fastai_kfold-cv.ipynb

{
 "cells": [
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "44132468",
   "metadata": {},
   "source": [
    "# Kitchenware Classifier Using KFold Cross Validation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "2b0010a5",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['sample_submission.csv', 'test.csv', 'models', 'train.csv', 'images', 'extra.csv']\n",
      "Python       : 3.9.15 | packaged by conda-forge | (main, Nov 22 2022, 15:55:03) \n",
      "FastAI       : 2.7.10\n",
      "Numpy        : 1.22.4\n",
      "Torch        : 1.13.1+cu117\n",
      "CUDA         : True\n"
     ]
    }
   ],
   "source": [
    "import gc\n",
    "import os\n",
    "import sys\n",
    "\n",
    "from collections import defaultdict\n",
    "from hashlib import md5\n",
    "\n",
    "import numpy as np\n",
    "import PIL\n",
    "import seaborn as sns\n",
    "import fastai\n",
    "import torch\n",
    "\n",
    "from fastai.vision.all import (\n",
    "    set_seed,\n",
    "    Path,\n",
    "    pd,\n",
    "    plt,\n",
    "    aug_transforms,\n",
    "    ImageDataLoaders,\n",
    "    vision_learner,\n",
    "    error_rate,\n",
    "    accuracy,\n",
    "    Resize,\n",
    "    minimum,\n",
    "    steep,\n",
    "    valley,\n",
    "    slide,\n",
    "    SaveModelCallback,\n",
    "    EarlyStoppingCallback,\n",
    "    Interpretation,\n",
    "    Learner,\n",
    "    MixUp,\n",
    "    cnn_learner,\n",
    "    partial,\n",
    "    DataBlock,\n",
    "    ImageBlock,\n",
    "    MultiCategoryBlock,\n",
    "    ColReader,\n",
    "    IndexSplitter,\n",
    "    FlipItem,\n",
    "    RandomErasing,\n",
    "    Normalize,\n",
    ")\n",
    "from iterstrat.ml_stratifiers import MultilabelStratifiedKFold\n",
    "from torch import device\n",
    "\n",
    "# import timm\n",
    "\n",
    "print(os.listdir(\"../kitchenware_classifier/data/\"))\n",
    "print(\"Python       :\", sys.version.split(\"\\n\")[0])\n",
    "print(\"FastAI       :\", fastai.__version__)\n",
    "print(\"Numpy        :\", np.__version__)\n",
    "print(\"Torch        :\", torch.__version__)\n",
    "print(\"CUDA         :\", torch.cuda.is_available())\n",
    "\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "76772191",
   "metadata": {},
   "source": [
    "## Load data and set variables"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "7e3b1daa",
   "metadata": {},
   "outputs": [],
   "source": [
    "SEED = 42\n",
    "set_seed(42)\n",
    "BATCH_SIZE = 8\n",
    "IMG_SIZE = 224\n",
    "\n",
    "PROJECT_DIR = Path(\"../kitchenware_classifier\")\n",
    "DATA_DIR = PROJECT_DIR / \"data\"\n",
    "IMG_DIR = DATA_DIR / \"images\"\n",
    "\n",
    "TRAIN_DF = pd.read_csv(DATA_DIR / \"train.csv\")\n",
    "TEST_DF = pd.read_csv(DATA_DIR / \"test.csv\")\n",
    "EXTRA_DF = pd.read_csv(DATA_DIR / \"extra.csv\")\n",
    "TRAIN_DF = pd.concat([TRAIN_DF, EXTRA_DF], ignore_index=True)\n",
    "\n",
    "TRAIN_DF.rename(columns={\"Id\": \"image\"}, inplace=True)\n",
    "TRAIN_DF[\"Id\"] = TRAIN_DF.loc[:, \"image\"]\n",
    "\n",
    "TEST_DF.rename(columns={\"Id\": \"image\"}, inplace=True)\n",
    "TEST_DF[\"Id\"] = TEST_DF.loc[:, \"image\"]\n",
    "\n",
    "TRAIN_DF[\"image\"] = TRAIN_DF[\"Id\"].map(lambda x: f\"{IMG_DIR.absolute()}/{x:0>4}.jpg\")\n",
    "TEST_DF[\"image\"] = TEST_DF[\"Id\"].map(lambda x: f\"{IMG_DIR}/{x:0>4}.jpg\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "ecdea5e7",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>image</th>\n",
