test_acc10_baseline.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import keras\n",
    "from keras.models import load_model\n",
    "import numpy as np\n",
    "from tqdm import tqdm"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "model_path = './models/basic_c_and_n_regression_hydrophobic_only.h5'\n",
    "\n",
    "df = pd.read_csv('./data/csv/test_acc10_300.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "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>lig_id</th>\n",
       "      <th>pro_id</th>\n",
       "      <th>dest</th>\n",
       "      <th>score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2701</td>\n",
       "      <td>2701</td>\n",
       "      <td>./processed_data/test_acc10/2701_pro_2701_lig.npy</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2701</td>\n",
       "      <td>2702</td>\n",
       "      <td>./processed_data/test_acc10/2702_pro_2701_lig.npy</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2701</td>\n",
       "      <td>2703</td>\n",
       "      <td>./processed_data/test_acc10/2703_pro_2701_lig.npy</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2701</td>\n",
       "      <td>2704</td>\n",
       "      <td>./processed_data/test_acc10/2704_pro_2701_lig.npy</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2701</td>\n",
       "      <td>2705</td>\n",
       "      <td>./processed_data/test_acc10/2705_pro_2701_lig.npy</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   lig_id  pro_id                                               dest  score\n",
       "0    2701    2701  ./processed_data/test_acc10/2701_pro_2701_lig.npy      1\n",
       "1    2701    2702  ./processed_data/test_acc10/2702_pro_2701_lig.npy      0\n",
       "2    2701    2703  ./processed_data/test_acc10/2703_pro_2701_lig.npy      0\n",
       "3    2701    2704  ./processed_data/test_acc10/2704_pro_2701_lig.npy      0\n",
       "4    2701    2705  ./processed_data/test_acc10/2705_pro_2701_lig.npy      0"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# group by ligand\n",
    "# for each group, compute top10 accuracy manually\n",
    "def test_model(model_path):\n",
    "    all_largest_ids = []\n",
    "    all_match_ids = []\n",
    "    model = load_model(model_path)\n",
    "    matches = 0\n",
    "    total = float(len(df.lig_id.unique()))\n",
    "    debug = False\n",
    "\n",
    "    for lig_id, grp in tqdm(df.groupby('lig_id')):\n",
    "        grp.reset_index(inplace=True)\n",
    "\n",
    "        match_id = grp.index[grp['pro_id']==lig_id]\n",
    "\n",
    "        # load all features\n",
    "        n_channels = 2\n",
    "        dims = (24,24,24)\n",
    "        X = np.empty((len(grp), *dims, n_channels))\n",
    "        for row in grp.itertuples():\n",
    "            X[row[0],] = np.load(row.dest)\n",
    "\n",
    "        probs = model.predict(X, batch_size=200)\n",
    "        probs = probs.flatten()\n",
    "\n",
    "        largest_first = list(reversed(np.argsort(probs).tolist()))\n",
    "        top10 = largest_first[:10]\n",
    "\n",
    "        if debug:\n",
