generate_tables.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "2a52b01a-febb-4462-ad47-7188d5675625",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Requirement already satisfied: pandas in ./.venv/lib/python3.9/site-packages (2.2.2)\n",
      "Requirement already satisfied: numpy>=1.22.4 in ./.venv/lib/python3.9/site-packages (from pandas) (1.26.4)\n",
      "Requirement already satisfied: python-dateutil>=2.8.2 in ./.venv/lib/python3.9/site-packages (from pandas) (2.9.0.post0)\n",
      "Requirement already satisfied: pytz>=2020.1 in ./.venv/lib/python3.9/site-packages (from pandas) (2024.1)\n",
      "Requirement already satisfied: tzdata>=2022.7 in ./.venv/lib/python3.9/site-packages (from pandas) (2024.1)\n",
      "Requirement already satisfied: six>=1.5 in ./.venv/lib/python3.9/site-packages (from python-dateutil>=2.8.2->pandas) (1.16.0)\n"
     ]
    }
   ],
   "source": [
    "!pip install pandas\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "9b6f5cf4-187c-4edb-af7e-dbad2fd6bd59",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "# Set display options\n",
    "pd.set_option('display.max_columns', None)  # Display all columns\n",
    "pd.set_option('display.width', 1000)         # Set display width\n",
    "pd.set_option('display.precision', 5)        # Set decimal precision\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "13607afb-31a0-4220-8626-30ab1af8f050",
   "metadata": {},
   "outputs": [],
   "source": [
    "df = pd.read_csv('results.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "d05f1986-9e41-4cec-98c8-c0c5f44a7ac0",
   "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>model</th>\n",
       "      <th>pruning</th>\n",
       "      <th>quantization</th>\n",
       "      <th>size (MB)</th>\n",
       "      <th>total</th>\n",
       "      <th>correct (top1)</th>\n",
       "      <th>correct (top5)</th>\n",
       "      <th>accuracy % (top1)</th>\n",
       "      <th>accuracy % (top5)</th>\n",
       "      <th>time (s)</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>lenet5</td>\n",
       "      <td>no_prune</td>\n",
       "      <td>no_quantize</td>\n",
       "      <td>0.18107</td>\n",
       "      <td>10000</td>\n",
       "      <td>9869</td>\n",
       "      <td>10000</td>\n",
       "      <td>98.690</td>\n",
       "      <td>100.000</td>\n",
       "      <td>1.59590</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>lenet5</td>\n",
       "      <td>no_prune</td>\n",
       "      <td>quanto_int4_quantize</td>\n",
       "      <td>0.03484</td>\n",
       "      <td>10000</td>\n",
       "      <td>8842</td>\n",
       "      <td>9928</td>\n",
       "      <td>88.420</td>\n",
       "      <td>99.280</td>\n",
       "      <td>1.81683</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>lenet5</td>\n",
       "      <td>no_prune</td>\n",
       "      <td>quanto_int8_quantize</td>\n",
       "      <td>0.05452</td>\n",
       "      <td>10000</td>\n",
       "      <td>9868</td>\n",
       "      <td>10000</td>\n",
       "      <td>98.680</td>\n",
       "      <td>100.000</td>\n",
       "      <td>1.58993</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>lenet5</td>\n",
       "      <td>no_prune</td>\n",
       "      <td>quanto_float8_quantize</td>\n",
       "      <td>0.05458</td>\n",
       "      <td>10000</td>\n",
       "      <td>9863</td>\n",
       "      <td>10000</td>\n",
       "      <td>98.630</td>\n",
       "      <td>100.000</td>\n",
       "      <td>1.57028</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>lenet5</td>\n",
       "      <td>l1_unstructured_prune</td>\n",
       "      <td>no_quantize</td>\n",
       "      <td>0.18107</td>\n",
       "      <td>10000</td>\n",
       "      <td>9860</td>\n",
       "      <td>9999</td>\n",
       "      <td>98.600</td>\n",
       "      <td>99.990</td>\n",
       "      <td>1.44367</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75</th>\n",
       "      <td>vgg16</td>\n",
       "      <td>l1_structured_prune_five_percent</td>\n",
       "      <td>quanto_float8_quantize</td>\n",
       "      <td>138.47895</td>\n",
       "      <td>50000</td>\n",
       "      <td>50</td>\n",
       "      <td>250</td>\n",
       "      <td>0.100</td>\n",
       "      <td>0.500</td>\n",
       "      <td>81.03596</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>76</th>\n",
       "      <td>vgg16</td>\n",
       "      <td>random_unstructured_prune</td>\n",
       "      <td>no_quantize</td>\n",
       "      <td>553.43963</td>\n",
       "      <td>50000</td>\n",
       "      <td>3566</td>\n",
       "      <td>10411</td>\n",
       "      <td>7.132</td>\n",
       "      <td>20.822</td>\n",
       "      <td>76.77526</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>77</th>\n",
       "      <td>vgg16</td>\n",
       "      <td>random_unstructured_prune</td>\n",
       "      <td>quanto_int4_quantize</td>\n",
       "      <td>77.15284</td>\n",
       "      <td>50000</td>\n",
       "      <td>214</td>\n",
       "      <td>818</td>\n",
       "      <td>0.428</td>\n",
       "      <td>1.636</td>\n",
       "      <td>82.04001</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>78</th>\n",
       "      <td>vgg16</td>\n",
       "      <td>random_unstructured_prune</td>\n",