       "      <th>label</th>\n",
       "      <th>Id</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>/home/clamytoe/Projects/kitchenware_classifier/notebooks/../kitchenware_classifier/data/images/0560.jpg</td>\n",
       "      <td>glass</td>\n",
       "      <td>560</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>/home/clamytoe/Projects/kitchenware_classifier/notebooks/../kitchenware_classifier/data/images/4675.jpg</td>\n",
       "      <td>cup</td>\n",
       "      <td>4675</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>/home/clamytoe/Projects/kitchenware_classifier/notebooks/../kitchenware_classifier/data/images/0875.jpg</td>\n",
       "      <td>glass</td>\n",
       "      <td>875</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>/home/clamytoe/Projects/kitchenware_classifier/notebooks/../kitchenware_classifier/data/images/4436.jpg</td>\n",
       "      <td>spoon</td>\n",
       "      <td>4436</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>/home/clamytoe/Projects/kitchenware_classifier/notebooks/../kitchenware_classifier/data/images/8265.jpg</td>\n",
       "      <td>plate</td>\n",
       "      <td>8265</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                                                                     image  \\\n",
       "0  /home/clamytoe/Projects/kitchenware_classifier/notebooks/../kitchenware_classifier/data/images/0560.jpg   \n",
       "1  /home/clamytoe/Projects/kitchenware_classifier/notebooks/../kitchenware_classifier/data/images/4675.jpg   \n",
       "2  /home/clamytoe/Projects/kitchenware_classifier/notebooks/../kitchenware_classifier/data/images/0875.jpg   \n",
       "3  /home/clamytoe/Projects/kitchenware_classifier/notebooks/../kitchenware_classifier/data/images/4436.jpg   \n",
       "4  /home/clamytoe/Projects/kitchenware_classifier/notebooks/../kitchenware_classifier/data/images/8265.jpg   \n",
       "\n",
       "   label    Id  \n",
       "0  glass   560  \n",
       "1    cup  4675  \n",
       "2  glass   875  \n",
       "3  spoon  4436  \n",
       "4  plate  8265  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>image</th>\n",
       "      <th>Id</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>../kitchenware_classifier/data/images/0678.jpg</td>\n",
       "      <td>678</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>../kitchenware_classifier/data/images/3962.jpg</td>\n",
       "      <td>3962</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>../kitchenware_classifier/data/images/9271.jpg</td>\n",
       "      <td>9271</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>../kitchenware_classifier/data/images/5133.jpg</td>\n",
       "      <td>5133</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>../kitchenware_classifier/data/images/8842.jpg</td>\n",
       "      <td>8842</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                            image    Id\n",
       "0  ../kitchenware_classifier/data/images/0678.jpg   678\n",
       "1  ../kitchenware_classifier/data/images/3962.jpg  3962\n",
       "2  ../kitchenware_classifier/data/images/9271.jpg  9271\n",
       "3  ../kitchenware_classifier/data/images/5133.jpg  5133\n",
       "4  ../kitchenware_classifier/data/images/8842.jpg  8842"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "display(TRAIN_DF.head())\n",
    "display(TEST_DF.head())"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2d902804",
   "metadata": {},
   "source": [