    "            print('ligid: ', lig_id)\n",
    "            print('match_id: ', match_id)\n",
    "            print('probs: ', probs)\n",
    "            print('top_10: ', top10)\n",
    "            print()\n",
    "            break\n",
    "\n",
    "        if match_id in top10:\n",
    "            matches += 1\n",
    "        all_match_ids.append(match_id)\n",
    "        all_largest_ids.append(largest_first)\n",
    "\n",
    "    print(matches/total)    \n",
    "    return all_match_ids, all_largest_ids"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "  0%|          | 0/300 [00:00<?, ?it/s]\u001b[A\n",
      "  0%|          | 1/300 [00:01<06:21,  1.27s/it]\u001b[A\n",
      "  1%|          | 2/300 [00:04<08:41,  1.75s/it]\u001b[A\n",
      "  1%|          | 3/300 [00:07<11:36,  2.35s/it]\u001b[A\n",
      "  1%|▏         | 4/300 [00:11<13:53,  2.82s/it]\u001b[A\n",
      "  2%|▏         | 5/300 [00:15<15:14,  3.10s/it]\u001b[A\n",
      "  2%|▏         | 6/300 [00:19<16:14,  3.31s/it]\u001b[A\n",
      "  2%|▏         | 7/300 [00:23<16:56,  3.47s/it]\u001b[A\n",
      "  3%|▎         | 8/300 [00:26<17:18,  3.56s/it]\u001b[A\n",
      "  3%|▎         | 9/300 [00:30<17:35,  3.63s/it]\u001b[A\n",
      "  3%|▎         | 10/300 [00:34<17:37,  3.65s/it]\u001b[A\n",
      "  4%|▎         | 11/300 [00:38<17:47,  3.69s/it]\u001b[A\n",
      "  4%|▍         | 12/300 [00:42<18:01,  3.75s/it]\u001b[A\n",
      "  4%|▍         | 13/300 [00:46<18:07,  3.79s/it]\u001b[A\n",
      "  5%|▍         | 14/300 [00:49<17:57,  3.77s/it]\u001b[A\n",
      "  5%|▌         | 15/300 [00:53<18:00,  3.79s/it]\u001b[A\n",
      "  5%|▌         | 16/300 [00:57<17:51,  3.77s/it]\u001b[A\n",
      "  6%|▌         | 17/300 [01:00<17:41,  3.75s/it]\u001b[A\n",
      "  6%|▌         | 18/300 [01:04<17:38,  3.75s/it]\u001b[A\n",
      "  6%|▋         | 19/300 [01:08<17:29,  3.73s/it]\u001b[A\n",
      "  7%|▋         | 20/300 [01:12<17:27,  3.74s/it]\u001b[A\n",
      "  7%|▋         | 21/300 [01:15<17:16,  3.72s/it]\u001b[A\n",
      "  7%|▋         | 22/300 [01:19<17:09,  3.70s/it]\u001b[A\n",
      "  8%|▊         | 23/300 [01:23<17:01,  3.69s/it]\u001b[A\n",
      "  8%|▊         | 24/300 [01:26<17:05,  3.71s/it]\u001b[A\n",
      "  8%|▊         | 25/300 [01:30<16:57,  3.70s/it]\u001b[A\n",
      "  9%|▊         | 26/300 [01:34<17:00,  3.72s/it]\u001b[A\n",
      "  9%|▉         | 27/300 [01:38<17:05,  3.76s/it]\u001b[A\n",
      "  9%|▉         | 28/300 [01:41<16:58,  3.74s/it]\u001b[A\n",
      " 10%|▉         | 29/300 [01:45<17:00,  3.77s/it]\u001b[A\n",
      " 10%|█         | 30/300 [01:49<16:57,  3.77s/it]\u001b[A\n",
      " 10%|█         | 31/300 [01:53<16:54,  3.77s/it]\u001b[A\n",
      " 11%|█         | 32/300 [01:57<16:54,  3.78s/it]\u001b[A\n",
      " 11%|█         | 33/300 [02:01<17:45,  3.99s/it]\u001b[A\n",
      " 11%|█▏        | 34/300 [02:05<17:25,  3.93s/it]\u001b[A\n",
      " 12%|█▏        | 35/300 [02:09<17:02,  3.86s/it]\u001b[A\n",
      " 12%|█▏        | 36/300 [02:13<17:06,  3.89s/it]\u001b[A\n",
      " 12%|█▏        | 37/300 [02:16<16:53,  3.86s/it]\u001b[A\n",
