIA647qiA6Hwxw+fJharWZjQHNzc8zOzlIqlVhfX+85X/W8ejA6gt3JnXpEgLUrgPWAarUapVIJYwwrKyuUy2U8zyMSieB5nlVnKom7majBkAXwkskkly5dYmpqynZiq9Wi0WggIiSTSdrtNjdv3iSXy3H+/HkuXLhgO1hHtMaYdMQGI6yulAA9RCWTScLhMKVSiUqlQrVaZX193aotdQLU8DebTVZWVrY7WRv+AF40GuXMmTOcPn3auoyLi4tcvXoV3/fJ5XLWA9Jjn376aatuXO+oXC5Tq9VoNBrWc1KVFUzueJ5HIpEgFotx/PhxkskkuVyO9fV1Xn75Zb797W/bRE+r1bK5Bg0q7hZ3JEFEvgT8PLBkjHm0u28c+CpwCpgFPmmMWdvtzfi+z8LCAuFwmGw2SzqdZnR0lIcffphIJMLExAThcJjZ2VnW1tYoFou8+OKLNowQCoWs8XZzzI1GA+jtfB3VrVbLjnwRYX5+3hr/UqlEoVDg4sWL1svyfZ9KpUKj0SCTyTA2NsbCwgKXL1+mXC5bdboTbEcSvgx8AfiKs+854EVjzOe79UbPAX+wo5b7wPd93n33XWq1Gg8//DCjo6NkMhnrsj766KNEo1EuX77MzMwMc3NzvPDCCzZiGovFbHpTvRagZ7athlptgSsdzWaTubk5O9IbjQYnT57k0qVL1iC3Wi1LwpkzZzh37hyvvvoq169ftzPrPSfBGPOfInIqsPvjwEe77/8O+A57QILneRw/fpwTJ07YPLLGeDzPs7p9ZGSEI0eOEI/HGRsbs+EOz/PIZDL2vRpg1w0FemxIcFacTqep1WrW6I6NjTEyMgJ0vLdQKESj0aBSqbC6usqtW7dYWlqyac9BbOygNuGIMWa++34BODLgdXqQTCZ58sknuXjxojWeOhELhULkcjnrLR07doyxsTEmJiY2hCCAnqybej8KlxjXLrhubKVSoV6vUygUWFxcpNFoUCgUqNVq5PN5FhYWWF5e5kc/+hHvvPMOuVyOSqWyryS4P8hs5fXstOTFnZSpD69hC9XtbudtFgVV1dC9x57OCQYEdfbtzilctaXHJpNJa8Sj0agNhaukqsraKRGDkrAoIkeNMfMichRY2uzAnZS8NBoNrl69iohw7NgxDh06RDweZ3Jykmq1yuzsLOVymbW1NSqVCqOjo2SzWTtK3XB3vwgr0JPwh9uEBDvPjbRGIhEymQwXL14kmUwyOTnJwsICU1NTHD9+nNdee42bN2+ysLDA4uIi1Wp1R505KAn/Avwq8Pnu9vkBr9ODdrtNoVBgZWXFlqyEw2GrkxuNBuVy2VY6pFIp8vk89Xrduq9uuNvtxH7ZM9eDUh9fQxs6wlOpFNlslmQyydjYGNlslvX1dTzP48SJEzz44IPk83nS6fTA0dTtuKj/QMcIT4rIu8Af0+n8r4nIb9CpKvvkjlvuA3VNDx8+bBM6GgfSeNL6+jqxWIxjx46RzWaZmJiw1RVu9i2Yb3B+T0+bbuIHeiXD932KxSJzc3M0Gg3m5uaoVqvkcjlyuRy+75PP57ly5QoLCwusrq5umBhuB9vxjj69yVc/s+PW7oBQKEQ6nWZsbIxoNGp980qlQrFYZHFxkUKhwEMPPcShQ4eYnJy08SQ3DNEvT6DvYWMq1CVKj9cA4OzsLEtLS/i+b7erq6usr69TKpWYn59nenraSuddIWE/ISIkEglGR0etOtAET6lUotlsUqlUbLBN1QnczhW4EzFFkAS3PaDHbuh13DC6zidUtSUSCVvTFIlEGB0dJZlMUqlUbKhkJxgqErR8UWfGGstx60NLpVJP7EY9IU096ohWXa85aOiVCrdiQxM00GsT9KWzYB0UIyMjNq8Qj8dZWloik8nY4N9OQxlDtVwqmEN29wM9M9x+Ix42jvbtfNdPQtxoq86U9Ts3oqrBP7fccqcYKkmA2x3glqPrVkeaiNgUo1sd50pC0DjrPkUwnB2s1FAJarValEolG8EVEVt3pKHyWCxGNpulVqsNVJs6dCQo3MyX68Govu1XXdHP4PZL6Ac/uwger6Fwdx7hTuSUDJWMQTJsQ0eCG+t3DWq73aZardqJUCQS6am66+d6Bt/3U12+7+N5Xk8HAz1BvlKpRCqVsgkcdQpcsrVKZBB1NFQ2QdFvNOmI1NCFdlK/c7ZrF4LFYe61XDvkGtx+Eqp2YlBJGEoSgA06WidkGiRzU5b9zoONtUT6ffD62kY/11Lb3oykYI7ivlBHiuAESuP45XLZVsFt9YO3KgLbzFZstg1m0NxzgxUbg2BoJSEIY0zP3MD1//V7d6vv3Vfwei6CKgZ6g4BqH4L2QOEa6p1i6CRhM/1sjKFWq1GpVGxFtu4PdnLQk+pntF3PyrULQadASyy1HLNSqWywR1qzdN/ZhCCChrCf/r3TaA/GkfoZ8H6SoC9XEoNtD9L5inuGBBfql7suKvSqj62MZFB6XFWk+l9Vjk7GANbW1mwqM7jMCjYGD7eLoSVhKzfTXZOwWVBus89bSUs/1eaSqepQK7T7tT8I7kiCiJwUkf8QkSsi8kMR+e3u/nER+TcRebu7HdvtzejkSWt6Npv5qiQEY0x6ja06drNr6jG+729QOZqN03XP7ro5xaBSANuTBB/4XWPMOeCngM+KyDlul72cAV7sft41tBNcoxqM/6ghDLqEW3lDLoLXCnpZ/aRBZ87FYrFngLjH3DUSjDHzxphXu++LwFt0Hi7ycTrlLnS3nxjoDroIjkpX3IOxm6Du3+x6/SQleMxmC0YUbhu1Ws2uINX9OlhcFblT7MhFlU790UXgMntY9qLT/mB+Vkeg1gG5CRetbNDjFMEYkDsT3mz0usf0kyZ1APL5PEDfZbz7QoKIjAD/BHzOGFMI/PBNy15kByUv/UIBKglu2nC77mm/2fJW52x2T0pCo9GgVqv1XcS4G2yLBBGJ0CHg740x3+ju3lbZi9lmyUs/PRwOh+0CEDWI+l0/Pb7Zdd1ruvv6GWzdp5M9kU7+oFar2XL9er3eN/F018IW0mnpb4G3jDF/6XylZS+wh2Uv/UZ0o9Gw0dM7HR/EZhO7rY51r62DQWfMWpm9l9iOJPw08CvAGyLyWnffH3KXyl6gV683m01b2eD++OCoc+1JP/dzq3b6SYabV0gkEvi+b6XRNcx7ge2UvLwEbNbanpe9BKEhbE3kw0YCoLdyQrfBGW3wWOhVPzrqg9fROYkuQNzKfR4EQxfAc70Nkc4zI7QGyF0Cq4870IBevwUgQc9nq84K2g5NYRpzO3ehwTtdNerer7vAfacYOhJcKAn5fL5HHWl8p16vW79dvZZgFNYlwS0CcNvQrbrKOvojkQjGGFugrCt+XPvkEjyoYR5qEmDzpIr65e5LO9E9tx8Jeo3Ntq4k6EOrYrFY32u5OfF7Pp+wWUxHi3VVH0Nv/F47XUvk9fFswUfzbNWebo0x9lka+qCTZDJJJpPpWRPttu+SdacZ+mYYGhJgo6+u74Mqptls2io89VaazaZVX8CmJPSbNetWJ4a6zk1f7tPG+hn5+8IwuzEcNbiaSgwmUHzfZ2ZmhldeecV2QLPZtOuOdXS6xG3mVbmflShVe6VSibW1NW7evMns7Cy5XM4u5VXHIPhUgHuaBNhY1rhZCFpjOPPz83a1jHpQQcKCEhB0Y91jgw8f0VDF6uqqdQzc+lVXQu8LEtwAXrCMUR9Oq8+q832fK1eusLi42BPIC8aWtkrObAUlQdssFossLS3ZNdHxeHyDSnJXj96VRSL7Ae1s9zHM2mmaXtQkeqvVYnp6munp6ffkXnX9nCuhun5aXdmdYihI0PhQvV63C7h936darbKysmIXYOz1k3kHQbvdZmlpiZmZGWsTCoWCXTgyyD0OBQla2BWLxVheXraPV45EIty6dcsaxkFWwdyNe52eniadTpNOp8lms+RyOW7dukUul6NWq+34mkOV6Fc9rp6Hekju4/cVO4mO7jVcD85dfBgsl9wu9vspL8tAGdj9A6b3H5Ps/r5/zPT5/4R9JQFARL5v+jxuZthxN+97qNTR+xUHJAwB3gsSvvgetLkXuGv3ve824QAbcaCOhgD7RoLcI/9OKPtYe2sRrDi7Gy86/7kzDTwERIH/Bc7tR9sD3OtR4Ce670fp/KviOeAvgOe6+58D/nyv2twvSbhn/p3Q7FPtrYv9IqHfvxMe36e2B4bcpdrbIA4M8yaQQO2t+53p6KQ9cyv3i4Rt/TvhsGCr2tvu91s+cm6n2C8S7L8TikiUzr8T/ss+tb0j7HftLbA/3lHXo3iajqcxDfzRe+0FbXGfj9NRNa8Dr3VfTwMTdFYkvQ38OzC+V20ezJiHAAeGeQhwQMIQ4ICEIcABCUOAAxKGAENPgoi0ROQ1EXlTRL4uIsk7HP8dEflw9/2siEzuz50OjqEnAagaYy6Yzr+YNIBn3+sbgs6kTkT2pP/uBRJcfBc4LSIfFZEXdKeIfEFEfm2rE0Xkd7rS9KaIfK677/Mi8lnnmD8Rkd/rvv99EXlZRF4XkT/t7jvVzYl8BXiT3lDMwLhnSBARD/g54I0Bzv0Q8Axwic7zOX5TRC7S+V8gd9XpJ4GvisjHgDN0QvAXgA+JyJPdY84Af2WMOW+MuTHgz+nBvUBCQjpLd78P3KQT19kpHgf+2RhTNsaUgG8ATxhjfgAcFpFjIvIYsGaMeQf4WPf1A+BV4Cydzge4YYz57139ogCGohb1DqgaYy64O0TEp3cAxXdx/a8DvwhM0ZEM6CwZ/jNjzF8H2j1Fp4JwT3EvSEI/3ADOiUhMRLLceT31d4FPiEhSRFLAL3T3QafjP0WHiK93930L+PVuTgEROS4ih/f4N1jcC5KwAcaYd0Tka3SM4wwdtbHV8a+KyJeB/+nu+puuKsIY80MRGQVuaebMGPOvIvLjwH91C45LwC8De/s8hS4OoqhDgHtVHd1XOCBhCHBAwhDggIQhwAEJQ4ADEoYAByQMAQ5IGAL8P57VSGcANJVEAAAAAElFTkSuQmCC\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAGEAAABwCAYAAAAOjB6dAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/bCgiHAAAACXBIWXMAAAsTAAALEwEAmpwYAAAQRklEQVR4nO2dbWxb13nHfw95+SaSoiiREinZjmNFsRw7jpMuixtvWIZlRdEsy7AN3Yps6IIB/TCs6D4MmIF+WIcBW7sPwwZs/WAswTxg2BJgCRb0S9AVTbN4yBY7c5O4RVJniRPHim1R1gvFt8vLsw+85/qIJiVaoqWb+v4BQvcennvPuc//PC/nOUeXopQiwM4itNMdCBCQ4AsEJPgAAQk+QECCDxCQ4ANsiQQR+byIvCMi50Xk+KA6dbtBNjtPEJEw8C7wy8BF4HXgS0qpHw2ue7cHtqIJPwucV0r9n1KqAfwr8MRgunV7wdrCtVPAR8b5ReCh9S4Qkdt9ej6vlMp3Fm6FhL4gIl8BvnKr2/mU4EK3wq2Q8DGw2zjf5ZatgVLqBHACAk3oha34hNeBGRG5U0SiwG8DLw6mW7cXNq0JSqmmiPwh8BIQBp5RSp0bWM9uI2w6RN1UY4E5OqOU+pnOwmDG7AMEJPgAAQk+QECCDxCQ4AMEJPgAAQk+QECCDxCQ4AMEJPgAAQk+QECCD3DLF3V2CiLiHXcmKUUEEUEpdcN3nfX6TXCGQiGvrZtNiv5UaIIWaidCoVDX73RZt2tupo5ZNxQKee3dLDbUBBF5BvgV4IpS6pBbNgo8C+wFPgC+qJS6dtOtDwgbjbzO7/V5Z3kkEuGOO+4glUqxd+9eisUi7777Lq+88gq2ba97/1ar1VdfuqEfTfhH4PMdZceB7ymlZoDvuee+wnojUgutU2CxWIwDBw5w7NgxnnzySY4fP87jjz9OLBbbsL1Wq9X1nv1gQ01QSr0iIns7ip8AHnGPTwIvA39y063fQvQa7SLC8PAwiUSCaDRKNBolFouRTqdJp9Ps37+ffD5POp1GKUUmk+HgwYOUSiUuXrxIrVbr2p6IYFkWuVyOWCzGwsICy8vLffV1s455Qik15x5/Akxs8j63DL0cZDgcZmZmhl27dpHL5cjn84yPj3PkyBHi8TjhcNgT6MrKCnv27OGpp57iwoULnDx5kkuXLt1wTxEhHA6TSqV4+OGHKRaLvPrqq5w9e7avvm45OlJKqfWWLXdqy4t2lOFwmGg0SigUwrIsYrEYk5OT7N69m2w26xGRy+WIx+M0Gg0cxwHaRCYSCSYnJ6lWq0Qika5thcNhhoaGSKVSjI6OMjY2xtDQUN993SwJl0WkqJSaE5EicKVXxZ3a8pJOp0mlUkxMTDAzM8PQ0BC7du0ilUpx7733Mjk5SaPRoF6vEwqFqFarVCoVarUajuOQSqVIJBKMjo5SKBQQEeLxeNe2MpkMhw4dYnR0lNnZWXK5HJlMpu++bpaEF4EvA990//77Ju8zcGgNSCQSpNNp8vk8d955J5lMhunpaYaHh5mdnaVQKLC0tMTCwgKNRoPV1VWazSbVahXHcUgkEoRCIaLRKJlMhkwmQzgcXtOWDmETiQSFQsETfjKZJBqN9j3P6CdE/RfaTjgnIheBP6Ut/OdE5Pdp7yr74s2La8N2CYVCa8K/XgiFQkQiEYaGhjh8+DD5fJ7p6Wlv5I+NjXmmyLIslpeXsW0b27ap1+u0Wi3C4TCO41CtVqnX6555mZub4+zZs7z33nuUy2WgHUVFIhHGx8cpFovs27ePxx57DMuyOH36NJcvX6ZUKjE1NUWlUuHatWvrktFPdPSlHl/90kbX3gx0SKk7q+25ts/rIRQKEYvFGB4e5oEHHuDuu+/m8OHD7N+/H8dxaDQa2LbN4uIitm2zsrLC0tIS4XDYm2CFw2Esy6Jer1Mul8lms7RaLa5evcrrr7/Oxx9/TKVSQUSIxWLE43EmJye57777mJ2d5dFHH6VSqfDCCy/w2muvkUgkmJiYoFQqsbi4uDUStgudJCilcBxn3RAzlUoxMjJCOp1mamqKdDrNPffcQz6fx7IslpaWcByHZrOJ4zi0Wi1EpKuD1W2l02lPY2q1GtFolD179pDJZBARKpUK0WgUy7IYHR0ll8th2zYvvvgiKysrXLhwgWq1ilKKZrNJo9HY+Nn9sPnLTA9sZHp0iDkxMcFdd93FgQMHKBaLPPjggyQSCer1Oo7jsLq6SrVa9SZR2rGGw2EikQjhcJhms0mz2aTVannCikQihEIhKpUK9XqdSCRCPB4nEomQSqW86xzHYWFhgVKpxFtvvcXTTz/NtWvXPDK1yarVaiwvL+vB1HXzl280oRdisRipVMoTQiwWY3p6mnw+78X6w8PDRKNRz3FqIrVfgesO28wndSNfzy/0d5o0y7K8+1+9epWVlRWuXbtGqVRiYWGBaDRKOp1mZGSEWCzmXVsqlbywt5dW+IKE9TKPepTncjmOHj1KNpslm82SSqW8kSwizM/PeyZAj8R0Ot31ntrhdyboWq0WlUoFpRSWZZFIJLzJW71eZ35+nnK5zHPPPceZM2e89sfGxjh69ChjY2M89NBDTExMUC6XqVQqnDlzhueff556vc5HH33UtT87QoIejZ0jE7jh78jICMVikUKhwL59+8jlcp5GrK6usry8jOM4XnyvyYzH41jW9cfrRnQ34vU9TKetI7RGo0G1WmV+ft6bOWsfNTExQaFQYHp6mkKhwPLyMuVymUuXLpHJZKhWqz3lse0+IRQKMTY2RjKZJJfLUSgUiMfjjIyMeBGKjr3j8TjZbJapqSni8Tj5fJ5oNKrv5d1XO3HAG7mRSIRIJOLZfH1N58g3fYI+V0qtGSiApxlKKc6fP8/Vq1c9UxWJRDwzpE2j9kXnz5/n1KlT2LbNiRMn/OETRIRkMsno6Ci7d+9mZmaGVCpFoVDwbK9lWQwPD5NMJolEIsRisTXRRr1ep9lseqGiFpYWvtYwPcCazSZwnSAzEjN9gL5OZ0O1IPW1mUyGaDTKxEQ7VaYTgPV63Zv0lctlyuWy9yzpdJp9+/Z5feiGbSUhkUhw8OBB9u/fz/j4OLlcjvHxcc9+ayFB2yFrR6hz+aZA9DxCC0trgm3bawTqOA6O43hJuW5mzySm04Hrv5ZleYLU5OrJno6WtAZZluVpiGkSe2FbSUilUhw7dowjR46wa9cur6OmELQ50ILWJmJNp11hanutIw8zn29qgx7pZoSjoxdNvK5rmh/Lsjyydd90/3Q7Zh90v0XE05JeSb81zzMY8fYH27b55JNP+PDDD7Ft28vvhMNh4vG498BawKYgO523ea4JM4VgClZfownXgtYkmIs82gSZ9+lcIDLNmHn/zjqtVstz5L4xR6urq5w6dYr333+fkZERCoWCN9MtFArEYjGy2SyxWMwbRdr5mcLrjOXh+pwglUp5JsCso80EsGb0dzpqnUuybXvNTLtzCVOPfh0iW5bl1TPLS6US586d67kYBNtMQqvVYnV1lYWFBZrNpifgSqUCtMNK27aJxWIkk0mGhoa8xJtO0nXTBlMjGo3GDaNU1zdNUWeGUwtZh7o632RGTJ3Ll5okwDOdOhurzdmVK1dYXFxcl4RtDVEty1LpdHqNrdV2WjtiHd/rnFAymSSbzRKNRhkZGSESiZBMJr36pmDhetpBk2DbNrVazQtDlVLUajWazSa2bdNoNGg2m166o1qtrjk3Z7paVlqTOmXXarUol8ueFimlWFpa4sqVK7RaLRYXF/0RomohaAGZU3nTeWazWYaHh0mn096ql16/HR4e9rREmxgzutHtKKW8rKg5yldXV6nVajQaDY8QvY6gR7ImqVu6QQ+ibs+2srLiPZeO6jYa6BtqgojsBv6J9jqyAk4opf52M9teQqGQ0kuNnTkco46XLtZ1tXboBJrOE5l2vdfuCtOMaOdo2nltTrS91+Xm/Tq1zSzXfdAOvlKpeNqltU+nQubm5rpqQj8kFIGiUuoNEUkDZ4BfA34PWFBKfVPar9nJKqXW3XGhs6g6Vax3OnTG7p2CazQaa6IbcxKlr+u082YCToe0Zsxvzge0bzL9jiZaZ1HNUNWcE+iPnlBqbdNEax/YarV4++23N2eO3F0Vc+7xioj8mPbLRTa97cVcqNFp5k6Y2U2TnG45J9Ppmtebo1RDX2u2YfZFRzVaayzLolKp3HAfE5oswMuYapNWrVZZXFxcN0V/Uz5B2vuP7gf+my1se9EdrNfrPc2IDk9NgWsN0aNUf0zBapizb3Mu0IlOJ1uv1z1fYsb25qjvTG+Y4bMu09dXq1WWlpYGQ4KIpIB/A/5IKbXckUDrue1FNtjy0sscag3oDCP1KDUnVt00odls3jDiO5dKTW0yhajnCdqxmk5ft6X7Zs7Q9fXajGpT2m2F0ERfJIhIhDYB/6yUet4t7mvbi9rklhczBjcJ7zZz7tHnfptaU18Lq1vqu7MfnfOQbtf3szWyn90WAjwN/Fgp9dfGV7d824v5QD/V6MyDdMmL/Bzt0PRN4Kz7+QIwRnsz8E+A/wBG+7iXus0/p7vJxRcL/bcRgre8+BUBCT5AQIIPEJDgAwQk+AABCT5AQIIPEJDgAwQk+AABCT5AQIIPEJDgA2z3bot5YNX9+2lDjq33+45uhduaRQUQkdPdMol+x63sd2COfICABB9gJ0g4sQNtDgK3rN/b7hMC3IjAHPkA20aCfEp+nVBEdovI90XkRyJyTkS+5paPish3ReQn7t/swBrdaIfEID60f3PnPWAfEAV+CNyzHW1voq9F4AH3OE37VxXvAf4KOO6WHwe+Nag2t0sTPjW/TqiUmlNKveEerwDm3tuTbrWTtDdFDwTbRUK3Xyec2qa2N41B7b3dCIFj7oHOvbfmd6ptkwYWVm4XCX39OqFfsN7eW/f7dV85d7PYLhI+Nb9O2MfeWxj03tttjDq+QDvSeA/4+k5HQev0c2B7b/v9BDNmHyBwzD5AQIIPEJDgAwQk+AABCT6A70kQka+72cw3ReSsiDw0wHs/IiLfGdT9NgtfvA2yF0Tks7R/xeQBpVRdRHK0s7A7DhGxlFK9X2J0E/C7JhSBeaVUHUApNa+UuiQiH4jIn4nIGyLylojMAohIUkSeEZH/EZH/FZEn3PK9IvKfbv03ROThzoZE5EH3mmkR+YyI/EBEzojIS0a64mUR+RsROQ18bWBPudMz1A1mrynaM9Z3gW8Dv+CWfwB81T3+A+Af3OO/AH7HPR5xr0sCQ0DcLZ/B/S9K2q+F+A7wMO13duwBIsB/AXm3zm8Bz7jHLwPfHvRz+tocKaXKIvIZ4OeBXwSeNVbldGLtDPDr7vHngF8VkT92z+O0BXsJ+DsROQI4wN1GMwdoL+J/ztWyQ8Ah4LvuP4iHcd/t4eLZwT1hG74mAUAp5dAegS+LyFtcT6LV3b8O159DgN9QSr1j3kNEvgFcBu6jbYLN13DN0SbrftpkCXBOKfXZHl1a3cLjdIWvfYKI7BeRGaPoCO3fa+iFl4CvuplQROR+tzwDzCmlWsDv0h7dGovAY8BfisgjwDtA3g0KEJGIiBzc8sOsA1+TQNsnnHQX3d+kvdb7jXXq/zltm/6miJxzz6HtT74sIj8EZukYzUqpy7SjsL+nrRG/CXzLrX+Wts+4ZQiyqD6A3zXhtkBAgg8QkOADBCT4AAEJPkBAgg8QkOADBCT4AP8PpHRg2j1cBo4AAAAASUVORK5CYII=\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "\u001b[34m\u001b[1mwandb\u001b[0m: \u001b[33mWARNING\u001b[0m Calling wandb.login() after wandb.init() has no effect.\n",
            "\u001b[34m\u001b[1mwandb\u001b[0m: \u001b[33mWARNING\u001b[0m Calling wandb.login() after wandb.init() has no effect.\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Create sweep with ID: zm71t1t4\n",
            "Sweep URL: https://wandb.ai/dl_sumit/Deep_learning_final1/sweeps/zm71t1t4\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "\u001b[34m\u001b[1mwandb\u001b[0m: Agent Starting Run: e8w563bi with config:\n",
            "\u001b[34m\u001b[1mwandb\u001b[0m: \tOptimization: RMS\n",
            "\u001b[34m\u001b[1mwandb\u001b[0m: \tactivation: sigmoid\n",
            "\u001b[34m\u001b[1mwandb\u001b[0m: \tepochs: 15\n",
            "\u001b[34m\u001b[1mwandb\u001b[0m: \thls: 5\n",
            "\u001b[34m\u001b[1mwandb\u001b[0m: \thn: 64\n",
            "\u001b[34m\u001b[1mwandb\u001b[0m: \tinitialization: xavier\n",
            "\u001b[34m\u001b[1mwandb\u001b[0m: \tlearning_rate: 0.01\n",
            "\u001b[34m\u001b[1mwandb\u001b[0m: \tsize_batch: 32\n",
            "\u001b[34m\u001b[1mwandb\u001b[0m: \u001b[33mWARNING\u001b[0m WANDB_NOTEBOOK_NAME should be a path to a notebook file, couldn't find Deep_learning_final1.ipynb.\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ],
            "text/html": [
              "Tracking run with wandb version 0.14.0"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ],
            "text/html": [
              "Run data is saved locally in <code>/content/wandb/run-20230319_093107-e8w563bi</code>"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ],
            "text/html": [
              "Syncing run <strong><a href='https://wandb.ai/dl_sumit/Deep_learning_final1/runs/e8w563bi' target=\"_blank\">snowy-sweep-1</a></strong> to <a href='https://wandb.ai/dl_sumit/Deep_learning_final1' target=\"_blank\">Weights & Biases</a> (<a href='https://wandb.me/run' target=\"_blank\">docs</a>)<br/>Sweep page: <a href='https://wandb.ai/dl_sumit/Deep_learning_final1/sweeps/zm71t1t4' target=\"_blank\">https://wandb.ai/dl_sumit/Deep_learning_final1/sweeps/zm71t1t4</a>"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ],
            "text/html": [
              " View project at <a href='https://wandb.ai/dl_sumit/Deep_learning_final1' target=\"_blank\">https://wandb.ai/dl_sumit/Deep_learning_final1</a>"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ],
            "text/html": [