    "## EDA"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "86bc0980",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 6875 entries, 0 to 6874\n",
      "Data columns (total 3 columns):\n",
      " #   Column  Non-Null Count  Dtype \n",
      "---  ------  --------------  ----- \n",
      " 0   image   6875 non-null   object\n",
      " 1   label   6875 non-null   object\n",
      " 2   Id      6875 non-null   object\n",
      "dtypes: object(3)\n",
      "memory usage: 161.3+ KB\n"
     ]
    }
   ],
   "source": [
    "TRAIN_DF.info()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b77febf2",
   "metadata": {},
   "source": [
    "### Check the class distributions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "e95858bd",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.countplot(data=TRAIN_DF, x=\"label\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f9fd4741",
   "metadata": {},
   "source": [
    "### Check for duplicate images"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "90213828",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "10677"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "hash_dict = defaultdict(list)\n",
    "for image in IMG_DIR.glob(\"*.jpg\"):\n",
    "    with image.open(\"rb\") as f:\n",
    "        img_hash = md5(f.read()).hexdigest()\n",
    "        hash_dict[img_hash].append(image)\n",
    "len(hash_dict)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "452ea9fe",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[Path('../kitchenware_classifier/data/images/7103.jpg'), Path('../kitchenware_classifier/data/images/5428.jpg')]\n",
      "[Path('../kitchenware_classifier/data/images/4234.jpg'), Path('../kitchenware_classifier/data/images/0237.jpg')]\n",
      "[Path('../kitchenware_classifier/data/images/4162.jpg'), Path('../kitchenware_classifier/data/images/3458.jpg')]\n",
      "[Path('../kitchenware_classifier/data/images/8532.jpg'), Path('../kitchenware_classifier/data/images/6833.jpg')]\n",
      "[Path('../kitchenware_classifier/data/images/7196.jpg'), Path('../kitchenware_classifier/data/images/7063.jpg')]\n",
      "[Path('../kitchenware_classifier/data/images/6624.jpg'), Path('../kitchenware_classifier/data/images/7469.jpg')]\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "12"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "duplicate_img = []\n",
    "for k, v in hash_dict.items():\n",
    "    if len(v) > 1:\n",
    "        if v[0].name != v[1].name:\n",
    "            duplicate_img.append(v[0])\n",
    "            duplicate_img.append(v[1])\n",
    "            print(v)\n",
    "len(duplicate_img)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "c7827d6e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1200x800 with 12 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(12, 8))\n",
    "\n",
    "for idx, img in enumerate(duplicate_img):\n",
    "    im = PIL.Image.open(img)\n",
    "    title = img.stem\n",
    "    plt.subplot(6, 2, idx + 1)\n",
    "    plt.imshow(im)\n",
    "    plt.title(title)\n",
    "    plt.axis(\"off\")\n",
    "plt.tight_layout()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "4968f44e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.00112391121101433"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "(len(duplicate_img) / len(hash_dict))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f2f9f1c7",
   "metadata": {},
   "source": [
    "> Since the amount of duplicate images are so insignificant, I'm not going to worry about them."
