{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "Ztmkso6EMYCM"
   },
   "source": [
    "# Setting up notebook for modelling"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 228
    },
    "colab_type": "code",
    "id": "-MzdfcwC5qDr",
    "outputId": "84441e96-ccd7-40d1-8bf2-c72863d945eb"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "rm: cannot remove '../root/.kaggle': No such file or directory\n",
      "Collecting efficientnet-pytorch\n",
      "  Downloading https://files.pythonhosted.org/packages/b8/cb/0309a6e3d404862ae4bc017f89645cf150ac94c14c88ef81d215c8e52925/efficientnet_pytorch-0.6.3.tar.gz\n",
      "Requirement already satisfied: torch in /usr/local/lib/python3.6/dist-packages (from efficientnet-pytorch) (1.4.0)\n",
      "Building wheels for collected packages: efficientnet-pytorch\n",
      "  Building wheel for efficientnet-pytorch (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
      "  Created wheel for efficientnet-pytorch: filename=efficientnet_pytorch-0.6.3-cp36-none-any.whl size=12422 sha256=9003408bbb509f71e6627632b0392ae0fda42b42a374e0c686c650687917b976\n",
      "  Stored in directory: /root/.cache/pip/wheels/42/1e/a9/2a578ba9ad04e776e80bf0f70d8a7f4c29ec0718b92d8f6ccd\n",
      "Successfully built efficientnet-pytorch\n",
      "Installing collected packages: efficientnet-pytorch\n",
      "Successfully installed efficientnet-pytorch-0.6.3\n"
     ]
    }
   ],
   "source": [
    "!rm -r ../root/.kaggle\n",
    "!mkdir ../root/.kaggle\n",
    "!cp kaggle.json ../root/.kaggle\n",
    "!chmod 600 ../root/.kaggle/kaggle.json\n",
    "!rm -r sample_data\n",
    "!pip install efficientnet-pytorch"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 69
    },
    "colab_type": "code",
    "id": "n5dq2guwMUVA",
    "outputId": "91b1bd17-9d83-41f1-c8a6-f0fab229c518"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Downloading pap-smear-datasets.zip to /content\n",
      "100% 6.44G/6.44G [02:23<00:00, 46.5MB/s]\n",
      "100% 6.44G/6.44G [02:23<00:00, 48.3MB/s]\n"
     ]
    }
   ],
   "source": [
    "!kaggle datasets download umangjpatel/pap-smear-datasets --unzip\n",
    "!rm -r herlev_pap_smear\n",
    "!rm -r sipakmed_fci_pap_smear"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "xJbKpakU6S6r"
   },
   "outputs": [],
   "source": [
    "from fastai import *\n",
    "from fastai.vision import *\n",
    "from fastai.vision.models import efficientnet\n",
    "from fastai.utils.ipython import *\n",
    "from fastai.callbacks.tracker import SaveModelCallback\n",
    "from sklearn.model_selection import StratifiedKFold"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 69
    },
    "colab_type": "code",
    "id": "WtLde_oj6Uqv",
    "outputId": "13209003-8524-4fc7-c5ef-dd175f729c44"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[PosixPath('.config'),\n",
       " PosixPath('kaggle.json'),\n",
       " PosixPath('sipakmed_wsi_pap_smear')]"
      ]
     },
     "execution_count": 4,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "path = Path(\".\")\n",
    "path.ls()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 332
    },
    "colab_type": "code",
    "id": "e0L5bOab7Li0",
    "outputId": "34b65b4e-7ecd-44ad-bfc3-153a4d29b929"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LabelLists;\n",
       "\n",
       "Train: LabelList (966 items)\n",
       "x: ImageList\n",
       "Image (3, 1536, 2048),Image (3, 1536, 2048),Image (3, 1536, 2048),Image (3, 1536, 2048),Image (3, 1536, 2048)\n",
       "y: CategoryList\n",
       "abnormal_Dyskeratotic,abnormal_Dyskeratotic,abnormal_Dyskeratotic,abnormal_Dyskeratotic,abnormal_Dyskeratotic\n",
       "Path: sipakmed_wsi_pap_smear;\n",
       "\n",
       "Valid: LabelList (0 items)\n",
       "x: ImageList\n",
       "\n",
       "y: CategoryList\n",
       "\n",
       "Path: sipakmed_wsi_pap_smear;\n",
       "\n",
       "Test: None"
      ]
     },
     "execution_count": 5,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_init = (ImageList.from_folder(path/\"sipakmed_wsi_pap_smear\")\n",
    "            .split_none()\n",
    "            .label_from_folder())\n",
    "data_init"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "qg07Zti59Sg5",
    "outputId": "e0896d6b-babe-4e03-9994-06fd3ca6c6c7"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "StratifiedKFold(n_splits=3, random_state=0, shuffle=True)"
      ]
     },
     "execution_count": 6,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "skf = StratifiedKFold(n_splits=3, shuffle=True, random_state=0)\n",
    "skf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "-r38w3l8ExI1"
   },
   "outputs": [],
   "source": [
    "!mkdir sipakmed_wsi_pap_smear/models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "Btupbe9q9V2M"
   },
   "outputs": [],
   "source": [
    "def model_callback(model, model_name):\n",
    "  return [SaveModelCallback(model, every=\"improvement\", monitor=\"accuracy\", name=model_name)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "v5Xkj0BSxLkz"
   },
   "outputs": [],
   "source": [
    "tfms = get_transforms(flip_vert=True, max_warp=0.0, max_rotate=60.0, max_zoom=1.0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 104
    },
    "colab_type": "code",
    "id": "4ec4oK-ongtj",
    "outputId": "6b694e57-239f-4d3f-9426-711eeb948311"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<function fastai.metrics.accuracy>,\n",
       " Precision(average='macro', pos_label=1, eps=1e-09),\n",
       " Recall(average='macro', pos_label=1, eps=1e-09),\n",
       " FBeta(average='macro', pos_label=1, eps=1e-09, beta=2),\n",
       " KappaScore(weights='quadratic')]"
      ]
     },
     "execution_count": 10,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "our_metrics = [accuracy, Precision(average=\"macro\"), Recall(average=\"macro\"), FBeta(average=\"macro\"), KappaScore(weights=\"quadratic\")]\n",
    "our_metrics"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "3BrARh-QmS2K"
   },
   "source": [
    "# K-Fold Training Process\n",
    "* Loop through fold indices\n",
    "* Load the fold data\n",
    "* Initialize the learner with the fold data (with freezing)\n",
    "* Train the learner\n",
    "* Fine-tune the learner\n",
    "* Repeat for different folds"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "FSqIh31zlw7m"
   },
   "outputs": [],
   "source": [
    "idxs = [[train_idx, val_idx] for train_idx, val_idx in skf.split(data_init.x.items, data_init.y.items)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "pBXCI83JNCdo"
   },
   "outputs": [],
   "source": [
    "fold1_idxs = idxs[0]\n",
    "fold_data = (ImageList.from_folder(path/\"sipakmed_wsi_pap_smear\")\n",
    "            .split_by_idxs(fold1_idxs[0], fold1_idxs[1])\n",
    "            .label_from_folder()\n",
    "            .transform(tfms, size=224)\n",
    "            .databunch(bs=16)\n",
    "            .normalize(imagenet_stats))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 121,
     "referenced_widgets": [
      "e5cdf98a8ef34dec9adc0fff5423a644",
      "2e23ad6a5dff415993e7b191ef1c11d6",
      "ea33c3bdffb347639b83e5639cb92046",
      "6f46df8e2309479989eabc39333473e3",
      "3665cf73253343f2a1c1479c1defc48f",
      "0b9017694ad44490b3867fffa5be15e7",
      "714202b45d904c27b5491f587474abbf",
      "04d841481b61417aab864a1a9c8872af"
     ]
    },
    "colab_type": "code",
    "id": "pQp0iKJ-NCg8",
    "outputId": "5adddf49-51c2-410c-b596-bcb79cc64181"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Downloading: \"https://github.com/lukemelas/EfficientNet-PyTorch/releases/download/1.0/efficientnet-b3-5fb5a3c3.pth\" to /root/.cache/torch/checkpoints/efficientnet-b3-5fb5a3c3.pth\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "e5cdf98a8ef34dec9adc0fff5423a644",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=49388949), HTML(value='')))"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Loaded pretrained weights for efficientnet-b3\n"
     ]
    }
   ],
   "source": [
    "learner = Learner(fold_data, efficientnet.EfficientNetB3(fold_data), metrics=our_metrics).to_fp16()\n",
    "learner = learner.split([learner.model._conv_stem, learner.model._blocks, learner.model._conv_head])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 458
    },
    "colab_type": "code",
    "id": "v6Fkw6SYPDAw",
    "outputId": "91069596-7d60-43cb-e6a7-162976b94318"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "    <div>\n",
       "        <style>\n",
       "            /* Turns off some styling */\n",
       "            progress {\n",
       "                /* gets rid of default border in Firefox and Opera. */\n",
       "                border: none;\n",
       "                /* Needs to be in here for Safari polyfill so background images work as expected. */\n",
       "                background-size: auto;\n",
       "            }\n",
       "            .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
       "                background: #F44336;\n",
       "            }\n",
       "        </style>\n",
       "      <progress value='2' class='' max='3', style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
       "      66.67% [2/3 00:39<00:19]\n",
       "    </div>\n",
       "    \n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>accuracy</th>\n",
       "      <th>precision</th>\n",
       "      <th>recall</th>\n",
       "      <th>f_beta</th>\n",
       "      <th>kappa_score</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>1.625695</td>\n",
       "      <td>#na#</td>\n",
       "      <td>00:20</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>4.709448</td>\n",
       "      <td>#na#</td>\n",
       "      <td>00:19</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table><p>\n",
       "\n",
       "    <div>\n",