              " View sweep at <a href='https://wandb.ai/dl_sumit/Deep_learning_final1/sweeps/zm71t1t4' target=\"_blank\">https://wandb.ai/dl_sumit/Deep_learning_final1/sweeps/zm71t1t4</a>"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ],
            "text/html": [
              " View run at <a href='https://wandb.ai/dl_sumit/Deep_learning_final1/runs/e8w563bi' target=\"_blank\">https://wandb.ai/dl_sumit/Deep_learning_final1/runs/e8w563bi</a>"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "\u001b[34m\u001b[1mwandb\u001b[0m: Ctrl + C detected. Stopping sweep.\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 0 Axes>"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Error in callback <function _WandbInit._pause_backend at 0x7fd176937c10> (for post_run_cell):\n"
          ]
        },
        {
          "output_type": "error",
          "ename": "BrokenPipeError",
          "evalue": "ignored",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mBrokenPipeError\u001b[0m                           Traceback (most recent call last)",
            "\u001b[0;32m/usr/local/lib/python3.9/dist-packages/backcall/backcall.py\u001b[0m in \u001b[0;36madapted\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m    102\u001b[0m                 \u001b[0mkwargs\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpop\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mname\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    103\u001b[0m \u001b[0;31m#            print(args, kwargs, unmatched_pos, cut_positional, unmatched_kw)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 104\u001b[0;31m             \u001b[0;32mreturn\u001b[0m \u001b[0mcallback\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    105\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    106\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0madapted\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.9/dist-packages/wandb/sdk/wandb_init.py\u001b[0m in \u001b[0;36m_pause_backend\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m    416\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mbackend\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0minterface\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    417\u001b[0m             \u001b[0mlogger\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0minfo\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"pausing backend\"\u001b[0m\u001b[0;34m)\u001b[0m  \u001b[0;31m# type: ignore\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 418\u001b[0;31m             \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mbackend\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0minterface\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpublish_pause\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    419\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    420\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0m_resume_backend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m->\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.9/dist-packages/wandb/sdk/interface/interface.py\u001b[0m in \u001b[0;36mpublish_pause\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m    663\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mpublish_pause\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m->\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    664\u001b[0m         \u001b[0mpause\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mpb\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mPauseRequest\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 665\u001b[0;31m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_publish_pause\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mpause\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    666\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    667\u001b[0m     \u001b[0;34m@\u001b[0m\u001b[0mabstractmethod\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.9/dist-packages/wandb/sdk/interface/interface_shared.py\u001b[0m in \u001b[0;36m_publish_pause\u001b[0;34m(self, pause)\u001b[0m\n\u001b[1;32m    338\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0m_publish_pause\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mpause\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0mpb\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mPauseRequest\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m->\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    339\u001b[0m         \u001b[0mrec\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_make_request\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mpause\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mpause\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 340\u001b[0;31m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_publish\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mrec\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    341\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    342\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0m_publish_resume\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mresume\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0mpb\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mResumeRequest\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m->\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.9/dist-packages/wandb/sdk/interface/interface_sock.py\u001b[0m in \u001b[0;36m_publish\u001b[0;34m(self, record, local)\u001b[0m\n\u001b[1;32m     49\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0m_publish\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mrecord\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0;34m\"pb.Record\"\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlocal\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0mOptional\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mbool\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m->\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     50\u001b[0m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_assign\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mrecord\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 51\u001b[0;31m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_sock_client\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msend_record_publish\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mrecord\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     52\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     53\u001b[0m     def _communicate_async(\n",