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "ddb9f571",
   "metadata": {},
   "source": [
    "## KFold validation\n",
    "\n",
    "I modified the code from this great article that I found on the subject: [fastai MultiLabel Classification using Kfold Cross Validation](https://medium.com/analytics-vidhya/fastai-multilabel-classification-using-kfold-cross-validation-706b5948e138)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "9341ced8",
   "metadata": {},
   "outputs": [],
   "source": [
    "strat_kfold = MultilabelStratifiedKFold(n_splits=3, random_state=42, shuffle=True)\n",
    "stfold = strat_kfold.split(TRAIN_DF.image.values, TRAIN_DF.iloc[:, 1:].values)\n",
    "TRAIN_DF[\"fold\"] = -1\n",
    "for i, (_, test_index) in enumerate(stfold):\n",
    "    TRAIN_DF.iloc[test_index, -1] = i"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "8fced7b7",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "TRAIN_DF.fold.value_counts().plot.bar();"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "f6224445",
   "metadata": {},
   "source": [
    "## Metrics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "5837e937",
   "metadata": {},
   "outputs": [],
   "source": [
    "def accuracy_multi(inp, targ, thresh=0.5, sigmoid=True):\n",
    "    \"Compute accuracy when `inp` and `targ` are the same size.\"\n",
    "    if sigmoid:\n",
    "        inp = inp.sigmoid()\n",
    "    return ((inp > thresh) == targ.bool()).float().mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "407485ec",
   "metadata": {},
   "outputs": [],
   "source": [
    "def F_score(output, label, threshold=0.2, beta=1):\n",
    "    prob = output > threshold\n",
    "    label = label > threshold\n",
    "\n",
    "    TP = (prob & label).sum(1).float()\n",
    "    TN = ((~prob) & (~label)).sum(1).float()\n",
    "    FP = (prob & (~label)).sum(1).float()\n",
    "    FN = ((~prob) & label).sum(1).float()\n",
    "\n",
    "    precision = torch.mean(TP / (TP + FP + 1e-12))\n",
    "    recall = torch.mean(TP / (TP + FN + 1e-12))\n",
    "    F2 = (1 + beta**2) * precision * recall / (beta**2 * precision + recall + 1e-12)\n",
    "\n",
    "    return F2.mean(0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "e543cfef",
   "metadata": {},
   "outputs": [],
   "source": [
    "mixup = MixUp(0.3)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "cb6682d3",
   "metadata": {},
   "source": [
    "The original code from the article was causing my system to run out of memory, so I had to play around with the values in order to get it to run all of th way through. In the end, I eded up reducing the bactch sizes by quite a bit."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "26550c45",
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_data(fold=0, size=224, bs=16):\n",
    "    blocks = (ImageBlock, MultiCategoryBlock)\n",
    "    idx_splitter = IndexSplitter(TRAIN_DF[TRAIN_DF.fold == fold].index)\n",
    "    xtrans = aug_transforms(\n",
    "        size=size,\n",
    "        do_flip=True,\n",
    "        flip_vert=True,\n",
    "        max_rotate=180.0,\n",
    "        max_zoom=1.1,\n",
    "        max_lighting=0.6,\n",
    "        max_warp=0.2,\n",
    "        p_affine=0.75,\n",
    "        p_lighting=0.75,\n",
    "        xtra_tfms=[\n",
    "            RandomErasing(p=0.5, sh=0.1, min_aspect=0.2, max_count=2),\n",
    "        ],\n",
    "    )\n",
    "    db = DataBlock(\n",
    "        blocks=blocks,\n",
    "        get_x=ColReader(0),                   # image paths\n",
    "        get_y=ColReader(1, label_delim=\" \"),  # label\n",
    "        splitter=idx_splitter,\n",
    "        item_tfms=[FlipItem(p=0.5), Resize(512, method=\"pad\")],\n",
    "        batch_tfms=[*xtrans, Normalize],\n",
    "    )\n",
    "\n",
    "    return db.dataloaders(TRAIN_DF, bs=bs)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "bb7ccde5",
   "metadata": {},
   "source": [
    "> **NOTE:** The `label_delim` is very important here. Originally I had removed it because it seemed like a way to split a string of texts into seperate classes, which is what it's used for. But without it. the actualy name of your class is split into individual letters! Not what I wanted!"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "1759477c",
   "metadata": {},
   "source": [
    "The article wasn't saving the best model, so I decided to add that here, so that at then end, the best model would be loaded and used for the final predictions. \n",
    "\n",
    "I also decided to optimize the learning rate and it allowed me to improve my score on Kaggle to **0.97777**!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "c8e4d164",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "\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:not([value]), progress:not([value])::-webkit-progress-bar {\n",
       "        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n",
       "    }\n",
       "    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