       "        <style>\n",
       "            /* Turns off some styling */\n",
       "            progress {\n",
       "                /* gets rid of default border in Firefox and Opera. */\n",
       "                border: none;\n",
       "                /* Needs to be in here for Safari polyfill so background images work as expected. */\n",
       "                background-size: auto;\n",
       "            }\n",
       "            .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
       "                background: #F44336;\n",
       "            }\n",
       "        </style>\n",
       "      <progress value='2' class='' max='40', style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
       "      5.00% [2/40 00:02<00:53 5.1559]\n",
       "    </div>\n",
       "    "
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "LR Finder is complete, type {learner_name}.recorder.plot() to see the graph.\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYIAAAEGCAYAAABo25JHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjMsIGh0\ndHA6Ly9tYXRwbG90bGliLm9yZy+AADFEAAAgAElEQVR4nO3deXRcd3338fd3RqPVWixb3hclzh7H\niWMnZGEJIdBAKUshlJzSJl0enkJDCgT6FPocWtJCS9lKyGlpWihLoSQQ6BNCaOKSBAjBCbZjy4nt\nBCexY8myLS8zkqyRNJr5Pn/MlaIIyZZt3bkj3c/rnDmZuXPnzveXseYzv9+993fN3RERkfhKRF2A\niIhES0EgIhJzCgIRkZhTEIiIxJyCQEQk5iqiLuBEzZ0711tbW6MuQ0RkWtm4ceNBd28Z77lpFwSt\nra1s2LAh6jJERKYVM9s90XMaGhIRiTkFgYhIzCkIRERiTkEgIhJzCgIRkZhTEIiIxJyCQEQk5hQE\nIiLTwBf+51f8fOfBULatIBARKXO9A0P844+f4Ze7DoeyfQWBiEiZe6ojgzusWtIYyvYVBCIiZW5r\nRwaACxY3hbJ9BYGISJlra8+wqLGalvqqULavIBARKXNt7WkuCGlYCBQEIiJlLdOXY9ehPlYtCWdY\nCBQEIiJl7cm9w/sH1CMQEYmltnYFgYhIrG3tSLO0uYbZdZWhvYeCQESkjLW1Z0LdPwAKAhGRsnX4\n6CDtR7KsCnFYCBQEIiJla+REshAPHQUFgYhI2WrbkwZgpXoEIiLx1NaR4fS5dTRUp0J9HwWBiEiZ\n2tqeCX1YCEIMAjOrNrPHzWyLmT1lZh8fZ50bzazLzDYHtz8Oqx4RkenkQHc/+7r7Qz9iCKAixG0P\nAFe7e6+ZpYBHzOxH7r5+zHp3uvtNIdYhIjLtDO8oDmvq6dFCCwJ3d6A3eJgKbh7W+4mIzCRt7RkS\nBuctbAj9vULdR2BmSTPbDBwA1rn7Y+Os9jYzazOz75rZ0jDrERGZLrZ2ZDhj3izqqsIcuCkKNQjc\nPe/uFwFLgEvNbOWYVX4AtLr7KmAd8LXxtmNm7zazDWa2oaurK8ySRUQi5+60tWdCuxDNWCU5asjd\n08BDwLVjlh9y94Hg4b8BayZ4/R3uvtbd17a0tIRbrIhIxDoz/RzsHSjJ/gEI96ihFjNrCu7XAK8F\ndoxZZ+Goh28CtodVj4jIdDE842ipgiDMwaeFwNfMLEkxcO5y93vN7FZgg7vfA9xsZm8ChoDDwI0h\n1iMiMi3s2NeNGZxbgh3FEO5RQ23A6nGWf2zU/Y8AHwmrBhGR6ejw0UEaa1JUp5IleT+dWSwiUmbS\nfTmaasKdVmI0BYGISJlJZ3M01oZ3IZqxFAQiImUmk83RqB6BiEh8ZfoGNTQkIhJn6hGIiMRYoeBk\nsjmaahUEIiKx1DMwRMFRj0BEJK66szlAQSAiElvpvmIQNOnwURGReEpnBwH1CEREYiuTHe4RKAhE\nRGJpZGhIPQIRkXga7hE0KAhEROIpk81RnUqUbOZRUBCIiJSVdN9gSXcUg4JARKSsZLI5mmpKd+go\nKAhERMpKui9HYwmPGAIFgYhIWSn1hHOgIBARKSvFoSEFgYhIbKX71CMQEYmtgaE82Vy+pGcVg4JA\nRKRsDJ9MVsrrFYOCQESkbGT6Sj8FNSgIRETKxsiEcwoCEZF4SqtHICISb1FMQQ0KAhGRspEeGRrS\nzmIRkVjK9A1iBvXVFSV9XwWBiEiZyGRzNFSnSCSspO+rIBARKRPpCOYZAgWBiEjZSPflSr6jGBQE\nIiJlI4qZR0FBICJSNhQEIiIxl8nOsKEhM6s2s8fNbIuZPWVmHx9nnSozu9PMdprZY2bWGlY9IiLl\nrFBw0n2DJT+HAMLtEQwAV7v7hcBFwLVmdtmYdf4IOOLuZwCfBz4VYj0iImWrd3CIgpd+egkIMQi8\nqDd4mApuPma1NwNfC+5/F3iNmZX2AFoRkTIwMvPoTBoaAjCzpJltBg4A69z9sTGrLAb2ALj7EJAB\n5oyznXeb2QYz29DV1RVmySIikRi5FsFM6hEAuHve3S8ClgCXmtnKk9zOHe6+1t3XtrS0TG2RIiJl\nIKopqKFERw25exp4CLh2zFMdwFIAM6sAGoFDpahJRKScDE9B3VTiq5NBuEcNtZhZU3C/BngtsGPM\navcANwT33w486O5j9yOIiMx46ewgEM3QUJhT3C0EvmZmSYqBc5e732tmtwIb3P0e4MvAN8xsJ3AY\neGeI9YiIlK2orkUAIQaBu7cBq8dZ/rFR9/uB68KqQURkusj05aisSFCdSpb8vXVmsYhIGUj35SLZ\nUQwKAhGRshDV9BKgIBARKQvp7GAkO4pBQSAiUhYy2SEaI5hnCBQEIiJlIdOnHoGISKyltY9ARCS+\nBocK9A3mddSQiEhcjUw4px6BiEg8ZSKcXgIUBCIikYtyCmpQEIiIRC7KmUdBQSAiErkor0UACgIR\nkcgN9wg0NCQiElPpoEfQoCAQEYmn7myO+uoKkgmL5P0VBCIiEUv3DUZ2VjEoCEREIpfJ5miKaMI5\nUBCIiEQunc1FtqMYFAQiIpHL9OUim14CFAQiIpHLqEcgIhJf7l6cglpBICISTz0DQ+QLrqOGRETi\nal+mH4AFjTWR1aAgEBGJUEc6C8CixurIaphUEJjZCjOrCu5fZWY3m1lTuKWJiMx8nelij2BRU/n3\nCO4G8mZ2BnAHsBT4VmhViYjExN50lmTCmFdfFVkNkw2CgrsPAW8FvujuHwYWhleWiEg87E1nmV9f\nRUUyupH6yb5zzsyuB24A7g2WRbeLW0RkhuhIZyMdFoLJB8EfAJcDn3D3583sNOAb4ZUlIhIPnZn+\nyIOgYjIrufs24GYAM5sN1Lv7p8IsTERkpisUnM5MltdfsCDSOiZ71NDDZtZgZs3AJuBfzexz4ZYm\nIjKzHewdIJd3Fk+ToaFGd+8Gfhv4uru/DLgmvLJERGa+vcHJZIsiPJkMJh8EFWa2EHgHL+4sFhGR\nU7B3+GSyadIjuBW4H3jW3X9pZqcDvwqvLBGRme/FIIjurGKYZBC4+3fcfZW7vyd4/Jy7v+1YrzGz\npWb2kJltM7OnzOzPxlnnKjPLmNnm4Paxk2uGiMj0szfdT21lMtIpqGGSRw2Z2RLgi8CVwaKfAX/m\n7u3HeNkQcIu7bzKzemCjma0LjkAa7Wfu/sYTLVxEZLrbG5xDYBbNReuHTXZo6N+Be4BFwe0HwbIJ\nuXunu28K7vcA24HFJ1+qiMjMsjeTZWGEk80Nm2wQtLj7v7v7UHD7KtAy2Tcxs1ZgNfDYOE9fbmZb\nzOxHZnb+BK9/t5ltMLMNXV1dk31bEZGytjfdH/mhozD5IDhkZu8ys2RwexdwaDIvNLNZFCete39w\nCOpom4Dl7n4hxaGn/xpvG+5+h7uvdfe1LS2Tzh8RkbLVn8tzsHcg8iOGYPJB8IcUDx3dB3QCbwdu\nPN6LzCxFMQS+6e7fG/u8u3e7e29w/z4gZWZzJ1mTiMi0NXxBmmkzNOTuu939Te7e4u7z3P0twPGO\nGjLgy8B2dx/3LGQzWxCsh5ldGtQzqZ6GiMh0tjdTPHS0HIaGJnXU0AQ+CPzjMZ6/Evg9YKuZbQ6W\nfRRYBuDuX6LYs3iPmQ0BWeCd7u6nUJOIyLSwtwwuSDPsVILgmMc7ufsjk1jnduD2U6hBRGRaGj6Z\nbMF0GRqagH65i4icpM5MlrmzKqlOJaMu5dg9AjPrYfwvfAOi78+IiExTHenor0Mw7JhB4O71pSpE\nRCRO9qaznNEyK+oygFMbGhIRkZPg7nSmsyyMeLK5YQoCEZES684OcXQwXxaHjoKCQESk5DrK5DoE\nwxQEIiIl1hmcTFYOZxWDgkBEpOSGzyHQ0JCISEx1pPtJJY25s6qiLgVQEIiIlFxnJsuCxmoSiWgv\nSDNMQSAiUmJ701kWNZbHsBAoCERESq5cLkgzTEEgIlJC+YKzr7t8ppcABYGISEkd6OknX/CyOasY\nFAQiIiW1t8xOJgMFgYhISXUEF6TRPgIRkZjqTJfXWcWgIBARKamOdJb6qgrqq1NRlzJCQSAiUkLP\n7O9hxbzyuA7BMAWBiEiJuDvbO3s4d2FD1KW8hIJARKREOjP9ZLI5zltYXhd/VBCIiJTI9s5uAPUI\nRETiajgIzlEQiIjE0/bOHpY11zKrqiLqUl5CQSAiUiLbO7s5t8z2D4CCQESkJPoGh3j+0FHOWVBe\nw0KgIBARKYmn9/XgXn47ikFBICJSEts7ewA4T0EgIhJPO/Z1M6uqgiWzy2eyuWEKAhGREtje2c05\nC+rL5jrFoykIRERC5u7sKMOpJYYpCEREQtZ+JEvPwJCCQEQkrraNTC1RfucQQIhBYGZLzewhM9tm\nZk+Z2Z+Ns46Z2W1mttPM2szs4rDqERGJyvbObszg7AXlGQRhnuc8BNzi7pvMrB7YaGbr3H3bqHVe\nD5wZ3F4G/HPwXxGRGWN7ZzenzamjtrK8ppYYFlqPwN073X1TcL8H2A4sHrPam4Gve9F6oMnMFoZV\nk4hIFMrxGgSjlWQfgZm1AquBx8Y8tRjYM+pxO78eFpjZu81sg5lt6OrqCqtMEZEp19Of44XDfWW7\nfwBKEARmNgu4G3i/u3efzDbc/Q53X+vua1taWqa2QBGRED29r3hGcWx7BGaWohgC33T3742zSgew\ndNTjJcEyEZEZYXucg8DMDPgysN3dPzfBavcAvx8cPXQZkHH3zrBqEhEpte2d3TRUV7CwsTrqUiYU\n5i7sK4HfA7aa2eZg2UeBZQDu/iXgPuANwE6gD/iDEOsRESm54jUIGij+Ni5PoQWBuz8CHLPl7u7A\nn4ZVg4hIlAoF5+l9Pbxj7dLjrxwhnVksIhKSJ/YcoW8wz/mLynf/ACgIRERC8/l1v2JOXSWvv6C8\nT49SEIiIhODRZw/yyM6DvOeqFWV3sfqxFAQiIlPM3fnM/U+zoKGad122POpyjktBICIyxR7ccYBN\nL6S5+TVnUp1KRl3OcSkIRESmUKHgfOaBZ1g+p5br1i6JupxJURCIiEyh+57sZHtnNx+45ixSyenx\nFTs9qhQRmQaG8gU+98AznDV/Fr914aKoy5k0BYGIyBT53qYOnjt4lFtedzbJMrxI/UQUBCIiU2Bv\nOssnf7Sd1cuaeN1586Mu54QoCERETtFQvsD7v72Z3FCBz73jorKeV2g85X2Wg4jINHDbgzt5fNdh\n/vF3LuK0uXVRl3PC1CMQETkFjz57kC8++CvevmYJb1n9axdYnBYUBCIiJ+lQ7wAfuHMzp82t4+Nv\nOj/qck6agkBE5CT05/J86DtbONKX44vXr6auzOcTOpbpW7mISIm5Oxt2H+Huje38sK2TnoEhbn3z\n+Zy/qDHq0k6JgkBE5DgKBecb63fzlZ8/z+5DfdRWJrl25QKuW7OUy1fMibq8U6YgEBE5hgPd/dzy\nnS387FcHubS1mZuvPpNrVy6Y1kNBY82cloiITLH/2bafP7+7jb7BIT751gu4/tKl0+4cgclQEIiI\njDEwlOcTP9zO13+xm/MWNnDb9Rdxxrz6qMsKjYJARGSUXL7ATd96gnXb9vPHLz+ND197NlUV5X9N\ngVOhIBARCeQLzi13bWHdtv18/E3nc8MVrVGXVBI6j0BEhOKRQR/93lbu2bKX/3PtObEJAVAQiIjg\n7tx67zbu3LCHm68+g/dctSLqkkpKQ0NTLF9wDvUOcKBngP5cnlzeyeUL5PIFEmbMrqukubaS2XUp\nZlVVTOoIhHzB6c/lGRgq0J/LMzhUwEc97+4U3BkcevG9BvMF8gVnqODk807enYQZdZVJaiqT1FZW\nUJ1KcHQgTyabo7s/R3c2R99gfuT1Q0Htg0MFBoJb8b2dmlSSmlSS2sok1ZVJmmoqaa5L0VRbSXNd\nJY01KaorklSlElRVJGbkkRYyc3z2gWf46qO7+KOXn8YHXntW1OWUXGyCoHdgiH2ZLH2DebKDebK5\nPP25PGZGfXUFDdUp6qsrqK2soD+Xp28wz9HBIY4ODJHJ5jjYM8DB3kEO9g5w6OggueCLdvjWN5hn\nf3c/B3sHKPjx6wFIJY3K4FJ2ZsbwV2Xei9t0f/F+VMygMln8Mq+sSFJVUax3YKj4/yiby+OTKK+q\nIjFyoY7h9c2Ky6tTSapTxW2nkgkSCSNhkDQjYcZAvsBAEIADQ4WR7VVWJEZeU3An78VQzBecZMJG\ngqoYekkSxksCtCJR/Ozrg8++oTpF86xK5tVX0VJfxZy6qml1cRE5OXf89Fluf2gn11+6lP/7m+fG\n8kdLbILg4acPcNO3njilbSQMmuuqmFNXSVWq+MWWNCOZMObMquS8hQ3Ma6ga+SKprawglUxQWWFU\nJBLk3Un3DXL4aI7DRwc4fDTHUL746374y9HxkW0mgu0Pf+ENf1lWViRIjPnHmkzYyHulkgkqEglS\nyeJ2hm/DgZUdfPFLfFZVkobqFA01KRprUtRWJklVJKhMFr9gj/dF6O705wpksjkOHx0stq9vkEw2\nR3+uMNKTGcjlKYxKDDOjUPCRXk5/8N+hfOElX+gFdxorU1TVV420HWAw6J0MBj2WhBlmjHwmQwUn\nm8tzsHeQvsE++nMFPHj/4T/0wXyBnv5inRN/3pU01VbSVJOiqbb4/6m2MkllMkmqwqhKJqiprKCl\nvvi5z2uoYn59NU21qVh+oUw33378BT553w7euGohf/uWC2L7mcUmCFYvm80Xr19dHNKoLP4CrUkl\nKbjT3Z+jp3+Inv4h+gaHqE4lqausoLaq+N+Gmgrmzqpidm2lfiGOYWbUBMNNCxqroy7npOTyBXr6\nh+jO5jh0dICunuLQXlfPAAd7B8hkc6T7cnRm+tne2TMyPDcQhNB4alJJljbXsHR2LUuba1k+p5aV\nixs5f1EDtZWx+bMraz9s6+Qj39/Kq85q4XPvuCjWf9ux+Re5uKmGxU01UZchZSiVTNBcV9y30XqC\nFxVxL/Y8hsPjQPcA+7v76Uhn2XO4jxcO97H+uUMcHcwDxV7GWfPrWbWkkbPm17OoqSa4VTO3ropE\njL+MSuknz3Tx/jufYM2y2XzpXWtGeppxFZsgEAmDmVFbWcHyORUsnzN+iLg7XT0DtLVnaGtPs6U9\nw7pt+7lrQ/tL1qusSHDanDpWzKtjRcssVrTM4qz59Zw1fxYVyXh/UZ2Kre0ZtnVmXtzH1zvIum37\nOWNePV++8RJqKmf2yWKToSAQCZmZMa+hmmvOq+aa4KLm7k53doiOdJa96Sx7M1naj2R5rquXHZ09\n3P/U/pGDBKpTCVYuauTCpU1cuLSJy05vZl799ByGK6XBoQKffeBp/uWnz40sm1VVwZxZlVy+Yg6f\netsqGmtSEVZYPhQEIhEwMxprUzTWpjhvUcOvPT84VGD3oaNs6+xmy54MW9rT/Mf63Xz5kecBOGdB\nPa88q4VXnDmXS1qbqU7pV+1oLxzq433ffoIte9L87suW8SevWkFLfZX+P03AfDLH/pWRtWvX+oYN\nG6IuQ6TkcvkC2zu7eWTnQX72zEE27j7CYL5AZUWCNctmc8WKOVy+Yg4XLm0iFeOhpB+2dfIXd7eB\nwafetoo3XLAw6pLKgpltdPe14z6nIBCZnvoGh3jsucP8fOdBfvHcIbZ1duMOtZVJrlgxl9ecO4+r\nz5nH/IaZPYx05Oggjz1/mPXPHeKx5w+zvbOb1cuauO2dq1naXBt1eWUjkiAws68AbwQOuPvKcZ6/\nCvh/wPPBou+5+63H266CQGR8xS/EQzyy8yAP7eiiI50FYOXiBq45dz5vXLVwRkylPJQv8MSeNA/u\nOMDDT3exvbMbKO5LWbN8Nq8+ex43XNEa617ReKIKglcCvcDXjxEEH3L3N57IdhUEIsfn7jyzv5cf\n79jPj7cfYNMLR3Av7lv4rQsX8VurFrFszvT5tZzLF3hwxwF+2NbJT57pIpPNkUwYa5fP5hVnzuWy\n0+ewaklT7A8DPZZjBUFoO4vd/adm1hrW9kVkYmbG2QvqOXtBPe+96gz2d/dz39ZOfrBlL5++/2k+\nff/TXHZ6M39w5Wlcc+78sj2Z6tmuXu765R7u3tTBwd4B5tRVcs2587n6nHm8/My5OupnioS6jyAI\ngnuP0SO4G2gH9lLsHTw1wXbeDbwbYNmyZWt2794dUsUiM9+ew33cs2Uv31y/m72ZfpY213DD5a1c\nt3Zp5F+s/bk8m144wmPPHeZnv+pi0wtpkgnj6nPm8Ttrl3LV2S06p+IkRbaz+DhB0AAU3L3XzN4A\nfMHdzzzeNjU0JDI1hvIFHti2n6/+fBeP7zpMbWWS3754MTdc3sqZ88Pbl5DJ5vjM/U9zuG+QhBUn\nGEyY0ZHOsvmFNIP5AgmD8xc18purFvLbFy/WeRNToCyDYJx1dwFr3f3gsdZTEIhMvSc7Mnz10V3c\ns2Uvg0MFrlgxhxuuaA1l2OiWu7bwX5s7WD6nFvfhadRhdm2Kl50+h8tOb2ZtazMN1Rr2mUqR7CM4\nHjNbAOx3dzezSyleJOdQVPWIxNnKxY185roL+cjrz+HODXv4j1/s5n9/YyOtc2p571Vn8JbVi6dk\nR+yDO/Zz96Z2bnr1GXzoN86egsplKoR51NB/AlcBc4H9wF8BKQB3/5KZ3QS8BxgCssAH3f3R421X\nPQKR8A0PG/3Twzt5sqObxU01/MmrTue6tUtP+uzcTDbH6z7/E5