            "\u001b[0;32m/usr/local/lib/python3.9/dist-packages/wandb/sdk/lib/sock_client.py\u001b[0m in \u001b[0;36msend_record_publish\u001b[0;34m(self, record)\u001b[0m\n\u001b[1;32m    219\u001b[0m         \u001b[0mserver_req\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mspb\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mServerRequest\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    220\u001b[0m         \u001b[0mserver_req\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mrecord_publish\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mCopyFrom\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mrecord\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 221\u001b[0;31m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msend_server_request\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mserver_req\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    222\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    223\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0m_extract_packet_bytes\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m->\u001b[0m \u001b[0mOptional\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mbytes\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.9/dist-packages/wandb/sdk/lib/sock_client.py\u001b[0m in \u001b[0;36msend_server_request\u001b[0;34m(self, msg)\u001b[0m\n\u001b[1;32m    153\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    154\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0msend_server_request\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mmsg\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0mAny\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m->\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 155\u001b[0;31m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_send_message\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mmsg\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    156\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    157\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0msend_server_response\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mmsg\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0mAny\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m->\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.9/dist-packages/wandb/sdk/lib/sock_client.py\u001b[0m in \u001b[0;36m_send_message\u001b[0;34m(self, msg)\u001b[0m\n\u001b[1;32m    150\u001b[0m         \u001b[0mheader\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mstruct\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpack\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"<BI\"\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mord\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"W\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mraw_size\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    151\u001b[0m         \u001b[0;32mwith\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_lock\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 152\u001b[0;31m             \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_sendall_with_error_handle\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mheader\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0mdata\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    153\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    154\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0msend_server_request\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mmsg\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0mAny\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m->\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.9/dist-packages/wandb/sdk/lib/sock_client.py\u001b[0m in \u001b[0;36m_sendall_with_error_handle\u001b[0;34m(self, data)\u001b[0m\n\u001b[1;32m    128\u001b[0m             \u001b[0mstart_time\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtime\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmonotonic\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    129\u001b[0m             \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 130\u001b[0;31m                 \u001b[0msent\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_sock\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdata\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    131\u001b[0m                 \u001b[0;31m# sent equal to 0 indicates a closed socket\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    132\u001b[0m                 \u001b[0;32mif\u001b[0m \u001b[0msent\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;36m0\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mBrokenPipeError\u001b[0m: [Errno 32] Broken pipe"
          ]
        }
      ],
      "source": [
        "#Terminologies\n",
        "#xtr----Xtrain, ytr----Ytrain, xte----Xtest, yte----Ytest\n",
        "#inn----Input Neuron, otn----Output Neuron, hls----Hidden Layers, Input 'u'---Input to the neuron, 'v'----Output from the neuron\n",
        "\n",
        "\n",
        "\n",
        "from keras.datasets import fashion_mnist\n",
        "import wandb\n",
        "import os\n",
        "import numpy as np \n",
        "import matplotlib.pyplot as plt # for data visualization\n",
        "import copy\n",
        "import math\n",
        "(xtr, ytr), (xte, yte) = fashion_mnist.load_data()\n",
        "class_names=['T-shirt/top', 'Trouser', 'Pullover', 'Dress', 'Coat', 'Sandal', 'Shirt', 'Sneaker', 'Bag', 'Ankle boot']\n",
        "\n",
        "#Plotting the images\n",
        "for k in range(9):\n",
        "  #Subplot definition \n",
        "  plt.subplot(330+1+k)\n",
        "  #Data plotting\n",
        "  plt.imshow(xtr[k], cmap=plt.get_cmap('gray'))                                        \n",
        "  plt.xlabel(class_names[ytr[k]])\n",
        "  plt.show()\n",
        "\n",
        "xtr = xtr.reshape(60000, 784)/255  #Data Normalization\n",
        "xte = xte.reshape(10000, 784)/255\n",
        "\n",
        "# relu function.\n",
        "def relu(x):\n",
        "    return np.maximum(0, x)\n",
        "\n",
        "# derivative of relu\n",
        "def drelu(x):\n",
        "    if x > 0:\n",
        "      return 1\n",
        "    elif x <= 0:\n",
        "      return 0\n",
        "\n",
        "# Definition of logistic function\n",
        "def sigmoid(x):\n",
        "    return 1.0 / (1.0 + np.exp(-x))\n",
        "\n",
        "\n",
        "# derivative of tanh function.\n",
        "def dtanh(x):\n",
        "    return 1.0 - np.square(np.tanh(x))\n",
        "\n",
        "\n",
        "# derivate of logistic function\n",
        "def dsigmoid(x):\n",
        "    \n",