      " 13%|█▎        | 38/300 [02:20<16:41,  3.82s/it]\u001b[A\n",
      " 13%|█▎        | 39/300 [02:24<16:36,  3.82s/it]\u001b[A\n",
      " 13%|█▎        | 40/300 [02:28<16:31,  3.82s/it]\u001b[A\n",
      " 14%|█▎        | 41/300 [02:31<16:24,  3.80s/it]\u001b[A\n",
      " 14%|█▍        | 42/300 [02:35<16:21,  3.80s/it]\u001b[A\n",
      " 14%|█▍        | 43/300 [02:39<16:12,  3.78s/it]\u001b[A\n",
      " 15%|█▍        | 44/300 [02:43<16:13,  3.80s/it]\u001b[A\n",
      " 15%|█▌        | 45/300 [02:47<16:02,  3.77s/it]\u001b[A\n",
      " 15%|█▌        | 46/300 [02:50<16:01,  3.78s/it]\u001b[A\n",
      " 16%|█▌        | 47/300 [02:54<15:56,  3.78s/it]\u001b[A\n",
      " 16%|█▌        | 48/300 [02:58<16:00,  3.81s/it]\u001b[A\n",
      " 16%|█▋        | 49/300 [03:02<16:03,  3.84s/it]\u001b[A\n",
      " 17%|█▋        | 50/300 [03:06<16:06,  3.87s/it]\u001b[A\n",
      " 17%|█▋        | 51/300 [03:10<16:02,  3.87s/it]\u001b[A\n",
      " 17%|█▋        | 52/300 [03:14<15:57,  3.86s/it]\u001b[A\n",
      " 18%|█▊        | 53/300 [03:17<15:52,  3.86s/it]\u001b[A\n",
      " 18%|█▊        | 54/300 [03:21<15:44,  3.84s/it]\u001b[A\n",
      " 18%|█▊        | 55/300 [03:25<15:46,  3.86s/it]\u001b[A\n",
      " 19%|█▊        | 56/300 [03:29<15:40,  3.86s/it]\u001b[A\n",
      " 19%|█▉        | 57/300 [03:33<15:37,  3.86s/it]\u001b[A\n",
      " 19%|█▉        | 58/300 [03:37<15:24,  3.82s/it]\u001b[A\n",
      " 20%|█▉        | 59/300 [03:40<15:26,  3.84s/it]\u001b[A\n",
      " 20%|██        | 60/300 [03:44<15:27,  3.86s/it]\u001b[A\n",
      " 20%|██        | 61/300 [03:48<15:27,  3.88s/it]\u001b[A\n",
      " 21%|██        | 62/300 [03:52<15:19,  3.86s/it]\u001b[A\n",
      " 21%|██        | 63/300 [03:56<15:13,  3.86s/it]\u001b[A\n",
      " 21%|██▏       | 64/300 [04:00<15:10,  3.86s/it]\u001b[A\n",
      " 22%|██▏       | 65/300 [04:04<15:04,  3.85s/it]\u001b[A\n",
      " 22%|██▏       | 66/300 [04:08<15:02,  3.86s/it]\u001b[A\n",
      " 22%|██▏       | 67/300 [04:11<14:59,  3.86s/it]\u001b[A\n",
      " 23%|██▎       | 68/300 [04:15<14:50,  3.84s/it]\u001b[A\n",
      " 23%|██▎       | 69/300 [04:19<14:48,  3.85s/it]\u001b[A\n",
      " 23%|██▎       | 70/300 [04:23<14:46,  3.86s/it]\u001b[A\n",
      " 24%|██▎       | 71/300 [04:27<14:40,  3.85s/it]\u001b[A\n",
      " 24%|██▍       | 72/300 [04:31<14:31,  3.82s/it]\u001b[A\n",
      " 24%|██▍       | 73/300 [04:34<14:33,  3.85s/it]\u001b[A\n",
      " 25%|██▍       | 74/300 [04:38<14:24,  3.83s/it]\u001b[A\n",
      " 25%|██▌       | 75/300 [04:42<14:20,  3.83s/it]\u001b[A\n",
      " 25%|██▌       | 76/300 [04:46<14:12,  3.81s/it]\u001b[A\n",
      " 26%|██▌       | 77/300 [04:49<14:02,  3.78s/it]\u001b[A\n",
      " 26%|██▌       | 78/300 [04:53<14:07,  3.82s/it]\u001b[A\n",
      " 26%|██▋       | 79/300 [04:57<14:03,  3.82s/it]\u001b[A\n",
      " 27%|██▋       | 80/300 [05:01<13:56,  3.80s/it]\u001b[A\n",
      " 27%|██▋       | 81/300 [05:05<13:51,  3.80s/it]\u001b[A\n",
      " 27%|██▋       | 82/300 [05:09<13:46,  3.79s/it]\u001b[A\n",
      " 28%|██▊       | 83/300 [05:12<13:45,  3.81s/it]\u001b[A\n",