       "      <td>quanto_int8_quantize</td>\n",
       "      <td>138.47883</td>\n",
       "      <td>50000</td>\n",
       "      <td>5464</td>\n",
       "      <td>14331</td>\n",
       "      <td>10.928</td>\n",
       "      <td>28.662</td>\n",
       "      <td>78.65669</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>79</th>\n",
       "      <td>vgg16</td>\n",
       "      <td>random_unstructured_prune</td>\n",
       "      <td>quanto_float8_quantize</td>\n",
       "      <td>138.47895</td>\n",
       "      <td>50000</td>\n",
       "      <td>5772</td>\n",
       "      <td>14060</td>\n",
       "      <td>11.544</td>\n",
       "      <td>28.120</td>\n",
       "      <td>81.21140</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>80 rows × 10 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "     model                           pruning            quantization  size (MB)  total  correct (top1)  correct (top5)  accuracy % (top1)  accuracy % (top5)  time (s)\n",
       "0   lenet5                          no_prune             no_quantize    0.18107  10000            9869           10000             98.690            100.000   1.59590\n",
       "1   lenet5                          no_prune    quanto_int4_quantize    0.03484  10000            8842            9928             88.420             99.280   1.81683\n",
       "2   lenet5                          no_prune    quanto_int8_quantize    0.05452  10000            9868           10000             98.680            100.000   1.58993\n",
       "3   lenet5                          no_prune  quanto_float8_quantize    0.05458  10000            9863           10000             98.630            100.000   1.57028\n",
       "4   lenet5             l1_unstructured_prune             no_quantize    0.18107  10000            9860            9999             98.600             99.990   1.44367\n",
       "..     ...                               ...                     ...        ...    ...             ...             ...                ...                ...       ...\n",
       "75   vgg16  l1_structured_prune_five_percent  quanto_float8_quantize  138.47895  50000              50             250              0.100              0.500  81.03596\n",
       "76   vgg16         random_unstructured_prune             no_quantize  553.43963  50000            3566           10411              7.132             20.822  76.77526\n",
       "77   vgg16         random_unstructured_prune    quanto_int4_quantize   77.15284  50000             214             818              0.428              1.636  82.04001\n",
       "78   vgg16         random_unstructured_prune    quanto_int8_quantize  138.47883  50000            5464           14331             10.928             28.662  78.65669\n",
       "79   vgg16         random_unstructured_prune  quanto_float8_quantize  138.47895  50000            5772           14060             11.544             28.120  81.21140\n",
       "\n",
       "[80 rows x 10 columns]"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "8302291f-e309-4c43-a5b4-5a526df9e5bc",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['no_prune', 'l1_unstructured_prune', 'l1_structured_prune_one_percent', 'l1_structured_prune_five_percent', 'random_unstructured_prune'] ['no_quantize', 'quanto_int4_quantize', 'quanto_int8_quantize', 'quanto_float8_quantize'] ['lenet5', 'lenet300', 'alexnet', 'vgg16'] no_prune no_quantize\n"
     ]
    }
   ],
   "source": [
    "prune_types = list(df.pruning.unique())\n",
    "quantization_types = list(df.quantization.unique())\n",
    "model_types = list(df.model.unique())\n",
    "no_prune_type = df.pruning.unique()[0]\n",
    "no_quantize_type = df.quantization.unique()[0]\n",
    "print(prune_types, quantization_types, model_types, no_prune_type, no_quantize_type)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "72ab771e-ed9e-4234-9280-8651f159c417",
   "metadata": {},
   "outputs": [],
   "source": [
    "def display_tables(key):\n",
    "    print(key)\n",
    "    for model in model_types:\n",
    "        model_df = df[df['model'] == model]\n",
    "        output_table = pd.DataFrame(index=prune_types, columns=quantization_types)\n",
    "        \n",
    "        base_model = model_df[(model_df['pruning'] == no_prune_type) & (model_df['quantization'] == no_quantize_type)]\n",
    "        base_model_size = base_model[key].iloc[0].astype(float)\n",
    "        print('model:', model, 'base model', key, base_model_size)\n",
    "        for prune_type in prune_types:\n",
    "            for quantization_type in quantization_types:\n",
    "                tuned_model = model_df[(model_df['pruning'] == prune_type) & (model_df['quantization'] == quantization_type)]\n",
    "                tuned_model_size = tuned_model[key].iloc[0].astype(float)\n",
    "                output_table.at[prune_type, quantization_type] = tuned_model_size\n",
    "        display(output_table)        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "3e6bea69-9025-4fd8-9659-4c50910ae8cb",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "size (MB)\n",
      "model: lenet5 base model size (MB) 0.18107\n"
     ]
    },
    {
     "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>no_quantize</th>\n",
       "      <th>quanto_int4_quantize</th>\n",
       "      <th>quanto_int8_quantize</th>\n",