        "    return sigmoid(x) * (1 - sigmoid(x))\n",
        "\n",
        "\n",
        "\n",
        "# Definition of an Activation function\n",
        "def activ(ini, hn, activation):\n",
        "    if activation == \"sigmoid\":\n",
        "        Oi = np.empty([hn, 1])\n",
        "        for k in range(hn):\n",
        "            Oi[k] = sigmoid(ini[k])\n",
        "        return Oi\n",
        "    elif activation == \"tanh\":\n",
        "        Oi = np.empty([hn, 1])\n",
        "        for k in range(hn):\n",
        "            Oi[k] = np.tanh(ini[k])\n",
        "        return Oi\n",
        "    elif activation == \"relu\":\n",
        "      Oi = np.empty([hn, 1])\n",
        "      for k in range(hn):\n",
        "            Oi[k] = relu(ini[k])\n",
        "      return Oi\n",
        "\n",
        "# Activation function derivative definition\n",
        "def deractiv(ini, hn, activation):\n",
        "    if activation == \"sigmoid\":\n",
        "        Oi = np.empty([hn, 1])\n",
        "        for k in range(hn):\n",
        "            Oi[k] = dsigmoid(ini[k])\n",
        "        return Oi\n",
        "    elif activation == \"tanh\":\n",
        "        Oi = np.empty([hn, 1])\n",
        "        for k in range(hn):\n",
        "            Oi[k] = dtanh(ini[k])\n",
        "        return Oi\n",
        "    elif activation == \"relu\":\n",
        "        Oi = np.empty([hn, 1])\n",
        "        for k in range(hn):\n",
        "            Oi[k] = drelu(ini[k])\n",
        "        return Oi\n",
        "\n",
        "# Output function\n",
        "def OUT(ul, otn):\n",
        "    outp = np.empty([otn, 1])\n",
        "    tot = 0\n",
        "    for k in range(otn):\n",
        "        tot = tot + np.exp(ul[k])\n",
        "    for k in range(otn):\n",
        "        outp[k] = np.exp(ul[k]) / tot\n",
        "    return outp\n",
        "\n",
        "# forward propogation  Returns the final probability distribution for each of the label.\n",
        "def forwardprop(xtr, hls, hn, inn, otn, W, b, u, v, la, activation):\n",
        "    v[0] = xtr[la].reshape(-1, 1)\n",
        "    for k in range(1, hls):\n",
        "        u[k] = np.dot(W[k], v[k-1]) + b[k]\n",
        "        v[k] = activ(u[k], hn, activation)\n",
        "    u[hls] = np.dot(W[hls], v[hls-1]) + b[hls]\n",
        "    y_predicted = OUT(u[hls], otn)\n",
        "    return y_predicted\n",
        "\n",
        "\n",
        "# Definition of an Initialization function, to intializer W's and b's\n",
        "def init(W, b, inn, hls, hn, otn, initialization):\n",
        "    if initialization == \"random\":\n",
        "          \n",
        "           W[1] = np.random.rand(hn,inn)*0.01\n",
        "           for k in range(2,hls):\n",
        "              \n",
        "            W[k] = (np.random.rand(hn, hn))*0.01   \n",
        "           for k in range(1,hls):\n",
        "              b[k] = np.zeros([hn, 1])\n",
        "        \n",
        "             \n",
        "              b[hls] = np.zeros([otn, 1])\n",
        "              W[hls] = np.random.rand(otn, hn)*0.01\n",
        "\n",
        "    if initialization == \"xavier\":\n",
        "           variance1 = 2.0 / (inn + otn)\n",
        "           std_dev1 = np.sqrt(variance1)\n",
        "           W[1] = np.random.normal(0.0, std_dev1, (hn, inn))\n",
        "    for k in range(2,hls):\n",
        "        variance2 = 2.0 / (hn + hn)\n",
        "        std_dev2 = np.sqrt(variance2)\n",
        "        W[k] = np.random.normal(0.0, std_dev2,(hn, hn))\n",
        "              \n",
        "    for k in range(1,hls):\n",
        "        b[k] = np.zeros([hn, 1])\n",
        "        variance3 = 2.0 / (otn + hn)\n",
        "        std_dev3 = np.sqrt(variance3)\n",
        "        W[hls] = np.random.normal(0.0, std_dev3, (otn, hn))\n",
        "        b[hls] = np.zeros([otn, 1])\n",
        "\n",
        "    return W, b\n",
        "\n",
        "# One hot vector\n",
        "#In order to classify one output at a time\n",
        "def hot(m, n):\n",
        "    h = np.zeros([n, 1])\n",
        "    h[m] = 1\n",
        "    return h\n",
        "\n",
        "#Backpropagation step, in order to find the partial derivatives of the loss function w.r.t different parametres\n",
        "def Backprop(ytr, hls, hn, inn, otn, W, b, u, v, gradW, gradb, la, y_predicted, activation, los):\n",
        "    grad_u = {}\n",
        "    grad_v = {}\n",
        "    value = np.empty([otn,1])\n",
        "    if los == \"cross_entropy\":\n",
        "        value = y_predicted - hot(ytr[la], otn)\n",
        "    elif los == \"squared_error\":\n",
        "        for k in range(otn):\n",
        "            value[k] = np.dot(np.transpose(np.multiply(y_predicted, y_predicted - hot(ytr[la], otn))), hot(k, otn) - np.ones([otn,1]) * y_predicted[k])\n",
        "    grad_u[hls] = value\n",
        "    for i in range(hls, 0, -1):\n",
        "        if i in gradW.keys():\n",
        "            gradW[i] = gradW[i] + np.dot(grad_u[i],np.transpose(v[i-1]))\n",
        "            gradb[i] = gradb[i]  + grad_u[i]\n",
        "        else:\n",
        "            gradW[i] = np.dot(grad_u[i],np.transpose(v[i-1]))\n",
        "            gradb[i] = grad_u[i]\n",
        "        grad_v[i-1] = np.dot(np.transpose(W[i]), grad_u[i])\n",
        "        if i != 1:\n",
        "            grad_u[i-1] = np.multiply(grad_v[i-1], deractiv(u[i-1], hn, activation).reshape(-1, 1))\n",
        "    return gradW, gradb\n",
        "\n",
        "# loss function : squared error loss, cross entropy\n",
        "def LOSS(y_predicted, ytr, los, la, otn):\n",
        "    if los == \"cross_entropy\":\n",
        "        return -np.log(y_predicted[ytr[la]])\n",
        "    elif los == \"squared_error\":\n",
        "        return np.square(y_predicted - hot(ytr[la], otn))\n",
        "\n",
        "# Computation of Accuracy with trained W's and b's\n",
        "def acc(xte, yte, xtr, ytr, W, b, hls, otn, hn, activation):\n",
        "    accuracy = 0\n",
        "    y_p = np.zeros(yte.shape)\n",
        "    for la in range(10000):\n",
        "        u = {}\n",
        "        v = {}\n",
        "        v[0] = xte[la].reshape(-1, 1)\n",
        "        for k in range(1, hls):\n",
        "            u[k] = np.dot(W[k], v[k-1]) + b[k]\n",
        "            v[k] = activ(u[k], hn, activation)\n",
        "        u[hls] = np.dot(W[hls], v[hls-1]) + b[hls]\n",
        "        y_predicted = OUT(u[hls], otn)\n",
        "        ix = 0\n",
        "        value = 0\n",