pqKrnnfVfO+AvClxudUCYiJ8Xd\nefiZLm5/cCcbdx+hua6Sy05v5uJls1nb2sx5Cxsm3VP40He28P0nOvj+e69g1ZKmkCuXscpyaEhE\nyp+Z8eqz53HVWS2sf+4wd/7yBTbsPsJ9W/cBxSvFrVrSyIVLihPiXbS0iSWza37tAi8P7TjAdze2\n86evXqEQKEPqEYjICdvf3c/G3UfYsOsIW9rTPNmRGbmMaHNdJWuWz+aS1tlc0trMsuZafvO2R2io\nqeAH73u5hoQioh6BiEyp+Q3VvOGChSMTuuXyBZ7e18OW9jSbdqfZuPsw67btB4rXpk6Yccfvr1EI\nlCkFgYicslQywcrFjaxc3Mjvvmw5AAd6+tmw6wi/3HWYcxc0aEiojCkIRCQU8+pf2muQ8qVztUVE\nYk5BICIScwoCEZGYUxCIiMScgkBEJOYUBCIiMacgEBGJOQWBiEjMTbu5hsysCxh9rcpGIDPOqmOX\nH+vxRPfnAse8UM4kTFTfiaw33nOTWaY2lk/7Jnr+eMsm095yaWNc/xbHW16ObVzu7i3jPuPu0/oG\n3DGZ5cd6fIz7G8Kq70TWG++5ySxTG8unfSfbxsm0t1zaGNe/xenYxrG3mTA09INJLj/W44nuT4XJ\nbu9Y64333GSWqY1TYyraN9Hzx1s22faeqrA+w/GWz7R/p+MtL/c2vsS0GxoqJTPb4BNM2zpTzPQ2\nzvT2gdo4U0TZxpnQIwjTHVEXUAIzvY0zvX2gNs4UkbVRPQIRkZhTj0BEJOYUBCIiMRebIDCzr5jZ\nATN78iReu8bMtprZTjO7zUZdmdvM3mdmO8zsKTP7h6mt+oRqnPL2mdlfm1mHmW0Obm+Y+spPqM5Q\nPsPg+VvMzM1s7tRVfOJC+hz/xszags/wATNbNPWVn1CdYbTx08HfYZuZfd/MIrscWkjtuy74jimY\n2dTvUD7V41anyw14JXAx8ORJvPZx4DLAgB8Brw+Wvxr4H6AqeDxvhrXvr4EPRf3ZhdnG4LmlwP0U\nT1ScO9PaCDSMWudm4EszsI2vAyqC+58CPjXD2ncucDbwMLB2qmuOTY/A3X8KHB69zMxWmNl/m9lG\nM/uZmZ0z9nVmtpDiH9J6L34iXwfeEjz9HuDv3X0geI8D4bZiYiG1r6yE2MbPA38ORH7kRBhtdPfu\nUavWEXE7Q2rjA+4+FKy6HlgSbismFlL7trv702HVHJsgmMAdwPvcfQ3wIeCfxllnMdA+6nF7sAzg\nLOAVZvaYmf3EzC4JtdoTd6rtA7gp6G5/xcxmh1fqSTulNprZm4EOd98SdqGn4JQ/RzP7hJntAX4X\n+FiItZ6sqfi3OuwPKf6aLidT2b4pF9uL15vZLOAK4DujhourTnAzFUAzxa7cJcBdZnZ6kOaRmqL2\n/TPwNxR/Qf4N8FmKf2Rl4VTbaGa1wEcpDiuUpSn6HHH3vwT+0sw+AtwE/NWUFXmKpqqNwbb+EhgC\nvjk11Z26qWxfWGIbBBR7Q2l3v2j0QjNLAhuDh/dQ/DIc3c1cAnQE99uB7wVf/I+bWYHixFFdYRY+\nSafcPnffP+p1/wrcG2bBJ+FU27gCOA3YEvyBLgE2mdml7r4v5Nonayr+nY72TeA+yigImKI2mtmN\nwBuB15TDj7FRpvoznHpR7VCJ4ga0MmoHDvAocF1w34ALJ3jd2B04bwiW/wlwa3D/LGAPwUl6M6R9\nC0et8wHg2zPtMxyzzi4i3lkc0ud45qh13gd8dwa28VpgG9ASddvCaN+o5x8mhJ3Fkf8PK+EH859A\nJ5Cj+Ev+jyj+GvxvYEvwj+hjE7x2LfAk8Cxw+/CXPVAJ/Efw3Cbg6hnWvm8AW4E2ir9YFpaqPaVq\n45h1Ig+CkD7Hu4PlbRQnJVs8A9u4k+IPsc3BLbIjo0Jq31uDbQ0A+4H7p7JmTTEhIhJzcT9qSEQk\n9hQEIiIxpyAQEYk5BYGISMwpCEREYk5BIDOCmfWW+P3+zczOm6Jt5YOZQZ80sx8cb+ZMM2sys/dO\nxXuLgK5QJjOEmfW6+6wp3F6FvziJWahG125mXwOecfdPHGP9VuBed19Zivpk5lOPQGYsM2sxs7vN\n7JfB7cpg+aVm9gsze8LMHjWzs4PlN5rZPWb2IPBjM7vKzB42s+8Gc91/c9T88A8PzwtvZr3BpG5b\nzGy9mc0Plq8IHm81s7+dZK/lF7w4Id4sM/uxmW0KtvHmYJ2/B1YEvYhPB+t+OGhjm5l9fAr/N0oM\nKAhkJvsC8Hl3vwR4G/BvwfIdwCvcfTXFmTg/Oeo1FwNvd/dXBY9XA+8HzgNOB64c533qgPXufiHw\nU+B/jXr/L7j7Bbx0VslxBXPPvIbiWdwA/cBb3f1iite++GwQRH8BPOvuF7n7h83sdcCZwKXARcAa\nM3vl8d5PZFicJ52Tme8a4FfgTKYAAAGySURBVLxRMz42BDNBNgJfM7MzKc6smhr1mnXuPnou+cfd\nvR3AzDZTnEPmkTHvM8iLE/JtBF4b3L+cF6978C3gMxPUWRNsezGwHVgXLDfgk8GXeiF4fv44r39d\ncHsieDyLYjD8dIL3E3kJBYHMZAngMnfvH73QzG4HHnL3twbj7Q+PevromG0MjLqfZ/y/mZy/uLNt\nonWOJevuFwXTYt8P/ClwG8VrB7QAa9w9Z2a7gOpxXm/A37n7v5zg+4oAGhqSme0BirNtAmBmw9MA\nN/Li9L43hvj+6ykOSQG883gru3sfxUtJ3mJmFRTrPBCEwKuB5cGqPUD9qJfeD/xh0NvBzBab2bwp\naoPEgIJAZopaM2sfdfsgxS/VtcEO1G0Upw0H+Afg78zsCcLtFb8f+KCZtQFnAJnjvcDdn6A4S+j1\nFK8dsNbMtgK/T3HfBu5+CPh5cLjpp939AYpDT78I1v0uLw0KkWPS4aMiIQmGerLu7mb2TuB6d3/z\n8V4nUmraRyASnjXA7cGRPmnK6DKfIqOpRyAiEnPaRyAiEnMKAhGRmFMQiIjEnIJARCTmFAQiIjH3\n/wFFkQjLr/eoVAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "learner.lr_find()\n",
    "learner.recorder.plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "colab_type": "code",
    "id": "JCDxaeVaPGY_",
    "outputId": "c56374d5-1c34-4e96-a95e-587c88ce015b"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>accuracy</th>\n",
       "      <th>precision</th>\n",
       "      <th>recall</th>\n",
       "      <th>f_beta</th>\n",
       "      <th>kappa_score</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>1.529235</td>\n",
       "      <td>1.560008</td>\n",
       "      <td>0.332298</td>\n",
       "      <td>0.387142</td>\n",
       "      <td>0.350793</td>\n",
       "      <td>0.319349</td>\n",
       "      <td>0.106924</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>1.345567</td>\n",
       "      <td>1.251083</td>\n",
       "      <td>0.605590</td>\n",
       "      <td>0.625240</td>\n",
       "      <td>0.644418</td>\n",
       "      <td>0.621799</td>\n",
       "      <td>0.415529</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>1.063855</td>\n",
       "      <td>0.759909</td>\n",
       "      <td>0.736025</td>\n",
       "      <td>0.745251</td>\n",
       "      <td>0.774227</td>\n",
       "      <td>0.754835</td>\n",
       "      <td>0.649526</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.754724</td>\n",
       "      <td>0.432532</td>\n",
       "      <td>0.841615</td>\n",
       "      <td>0.846137</td>\n",
       "      <td>0.864741</td>\n",
       "      <td>0.857862</td>\n",
       "      <td>0.813568</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.499305</td>\n",
       "      <td>0.350550</td>\n",
       "      <td>0.875776</td>\n",
       "      <td>0.891220</td>\n",
       "      <td>0.895502</td>\n",
       "      <td>0.892029</td>\n",
       "      <td>0.858935</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.372624</td>\n",
       "      <td>0.337715</td>\n",
       "      <td>0.881988</td>\n",
       "      <td>0.898450</td>\n",
       "      <td>0.899669</td>\n",
       "      <td>0.897659</td>\n",
       "      <td>0.912594</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.292538</td>\n",
       "      <td>0.362402</td>\n",
       "      <td>0.885093</td>\n",
       "      <td>0.887552</td>\n",
       "      <td>0.903978</td>\n",
       "      <td>0.899129</td>\n",
       "      <td>0.883105</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>0.276697</td>\n",
       "      <td>0.336655</td>\n",
       "      <td>0.916149</td>\n",
       "      <td>0.919764</td>\n",
       "      <td>0.929295</td>\n",
       "      <td>0.927108</td>\n",
       "      <td>0.917621</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>0.259163</td>\n",
       "      <td>0.405612</td>\n",
       "      <td>0.897516</td>\n",
       "      <td>0.915889</td>\n",
       "      <td>0.920945</td>\n",
       "      <td>0.918142</td>\n",
       "      <td>0.942077</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>0.215705</td>\n",
       "      <td>0.261663</td>\n",
       "      <td>0.925466</td>\n",
       "      <td>0.930461</td>\n",
       "      <td>0.938152</td>\n",
       "      <td>0.935876</td>\n",
       "      <td>0.940589</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>0.185297</td>\n",
       "      <td>0.383690</td>\n",
       "      <td>0.909938</td>\n",
       "      <td>0.916674</td>\n",
       "      <td>0.929446</td>\n",
       "      <td>0.925554</td>\n",
       "      <td>0.926687</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>0.174605</td>\n",
       "      <td>0.395693</td>\n",
       "      <td>0.916149</td>\n",
       "      <td>0.923871</td>\n",
       "      <td>0.932945</td>\n",
       "      <td>0.929694</td>\n",
       "      <td>0.921255</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>0.149097</td>\n",
       "      <td>0.320496</td>\n",
       "      <td>0.916149</td>\n",
       "      <td>0.923900</td>\n",
       "      <td>0.929564</td>\n",
       "      <td>0.928111</td>\n",
       "      <td>0.929999</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>0.164206</td>\n",
       "      <td>0.380259</td>\n",
       "      <td>0.922360</td>\n",
       "      <td>0.929167</td>\n",
       "      <td>0.936160</td>\n",
       "      <td>0.933499</td>\n",
       "      <td>0.936411</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>0.155053</td>\n",
       "      <td>0.277581</td>\n",
       "      <td>0.925466</td>\n",
       "      <td>0.937633</td>\n",
       "      <td>0.934374</td>\n",
       "      <td>0.934726</td>\n",
       "      <td>0.939905</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>0.137068</td>\n",
       "      <td>0.255496</td>\n",
       "      <td>0.913043</td>\n",
       "      <td>0.934406</td>\n",
       "      <td>0.922675</td>\n",
       "      <td>0.923669</td>\n",
       "      <td>0.958379</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>0.126534</td>\n",
       "      <td>0.237955</td>\n",
       "      <td>0.931677</td>\n",
       "      <td>0.939703</td>\n",
       "      <td>0.940017</td>\n",
       "      <td>0.939641</td>\n",
       "      <td>0.943204</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>0.105155</td>\n",
       "      <td>0.242772</td>\n",
       "      <td>0.928571</td>\n",
       "      <td>0.935950</td>\n",
       "      <td>0.941437</td>\n",
       "      <td>0.940233</td>\n",
       "      <td>0.943920</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>0.081622</td>\n",
       "      <td>0.218551</td>\n",
       "      <td>0.937888</td>\n",
       "      <td>0.944087</td>\n",
       "      <td>0.948374</td>\n",
       "      <td>0.947078</td>\n",
       "      <td>0.948131</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>0.081521</td>\n",
       "      <td>0.241942</td>\n",
       "      <td>0.931677</td>\n",
       "      <td>0.938926</td>\n",
       "      <td>0.942866</td>\n",
       "      <td>0.941803</td>\n",
       "      <td>0.952882</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>20</td>\n",
       "      <td>0.075961</td>\n",
       "      <td>0.219314</td>\n",
       "      <td>0.940994</td>\n",
       "      <td>0.943262</td>\n",
       "      <td>0.954374</td>\n",
       "      <td>0.951763</td>\n",
       "      <td>0.955047</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>21</td>\n",
       "      <td>0.077337</td>\n",
       "      <td>0.243636</td>\n",
       "      <td>0.950311</td>\n",
       "      <td>0.958428</td>\n",
       "      <td>0.961969</td>\n",
       "      <td>0.961100</td>\n",
       "      <td>0.970194</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>22</td>\n",
       "      <td>0.054068</td>\n",