      " 28%|██▊       | 84/300 [05:16<13:41,  3.80s/it]\u001b[A\n",
      " 28%|██▊       | 85/300 [05:20<13:41,  3.82s/it]\u001b[A\n",
      " 29%|██▊       | 86/300 [05:24<13:41,  3.84s/it]\u001b[A\n",
      " 29%|██▉       | 87/300 [05:28<13:38,  3.84s/it]\u001b[A\n",
      " 29%|██▉       | 88/300 [05:31<13:23,  3.79s/it]\u001b[A\n",
      " 30%|██▉       | 89/300 [05:35<13:22,  3.80s/it]\u001b[A\n",
      " 30%|███       | 90/300 [05:39<13:21,  3.82s/it]\u001b[A\n",
      " 30%|███       | 91/300 [05:43<13:24,  3.85s/it]\u001b[A\n",
      " 31%|███       | 92/300 [05:47<13:17,  3.83s/it]\u001b[A\n",
      " 31%|███       | 93/300 [05:51<13:12,  3.83s/it]\u001b[A\n",
      " 31%|███▏      | 94/300 [05:55<13:10,  3.84s/it]\u001b[A\n",
      " 32%|███▏      | 95/300 [05:59<13:14,  3.88s/it]\u001b[A\n",
      " 32%|███▏      | 96/300 [06:02<13:08,  3.86s/it]\u001b[A\n",
      " 32%|███▏      | 97/300 [06:06<13:07,  3.88s/it]\u001b[A\n",
      " 33%|███▎      | 98/300 [06:10<12:56,  3.84s/it]\u001b[A\n",
      " 33%|███▎      | 99/300 [06:14<12:50,  3.83s/it]\u001b[A\n",
      " 33%|███▎      | 100/300 [06:18<12:49,  3.85s/it]\u001b[A\n",
      " 34%|███▎      | 101/300 [06:22<12:44,  3.84s/it]\u001b[A\n",
      " 34%|███▍      | 102/300 [06:25<12:39,  3.84s/it]\u001b[A\n",
      " 34%|███▍      | 103/300 [06:29<12:28,  3.80s/it]\u001b[A\n",
      " 35%|███▍      | 104/300 [06:33<12:23,  3.79s/it]\u001b[A\n",
      " 35%|███▌      | 105/300 [06:37<12:24,  3.82s/it]\u001b[A\n",
      " 35%|███▌      | 106/300 [06:41<12:19,  3.81s/it]\u001b[A\n",
      " 36%|███▌      | 107/300 [06:44<12:16,  3.82s/it]\u001b[A\n",
      " 36%|███▌      | 108/300 [06:48<12:08,  3.79s/it]\u001b[A\n",
      " 36%|███▋      | 109/300 [06:52<12:10,  3.82s/it]\u001b[A\n",
      " 37%|███▋      | 110/300 [06:56<12:05,  3.82s/it]\u001b[A\n",
      " 37%|███▋      | 111/300 [07:00<12:02,  3.82s/it]\u001b[A\n",
      " 37%|███▋      | 112/300 [07:04<12:04,  3.85s/it]\u001b[A\n",
      " 38%|███▊      | 113/300 [07:07<12:01,  3.86s/it]\u001b[A\n",
      " 38%|███▊      | 114/300 [07:11<11:56,  3.85s/it]\u001b[A\n",
      " 38%|███▊      | 115/300 [07:15<11:45,  3.81s/it]\u001b[A\n",
      " 39%|███▊      | 116/300 [07:19<11:49,  3.86s/it]\u001b[A\n",
      " 39%|███▉      | 117/300 [07:23<11:45,  3.85s/it]\u001b[A\n",
      " 39%|███▉      | 118/300 [07:27<11:41,  3.85s/it]\u001b[A\n",
      " 40%|███▉      | 119/300 [07:30<11:36,  3.85s/it]\u001b[A\n",
      " 40%|████      | 120/300 [07:34<11:31,  3.84s/it]\u001b[A\n",
      " 40%|████      | 121/300 [07:38<11:31,  3.86s/it]\u001b[A\n",
      " 41%|████      | 122/300 [07:42<11:27,  3.86s/it]\u001b[A\n",
      " 41%|████      | 123/300 [07:46<11:29,  3.89s/it]\u001b[A\n",
      " 41%|████▏     | 124/300 [07:50<11:26,  3.90s/it]\u001b[A\n",
      " 42%|████▏     | 125/300 [07:54<11:20,  3.89s/it]\u001b[A\n",
      " 42%|████▏     | 126/300 [07:58<11:18,  3.90s/it]\u001b[A\n",
      " 42%|████▏     | 127/300 [08:02<11:16,  3.91s/it]\u001b[A\n",
      " 43%|████▎     | 128/300 [08:06<11:15,  3.93s/it]\u001b[A\n",
      " 43%|████▎     | 129/300 [08:10<11:10,  3.92s/it]\u001b[A\n",