        "        for k in range(np.shape(y_predicted)[0]):\n",
        "            if value < y_predicted[k]:\n",
        "                ix = k\n",
        "                value = y_predicted[k]\n",
        "        y_p[la] = ix\n",
        "        accuracy += int(ix == yte[la])\n",
        "    print(\"accuracy is \", accuracy/100, \"%\")\n",
        "    wandb.log({\"conf_mat\" : wandb.plot.confusion_matrix(probs = None, preds = y_p, y_true = yte, class_names = class_names)})\n",
        "    wandb.log({\"test_acc\": accuracy / 100})\n",
        "    acc_vali = 0\n",
        "    vali_total_loss = 0 \n",
        "    for la in range(54000, 60000):\n",
        "        u = {}\n",
        "        v = {}\n",
        "        y_predicted = forwardprop(xtr, hls, hn, inn, otn, W, b, u, v, la, activation)\n",
        "        loss_vali = LOSS(y_predicted, ytr, los, la, otn)\n",
        "        vali_total_loss = vali_total_loss + loss_vali\n",
        "        ix = 0\n",
        "        value = 0\n",
        "        for k in range(np.shape(y_predicted)[0]):\n",
        "            if value < y_predicted[k]:\n",
        "                ix = k\n",
        "                value = y_predicted[k]\n",
        "        acc_vali = acc_vali + int(ix == ytr[la])\n",
        "    print(\"Validation Accuracy is \", acc_vali/60, \"%\")\n",
        "    print(\"Total Validation Loss is \", vali_total_loss)\n",
        "    wandb.log({\"val_acc\" : acc_vali/60})\n",
        "    wandb.log({\"val_loss\" : vali_total_loss})\n",
        "\n",
        "# Different Optimization Methods i.e. Momentum based Gradient Descent Method Definition\n",
        "def momentumbasedgd(xtr, ytr, hls, hn, inn, otn, W, b, u, v, learning_rate, epochs, size_batch, set_train, activation, los):\n",
        "        gamma  = 0.9\n",
        "        W_prev = copy.deepcopy(W)\n",
        "        b_prev = copy.deepcopy(b)\n",
        "        \n",
        "        for p in range(epochs):\n",
        "            gradW = {}\n",
        "            gradb = {}\n",
        "            loss_total = 0\n",
        "            for la in range(set_train):\n",
        "                y_predicted = forwardprop(xtr, hls, hn, inn, otn, W, b, u, v, la, activation)\n",
        "                loss_current = LOSS(y_predicted, ytr, los, la, otn)\n",
        "                loss_total = loss_total + loss_current\n",
        "                Backprop(ytr, hls, hn, inn, otn, W, b, u, v, gradW, gradb, la, y_predicted, activation, los)\n",
        "                if la % size_batch == 0:\n",
        "                    for k in range(1, hls+1):\n",
        "                        update_W = gamma * W_prev[k] * int(p != 0) + learning_rate * gradW[k]\n",
        "                        W[k] = W[k] - update_W\n",
        "                        W_prev[k] =  copy.deepcopy(update_W)\n",
        "                    for k in range(1, hls):    \n",
        "                        update_b = gamma * b_prev[k] * int(p != 0) + learning_rate * gradb[k]\n",
        "                        b[k] = b[k] - update_b\n",
        "                        b_prev[k] =  copy.deepcopy(update_b)\n",
        "                    gradW = {}\n",
        "                    gradb = {}\n",
        "            \n",
        "            wandb.log({\"loss\": loss_total})\n",
        "            wandb.log({\"epoch\": p})\n",
        "            acc(xte, yte, xtr, ytr, W, b, hls, otn, hn, activation)\n",
        "\n",
        "# Different Optimization methods i.e., RMSprop\n",
        "def RMS(xtr, ytr, hls, hn, inn, otn, W, b, u, v, learning_rate, epochs, size_batch, set_train, activation, los): \n",
        "         Ep = 1e-8\n",
        "         B = 0.1\n",
        "         B1 = 0.9\n",
        "         update_b = {}\n",
        "         update_W = {}\n",
        "         for p in range(epochs):\n",
        "            gradW = {}\n",
        "            gradb = {}\n",
        "            loss_total = 0\n",
        "            for la in range(set_train):\n",
        "                y_predicted = forwardprop(xtr, hls, hn, inn, otn, W, b, u, v, la, activation)\n",
        "                loss_current = LOSS(y_predicted, ytr, los, la, otn)\n",
        "                loss_total = loss_total + loss_current\n",
        "                Backprop(ytr, hls, hn, inn, otn, W, b, u, v, gradW, gradb, la, y_predicted, activation, los)\n",
        "                if la % size_batch == 0:\n",
        "                    for k in range(1, hls+1):\n",
        "                        if(p==0):\n",
        "                            update_W[k] = np.zeros(gradW[k].shape, dtype=float)\n",
        "                            update_b[k] = np.zeros(gradb[k].shape, dtype=float)\n",
        "\n",
        "                    for k in range(1, hls+1):\n",
        "                        update_W[k] = B1 * update_W[k] + (1 - B1) * np.square(gradW[k])\n",
        "                        W[k] = W[k]- (learning_rate / np.sqrt(update_W[k] + Ep)) * gradW[k]\n",
        "\n",
        "                    for k in range(1, hls):\n",
        "                        update_b[k] = B1 * update_b[k] + (1 - B1) * np.square(gradb[k])\n",
        "                        b[k] -= (learning_rate / np.sqrt(update_b[k] + Ep)) * gradb[k]\n",
        "                    gradW = {}\n",
        "                    gradb = {}\n",
        "           \n",
        "                    wandb.log({\"loss\": loss_total})\n",
        "                    wandb.log({\"epoch\": p})\n",
        "            acc(xte, yte, xtr, ytr, W, b, hls, otn, hn, activation)\n",
        "\n",
        "# Different Optimization methods i.e., Nesterov Accelerated gradient descent\n",
        "def Nesterovagd(xtr, ytr, hls, hn, inn, otn, W, b, u, v, learning_rate, epochs, size_batch, set_train, activation, los): \n",
        "         W_prev = copy.deepcopy(W)\n",
        "         b_prev = copy.deepcopy(b)\n",
        "         gamma  = 0.9\n",
        "         for p in range(epochs):\n",
        "            gradW = {}\n",
        "            gradb = {}\n",
        "            loss_total = 0\n",
        "            for la in range(set_train):\n",
        "                if la % size_batch == 0:\n",
        "                    for k in range(1, hls+1):\n",
        "                        update_W = gamma * W_prev[k] * int(p != 0)\n",
        "                        W[k] = W[k] - update_W\n",
        "                    for k in range(1, hls):    \n",
        "                        update_b = gamma * b_prev[k] * int(p != 0)\n",