       "      <td>0.250103</td>\n",
       "      <td>0.950311</td>\n",
       "      <td>0.958052</td>\n",
       "      <td>0.962278</td>\n",
       "      <td>0.961184</td>\n",
       "      <td>0.970283</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>23</td>\n",
       "      <td>0.050315</td>\n",
       "      <td>0.229084</td>\n",
       "      <td>0.953416</td>\n",
       "      <td>0.958954</td>\n",
       "      <td>0.963882</td>\n",
       "      <td>0.962513</td>\n",
       "      <td>0.964100</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>24</td>\n",
       "      <td>0.033063</td>\n",
       "      <td>0.228647</td>\n",
       "      <td>0.956522</td>\n",
       "      <td>0.961210</td>\n",
       "      <td>0.966413</td>\n",
       "      <td>0.965033</td>\n",
       "      <td>0.965008</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>25</td>\n",
       "      <td>0.032114</td>\n",
       "      <td>0.234541</td>\n",
       "      <td>0.953416</td>\n",
       "      <td>0.958547</td>\n",
       "      <td>0.964191</td>\n",
       "      <td>0.962750</td>\n",
       "      <td>0.964121</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>26</td>\n",
       "      <td>0.026368</td>\n",
       "      <td>0.222886</td>\n",
       "      <td>0.953416</td>\n",
       "      <td>0.958954</td>\n",
       "      <td>0.963882</td>\n",
       "      <td>0.962513</td>\n",
       "      <td>0.964100</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>27</td>\n",
       "      <td>0.027654</td>\n",
       "      <td>0.216881</td>\n",
       "      <td>0.953416</td>\n",
       "      <td>0.958954</td>\n",
       "      <td>0.963882</td>\n",
       "      <td>0.962513</td>\n",
       "      <td>0.964100</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>28</td>\n",
       "      <td>0.039466</td>\n",
       "      <td>0.217479</td>\n",
       "      <td>0.956522</td>\n",
       "      <td>0.961210</td>\n",
       "      <td>0.966413</td>\n",
       "      <td>0.965033</td>\n",
       "      <td>0.965008</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>29</td>\n",
       "      <td>0.053051</td>\n",
       "      <td>0.219230</td>\n",
       "      <td>0.953416</td>\n",
       "      <td>0.958954</td>\n",
       "      <td>0.963882</td>\n",
       "      <td>0.962513</td>\n",
       "      <td>0.964100</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Better model found at epoch 0 with accuracy value: 0.3322981297969818.\n",
      "Better model found at epoch 1 with accuracy value: 0.6055900454521179.\n",
      "Better model found at epoch 2 with accuracy value: 0.7360248565673828.\n",
      "Better model found at epoch 3 with accuracy value: 0.8416149020195007.\n",
      "Better model found at epoch 4 with accuracy value: 0.8757764101028442.\n",
      "Better model found at epoch 5 with accuracy value: 0.8819875717163086.\n",
      "Better model found at epoch 6 with accuracy value: 0.8850931525230408.\n",
      "Better model found at epoch 7 with accuracy value: 0.9161490797996521.\n",
      "Better model found at epoch 9 with accuracy value: 0.9254658222198486.\n",
      "Better model found at epoch 16 with accuracy value: 0.9316770434379578.\n",
      "Better model found at epoch 18 with accuracy value: 0.9378882050514221.\n",
      "Better model found at epoch 20 with accuracy value: 0.9409937858581543.\n",
      "Better model found at epoch 21 with accuracy value: 0.9503105878829956.\n",
      "Better model found at epoch 23 with accuracy value: 0.9534161686897278.\n",
      "Better model found at epoch 24 with accuracy value: 0.95652174949646.\n"
     ]
    }
   ],
   "source": [
    "learner.fit_one_cycle(30, max_lr=slice(2e-03), callbacks=model_callback(learner, \"best-effb3-fold1-stage1\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 426
    },
    "colab_type": "code",
    "id": "Jiwzs8IuPYKa",
    "outputId": "9e40e2c0-85ae-4255-fac8-c1a9004d5258"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "    <div>\n",
       "        <style>\n",
       "            /* Turns off some styling */\n",
       "            progress {\n",
       "                /* gets rid of default border in Firefox and Opera. */\n",
       "                border: none;\n",
       "                /* Needs to be in here for Safari polyfill so background images work as expected. */\n",
       "                background-size: auto;\n",
       "            }\n",
       "            .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
       "                background: #F44336;\n",
       "            }\n",
       "        </style>\n",
       "      <progress value='1' class='' max='3', style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
       "      33.33% [1/3 00:20<00:40]\n",
       "    </div>\n",
       "    \n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>accuracy</th>\n",
       "      <th>precision</th>\n",
       "      <th>recall</th>\n",
       "      <th>f_beta</th>\n",
       "      <th>kappa_score</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.035085</td>\n",
       "      <td>#na#</td>\n",
       "      <td>00:20</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table><p>\n",
       "\n",
       "    <div>\n",
       "        <style>\n",
       "            /* Turns off some styling */\n",
       "            progress {\n",
       "                /* gets rid of default border in Firefox and Opera. */\n",
       "                border: none;\n",
       "                /* Needs to be in here for Safari polyfill so background images work as expected. */\n",
       "                background-size: auto;\n",
       "            }\n",
       "            .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
       "                background: #F44336;\n",
       "            }\n",
       "        </style>\n",
       "      <progress value='14' class='' max='40', style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
       "      35.00% [14/40 00:08<00:16 0.0569]\n",
       "    </div>\n",
       "    "
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "LR Finder is complete, type {learner_name}.recorder.plot() to see the graph.\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZUAAAEGCAYAAACtqQjWAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n5v8Bxq779pLpPpAAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "learner.load(\"best-effb3-fold1-stage1\")\n",
    "learner.unfreeze()\n",
    "learner.lr_find()\n",
    "learner.recorder.plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "colab_type": "code",
    "id": "dZLPVIEESvdr",
    "outputId": "a7591f92-f6d2-4196-a8ab-851aa5fc6635"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>accuracy</th>\n",
       "      <th>precision</th>\n",
       "      <th>recall</th>\n",
       "      <th>f_beta</th>\n",
       "      <th>kappa_score</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.038885</td>\n",
       "      <td>0.227551</td>\n",
       "      <td>0.956522</td>\n",
       "      <td>0.961210</td>\n",
       "      <td>0.966413</td>\n",
       "      <td>0.965033</td>\n",
       "      <td>0.965008</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.050994</td>\n",
       "      <td>0.227400</td>\n",
       "      <td>0.956522</td>\n",
       "      <td>0.961210</td>\n",
       "      <td>0.966413</td>\n",
       "      <td>0.965033</td>\n",
       "      <td>0.965008</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.038471</td>\n",
       "      <td>0.224671</td>\n",
       "      <td>0.956522</td>\n",
       "      <td>0.961210</td>\n",
       "      <td>0.966413</td>\n",
       "      <td>0.965033</td>\n",
       "      <td>0.965008</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.043162</td>\n",
       "      <td>0.223609</td>\n",
       "      <td>0.956522</td>\n",
       "      <td>0.961210</td>\n",
       "      <td>0.966413</td>\n",
       "      <td>0.965033</td>\n",
       "      <td>0.965008</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.048894</td>\n",
       "      <td>0.223055</td>\n",
       "      <td>0.953416</td>\n",
       "      <td>0.958954</td>\n",
       "      <td>0.963882</td>\n",
       "      <td>0.962513</td>\n",
       "      <td>0.964100</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.054109</td>\n",
       "      <td>0.230694</td>\n",
       "      <td>0.950311</td>\n",
       "      <td>0.956235</td>\n",
       "      <td>0.961660</td>\n",
       "      <td>0.960237</td>\n",
       "      <td>0.963213</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.034101</td>\n",
       "      <td>0.215827</td>\n",
       "      <td>0.950311</td>\n",
       "      <td>0.954597</td>\n",
       "      <td>0.961350</td>\n",
       "      <td>0.959469</td>\n",
       "      <td>0.956220</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>0.037242</td>\n",
       "      <td>0.221526</td>\n",
       "      <td>0.953416</td>\n",
       "      <td>0.958363</td>\n",
       "      <td>0.964501</td>\n",
       "      <td>0.962957</td>\n",
       "      <td>0.964142</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>0.035228</td>\n",
       "      <td>0.210566</td>\n",
       "      <td>0.956522</td>\n",
       "      <td>0.961210</td>\n",
       "      <td>0.966413</td>\n",
       "      <td>0.965033</td>\n",
       "      <td>0.965008</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>0.028013</td>\n",
       "      <td>0.207686</td>\n",
       "      <td>0.956522</td>\n",
       "      <td>0.961210</td>\n",
       "      <td>0.966413</td>\n",
       "      <td>0.965033</td>\n",
       "      <td>0.965008</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>0.042250</td>\n",
       "      <td>0.213870</td>\n",
       "      <td>0.950311</td>\n",
       "      <td>0.956204</td>\n",