      " 43%|████▎     | 130/300 [08:14<11:08,  3.93s/it]\u001b[A\n",
      " 44%|████▎     | 131/300 [08:17<11:01,  3.91s/it]\u001b[A\n",
      " 44%|████▍     | 132/300 [08:21<10:53,  3.89s/it]\u001b[A\n",
      " 44%|████▍     | 133/300 [08:25<10:43,  3.85s/it]\u001b[A\n",
      " 45%|████▍     | 134/300 [08:29<10:38,  3.85s/it]\u001b[A\n",
      " 45%|████▌     | 135/300 [08:33<10:35,  3.85s/it]\u001b[A\n",
      " 45%|████▌     | 136/300 [08:37<10:33,  3.87s/it]\u001b[A\n",
      " 46%|████▌     | 137/300 [08:40<10:27,  3.85s/it]\u001b[A\n",
      " 46%|████▌     | 138/300 [08:44<10:18,  3.82s/it]\u001b[A\n",
      " 46%|████▋     | 139/300 [08:48<10:15,  3.83s/it]\u001b[A\n",
      " 47%|████▋     | 140/300 [08:52<10:06,  3.79s/it]\u001b[A\n",
      " 47%|████▋     | 141/300 [08:55<10:01,  3.78s/it]\u001b[A\n",
      " 47%|████▋     | 142/300 [08:59<09:56,  3.77s/it]\u001b[A\n",
      " 48%|████▊     | 143/300 [09:03<09:59,  3.82s/it]\u001b[A\n",
      " 48%|████▊     | 144/300 [09:07<09:55,  3.82s/it]\u001b[A\n",
      " 48%|████▊     | 145/300 [09:11<09:53,  3.83s/it]\u001b[A\n",
      " 49%|████▊     | 146/300 [09:15<09:49,  3.83s/it]\u001b[A\n",
      " 49%|████▉     | 147/300 [09:19<09:49,  3.85s/it]\u001b[A\n",
      " 49%|████▉     | 148/300 [09:22<09:43,  3.84s/it]\u001b[A\n",
      " 50%|████▉     | 149/300 [09:26<09:35,  3.81s/it]\u001b[A\n",
      " 50%|█████     | 150/300 [09:30<09:34,  3.83s/it]\u001b[A\n",
      " 50%|█████     | 151/300 [09:34<09:26,  3.80s/it]\u001b[A\n",
      " 51%|█████     | 152/300 [09:37<09:22,  3.80s/it]\u001b[A\n",
      " 51%|█████     | 153/300 [09:41<09:20,  3.81s/it]\u001b[A\n",
      " 51%|█████▏    | 154/300 [09:45<09:12,  3.78s/it]\u001b[A\n",
      " 52%|█████▏    | 155/300 [09:49<09:10,  3.80s/it]\u001b[A\n",
      " 52%|█████▏    | 156/300 [09:53<09:06,  3.80s/it]\u001b[A\n",
      " 52%|█████▏    | 157/300 [09:57<09:04,  3.81s/it]\u001b[A\n",
      " 53%|█████▎    | 158/300 [10:00<09:00,  3.81s/it]\u001b[A\n",
      " 53%|█████▎    | 159/300 [10:04<08:55,  3.80s/it]\u001b[A\n",
      " 53%|█████▎    | 160/300 [10:08<08:53,  3.81s/it]\u001b[A\n",
      " 54%|█████▎    | 161/300 [10:12<08:52,  3.83s/it]\u001b[A\n",
      " 54%|█████▍    | 162/300 [10:16<08:48,  3.83s/it]\u001b[A\n",
      " 54%|█████▍    | 163/300 [10:19<08:45,  3.83s/it]\u001b[A\n",
      " 55%|█████▍    | 164/300 [10:23<08:38,  3.81s/it]\u001b[A\n",
      " 55%|█████▌    | 165/300 [10:27<08:34,  3.81s/it]\u001b[A\n",
      " 55%|█████▌    | 166/300 [10:31<08:24,  3.76s/it]\u001b[A\n",
      " 56%|█████▌    | 167/300 [10:34<08:22,  3.78s/it]\u001b[A\n",
      " 56%|█████▌    | 168/300 [10:38<08:17,  3.77s/it]\u001b[A\n",
      " 56%|█████▋    | 169/300 [10:42<08:13,  3.77s/it]\u001b[A\n",
      " 57%|█████▋    | 170/300 [10:46<08:11,  3.78s/it]\u001b[A\n",
      " 57%|█████▋    | 171/300 [10:50<08:09,  3.80s/it]\u001b[A\n",
      " 57%|█████▋    | 172/300 [10:53<08:06,  3.80s/it]\u001b[A\n",
      " 58%|█████▊    | 173/300 [10:57<08:02,  3.80s/it]\u001b[A\n",
      " 58%|█████▊    | 174/300 [11:01<07:59,  3.80s/it]\u001b[A\n",