        "                        b[k] = b[k] - update_b\n",
        "                y_predicted = forwardprop(xtr, hls, hn, inn, otn, W, b, u, v, la, activation)\n",
        "                loss_current = LOSS(y_predicted, ytr, los, la, otn)\n",
        "                loss_total += loss_current\n",
        "                Backprop(ytr, hls, hn, inn, otn, W, b, u, v, gradW, gradb, la, y_predicted, activation, los)\n",
        "                if la % size_batch == 0:\n",
        "                    for k in range(1, hls+1):\n",
        "                        update_W = gamma * W_prev[k] * int(p != 0) + learning_rate * gradW[k]\n",
        "                        W[k] -= update_W\n",
        "                        W_prev[k]=  copy.deepcopy(update_W)\n",
        "                    for k in range(1, hls):    \n",
        "                        update_b = gamma * b_prev[k] * int(p != 0) + learning_rate * gradb[k]\n",
        "                        b[k] -= update_b\n",
        "                        b_prev[k] = copy.deepcopy(update_b)\n",
        "                    gradW = {}\n",
        "                    gradb = {}\n",
        "            wandb.log({\"epoch\": p})\n",
        "            wandb.log({\"loss\": loss_total})\n",
        "            acc(xte, yte, xtr, ytr, W, b, hls, otn, hn, activation)  \n",
        "           \n",
        "\n",
        "# Different Optimization Methods i.e. Stochastic Gradient Descent Method Definition\n",
        "def stochasticgd(xtr, ytr, hls, hn, inn, otn, W, b, u, v, learning_rate, epochs, size_batch, set_train, activation, los):\n",
        "          for p in range(epochs):\n",
        "            gradW = {}\n",
        "            gradb = {}\n",
        "            loss_total = 0\n",
        "            for la in range(set_train):\n",
        "                y_predicted = forwardprop(xtr, hls, hn, inn, otn, W, b, u, v, la, activation)\n",
        "                loss_current = LOSS(y_predicted, ytr, los, la, otn)\n",
        "                loss_total = loss_total + loss_current\n",
        "                Backprop(ytr, hls, hn, inn, otn, W, b, u, v, gradW, gradb, la, y_predicted, activation, los)\n",
        "                if la % size_batch == 0:\n",
        "                    for k in range(1, hls+1):\n",
        "                        W[k] = W[k] - learning_rate * gradW[k]\n",
        "\n",
        "                    for k in range(1, hls):    \n",
        "                        b[k] = b[k] - learning_rate * gradb[k]\n",
        "                    gradW = {}\n",
        "                    gradb = {}\n",
        "            \n",
        "            wandb.log({\"loss\": loss_total})\n",
        "            wandb.log({\"epoch\": p})\n",
        "            acc(xte, yte, xtr, ytr, W, b, hls, otn, hn, activation)\n",
        "\n",
        "\n",
        "\n",
        "# Different Optimization Methods definitions i.e. Nesterov Adaptive Movement Estimation  \n",
        "def Nesterovadapt(xtr, ytr, hls, hn, inn, otn, W, b, u, v, learning_rate, epochs, size_batch, set_train, activation, los): \n",
        "        B1 =  0.9\n",
        "        Ep = 1e-8\n",
        "        B2 = 0.999\n",
        "        updatem_W = {}\n",
        "        updatem_b = {}\n",
        "        updatev_W = {}\n",
        "        updatev_b = {}\n",
        "        for p in range(epochs):\n",
        "            gradW = {}\n",
        "            gradb = {}\n",
        "            loss_total = 0\n",
        "            for la in range(set_train):\n",
        "                y_predicted = forwardprop(xtr, hls, hn, inn, otn, W, b, u, v, la, activation)\n",
        "                loss_current = LOSS(y_predicted, ytr, los, la, otn)\n",
        "                loss_total = loss_total + loss_current\n",
        "                gradW, gradb = Backprop(ytr, hls, hn, inn, otn, W, b, u, v, gradW, gradb, la, y_predicted, activation, los)\n",
        "                if la % size_batch == 0:\n",
        "                    for k in range(1, hls+1):\n",
        "                        if(p == 0):\n",
        "                            updatem_W[k] = np.zeros(gradW[k].shape, dtype=float)\n",
        "                            updatev_W[k] = np.zeros(gradW[k].shape, dtype=float)\n",
        "                            updatem_b[k] = np.zeros(gradb[k].shape, dtype=float)\n",
        "                            updatev_b[k] = np.zeros(gradb[k].shape, dtype=float)\n",
        "                    for k in range(1, hls+1):\n",
        "                        updatem_W[k] = B1 * updatem_W[k] + (1 - B1) * gradW[k]\n",
        "                        updatem_W_cap = updatem_W[k] / (1 - math.pow(B1, (la / size_batch) + 1))\n",
        "                        updatev_W[k] = B2 * updatev_W[k] + (1 - B2) * np.square(gradW[k])\n",
        "                        updatev_W_cap = updatev_W[k] / (1 - math.pow(B2,(la / size_batch) + 1))\n",
        "                        W[k] = W[k] - (learning_rate / np.sqrt(updatev_W_cap + Ep)) * (B1 * updatem_W_cap + ((1 - B1) / (1 - math.pow(B1,(la / size_batch) + 1)) * gradW[k])) \n",
        "                    for k in range(1, hls):\n",
        "                        updatem_b[k] = B1 * updatem_b[k] + (1 - B1) * gradb[k]\n",
        "                        updatem_b_cap = updatem_b[k] / (1 - math.pow(B1, (la / size_batch) + 1))\n",
        "                        updatev_b[k] = B2 * updatev_b[k] + (1 - B2) * np.square(gradb[k])\n",
        "                        updatev_b_cap = updatev_b[k] / (1 - math.pow(B2, (la / size_batch) + 1))\n",
        "                        b[k] = b[k] - (learning_rate / np.sqrt(updatev_b_cap + Ep)) * (B1 * updatem_b_cap + ((1 - B1) / (1 - math.pow(B1,(la / size_batch) + 1)) * gradb[k]))\n",
        "                    gradW = {}\n",
        "                    gradb = {}\n",
        "            \n",
        "            wandb.log({\"loss\": loss_total})\n",
        "            wandb.log({\"epoch\": p})\n",
        "            acc(xte, yte, xtr, ytr, W, b, hls, otn, hn, activation)\n",
        "\n",
        "#Different Optimization methods i.e., Adaptive Movement Estimation  \n",
        "def adaptivem(xtr, ytr, hls, hn, inn, otn, W, b, u, v, learning_rate, epochs, size_batch, set_train, activation, los): \n",
        "        updatev_W={}\n",
        "        updatem_W={}       \n",
        "        updatev_b={}\n",
        "        updatem_b={}\n",
        "        Ep  = 1e-8\n",
        "        B1  =  0.9\n",
        "        B2 =  0.999\n",
        "        for p in range(epochs):\n",
        "            gradW = {}\n",