       "      <td>0.961215</td>\n",
       "      <td>0.959933</td>\n",
       "      <td>0.963126</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>0.032029</td>\n",
       "      <td>0.208937</td>\n",
       "      <td>0.953416</td>\n",
       "      <td>0.958493</td>\n",
       "      <td>0.963747</td>\n",
       "      <td>0.962449</td>\n",
       "      <td>0.964035</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>0.023592</td>\n",
       "      <td>0.209541</td>\n",
       "      <td>0.950311</td>\n",
       "      <td>0.955807</td>\n",
       "      <td>0.961834</td>\n",
       "      <td>0.960375</td>\n",
       "      <td>0.963168</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>0.030212</td>\n",
       "      <td>0.200931</td>\n",
       "      <td>0.953416</td>\n",
       "      <td>0.958287</td>\n",
       "      <td>0.964056</td>\n",
       "      <td>0.962669</td>\n",
       "      <td>0.964056</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>0.025118</td>\n",
       "      <td>0.210588</td>\n",
       "      <td>0.953416</td>\n",
       "      <td>0.958493</td>\n",
       "      <td>0.963747</td>\n",
       "      <td>0.962449</td>\n",
       "      <td>0.964035</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>0.016788</td>\n",
       "      <td>0.218397</td>\n",
       "      <td>0.953416</td>\n",
       "      <td>0.958493</td>\n",
       "      <td>0.963747</td>\n",
       "      <td>0.962449</td>\n",
       "      <td>0.964035</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>0.027053</td>\n",
       "      <td>0.226074</td>\n",
       "      <td>0.956522</td>\n",
       "      <td>0.961210</td>\n",
       "      <td>0.966413</td>\n",
       "      <td>0.965033</td>\n",
       "      <td>0.965008</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>0.021930</td>\n",
       "      <td>0.219676</td>\n",
       "      <td>0.953416</td>\n",
       "      <td>0.958493</td>\n",
       "      <td>0.963747</td>\n",
       "      <td>0.962449</td>\n",
       "      <td>0.964035</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>0.022112</td>\n",
       "      <td>0.222444</td>\n",
       "      <td>0.956522</td>\n",
       "      <td>0.961210</td>\n",
       "      <td>0.966413</td>\n",
       "      <td>0.965033</td>\n",
       "      <td>0.965008</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>0.024521</td>\n",
       "      <td>0.221684</td>\n",
       "      <td>0.953416</td>\n",
       "      <td>0.958493</td>\n",
       "      <td>0.963747</td>\n",
       "      <td>0.962449</td>\n",
       "      <td>0.964035</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>20</td>\n",
       "      <td>0.018080</td>\n",
       "      <td>0.223894</td>\n",
       "      <td>0.956522</td>\n",
       "      <td>0.961210</td>\n",
       "      <td>0.966413</td>\n",
       "      <td>0.965033</td>\n",
       "      <td>0.965008</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>21</td>\n",
       "      <td>0.031532</td>\n",
       "      <td>0.222188</td>\n",
       "      <td>0.956522</td>\n",
       "      <td>0.961210</td>\n",
       "      <td>0.966413</td>\n",
       "      <td>0.965033</td>\n",
       "      <td>0.965008</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>22</td>\n",
       "      <td>0.024039</td>\n",
       "      <td>0.221843</td>\n",
       "      <td>0.956522</td>\n",
       "      <td>0.961210</td>\n",
       "      <td>0.966413</td>\n",
       "      <td>0.965033</td>\n",
       "      <td>0.965008</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>23</td>\n",
       "      <td>0.020910</td>\n",
       "      <td>0.218137</td>\n",
       "      <td>0.956522</td>\n",
       "      <td>0.961210</td>\n",
       "      <td>0.966413</td>\n",
       "      <td>0.965033</td>\n",
       "      <td>0.965008</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>24</td>\n",
       "      <td>0.030734</td>\n",
       "      <td>0.217057</td>\n",
       "      <td>0.956522</td>\n",
       "      <td>0.961210</td>\n",
       "      <td>0.966413</td>\n",
       "      <td>0.965033</td>\n",
       "      <td>0.965008</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>25</td>\n",
       "      <td>0.024385</td>\n",
       "      <td>0.217179</td>\n",
       "      <td>0.956522</td>\n",
       "      <td>0.961210</td>\n",
       "      <td>0.966413</td>\n",
       "      <td>0.965033</td>\n",
       "      <td>0.965008</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>26</td>\n",
       "      <td>0.023683</td>\n",
       "      <td>0.216945</td>\n",
       "      <td>0.956522</td>\n",
       "      <td>0.961210</td>\n",
       "      <td>0.966413</td>\n",
       "      <td>0.965033</td>\n",
       "      <td>0.965008</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>27</td>\n",
       "      <td>0.020271</td>\n",
       "      <td>0.217216</td>\n",
       "      <td>0.956522</td>\n",
       "      <td>0.961210</td>\n",
       "      <td>0.966413</td>\n",
       "      <td>0.965033</td>\n",
       "      <td>0.965008</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>28</td>\n",
       "      <td>0.019981</td>\n",
       "      <td>0.219187</td>\n",
       "      <td>0.953416</td>\n",
       "      <td>0.958954</td>\n",
       "      <td>0.963882</td>\n",
       "      <td>0.962513</td>\n",
       "      <td>0.964100</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>29</td>\n",
       "      <td>0.012691</td>\n",
       "      <td>0.218887</td>\n",
       "      <td>0.953416</td>\n",
       "      <td>0.958954</td>\n",
       "      <td>0.963882</td>\n",
       "      <td>0.962513</td>\n",
       "      <td>0.964100</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Better model found at epoch 0 with accuracy value: 0.95652174949646.\n"
     ]
    }
   ],
   "source": [
    "learner.fit_one_cycle(30, max_lr=slice(1.3e-06, 2e-04), callbacks=model_callback(learner, \"best-effb3-fold1-stage2\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "CilAWIJBTAEl"
   },
   "outputs": [],
   "source": [
    "fold2_idxs = idxs[1]\n",
    "fold_data = (ImageList.from_folder(path/\"sipakmed_wsi_pap_smear\")\n",
    "            .split_by_idxs(fold2_idxs[0], fold2_idxs[1])\n",
    "            .label_from_folder()\n",
    "            .transform(tfms, size=224)\n",
    "            .databunch(bs=16)\n",
    "            .normalize(imagenet_stats))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 426
    },
    "colab_type": "code",
    "id": "13j-dSNuWitL",
    "outputId": "c8d1b0d9-9bfd-43b5-c8fd-9595c060f171"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "    <div>\n",
       "        <style>\n",
       "            /* Turns off some styling */\n",
       "            progress {\n",
       "                /* gets rid of default border in Firefox and Opera. */\n",
       "                border: none;\n",
       "                /* Needs to be in here for Safari polyfill so background images work as expected. */\n",
       "                background-size: auto;\n",
       "            }\n",
       "            .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
       "                background: #F44336;\n",
       "            }\n",
       "        </style>\n",
       "      <progress value='1' class='' max='3', style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
       "      33.33% [1/3 00:19<00:38]\n",
       "    </div>\n",
       "    \n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>accuracy</th>\n",
       "      <th>precision</th>\n",
       "      <th>recall</th>\n",
       "      <th>f_beta</th>\n",
       "      <th>kappa_score</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.198492</td>\n",
       "      <td>#na#</td>\n",
       "      <td>00:19</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table><p>\n",
       "\n",
       "    <div>\n",
       "        <style>\n",
       "            /* Turns off some styling */\n",
       "            progress {\n",
       "                /* gets rid of default border in Firefox and Opera. */\n",
       "                border: none;\n",
       "                /* Needs to be in here for Safari polyfill so background images work as expected. */\n",
       "                background-size: auto;\n",
       "            }\n",
       "            .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
       "                background: #F44336;\n",
       "            }\n",
       "        </style>\n",
       "      <progress value='38' class='' max='40', style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
       "      95.00% [38/40 00:19<00:01 0.5720]\n",
       "    </div>\n",
       "    "
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "LR Finder is complete, type {learner_name}.recorder.plot() to see the graph.\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZEAAAEGCAYAAACkQqisAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "learner.load(\"best-effb3-fold1-stage2\")\n",
    "learner.data = fold_data\n",
    "learner.freeze()\n",
    "learner = to_fp16(learner)\n",
    "learner.lr_find()\n",
    "learner.recorder.plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "colab_type": "code",
    "id": "79lTkQv2Wr9u",
    "outputId": "cebe667c-bb1d-4567-e7f1-4a08467562de"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>accuracy</th>\n",
       "      <th>precision</th>\n",
       "      <th>recall</th>\n",
       "      <th>f_beta</th>\n",
       "      <th>kappa_score</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.157517</td>\n",
       "      <td>0.000876</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.205018</td>\n",
       "      <td>0.000850</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.215622</td>\n",
       "      <td>0.000841</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.217041</td>\n",