       "      <th>quanto_float8_quantize</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>no_prune</th>\n",
       "      <td>0.18107</td>\n",
       "      <td>0.03484</td>\n",
       "      <td>0.05452</td>\n",
       "      <td>0.05458</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_unstructured_prune</th>\n",
       "      <td>0.18107</td>\n",
       "      <td>0.03484</td>\n",
       "      <td>0.05452</td>\n",
       "      <td>0.05458</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_structured_prune_one_percent</th>\n",
       "      <td>0.18107</td>\n",
       "      <td>0.03484</td>\n",
       "      <td>0.05452</td>\n",
       "      <td>0.05458</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_structured_prune_five_percent</th>\n",
       "      <td>0.18107</td>\n",
       "      <td>0.03484</td>\n",
       "      <td>0.05452</td>\n",
       "      <td>0.05458</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>random_unstructured_prune</th>\n",
       "      <td>0.18107</td>\n",
       "      <td>0.03484</td>\n",
       "      <td>0.05452</td>\n",
       "      <td>0.05458</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 no_quantize quanto_int4_quantize quanto_int8_quantize quanto_float8_quantize\n",
       "no_prune                             0.18107              0.03484              0.05452                0.05458\n",
       "l1_unstructured_prune                0.18107              0.03484              0.05452                0.05458\n",
       "l1_structured_prune_one_percent      0.18107              0.03484              0.05452                0.05458\n",
       "l1_structured_prune_five_percent     0.18107              0.03484              0.05452                0.05458\n",
       "random_unstructured_prune            0.18107              0.03484              0.05452                0.05458"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "model: lenet300 base model size (MB) 1.068842\n"
     ]
    },
    {
     "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>no_quantize</th>\n",
       "      <th>quanto_int4_quantize</th>\n",
       "      <th>quanto_int8_quantize</th>\n",
       "      <th>quanto_float8_quantize</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>no_prune</th>\n",
       "      <td>1.06884</td>\n",
       "      <td>0.15376</td>\n",
       "      <td>0.27474</td>\n",
       "      <td>0.27486</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_unstructured_prune</th>\n",
       "      <td>1.06884</td>\n",
       "      <td>0.15376</td>\n",
       "      <td>0.27474</td>\n",
       "      <td>0.27486</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_structured_prune_one_percent</th>\n",
       "      <td>1.06884</td>\n",
       "      <td>0.15376</td>\n",
       "      <td>0.27474</td>\n",
       "      <td>0.27486</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_structured_prune_five_percent</th>\n",
       "      <td>1.06884</td>\n",
       "      <td>0.15376</td>\n",
       "      <td>0.27474</td>\n",
       "      <td>0.27486</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>random_unstructured_prune</th>\n",
       "      <td>1.06884</td>\n",
       "      <td>0.15376</td>\n",
       "      <td>0.27474</td>\n",
       "      <td>0.27486</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 no_quantize quanto_int4_quantize quanto_int8_quantize quanto_float8_quantize\n",
       "no_prune                             1.06884              0.15376              0.27474                0.27486\n",
       "l1_unstructured_prune                1.06884              0.15376              0.27474                0.27486\n",
       "l1_structured_prune_one_percent      1.06884              0.15376              0.27474                0.27486\n",
       "l1_structured_prune_five_percent     1.06884              0.15376              0.27474                0.27486\n",
       "random_unstructured_prune            1.06884              0.15376              0.27474                0.27486"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "model: alexnet base model size (MB) 244.408362\n"
     ]
    },
    {
     "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>no_quantize</th>\n",
       "      <th>quanto_int4_quantize</th>\n",
       "      <th>quanto_int8_quantize</th>\n",
       "      <th>quanto_float8_quantize</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>no_prune</th>\n",
       "      <td>244.40836</td>\n",
       "      <td>34.98436</td>\n",
       "      <td>61.18716</td>\n",
       "      <td>61.18723</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_unstructured_prune</th>\n",
       "      <td>244.40836</td>\n",
       "      <td>34.98436</td>\n",
       "      <td>61.18716</td>\n",
       "      <td>61.18723</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_structured_prune_one_percent</th>\n",
       "      <td>244.40836</td>\n",
       "      <td>34.98436</td>\n",
       "      <td>61.18716</td>\n",
       "      <td>61.18723</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_structured_prune_five_percent</th>\n",
       "      <td>244.40836</td>\n",
       "      <td>34.98436</td>\n",
       "      <td>61.18716</td>\n",
       "      <td>61.18723</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>random_unstructured_prune</th>\n",
       "      <td>244.40836</td>\n",