        "            gradb = {}\n",
        "            loss_total = 0\n",
        "            for la in range(set_train):\n",
        "                y_predicted = forwardprop(xtr, hls, hn, inn, otn, W, b, u, v, la, activation)\n",
        "                loss_current = LOSS(y_predicted, ytr, los, la, otn)\n",
        "                loss_total = loss_total + loss_current\n",
        "                Backprop(ytr, hls, hn, inn, otn, W, b, u, v, gradW, gradb, la, y_predicted, activation, los)\n",
        "                if la % size_batch == 0:\n",
        "                    for k in range(1, hls+1):\n",
        "                        if(p == 0):\n",
        "                            updatem_W[k] = np.zeros(gradW[k].shape, dtype=float)\n",
        "                            updatem_b[k] = np.zeros(gradb[k].shape, dtype=float)\n",
        "                            updatev_W[k] = np.zeros(gradW[k].shape, dtype=float)\n",
        "                            updatev_b[k] = np.zeros(gradb[k].shape, dtype=float)\n",
        "                    for k in range(1, hls+1):\n",
        "                        updatem_W[k] = B1 * updatem_W[k] + (1 - B1) * gradW[k]\n",
        "                        updatem_W_cap = updatem_W[k] / (1 - math.pow(B1, (la / size_batch) + 1))\n",
        "                        updatev_W[k] = B2 * updatev_W[k] + (1 - B2) * np.square(gradW[k])\n",
        "                        updatev_W_cap = updatev_W[k] / (1 - math.pow(B2, (la / size_batch) + 1))   \n",
        "                        W[k] = W[k] - (learning_rate / np.sqrt(updatev_W_cap + Ep)) * updatem_W_cap\n",
        "\n",
        "                    for k in range(1, hls):\n",
        "                        updatem_b[k] = B1 * updatem_b[k] + (1 - B1) * gradb[k]\n",
        "                        updatem_b_cap = updatem_b[k] / (1-math.pow(B1,(la / size_batch) + 1))\n",
        "                        updatev_b[k] = B2 * updatev_b[k] +(1 - B2) * np.square(gradb[k])\n",
        "                        updatev_b_cap = updatev_b[k] / (1-math.pow(B2, (la / size_batch) + 1))\n",
        "                        b[k] = b[k] - (learning_rate / np.sqrt(updatev_b_cap + Ep)) * updatem_b_cap\n",
        "                    gradW = {}\n",
        "                    gradb = {}\n",
        "            \n",
        "            wandb.log({\"loss\": loss_total})\n",
        "            wandb.log({\"epoch\": p})\n",
        "            acc(xte, yte, xtr, ytr, W, b, hls, otn, hn, activation)     \n",
        "\n",
        "\n",
        "# complete training of the dataset\n",
        "def trning(xtr, ytr, hls, hn, inn, otn, W, b, u, v, learning_rate, epochs, size_batch, Optimization, set_train, activation, los, initialization):\n",
        "    init(W, b,inn, hls, hn, otn, initialization)\n",
        "    if Optimization == \"stochasticgd\":\n",
        "        stochasticgd(xtr, ytr, hls, hn, inn, otn, W, b, u, v, learning_rate, epochs, size_batch, set_train, activation, los)\n",
        "    elif Optimization == \"momentumbasedgd\":\n",
        "          momentumbasedgd(xtr, ytr, hls, hn, inn, otn, W, b, u, v, learning_rate, epochs, size_batch, set_train, activation, los)\n",
        "    elif Optimization == \"Nesterovadapt\":\n",
        "          Nesterovadapt(xtr, ytr, hls, hn, inn, otn, W, b, u, v, learning_rate, epochs, size_batch, set_train, activation, los)\n",
        "    elif Optimization == \"RMS\":\n",
        "          RMS(xtr, ytr, hls, hn, inn, otn, W, b, u, v, learning_rate, epochs, size_batch, set_train, activation, los)\n",
        "    elif Optimization == \"adaptivem\":\n",
        "          adaptivem(xtr, ytr, hls, hn, inn, otn, W, b, u, v, learning_rate, epochs, size_batch, set_train, activation, los)\n",
        "    elif Optimization == \"nesterovagd\":\n",
        "          Nesterovagd(xtr, ytr, hls, hn, inn, otn, W, b, u, v, learning_rate, epochs, size_batch, set_train, activation, los)\n",
        "    \n",
        "\n",
        "# Sweep config for wandb plotting\n",
        "\n",
        "sweep_config = {\n",
        "    'name'  : \"Sumit Sharma\", \n",
        "    'method': 'random', \n",
        "    'metric': {\n",
        "      'name': 'val_acc',\n",
        "      'goal': 'maximize'   \n",
        "    },\n",
        "\n",
        "    'parameters': {\n",
        "\n",
        "        'hls': {\n",
        "            'values': [3,4, 5]\n",
        "        },\n",
        "        'epochs': {\n",
        "            'values': [10, 15]\n",
        "        },\n",
        "        'hn': {\n",
        "            'values': [32, 64,128]\n",
        "        },\n",
        "        'learning_rate': {\n",
        "            'values': [1e-2,1e-3,5e-3]\n",
        "        },\n",
        "        'initialization': {\n",
        "            'values': [\"random\",\"xavier\"]\n",
        "        },\n",
        "        'size_batch': {\n",
        "            'values': [32,64]\n",
        "        },\n",
        "        'Optimization': {\n",
        "            'values': [\"momentumbasedgd\",\"stochasticgd\",\"Nesterovadapt\",\"RMS\",\"adaptivem\",\"nesterovagd\"]\n",
        "        },\n",
        "        'activation': {\n",
        "            'values': [\"sigmoid\",\"relu\",\"tanh\"]\n",
        "        }\n",
        "    }\n",
        "}\n",
        "\n",
        "sweep_id = wandb.sweep(sweep_config, project = \"Deep_learning_final1\")\n",
        "\n",
        "#Initialization for neurons and training\n",
        "otn = 10\n",
        "set_train = 54000\n",
        "inn = 784\n",
        "u = {}\n",
        "v = {}\n",
        "W = {}\n",
        "b = {}\n",
        "los = \"cross_entropy\"\n",
        "os.environ['WANDB_NOTEBOOK_NAME'] = 'Deep_learning_final1.ipynb'\n",
        "wandb.login()\n",
        "\n",
        "def execute():\n",
        "    wandb.init(config = sweep_config)\n",
        "    wandb.log({\"images\" : wandb.Image(plt)})\n",
        "    config = wandb.config\n",
        "    wandb.run.name = \"hl_{}_hn_{}_bs_{}_ac_{}_learning_rate_{}_init_{}_\".format(config.hls, config.hn, config.size_batch , config.activation ,config.learning_rate, config.initialization)\n",
        "    trning(xtr, ytr, config.hls, config.hn, inn, otn, W, b, u, v, config.learning_rate, config.epochs, config.size_batch, config.Optimization, set_train, config.activation, los, config.initialization)\n",
        "    \n",
        "wandb.agent(sweep_id, execute)\n"
      ]
    }
  ],
  "metadata": {
    "colab": {
      "provenance": []
    },
    "gpuClass": "standard",
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}