       "      <td>0.000828</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.207685</td>\n",
       "      <td>0.000874</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.242993</td>\n",
       "      <td>0.000861</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.235516</td>\n",
       "      <td>0.000916</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>0.198069</td>\n",
       "      <td>0.000932</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>0.211231</td>\n",
       "      <td>0.000933</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>0.218841</td>\n",
       "      <td>0.000910</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>0.209826</td>\n",
       "      <td>0.000925</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>0.198689</td>\n",
       "      <td>0.000894</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>0.190199</td>\n",
       "      <td>0.000971</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>0.207259</td>\n",
       "      <td>0.000989</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>0.200508</td>\n",
       "      <td>0.001019</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>0.182541</td>\n",
       "      <td>0.001049</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>0.205254</td>\n",
       "      <td>0.001029</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>0.169524</td>\n",
       "      <td>0.001047</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>0.232755</td>\n",
       "      <td>0.000994</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>0.223920</td>\n",
       "      <td>0.000979</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>20</td>\n",
       "      <td>0.200582</td>\n",
       "      <td>0.000967</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>21</td>\n",
       "      <td>0.170454</td>\n",
       "      <td>0.000980</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>22</td>\n",
       "      <td>0.174318</td>\n",
       "      <td>0.000983</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>23</td>\n",
       "      <td>0.211107</td>\n",
       "      <td>0.000985</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>24</td>\n",
       "      <td>0.235892</td>\n",
       "      <td>0.001087</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>25</td>\n",
       "      <td>0.231691</td>\n",
       "      <td>0.001066</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>26</td>\n",
       "      <td>0.217821</td>\n",
       "      <td>0.001070</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>27</td>\n",
       "      <td>0.201027</td>\n",
       "      <td>0.001055</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>28</td>\n",
       "      <td>0.219495</td>\n",
       "      <td>0.001124</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>29</td>\n",
       "      <td>0.208078</td>\n",
       "      <td>0.001114</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Better model found at epoch 0 with accuracy value: 1.0.\n"
     ]
    }
   ],
   "source": [
    "learner.fit_one_cycle(30, max_lr=slice(7e-06), callbacks=model_callback(learner, \"best-effb3-fold2-stage1\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 426
    },
    "colab_type": "code",
    "id": "aB1A8s4uXawp",
    "outputId": "7d1a85f6-0d68-4e0b-d860-6f4b27afe65b"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "    <div>\n",
       "        <style>\n",
       "            /* Turns off some styling */\n",
       "            progress {\n",
       "                /* gets rid of default border in Firefox and Opera. */\n",
       "                border: none;\n",
       "                /* Needs to be in here for Safari polyfill so background images work as expected. */\n",
       "                background-size: auto;\n",
       "            }\n",
       "            .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
       "                background: #F44336;\n",
       "            }\n",
       "        </style>\n",
       "      <progress value='1' class='' max='3', style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
       "      33.33% [1/3 00:19<00:39]\n",
       "    </div>\n",
       "    \n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>accuracy</th>\n",
       "      <th>precision</th>\n",
       "      <th>recall</th>\n",
       "      <th>f_beta</th>\n",
       "      <th>kappa_score</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.172603</td>\n",
       "      <td>#na#</td>\n",
       "      <td>00:19</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table><p>\n",
       "\n",
       "    <div>\n",
       "        <style>\n",
       "            /* Turns off some styling */\n",
       "            progress {\n",
       "                /* gets rid of default border in Firefox and Opera. */\n",
       "                border: none;\n",
       "                /* Needs to be in here for Safari polyfill so background images work as expected. */\n",
       "                background-size: auto;\n",
       "            }\n",
       "            .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
       "                background: #F44336;\n",
       "            }\n",
       "        </style>\n",
       "      <progress value='23' class='' max='40', style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
       "      57.50% [23/40 00:12<00:09 0.4466]\n",
       "    </div>\n",
       "    "
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "LR Finder is complete, type {learner_name}.recorder.plot() to see the graph.\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYgAAAEGCAYAAAB/+QKOAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "learner.load(\"best-effb3-fold2-stage1\")\n",
    "learner.unfreeze()\n",
    "learner.lr_find()\n",
    "learner.recorder.plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "colab_type": "code",
    "id": "0y8UaU8Xa312",
    "outputId": "b718a4d7-cc25-4420-a546-694965c66975"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>accuracy</th>\n",
       "      <th>precision</th>\n",
       "      <th>recall</th>\n",
       "      <th>f_beta</th>\n",
       "      <th>kappa_score</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.227533</td>\n",
       "      <td>0.000847</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.181532</td>\n",
       "      <td>0.000862</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.199757</td>\n",
       "      <td>0.000852</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.183205</td>\n",
       "      <td>0.000865</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.196484</td>\n",
       "      <td>0.000860</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.205374</td>\n",
       "      <td>0.000872</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.194905</td>\n",
       "      <td>0.000895</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>0.214245</td>\n",
       "      <td>0.000895</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>0.200390</td>\n",
       "      <td>0.000907</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>0.209829</td>\n",
       "      <td>0.000911</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>0.215189</td>\n",
       "      <td>0.000885</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>0.187583</td>\n",
       "      <td>0.000869</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>0.208934</td>\n",
       "      <td>0.000943</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>0.179575</td>\n",
       "      <td>0.000988</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>0.202407</td>\n",
       "      <td>0.000950</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>0.181094</td>\n",
       "      <td>0.000966</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>0.188188</td>\n",
       "      <td>0.001009</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>0.196274</td>\n",
       "      <td>0.000990</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>0.204076</td>\n",
       "      <td>0.000948</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>0.181915</td>\n",
       "      <td>0.000969</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>20</td>\n",
       "      <td>0.218741</td>\n",
       "      <td>0.000991</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>21</td>\n",
       "      <td>0.180612</td>\n",
       "      <td>0.001006</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>22</td>\n",
       "      <td>0.179668</td>\n",
       "      <td>0.000946</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>23</td>\n",
       "      <td>0.195974</td>\n",
       "      <td>0.000922</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>24</td>\n",
       "      <td>0.226178</td>\n",
       "      <td>0.000968</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>25</td>\n",
       "      <td>0.191295</td>\n",
       "      <td>0.000956</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>26</td>\n",
       "      <td>0.191618</td>\n",
       "      <td>0.000903</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>27</td>\n",
       "      <td>0.196646</td>\n",
       "      <td>0.000941</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>28</td>\n",
       "      <td>0.182711</td>\n",
       "      <td>0.000993</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>29</td>\n",
       "      <td>0.173328</td>\n",
       "      <td>0.000988</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Better model found at epoch 0 with accuracy value: 1.0.\n"
     ]
    }
   ],
   "source": [
    "learner.fit_one_cycle(30, max_lr=slice(7e-07, 7e-07), callbacks=model_callback(learner, \"best-effb3-fold2-stage2\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "nR8Bwe1TenBv"