       "      <td>34.98436</td>\n",
       "      <td>61.18716</td>\n",
       "      <td>61.18723</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 no_quantize quanto_int4_quantize quanto_int8_quantize quanto_float8_quantize\n",
       "no_prune                           244.40836             34.98436             61.18716               61.18723\n",
       "l1_unstructured_prune              244.40836             34.98436             61.18716               61.18723\n",
       "l1_structured_prune_one_percent    244.40836             34.98436             61.18716               61.18723\n",
       "l1_structured_prune_five_percent   244.40836             34.98436             61.18716               61.18723\n",
       "random_unstructured_prune          244.40836             34.98436             61.18716               61.18723"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "model: vgg16 base model size (MB) 553.439626\n"
     ]
    },
    {
     "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>no_quantize</th>\n",
       "      <th>quanto_int4_quantize</th>\n",
       "      <th>quanto_int8_quantize</th>\n",
       "      <th>quanto_float8_quantize</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>no_prune</th>\n",
       "      <td>553.43963</td>\n",
       "      <td>77.15284</td>\n",
       "      <td>138.47883</td>\n",
       "      <td>138.47895</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_unstructured_prune</th>\n",
       "      <td>553.43963</td>\n",
       "      <td>77.15284</td>\n",
       "      <td>138.47883</td>\n",
       "      <td>138.47895</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_structured_prune_one_percent</th>\n",
       "      <td>553.43963</td>\n",
       "      <td>77.15284</td>\n",
       "      <td>138.47883</td>\n",
       "      <td>138.47895</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_structured_prune_five_percent</th>\n",
       "      <td>553.43963</td>\n",
       "      <td>77.15284</td>\n",
       "      <td>138.47883</td>\n",
       "      <td>138.47895</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>random_unstructured_prune</th>\n",
       "      <td>553.43963</td>\n",
       "      <td>77.15284</td>\n",
       "      <td>138.47883</td>\n",
       "      <td>138.47895</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 no_quantize quanto_int4_quantize quanto_int8_quantize quanto_float8_quantize\n",
       "no_prune                           553.43963             77.15284            138.47883              138.47895\n",
       "l1_unstructured_prune              553.43963             77.15284            138.47883              138.47895\n",
       "l1_structured_prune_one_percent    553.43963             77.15284            138.47883              138.47895\n",
       "l1_structured_prune_five_percent   553.43963             77.15284            138.47883              138.47895\n",
       "random_unstructured_prune          553.43963             77.15284            138.47883              138.47895"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "display_tables('size (MB)')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "cec0a434-e8af-4740-970b-ddc17ca956c0",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "time (s)\n",
      "model: lenet5 base model time (s) 1.595897912979126\n"
     ]
    },
    {
     "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>no_quantize</th>\n",
       "      <th>quanto_int4_quantize</th>\n",
       "      <th>quanto_int8_quantize</th>\n",
       "      <th>quanto_float8_quantize</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>no_prune</th>\n",
       "      <td>1.5959</td>\n",
       "      <td>1.81683</td>\n",
       "      <td>1.58993</td>\n",
       "      <td>1.57028</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_unstructured_prune</th>\n",
       "      <td>1.44367</td>\n",
       "      <td>1.82981</td>\n",
       "      <td>1.56427</td>\n",
       "      <td>1.63036</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_structured_prune_one_percent</th>\n",
       "      <td>1.50232</td>\n",
       "      <td>1.77563</td>\n",
       "      <td>1.67182</td>\n",
       "      <td>1.60231</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_structured_prune_five_percent</th>\n",
       "      <td>1.40963</td>\n",
       "      <td>1.81139</td>\n",
       "      <td>1.60057</td>\n",
       "      <td>1.66042</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>random_unstructured_prune</th>\n",
       "      <td>1.44456</td>\n",
       "      <td>1.81929</td>\n",
       "      <td>1.66235</td>\n",
       "      <td>1.65053</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 no_quantize quanto_int4_quantize quanto_int8_quantize quanto_float8_quantize\n",
       "no_prune                              1.5959              1.81683              1.58993                1.57028\n",
       "l1_unstructured_prune                1.44367              1.82981              1.56427                1.63036\n",
       "l1_structured_prune_one_percent      1.50232              1.77563              1.67182                1.60231\n",
       "l1_structured_prune_five_percent     1.40963              1.81139              1.60057                1.66042\n",