      " 58%|█████▊    | 175/300 [11:05<07:54,  3.79s/it]\u001b[A\n",
      " 59%|█████▊    | 176/300 [11:09<07:52,  3.81s/it]\u001b[A\n",
      " 59%|█████▉    | 177/300 [11:12<07:47,  3.80s/it]\u001b[A\n",
      " 59%|█████▉    | 178/300 [11:16<07:46,  3.82s/it]\u001b[A\n",
      " 60%|█████▉    | 179/300 [11:20<07:39,  3.80s/it]\u001b[A\n",
      " 60%|██████    | 180/300 [11:24<07:35,  3.80s/it]\u001b[A\n",
      " 60%|██████    | 181/300 [11:28<07:31,  3.79s/it]\u001b[A\n",
      " 61%|██████    | 182/300 [11:31<07:26,  3.79s/it]\u001b[A\n",
      " 61%|██████    | 183/300 [11:35<07:25,  3.81s/it]\u001b[A\n",
      " 61%|██████▏   | 184/300 [11:39<07:22,  3.81s/it]\u001b[A\n",
      " 62%|██████▏   | 185/300 [11:43<07:20,  3.83s/it]\u001b[A\n",
      " 62%|██████▏   | 186/300 [11:47<07:18,  3.84s/it]\u001b[A\n",
      " 62%|██████▏   | 187/300 [11:51<07:11,  3.82s/it]\u001b[A\n",
      " 63%|██████▎   | 188/300 [11:54<07:07,  3.82s/it]\u001b[A\n",
      " 63%|██████▎   | 189/300 [11:58<07:04,  3.82s/it]\u001b[A\n",
      " 63%|██████▎   | 190/300 [12:03<07:22,  4.02s/it]\u001b[A\n",
      " 64%|██████▎   | 191/300 [12:07<07:10,  3.95s/it]\u001b[A\n",
      " 64%|██████▍   | 192/300 [12:10<07:03,  3.92s/it]\u001b[A\n",
      " 64%|██████▍   | 193/300 [12:14<06:57,  3.90s/it]\u001b[A\n",
      " 65%|██████▍   | 194/300 [12:18<06:52,  3.89s/it]\u001b[A\n",
      " 65%|██████▌   | 195/300 [12:22<06:47,  3.88s/it]\u001b[A\n",
      " 65%|██████▌   | 196/300 [12:26<06:42,  3.87s/it]\u001b[A\n",
      " 66%|██████▌   | 197/300 [12:30<06:37,  3.86s/it]\u001b[A\n",
      " 66%|██████▌   | 198/300 [12:34<06:36,  3.88s/it]\u001b[A\n",
      " 66%|██████▋   | 199/300 [12:38<06:35,  3.91s/it]\u001b[A\n",
      " 67%|██████▋   | 200/300 [12:41<06:28,  3.88s/it]\u001b[A\n",
      " 67%|██████▋   | 201/300 [12:45<06:26,  3.90s/it]\u001b[A\n",
      " 67%|██████▋   | 202/300 [12:49<06:22,  3.90s/it]\u001b[A\n",
      " 68%|██████▊   | 203/300 [12:53<06:18,  3.90s/it]\u001b[A\n",
      " 68%|██████▊   | 204/300 [12:57<06:12,  3.88s/it]\u001b[A\n",
      " 68%|██████▊   | 205/300 [13:01<06:06,  3.86s/it]\u001b[A\n",
      " 69%|██████▊   | 206/300 [13:05<06:02,  3.86s/it]\u001b[A\n",
      " 69%|██████▉   | 207/300 [13:08<05:59,  3.86s/it]\u001b[A\n",
      " 69%|██████▉   | 208/300 [13:12<05:55,  3.86s/it]\u001b[A\n",
      " 70%|██████▉   | 209/300 [13:16<05:52,  3.88s/it]\u001b[A\n",
      " 70%|███████   | 210/300 [13:20<05:47,  3.86s/it]\u001b[A\n",
      " 70%|███████   | 211/300 [13:24<05:42,  3.85s/it]\u001b[A\n",
      " 71%|███████   | 212/300 [13:28<05:38,  3.85s/it]\u001b[A\n",
      " 71%|███████   | 213/300 [13:32<05:35,  3.85s/it]\u001b[A\n",
      " 71%|███████▏  | 214/300 [13:35<05:32,  3.86s/it]\u001b[A\n",
      " 72%|███████▏  | 215/300 [13:39<05:28,  3.87s/it]\u001b[A\n",
      " 72%|███████▏  | 216/300 [13:43<05:28,  3.91s/it]\u001b[A\n",
      " 72%|███████▏  | 217/300 [13:47<05:23,  3.90s/it]\u001b[A\n",
      " 73%|███████▎  | 218/300 [13:51<05:21,  3.93s/it]\u001b[A\n",
      " 73%|███████▎  | 219/300 [13:55<05:16,  3.90s/it]\u001b[A\n",
      " 73%|███████▎  | 220/300 [13:59<05:12,  3.91s/it]\u001b[A\n",