   },
   "outputs": [],
   "source": [
    "fold3_idxs = idxs[2]\n",
    "fold_data = (ImageList.from_folder(path/\"sipakmed_wsi_pap_smear\")\n",
    "            .split_by_idxs(fold3_idxs[0], fold3_idxs[1])\n",
    "            .label_from_folder()\n",
    "            .transform(tfms, size=224)\n",
    "            .databunch(bs=16)\n",
    "            .normalize(imagenet_stats))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 426
    },
    "colab_type": "code",
    "id": "U2-9q3-NbNIR",
    "outputId": "a5ede3b6-347f-4680-b904-af905b94d302"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "    <div>\n",
       "        <style>\n",
       "            /* Turns off some styling */\n",
       "            progress {\n",
       "                /* gets rid of default border in Firefox and Opera. */\n",
       "                border: none;\n",
       "                /* Needs to be in here for Safari polyfill so background images work as expected. */\n",
       "                background-size: auto;\n",
       "            }\n",
       "            .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
       "                background: #F44336;\n",
       "            }\n",
       "        </style>\n",
       "      <progress value='1' class='' max='3', style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
       "      33.33% [1/3 00:18<00:37]\n",
       "    </div>\n",
       "    \n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>accuracy</th>\n",
       "      <th>precision</th>\n",
       "      <th>recall</th>\n",
       "      <th>f_beta</th>\n",
       "      <th>kappa_score</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.159670</td>\n",
       "      <td>#na#</td>\n",
       "      <td>00:18</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table><p>\n",
       "\n",
       "    <div>\n",
       "        <style>\n",
       "            /* Turns off some styling */\n",
       "            progress {\n",
       "                /* gets rid of default border in Firefox and Opera. */\n",
       "                border: none;\n",
       "                /* Needs to be in here for Safari polyfill so background images work as expected. */\n",
       "                background-size: auto;\n",
       "            }\n",
       "            .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
       "                background: #F44336;\n",
       "            }\n",
       "        </style>\n",
       "      <progress value='36' class='' max='40', style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
       "      90.00% [36/40 00:17<00:01 0.4626]\n",
       "    </div>\n",
       "    "
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "LR Finder is complete, type {learner_name}.recorder.plot() to see the graph.\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYgAAAEGCAYAAAB/+QKOAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "learner.purge()\n",
    "learner.load(\"best-effb3-fold2-stage2\")\n",
    "learner.data = fold_data\n",
    "learner.freeze()\n",
    "learner = to_fp16(learner)\n",
    "learner.lr_find()\n",
    "learner.recorder.plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "colab_type": "code",
    "id": "4aS-gWGtes2j",
    "outputId": "7f975b66-d828-438d-a2ca-0ec6220185c1"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>accuracy</th>\n",
       "      <th>precision</th>\n",
       "      <th>recall</th>\n",
       "      <th>f_beta</th>\n",
       "      <th>kappa_score</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.236722</td>\n",
       "      <td>0.000536</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.259883</td>\n",
       "      <td>0.000710</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.215287</td>\n",
       "      <td>0.000830</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:26</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.178568</td>\n",
       "      <td>0.001315</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.154813</td>\n",
       "      <td>0.004105</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.154491</td>\n",
       "      <td>0.011312</td>\n",
       "      <td>0.993789</td>\n",
       "      <td>0.995122</td>\n",
       "      <td>0.995556</td>\n",
       "      <td>0.995434</td>\n",
       "      <td>0.998162</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.133067</td>\n",
       "      <td>0.017016</td>\n",
       "      <td>0.993789</td>\n",
       "      <td>0.995122</td>\n",
       "      <td>0.995556</td>\n",
       "      <td>0.995434</td>\n",
       "      <td>0.998162</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>0.126992</td>\n",
       "      <td>0.017083</td>\n",
       "      <td>0.990683</td>\n",
       "      <td>0.992771</td>\n",
       "      <td>0.993333</td>\n",
       "      <td>0.993142</td>\n",
       "      <td>0.997244</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>0.144840</td>\n",
       "      <td>0.020833</td>\n",
       "      <td>0.990683</td>\n",
       "      <td>0.992771</td>\n",
       "      <td>0.993333</td>\n",
       "      <td>0.993142</td>\n",
       "      <td>0.997244</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>0.108641</td>\n",
       "      <td>0.016100</td>\n",
       "      <td>0.990683</td>\n",
       "      <td>0.992771</td>\n",
       "      <td>0.993333</td>\n",
       "      <td>0.993142</td>\n",
       "      <td>0.997244</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>0.096432</td>\n",
       "      <td>0.009042</td>\n",
       "      <td>0.993789</td>\n",
       "      <td>0.995122</td>\n",
       "      <td>0.995556</td>\n",
       "      <td>0.995434</td>\n",
       "      <td>0.998162</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>0.088213</td>\n",
       "      <td>0.009045</td>\n",
       "      <td>0.993789</td>\n",
       "      <td>0.995122</td>\n",
       "      <td>0.995556</td>\n",
       "      <td>0.995434</td>\n",
       "      <td>0.998162</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>0.122962</td>\n",
       "      <td>0.015409</td>\n",
       "      <td>0.993789</td>\n",
       "      <td>0.995122</td>\n",
       "      <td>0.995556</td>\n",
       "      <td>0.995434</td>\n",
       "      <td>0.998162</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>0.114125</td>\n",
       "      <td>0.021976</td>\n",
       "      <td>0.993789</td>\n",
       "      <td>0.995122</td>\n",
       "      <td>0.995556</td>\n",
       "      <td>0.995434</td>\n",
       "      <td>0.998162</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>0.098468</td>\n",
       "      <td>0.006065</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>0.093015</td>\n",
       "      <td>0.011731</td>\n",
       "      <td>0.996894</td>\n",
       "      <td>0.997531</td>\n",
       "      <td>0.997778</td>\n",
       "      <td>0.997720</td>\n",
       "      <td>0.999081</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>0.078773</td>\n",
       "      <td>0.011778</td>\n",
       "      <td>0.996894</td>\n",
       "      <td>0.997531</td>\n",
       "      <td>0.997778</td>\n",
       "      <td>0.997720</td>\n",
       "      <td>0.999081</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>0.069305</td>\n",
       "      <td>0.006894</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>0.070586</td>\n",
       "      <td>0.006269</td>\n",
       "      <td>0.996894</td>\n",
       "      <td>0.997531</td>\n",
       "      <td>0.997778</td>\n",
       "      <td>0.997720</td>\n",
       "      <td>0.999081</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>0.075973</td>\n",
       "      <td>0.006070</td>\n",
       "      <td>0.996894</td>\n",
       "      <td>0.997531</td>\n",
       "      <td>0.997778</td>\n",
       "      <td>0.997720</td>\n",
       "      <td>0.999081</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>20</td>\n",
       "      <td>0.057336</td>\n",
       "      <td>0.005463</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>21</td>\n",
       "      <td>0.069509</td>\n",
       "      <td>0.005292</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:26</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>22</td>\n",
       "      <td>0.063864</td>\n",
       "      <td>0.005111</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>23</td>\n",
       "      <td>0.069831</td>\n",
       "      <td>0.005314</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>24</td>\n",
       "      <td>0.057456</td>\n",
       "      <td>0.005249</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>25</td>\n",
       "      <td>0.051420</td>\n",
       "      <td>0.005212</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>26</td>\n",
       "      <td>0.060349</td>\n",
       "      <td>0.004648</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>27</td>\n",
       "      <td>0.051649</td>\n",
       "      <td>0.004600</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>28</td>\n",
       "      <td>0.058131</td>\n",
       "      <td>0.004689</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>29</td>\n",
       "      <td>0.059828</td>\n",
       "      <td>0.004949</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Better model found at epoch 0 with accuracy value: 1.0.\n"
     ]
    }
   ],
   "source": [
    "learner.fit_one_cycle(30, max_lr=slice(1e-03), callbacks=model_callback(learner, \"best-effb3-fold3-stage1\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 426
    },
    "colab_type": "code",
    "id": "WQjTXqV9fPW1",
    "outputId": "66663d5b-3f41-44b0-c199-ca1fbfeaa0d8"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "    <div>\n",
       "        <style>\n",
       "            /* Turns off some styling */\n",
       "            progress {\n",