       "random_unstructured_prune            1.44456              1.81929              1.66235                1.65053"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "model: lenet300 base model time (s) 1.5003230571746826\n"
     ]
    },
    {
     "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>no_quantize</th>\n",
       "      <th>quanto_int4_quantize</th>\n",
       "      <th>quanto_int8_quantize</th>\n",
       "      <th>quanto_float8_quantize</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>no_prune</th>\n",
       "      <td>1.50032</td>\n",
       "      <td>1.68077</td>\n",
       "      <td>1.52841</td>\n",
       "      <td>1.47726</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_unstructured_prune</th>\n",
       "      <td>1.49622</td>\n",
       "      <td>1.63178</td>\n",
       "      <td>1.55171</td>\n",
       "      <td>1.59309</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_structured_prune_one_percent</th>\n",
       "      <td>1.43131</td>\n",
       "      <td>1.66544</td>\n",
       "      <td>1.57773</td>\n",
       "      <td>1.5597</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_structured_prune_five_percent</th>\n",
       "      <td>1.44365</td>\n",
       "      <td>1.68328</td>\n",
       "      <td>1.56245</td>\n",
       "      <td>1.46811</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>random_unstructured_prune</th>\n",
       "      <td>1.45635</td>\n",
       "      <td>1.68194</td>\n",
       "      <td>1.55521</td>\n",
       "      <td>1.58358</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 no_quantize quanto_int4_quantize quanto_int8_quantize quanto_float8_quantize\n",
       "no_prune                             1.50032              1.68077              1.52841                1.47726\n",
       "l1_unstructured_prune                1.49622              1.63178              1.55171                1.59309\n",
       "l1_structured_prune_one_percent      1.43131              1.66544              1.57773                 1.5597\n",
       "l1_structured_prune_five_percent     1.44365              1.68328              1.56245                1.46811\n",
       "random_unstructured_prune            1.45635              1.68194              1.55521                1.58358"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "model: alexnet base model time (s) 35.920411586761475\n"
     ]
    },
    {
     "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>no_quantize</th>\n",
       "      <th>quanto_int4_quantize</th>\n",
       "      <th>quanto_int8_quantize</th>\n",
       "      <th>quanto_float8_quantize</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>no_prune</th>\n",
       "      <td>35.92041</td>\n",
       "      <td>36.81441</td>\n",
       "      <td>36.33962</td>\n",
       "      <td>36.59337</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_unstructured_prune</th>\n",
       "      <td>35.91069</td>\n",
       "      <td>36.93813</td>\n",
       "      <td>36.28348</td>\n",
       "      <td>36.62818</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_structured_prune_one_percent</th>\n",
       "      <td>35.75678</td>\n",
       "      <td>36.84587</td>\n",
       "      <td>36.41082</td>\n",
       "      <td>36.51584</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_structured_prune_five_percent</th>\n",
       "      <td>35.85509</td>\n",
       "      <td>36.79347</td>\n",
       "      <td>36.20274</td>\n",
       "      <td>36.60355</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>random_unstructured_prune</th>\n",
       "      <td>35.90395</td>\n",
       "      <td>37.158</td>\n",
       "      <td>36.36515</td>\n",
       "      <td>36.50318</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 no_quantize quanto_int4_quantize quanto_int8_quantize quanto_float8_quantize\n",
       "no_prune                            35.92041             36.81441             36.33962               36.59337\n",
       "l1_unstructured_prune               35.91069             36.93813             36.28348               36.62818\n",
       "l1_structured_prune_one_percent     35.75678             36.84587             36.41082               36.51584\n",
       "l1_structured_prune_five_percent    35.85509             36.79347             36.20274               36.60355\n",
       "random_unstructured_prune           35.90395               37.158             36.36515               36.50318"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "model: vgg16 base model time (s) 76.71681356430054\n"
     ]
    },
    {
     "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>no_quantize</th>\n",
       "      <th>quanto_int4_quantize</th>\n",
       "      <th>quanto_int8_quantize</th>\n",
       "      <th>quanto_float8_quantize</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>no_prune</th>\n",
       "      <td>76.71681</td>\n",
       "      <td>81.97319</td>\n",
       "      <td>78.6547</td>\n",
       "      <td>81.20747</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_unstructured_prune</th>\n",
       "      <td>76.74276</td>\n",
       "      <td>82.04073</td>\n",
       "      <td>78.52865</td>\n",
       "      <td>81.23339</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_structured_prune_one_percent</th>\n",
       "      <td>76.75068</td>\n",
       "      <td>82.01412</td>\n",