       "        background: #F44336;\n",
       "    }\n",
       "</style>\n"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "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_multi</th>\n",
       "      <th>F_score</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.742832</td>\n",
       "      <td>0.332222</td>\n",
       "      <td>0.567699</td>\n",
       "      <td>0.850564</td>\n",
       "      <td>00:47</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.448280</td>\n",
       "      <td>0.140894</td>\n",
       "      <td>0.899287</td>\n",
       "      <td>0.932870</td>\n",
       "      <td>00:48</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.262894</td>\n",
       "      <td>0.064578</td>\n",
       "      <td>0.978985</td>\n",
       "      <td>0.918298</td>\n",
       "      <td>00:49</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.211160</td>\n",
       "      <td>0.055536</td>\n",
       "      <td>0.977822</td>\n",
       "      <td>0.929846</td>\n",
       "      <td>00:49</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.200247</td>\n",
       "      <td>0.050648</td>\n",
       "      <td>0.982330</td>\n",
       "      <td>0.927354</td>\n",
       "      <td>00:49</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.191389</td>\n",
       "      <td>0.049232</td>\n",
       "      <td>0.980803</td>\n",
       "      <td>0.937275</td>\n",
       "      <td>00:49</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.175267</td>\n",
       "      <td>0.045705</td>\n",
       "      <td>0.981239</td>\n",
       "      <td>0.947857</td>\n",
       "      <td>00:49</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>0.179084</td>\n",
       "      <td>0.046314</td>\n",
       "      <td>0.982257</td>\n",
       "      <td>0.943277</td>\n",
       "      <td>00:49</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "No improvement since epoch 6: early stopping\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "\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:not([value]), progress:not([value])::-webkit-progress-bar {\n",
       "        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n",
       "    }\n",
       "    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
       "        background: #F44336;\n",
       "    }\n",
       "</style>\n"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "lrs=SuggestedLRs(minimum=3.311311274956097e-07, steep=1.9054607491852948e-06, valley=0.0020892962347716093, slide=0.0030199517495930195)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "\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:not([value]), progress:not([value])::-webkit-progress-bar {\n",
       "        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n",
       "    }\n",
       "    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
       "        background: #F44336;\n",
       "    }\n",
       "</style>\n"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "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_multi</th>\n",
       "      <th>F_score</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.199999</td>\n",
       "      <td>0.052038</td>\n",
       "      <td>0.977094</td>\n",
       "      <td>0.938127</td>\n",
       "      <td>02:18</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Better model found at epoch 0 with accuracy_multi value: 0.9770942330360413.\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "\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:not([value]), progress:not([value])::-webkit-progress-bar {\n",
       "        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n",
       "    }\n",
       "    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
       "        background: #F44336;\n",
       "    }\n",
       "</style>\n"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "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_multi</th>\n",
       "      <th>F_score</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.187667</td>\n",
       "      <td>0.043779</td>\n",
       "      <td>0.982984</td>\n",
       "      <td>0.934983</td>\n",
       "      <td>06:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.168828</td>\n",
       "      <td>0.040531</td>\n",
       "      <td>0.983203</td>\n",
       "      <td>0.950143</td>\n",
       "      <td>06:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.173600</td>\n",
       "      <td>0.041492</td>\n",
       "      <td>0.982403</td>\n",
       "      <td>0.958544</td>\n",
       "      <td>06:07</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Better model found at epoch 0 with accuracy_multi value: 0.9829843044281006.\n",
      "Better model found at epoch 1 with accuracy_multi value: 0.9832028746604919.\n",
      "No improvement since epoch 1: early stopping\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "\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:not([value]), progress:not([value])::-webkit-progress-bar {\n",