      " 74%|███████▎  | 221/300 [14:03<05:07,  3.89s/it]\u001b[A\n",
      " 74%|███████▍  | 222/300 [14:07<05:03,  3.89s/it]\u001b[A\n",
      " 74%|███████▍  | 223/300 [14:11<05:01,  3.92s/it]\u001b[A\n",
      " 75%|███████▍  | 224/300 [14:15<04:54,  3.87s/it]\u001b[A\n",
      " 75%|███████▌  | 225/300 [14:18<04:53,  3.91s/it]\u001b[A\n",
      " 75%|███████▌  | 226/300 [14:22<04:47,  3.89s/it]\u001b[A\n",
      " 76%|███████▌  | 227/300 [14:26<04:42,  3.88s/it]\u001b[A\n",
      " 76%|███████▌  | 228/300 [14:30<04:39,  3.88s/it]\u001b[A\n",
      " 76%|███████▋  | 229/300 [14:34<04:35,  3.88s/it]\u001b[A\n",
      " 77%|███████▋  | 230/300 [14:38<04:32,  3.89s/it]\u001b[A\n",
      " 77%|███████▋  | 231/300 [14:42<04:30,  3.92s/it]\u001b[A\n",
      " 77%|███████▋  | 232/300 [14:46<04:25,  3.91s/it]\u001b[A\n",
      " 78%|███████▊  | 233/300 [14:50<04:21,  3.90s/it]\u001b[A\n",
      " 78%|███████▊  | 234/300 [14:54<04:17,  3.90s/it]\u001b[A\n",
      " 78%|███████▊  | 235/300 [14:57<04:12,  3.88s/it]\u001b[A\n",
      " 79%|███████▊  | 236/300 [15:01<04:07,  3.87s/it]\u001b[A\n",
      " 79%|███████▉  | 237/300 [15:05<04:03,  3.86s/it]\u001b[A\n",
      " 79%|███████▉  | 238/300 [15:10<04:21,  4.22s/it]\u001b[A\n",
      " 80%|███████▉  | 239/300 [15:14<04:08,  4.08s/it]\u001b[A\n",
      " 80%|████████  | 240/300 [15:18<03:59,  3.99s/it]\u001b[A\n",
      " 80%|████████  | 241/300 [15:21<03:51,  3.92s/it]\u001b[A\n",
      " 81%|████████  | 242/300 [15:25<03:49,  3.96s/it]\u001b[A\n",
      " 81%|████████  | 243/300 [15:29<03:42,  3.90s/it]\u001b[A\n",
      " 81%|████████▏ | 244/300 [15:33<03:36,  3.86s/it]\u001b[A\n",
      " 82%|████████▏ | 245/300 [15:37<03:33,  3.88s/it]\u001b[A\n",
      " 82%|████████▏ | 246/300 [15:41<03:30,  3.90s/it]\u001b[A\n",
      " 82%|████████▏ | 247/300 [15:45<03:26,  3.90s/it]\u001b[A\n",
      " 83%|████████▎ | 248/300 [15:49<03:22,  3.90s/it]\u001b[A\n",
      " 83%|████████▎ | 249/300 [15:52<03:15,  3.84s/it]\u001b[A\n",
      " 83%|████████▎ | 250/300 [15:56<03:11,  3.83s/it]\u001b[A\n",
      " 84%|████████▎ | 251/300 [16:00<03:06,  3.81s/it]\u001b[A\n",
      " 84%|████████▍ | 252/300 [16:04<03:03,  3.82s/it]\u001b[A\n",
      " 84%|████████▍ | 253/300 [16:08<02:59,  3.81s/it]\u001b[A\n",
      " 85%|████████▍ | 254/300 [16:11<02:54,  3.78s/it]\u001b[A\n",
      " 85%|████████▌ | 255/300 [16:15<02:50,  3.80s/it]\u001b[A\n",
      " 85%|████████▌ | 256/300 [16:19<02:49,  3.85s/it]\u001b[A\n",
      " 86%|████████▌ | 257/300 [16:23<02:45,  3.85s/it]\u001b[A\n",
      " 86%|████████▌ | 258/300 [16:27<02:41,  3.84s/it]\u001b[A\n",
      " 86%|████████▋ | 259/300 [16:31<02:37,  3.84s/it]\u001b[A\n",
      " 87%|████████▋ | 260/300 [16:34<02:33,  3.85s/it]\u001b[A\n",
      " 87%|████████▋ | 261/300 [16:38<02:29,  3.85s/it]\u001b[A\n",
      " 87%|████████▋ | 262/300 [16:42<02:26,  3.84s/it]\u001b[A\n",
      " 88%|████████▊ | 263/300 [16:46<02:21,  3.83s/it]\u001b[A\n",
      " 88%|████████▊ | 264/300 [16:50<02:18,  3.85s/it]\u001b[A\n",
      " 88%|████████▊ | 265/300 [16:54<02:14,  3.83s/it]\u001b[A\n",