       "      <td>78.60611</td>\n",
       "      <td>81.21474</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_structured_prune_five_percent</th>\n",
       "      <td>76.651</td>\n",
       "      <td>82.03824</td>\n",
       "      <td>78.54475</td>\n",
       "      <td>81.03596</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>random_unstructured_prune</th>\n",
       "      <td>76.77526</td>\n",
       "      <td>82.04001</td>\n",
       "      <td>78.65669</td>\n",
       "      <td>81.2114</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 no_quantize quanto_int4_quantize quanto_int8_quantize quanto_float8_quantize\n",
       "no_prune                            76.71681             81.97319              78.6547               81.20747\n",
       "l1_unstructured_prune               76.74276             82.04073             78.52865               81.23339\n",
       "l1_structured_prune_one_percent     76.75068             82.01412             78.60611               81.21474\n",
       "l1_structured_prune_five_percent      76.651             82.03824             78.54475               81.03596\n",
       "random_unstructured_prune           76.77526             82.04001             78.65669                81.2114"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "display_tables('time (s)')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "c081b0ba-ba64-400a-8026-76dd675bcb0c",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "accuracy % (top1)\n",
      "model: lenet5 base model accuracy % (top1) 98.69\n"
     ]
    },
    {
     "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>no_quantize</th>\n",
       "      <th>quanto_int4_quantize</th>\n",
       "      <th>quanto_int8_quantize</th>\n",
       "      <th>quanto_float8_quantize</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>no_prune</th>\n",
       "      <td>98.69</td>\n",
       "      <td>88.42</td>\n",
       "      <td>98.68</td>\n",
       "      <td>98.63</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_unstructured_prune</th>\n",
       "      <td>98.6</td>\n",
       "      <td>94.12</td>\n",
       "      <td>98.63</td>\n",
       "      <td>98.63</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_structured_prune_one_percent</th>\n",
       "      <td>98.69</td>\n",
       "      <td>88.42</td>\n",
       "      <td>98.68</td>\n",
       "      <td>11.35</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_structured_prune_five_percent</th>\n",
       "      <td>98.75</td>\n",
       "      <td>88.5</td>\n",
       "      <td>98.73</td>\n",
       "      <td>11.35</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>random_unstructured_prune</th>\n",
       "      <td>95.58</td>\n",
       "      <td>83.8</td>\n",
       "      <td>96.54</td>\n",
       "      <td>91.89</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 no_quantize quanto_int4_quantize quanto_int8_quantize quanto_float8_quantize\n",
       "no_prune                               98.69                88.42                98.68                  98.63\n",
       "l1_unstructured_prune                   98.6                94.12                98.63                  98.63\n",
       "l1_structured_prune_one_percent        98.69                88.42                98.68                  11.35\n",
       "l1_structured_prune_five_percent       98.75                 88.5                98.73                  11.35\n",
       "random_unstructured_prune              95.58                 83.8                96.54                  91.89"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "model: lenet300 base model accuracy % (top1) 97.9\n"
     ]
    },
    {
     "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>no_quantize</th>\n",
       "      <th>quanto_int4_quantize</th>\n",
       "      <th>quanto_int8_quantize</th>\n",
       "      <th>quanto_float8_quantize</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>no_prune</th>\n",
       "      <td>97.9</td>\n",
       "      <td>96.95</td>\n",
       "      <td>97.9</td>\n",
       "      <td>97.91</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_unstructured_prune</th>\n",
       "      <td>97.9</td>\n",
       "      <td>97.28</td>\n",
       "      <td>97.93</td>\n",
       "      <td>97.95</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_structured_prune_one_percent</th>\n",
       "      <td>97.9</td>\n",
       "      <td>96.95</td>\n",
       "      <td>97.9</td>\n",
       "      <td>11.35</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_structured_prune_five_percent</th>\n",
       "      <td>97.9</td>\n",
       "      <td>96.95</td>\n",
       "      <td>97.9</td>\n",
       "      <td>11.35</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>random_unstructured_prune</th>\n",
       "      <td>96.37</td>\n",
       "      <td>93.45</td>\n",
       "      <td>96.93</td>\n",
       "      <td>96.63</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 no_quantize quanto_int4_quantize quanto_int8_quantize quanto_float8_quantize\n",
       "no_prune                                97.9                96.95                 97.9                  97.91\n",
       "l1_unstructured_prune                   97.9                97.28                97.93                  97.95\n",