       "        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n",
       "    }\n",
       "    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
       "        background: #F44336;\n",
       "    }\n",
       "</style>\n"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "\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:not([value]), progress:not([value])::-webkit-progress-bar {\n",
       "        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n",
       "    }\n",
       "    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
       "        background: #F44336;\n",
       "    }\n",
       "</style>\n"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "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_multi</th>\n",
       "      <th>F_score</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.749024</td>\n",
       "      <td>0.362973</td>\n",
       "      <td>0.512804</td>\n",
       "      <td>0.823005</td>\n",
       "      <td>00:49</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.459727</td>\n",
       "      <td>0.127862</td>\n",
       "      <td>0.911392</td>\n",
       "      <td>0.939230</td>\n",
       "      <td>00:49</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.268197</td>\n",
       "      <td>0.059542</td>\n",
       "      <td>0.978466</td>\n",
       "      <td>0.923698</td>\n",
       "      <td>00:49</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.210407</td>\n",
       "      <td>0.054505</td>\n",
       "      <td>0.979194</td>\n",
       "      <td>0.927755</td>\n",
       "      <td>00:49</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.198752</td>\n",
       "      <td>0.046700</td>\n",
       "      <td>0.982686</td>\n",
       "      <td>0.942269</td>\n",
       "      <td>00:49</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.187547</td>\n",
       "      <td>0.045462</td>\n",
       "      <td>0.980794</td>\n",
       "      <td>0.947401</td>\n",
       "      <td>00:49</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.175918</td>\n",
       "      <td>0.045117</td>\n",
       "      <td>0.981885</td>\n",
       "      <td>0.941614</td>\n",
       "      <td>00:49</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>0.171502</td>\n",
       "      <td>0.044164</td>\n",
       "      <td>0.981595</td>\n",
       "      <td>0.945109</td>\n",
       "      <td>00:49</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>0.175151</td>\n",
       "      <td>0.043920</td>\n",
       "      <td>0.982322</td>\n",
       "      <td>0.946418</td>\n",
       "      <td>00:49</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>0.173692</td>\n",
       "      <td>0.043385</td>\n",
       "      <td>0.982395</td>\n",
       "      <td>0.947838</td>\n",
       "      <td>00:49</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "\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:not([value]), progress:not([value])::-webkit-progress-bar {\n",
       "        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n",
       "    }\n",
       "    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
       "        background: #F44336;\n",
       "    }\n",
       "</style>\n"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "lrs=SuggestedLRs(minimum=7.585775847473997e-08, steep=9.12010818865383e-07, valley=0.0004786300996784121, slide=0.0020892962347716093)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "\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:not([value]), progress:not([value])::-webkit-progress-bar {\n",
       "        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n",
       "    }\n",
       "    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
       "        background: #F44336;\n",
       "    }\n",
       "</style>\n"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "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_multi</th>\n",
       "      <th>F_score</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.179669</td>\n",
       "      <td>0.053725</td>\n",
       "      <td>0.979631</td>\n",
       "      <td>0.929821</td>\n",
       "      <td>02:19</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Better model found at epoch 0 with accuracy_multi value: 0.9796305894851685.\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "\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:not([value]), progress:not([value])::-webkit-progress-bar {\n",
       "        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n",
       "    }\n",
       "    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
       "        background: #F44336;\n",
       "    }\n",
       "</style>\n"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "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_multi</th>\n",
       "      <th>F_score</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.188371</td>\n",
       "      <td>0.048567</td>\n",
       "      <td>0.979267</td>\n",
       "      <td>0.940504</td>\n",