      " 89%|████████▊ | 266/300 [16:58<02:11,  3.86s/it]\u001b[A\n",
      " 89%|████████▉ | 267/300 [17:01<02:06,  3.83s/it]\u001b[A\n",
      " 89%|████████▉ | 268/300 [17:05<02:02,  3.84s/it]\u001b[A\n",
      " 90%|████████▉ | 269/300 [17:09<01:59,  3.84s/it]\u001b[A\n",
      " 90%|█████████ | 270/300 [17:13<01:54,  3.82s/it]\u001b[A\n",
      " 90%|█████████ | 271/300 [17:17<01:50,  3.80s/it]\u001b[A\n",
      " 91%|█████████ | 272/300 [17:20<01:46,  3.82s/it]\u001b[A\n",
      " 91%|█████████ | 273/300 [17:24<01:43,  3.82s/it]\u001b[A\n",
      " 91%|█████████▏| 274/300 [17:28<01:39,  3.82s/it]\u001b[A\n",
      " 92%|█████████▏| 275/300 [17:32<01:35,  3.84s/it]\u001b[A\n",
      " 92%|█████████▏| 276/300 [17:36<01:31,  3.83s/it]\u001b[A\n",
      " 92%|█████████▏| 277/300 [17:39<01:27,  3.82s/it]\u001b[A\n",
      " 93%|█████████▎| 278/300 [17:43<01:24,  3.85s/it]\u001b[A\n",
      " 93%|█████████▎| 279/300 [17:47<01:20,  3.82s/it]\u001b[A\n",
      " 93%|█████████▎| 280/300 [17:51<01:16,  3.83s/it]\u001b[A\n",
      " 94%|█████████▎| 281/300 [17:55<01:12,  3.83s/it]\u001b[A\n",
      " 94%|█████████▍| 282/300 [17:59<01:08,  3.82s/it]\u001b[A\n",
      " 94%|█████████▍| 283/300 [18:02<01:04,  3.82s/it]\u001b[A\n",
      " 95%|█████████▍| 284/300 [18:06<01:01,  3.85s/it]\u001b[A\n",
      " 95%|█████████▌| 285/300 [18:10<00:57,  3.83s/it]\u001b[A\n",
      " 95%|█████████▌| 286/300 [18:14<00:54,  3.86s/it]\u001b[A\n",
      " 96%|█████████▌| 287/300 [18:18<00:50,  3.86s/it]\u001b[A\n",
      " 96%|█████████▌| 288/300 [18:22<00:46,  3.87s/it]\u001b[A\n",
      " 96%|█████████▋| 289/300 [18:26<00:42,  3.83s/it]\u001b[A\n",
      " 97%|█████████▋| 290/300 [18:29<00:38,  3.82s/it]\u001b[A\n",
      " 97%|█████████▋| 291/300 [18:33<00:34,  3.85s/it]\u001b[A\n",
      " 97%|█████████▋| 292/300 [18:37<00:30,  3.86s/it]\u001b[A\n",
      " 98%|█████████▊| 293/300 [18:41<00:26,  3.84s/it]\u001b[A\n",
      " 98%|█████████▊| 294/300 [18:45<00:23,  3.85s/it]\u001b[A\n",
      " 98%|█████████▊| 295/300 [18:49<00:19,  3.83s/it]\u001b[A\n",
      " 99%|█████████▊| 296/300 [18:52<00:15,  3.81s/it]\u001b[A\n",
      " 99%|█████████▉| 297/300 [18:56<00:11,  3.79s/it]\u001b[A\n",
      " 99%|█████████▉| 298/300 [19:00<00:07,  3.78s/it]\u001b[A\n",
      "100%|█████████▉| 299/300 [19:04<00:03,  3.77s/it]\u001b[A\n",
      "100%|██████████| 300/300 [19:08<00:00,  3.82s/it]\u001b[A\n",
      "\u001b[A"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.8866666666666667\n"
     ]
    }
   ],
   "source": [
    "all_match, all_largest = test_model(model_path)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.7\n"
     ]
    }
   ],
   "source": [
    "paired = list(zip(all_match, all_largest))\n",
    "\n",
    "acc_3_matches = 0\n",
    "total = len(paired)\n",
    "for match, largest in paired:\n",
    "    top3 = largest[:3]\n",
    "    if match in top3:\n",
    "        acc_3_matches +=1\n",
    "print(acc_3_matches/total)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "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": "3.6.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}