       "l1_structured_prune_one_percent         97.9                96.95                 97.9                  11.35\n",
       "l1_structured_prune_five_percent        97.9                96.95                 97.9                  11.35\n",
       "random_unstructured_prune              96.37                93.45                96.93                  96.63"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "model: alexnet base model accuracy % (top1) 56.55\n"
     ]
    },
    {
     "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>no_quantize</th>\n",
       "      <th>quanto_int4_quantize</th>\n",
       "      <th>quanto_int8_quantize</th>\n",
       "      <th>quanto_float8_quantize</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>no_prune</th>\n",
       "      <td>56.55</td>\n",
       "      <td>0.346</td>\n",
       "      <td>56.432</td>\n",
       "      <td>56.356</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_unstructured_prune</th>\n",
       "      <td>56.14</td>\n",
       "      <td>0.556</td>\n",
       "      <td>56.046</td>\n",
       "      <td>55.888</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_structured_prune_one_percent</th>\n",
       "      <td>54.486</td>\n",
       "      <td>0.348</td>\n",
       "      <td>54.39</td>\n",
       "      <td>0.1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_structured_prune_five_percent</th>\n",
       "      <td>44.952</td>\n",
       "      <td>0.264</td>\n",
       "      <td>44.804</td>\n",
       "      <td>0.1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>random_unstructured_prune</th>\n",
       "      <td>7.946</td>\n",
       "      <td>0.3</td>\n",
       "      <td>6.804</td>\n",
       "      <td>10.886</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 no_quantize quanto_int4_quantize quanto_int8_quantize quanto_float8_quantize\n",
       "no_prune                               56.55                0.346               56.432                 56.356\n",
       "l1_unstructured_prune                  56.14                0.556               56.046                 55.888\n",
       "l1_structured_prune_one_percent       54.486                0.348                54.39                    0.1\n",
       "l1_structured_prune_five_percent      44.952                0.264               44.804                    0.1\n",
       "random_unstructured_prune              7.946                  0.3                6.804                 10.886"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "model: vgg16 base model accuracy % (top1) 71.58200000000001\n"
     ]
    },
    {
     "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>no_quantize</th>\n",
       "      <th>quanto_int4_quantize</th>\n",
       "      <th>quanto_int8_quantize</th>\n",
       "      <th>quanto_float8_quantize</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>no_prune</th>\n",
       "      <td>71.582</td>\n",
       "      <td>4.934</td>\n",
       "      <td>71.606</td>\n",
       "      <td>71.492</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_unstructured_prune</th>\n",
       "      <td>71.448</td>\n",
       "      <td>8.26</td>\n",
       "      <td>71.44</td>\n",
       "      <td>71.232</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_structured_prune_one_percent</th>\n",
       "      <td>70.396</td>\n",
       "      <td>5.264</td>\n",
       "      <td>70.34</td>\n",
       "      <td>0.1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_structured_prune_five_percent</th>\n",
       "      <td>61.54</td>\n",
       "      <td>2.196</td>\n",
       "      <td>61.444</td>\n",
       "      <td>0.1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>random_unstructured_prune</th>\n",
       "      <td>7.132</td>\n",
       "      <td>0.428</td>\n",
       "      <td>10.928</td>\n",
       "      <td>11.544</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 no_quantize quanto_int4_quantize quanto_int8_quantize quanto_float8_quantize\n",
       "no_prune                              71.582                4.934               71.606                 71.492\n",
       "l1_unstructured_prune                 71.448                 8.26                71.44                 71.232\n",
       "l1_structured_prune_one_percent       70.396                5.264                70.34                    0.1\n",
       "l1_structured_prune_five_percent       61.54                2.196               61.444                    0.1\n",
       "random_unstructured_prune              7.132                0.428               10.928                 11.544"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "display_tables('accuracy % (top1)')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "4ed883cb-c58f-43f7-87da-99389482e4c1",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "accuracy % (top5)\n",
      "model: lenet5 base model accuracy % (top5) 100.0\n"
     ]
    },
    {
     "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>no_quantize</th>\n",
       "      <th>quanto_int4_quantize</th>\n",
       "      <th>quanto_int8_quantize</th>\n",
       "      <th>quanto_float8_quantize</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>no_prune</th>\n",
       "      <td>100.0</td>\n",
       "      <td>99.28</td>\n",
       "      <td>100.0</td>\n",