       "      <td>06:07</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.184169</td>\n",
       "      <td>0.047059</td>\n",
       "      <td>0.981595</td>\n",
       "      <td>0.945106</td>\n",
       "      <td>06:07</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.173576</td>\n",
       "      <td>0.045270</td>\n",
       "      <td>0.982322</td>\n",
       "      <td>0.949471</td>\n",
       "      <td>06:07</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.166856</td>\n",
       "      <td>0.042848</td>\n",
       "      <td>0.983122</td>\n",
       "      <td>0.951099</td>\n",
       "      <td>06:07</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.171654</td>\n",
       "      <td>0.043523</td>\n",
       "      <td>0.984359</td>\n",
       "      <td>0.947614</td>\n",
       "      <td>06:08</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Better model found at epoch 0 with accuracy_multi value: 0.9792668223381042.\n",
      "Better model found at epoch 1 with accuracy_multi value: 0.9815947413444519.\n",
      "Better model found at epoch 2 with accuracy_multi value: 0.9823222160339355.\n",
      "Better model found at epoch 3 with accuracy_multi value: 0.9831223487854004.\n",
      "Better model found at epoch 4 with accuracy_multi value: 0.9843591451644897.\n",
      "No improvement since epoch 3: early stopping\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "\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:not([value]), progress:not([value])::-webkit-progress-bar {\n",
       "        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n",
       "    }\n",
       "    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
       "        background: #F44336;\n",
       "    }\n",
       "</style>\n"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "\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:not([value]), progress:not([value])::-webkit-progress-bar {\n",
       "        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n",
       "    }\n",
       "    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
       "        background: #F44336;\n",
       "    }\n",
       "</style>\n"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "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_multi</th>\n",
       "      <th>F_score</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.757520</td>\n",
       "      <td>0.312974</td>\n",
       "      <td>0.583115</td>\n",
       "      <td>0.850809</td>\n",
       "      <td>00:48</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.467646</td>\n",
       "      <td>0.124635</td>\n",
       "      <td>0.916667</td>\n",
       "      <td>0.933606</td>\n",
       "      <td>00:50</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.276081</td>\n",
       "      <td>0.057282</td>\n",
       "      <td>0.976222</td>\n",
       "      <td>0.936824</td>\n",
       "      <td>00:49</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.222573</td>\n",
       "      <td>0.049406</td>\n",
       "      <td>0.977894</td>\n",
       "      <td>0.938693</td>\n",
       "      <td>00:49</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.204121</td>\n",
       "      <td>0.049408</td>\n",
       "      <td>0.981748</td>\n",
       "      <td>0.939351</td>\n",
       "      <td>00:49</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "No improvement since epoch 3: early stopping\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "\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:not([value]), progress:not([value])::-webkit-progress-bar {\n",
       "        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n",
       "    }\n",
       "    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
       "        background: #F44336;\n",
       "    }\n",
       "</style>\n"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "lrs=SuggestedLRs(minimum=6.309573450380412e-08, steep=2.2908675418875646e-06, valley=0.0010000000474974513, slide=0.001737800776027143)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "\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:not([value]), progress:not([value])::-webkit-progress-bar {\n",
       "        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n",
       "    }\n",
       "    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
       "        background: #F44336;\n",
       "    }\n",
       "</style>\n"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "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_multi</th>\n",
       "      <th>F_score</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.203241</td>\n",
       "      <td>0.054311</td>\n",
       "      <td>0.978330</td>\n",
       "      <td>0.938666</td>\n",
       "      <td>02:18</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Better model found at epoch 0 with accuracy_multi value: 0.9783303737640381.\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "\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:not([value]), progress:not([value])::-webkit-progress-bar {\n",