       "                /* gets rid of default border in Firefox and Opera. */\n",
       "                border: none;\n",
       "                /* Needs to be in here for Safari polyfill so background images work as expected. */\n",
       "                background-size: auto;\n",
       "            }\n",
       "            .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
       "                background: #F44336;\n",
       "            }\n",
       "        </style>\n",
       "      <progress value='1' class='' max='3', style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
       "      33.33% [1/3 00:19<00:38]\n",
       "    </div>\n",
       "    \n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>accuracy</th>\n",
       "      <th>precision</th>\n",
       "      <th>recall</th>\n",
       "      <th>f_beta</th>\n",
       "      <th>kappa_score</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.249856</td>\n",
       "      <td>#na#</td>\n",
       "      <td>00:19</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table><p>\n",
       "\n",
       "    <div>\n",
       "        <style>\n",
       "            /* Turns off some styling */\n",
       "            progress {\n",
       "                /* gets rid of default border in Firefox and Opera. */\n",
       "                border: none;\n",
       "                /* Needs to be in here for Safari polyfill so background images work as expected. */\n",
       "                background-size: auto;\n",
       "            }\n",
       "            .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
       "                background: #F44336;\n",
       "            }\n",
       "        </style>\n",
       "      <progress value='27' class='' max='40', style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
       "      67.50% [27/40 00:14<00:06 0.6931]\n",
       "    </div>\n",
       "    "
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "LR Finder is complete, type {learner_name}.recorder.plot() to see the graph.\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYoAAAEGCAYAAAB7DNKzAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n8kgThVJKKY80USillPJIE4VSSimP/j+j6Gjw2fKVvQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "learner.load(\"best-effb3-fold3-stage1\")\n",
    "learner.unfreeze()\n",
    "learner.lr_find()\n",
    "learner.recorder.plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 714
    },
    "colab_type": "code",
    "id": "EGL9-RX6ibhO",
    "outputId": "3985bc1b-8fd2-4c18-8f5f-222e5d4114be"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>accuracy</th>\n",
       "      <th>precision</th>\n",
       "      <th>recall</th>\n",
       "      <th>f_beta</th>\n",
       "      <th>kappa_score</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.210297</td>\n",
       "      <td>0.000544</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.173572</td>\n",
       "      <td>0.000771</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.156403</td>\n",
       "      <td>0.001983</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.129866</td>\n",
       "      <td>0.002633</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.122232</td>\n",
       "      <td>0.001482</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.127294</td>\n",
       "      <td>0.002049</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.101492</td>\n",
       "      <td>0.002653</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>0.085448</td>\n",
       "      <td>0.002352</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>0.081437</td>\n",
       "      <td>0.001773</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>0.075370</td>\n",
       "      <td>0.001442</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>0.051281</td>\n",
       "      <td>0.001601</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>0.054204</td>\n",
       "      <td>0.001603</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>0.068608</td>\n",
       "      <td>0.002185</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>0.049463</td>\n",
       "      <td>0.001935</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>0.038921</td>\n",
       "      <td>0.001626</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>0.035674</td>\n",
       "      <td>0.001651</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>0.038160</td>\n",
       "      <td>0.001442</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>0.030349</td>\n",
       "      <td>0.001427</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>0.031969</td>\n",
       "      <td>0.001420</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>0.044353</td>\n",
       "      <td>0.001475</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Better model found at epoch 0 with accuracy value: 1.0.\n"
     ]
    }
   ],
   "source": [
    "learner.fit_one_cycle(20, max_lr=slice(5e-04), callbacks=model_callback(learner, \"best-effb3-fold3-stage2\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "PxEsi-THjHK7"
   },
   "outputs": [],
   "source": [
    "all_data = (ImageList.from_folder(path/\"sipakmed_wsi_pap_smear\")\n",
    "            .split_none()\n",
    "            .label_from_folder()\n",
    "            .transform(tfms, size=224)\n",
    "            .databunch(bs=16)\n",
    "            .normalize(imagenet_stats))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 527
    },
    "colab_type": "code",
    "id": "6XWhajuYjRzU",
    "outputId": "c8459f54-d36b-48d7-939a-73a8dff1e1af"
   },
   "outputs": [
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAe4AAAH+CAYAAABawe3NAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nIiJiCknhjoiImEJSuCMiIqaQ/we6ekYDXSWOeQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 504x504 with 1 Axes>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "interp = ClassificationInterpretation.from_learner(learner, ds_type=DatasetType.Train)\n",
    "interp.plot_confusion_matrix(figsize=(7, 7))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 527
    },
    "colab_type": "code",
    "id": "XBEf5WhVktFJ",
    "outputId": "9685f33a-e849-4bc9-c7ab-39757a987594"
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAe4AAAH+CAYAAABawe3NAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nxCyShjsiImIWScMdERExi/wfK2LFoqSiUGwAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 504x504 with 1 Axes>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "interp.plot_confusion_matrix(figsize=(7, 7), normalize=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 121
    },
    "colab_type": "code",
    "id": "WPXVp7wMjdFO",
    "outputId": "efda31be-9f29-4ff4-96d3-0671bf050e01"
   },
   "outputs": [
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "[0.09148381,\n",
       " tensor(0.9781),\n",
       " tensor(0.9809),\n",
       " tensor(0.9828),\n",
       " tensor(0.9824),\n",
       " tensor(0.9858)]"
      ]
     },
     "execution_count": 49,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "val_learner = to_fp32(learner)\n",
    "test_mets = [accuracy, Precision(average=\"macro\"), Recall(average=\"macro\"), FBeta(average=\"macro\"), KappaScore(weights=\"quadratic\")]\n",
    "val_learner.validate(dl=all_data.train_dl, metrics=test_mets)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "dx2g0jsRlTZF"
   },
   "outputs": [],
   "source": [
    "learner.export()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "BGeH_mMdmA-d"
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "collapsed_sections": [],
   "machine_shape": "hm",
   "name": "Sipakmed K-Fold CV in fastai.ipynb",
   "provenance": []
  },
  "kernelspec": {
   "display_name": "Python 3",
   "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": "3.7.1"
  },
  "widgets": {
   "application/vnd.jupyter.widget-state+json": {
    "04d841481b61417aab864a1a9c8872af": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "0b9017694ad44490b3867fffa5be15e7": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "2e23ad6a5dff415993e7b191ef1c11d6": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "3665cf73253343f2a1c1479c1defc48f": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "ProgressStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "ProgressStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "bar_color": null,
      "description_width": ""
     }
    },
    "6f46df8e2309479989eabc39333473e3": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_04d841481b61417aab864a1a9c8872af",
      "placeholder": "​",
      "style": "IPY_MODEL_714202b45d904c27b5491f587474abbf",
      "value": "100% 47.1M/47.1M [00:04&lt;00:00, 11.9MB/s]"
     }
    },
    "714202b45d904c27b5491f587474abbf": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "e5cdf98a8ef34dec9adc0fff5423a644": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HBoxModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HBoxModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HBoxView",
      "box_style": "",
      "children": [
       "IPY_MODEL_ea33c3bdffb347639b83e5639cb92046",
       "IPY_MODEL_6f46df8e2309479989eabc39333473e3"
      ],
      "layout": "IPY_MODEL_2e23ad6a5dff415993e7b191ef1c11d6"
     }
    },
    "ea33c3bdffb347639b83e5639cb92046": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "IntProgressModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "IntProgressModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "ProgressView",
      "bar_style": "success",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_0b9017694ad44490b3867fffa5be15e7",
      "max": 49388949,
      "min": 0,
      "orientation": "horizontal",
      "style": "IPY_MODEL_3665cf73253343f2a1c1479c1defc48f",
      "value": 49388949
     }
    }
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}