       "        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n",
       "    }\n",
       "    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
       "        background: #F44336;\n",
       "    }\n",
       "</style>\n"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "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_multi</th>\n",
       "      <th>F_score</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.193026</td>\n",
       "      <td>0.046385</td>\n",
       "      <td>0.984366</td>\n",
       "      <td>0.944583</td>\n",
       "      <td>06:06</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.178738</td>\n",
       "      <td>0.041551</td>\n",
       "      <td>0.985602</td>\n",
       "      <td>0.946335</td>\n",
       "      <td>06:06</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.179688</td>\n",
       "      <td>0.040350</td>\n",
       "      <td>0.986257</td>\n",
       "      <td>0.955712</td>\n",
       "      <td>06:06</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.168384</td>\n",
       "      <td>0.040447</td>\n",
       "      <td>0.984802</td>\n",
       "      <td>0.957893</td>\n",
       "      <td>06:06</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Better model found at epoch 0 with accuracy_multi value: 0.9843660593032837.\n",
      "Better model found at epoch 1 with accuracy_multi value: 0.9856023192405701.\n",
      "Better model found at epoch 2 with accuracy_multi value: 0.9862567782402039.\n",
      "No improvement since epoch 2: early stopping\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "\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:not([value]), progress:not([value])::-webkit-progress-bar {\n",
       "        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n",
       "    }\n",
       "    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
       "        background: #F44336;\n",
       "    }\n",
       "</style>\n"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "all_preds = []\n",
    "\n",
    "for i in range(3):\n",
    "    dls = get_data(i, 256, 32)\n",
    "    learn = vision_learner(\n",
    "        dls,\n",
    "        \"convnext_nano\",\n",
    "        metrics=[\n",
    "            partial(accuracy_multi, thresh=0.2),\n",
    "            partial(F_score, threshold=0.2),\n",
    "        ],\n",
    "        cbs=mixup,\n",
    "    )  # .to_fp16()\n",
    "\n",
    "    learn.fit_one_cycle(10, cbs=EarlyStoppingCallback(monitor=\"valid_loss\"))\n",
    "    learn.dls = get_data(i, 512, 16)\n",
    "\n",
    "    lrs = learn.lr_find(suggest_funcs=(minimum, steep, valley, slide))\n",
    "    print(f\"{lrs=}\")\n",
    "\n",
    "    learn.fine_tune(10, lrs.valley, cbs=[\n",
    "        EarlyStoppingCallback(monitor=\"valid_loss\"),\n",
    "        SaveModelCallback(\n",
    "            monitor=\"accuracy_multi\",\n",
    "            min_delta=0.0001,\n",
    "            fname=\"best-cv_model\",\n",
    "        ),\n",
    "    ])\n",
    "    m = f\"{i}-best.pkl\"\n",
    "    learn.export(m)\n",
    "    tst_dl = learn.dls.test_dl(TEST_DF)\n",
    "    preds, _ = learn.get_preds(dl=tst_dl)\n",
    "    all_preds.append(preds)\n",
    "\n",
    "    # free up memory\n",
    "    del learn\n",
    "    torch.cuda.empty_cache()\n",
    "    gc.collect()"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "7bec66e9",
   "metadata": {},
   "source": [
    "Unlike the article, we are not trying to classify multiple objects in a single image, so I am not using the threshold variable. Instead I'm taking the class with the highest prediction."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "914847ec",
   "metadata": {},
   "outputs": [],
   "source": [
    "preds = np.mean(np.stack(all_preds), axis=0)\n",
    "\n",
    "k = dls.vocab\n",
    "preds[0]\n",
    "labelled_preds = [\n",
    "    \" \".join([k[i] for i, p in enumerate(pred) if p == max(pred)]) for pred in preds\n",
    "]\n",
    "TEST_DF[\"label\"] = labelled_preds\n",
    "TEST_DF.drop(['image'], axis=1, inplace=True)\n",
    "TEST_DF.to_csv(\"submission.csv\", index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "10c2bbad",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Id,label\n",
      "678,spoon\n",
      "3962,knife\n",
      "9271,fork\n",
      "5133,plate\n",
      "8842,fork\n",
      "8256,fork\n",
      "4622,spoon\n",
      "2775,fork\n",
      "5900,knife\n"
     ]
    }
   ],
   "source": [
    "!head -n 10 submission.csv"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "623ec9e7",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "py39",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.15 | packaged by conda-forge | (main, Nov 22 2022, 15:55:03) \n[GCC 10.4.0]"
  },
  "vscode": {
   "interpreter": {
    "hash": "e02e9884b3f6679bcf8253a6deab40bf4cc99ad8d461d7e4c2e9bca419f3f347"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}