       "      <td>100.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_unstructured_prune</th>\n",
       "      <td>99.99</td>\n",
       "      <td>99.81</td>\n",
       "      <td>99.99</td>\n",
       "      <td>99.99</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_structured_prune_one_percent</th>\n",
       "      <td>100.0</td>\n",
       "      <td>99.28</td>\n",
       "      <td>100.0</td>\n",
       "      <td>51.39</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_structured_prune_five_percent</th>\n",
       "      <td>100.0</td>\n",
       "      <td>99.36</td>\n",
       "      <td>100.0</td>\n",
       "      <td>51.39</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>random_unstructured_prune</th>\n",
       "      <td>99.85</td>\n",
       "      <td>99.38</td>\n",
       "      <td>99.95</td>\n",
       "      <td>99.95</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 no_quantize quanto_int4_quantize quanto_int8_quantize quanto_float8_quantize\n",
       "no_prune                               100.0                99.28                100.0                  100.0\n",
       "l1_unstructured_prune                  99.99                99.81                99.99                  99.99\n",
       "l1_structured_prune_one_percent        100.0                99.28                100.0                  51.39\n",
       "l1_structured_prune_five_percent       100.0                99.36                100.0                  51.39\n",
       "random_unstructured_prune              99.85                99.38                99.95                  99.95"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "model: lenet300 base model accuracy % (top5) 99.98\n"
     ]
    },
    {
     "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>no_quantize</th>\n",
       "      <th>quanto_int4_quantize</th>\n",
       "      <th>quanto_int8_quantize</th>\n",
       "      <th>quanto_float8_quantize</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>no_prune</th>\n",
       "      <td>99.98</td>\n",
       "      <td>99.96</td>\n",
       "      <td>99.98</td>\n",
       "      <td>99.98</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_unstructured_prune</th>\n",
       "      <td>99.98</td>\n",
       "      <td>99.96</td>\n",
       "      <td>99.98</td>\n",
       "      <td>99.98</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_structured_prune_one_percent</th>\n",
       "      <td>99.98</td>\n",
       "      <td>99.96</td>\n",
       "      <td>99.98</td>\n",
       "      <td>51.39</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_structured_prune_five_percent</th>\n",
       "      <td>99.98</td>\n",
       "      <td>99.96</td>\n",
       "      <td>99.98</td>\n",
       "      <td>51.39</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>random_unstructured_prune</th>\n",
       "      <td>99.94</td>\n",
       "      <td>99.88</td>\n",
       "      <td>99.95</td>\n",
       "      <td>99.95</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 no_quantize quanto_int4_quantize quanto_int8_quantize quanto_float8_quantize\n",
       "no_prune                               99.98                99.96                99.98                  99.98\n",
       "l1_unstructured_prune                  99.98                99.96                99.98                  99.98\n",
       "l1_structured_prune_one_percent        99.98                99.96                99.98                  51.39\n",
       "l1_structured_prune_five_percent       99.98                99.96                99.98                  51.39\n",
       "random_unstructured_prune              99.94                99.88                99.95                  99.95"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "model: alexnet base model accuracy % (top5) 79.086\n"
     ]
    },
    {
     "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>no_quantize</th>\n",
       "      <th>quanto_int4_quantize</th>\n",
       "      <th>quanto_int8_quantize</th>\n",
       "      <th>quanto_float8_quantize</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>no_prune</th>\n",
       "      <td>79.086</td>\n",
       "      <td>1.33</td>\n",
       "      <td>79.122</td>\n",
       "      <td>78.964</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_unstructured_prune</th>\n",
       "      <td>78.818</td>\n",
       "      <td>1.796</td>\n",
       "      <td>78.824</td>\n",
       "      <td>78.574</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_structured_prune_one_percent</th>\n",
       "      <td>77.248</td>\n",
       "      <td>1.398</td>\n",
       "      <td>77.2</td>\n",
       "      <td>0.5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_structured_prune_five_percent</th>\n",
       "      <td>68.268</td>\n",
       "      <td>1.106</td>\n",
       "      <td>68.15</td>\n",
       "      <td>0.5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>random_unstructured_prune</th>\n",
       "      <td>20.208</td>\n",
       "      <td>1.156</td>\n",
       "      <td>18.104</td>\n",
       "      <td>25.48</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 no_quantize quanto_int4_quantize quanto_int8_quantize quanto_float8_quantize\n",
       "no_prune                              79.086                 1.33               79.122                 78.964\n",
       "l1_unstructured_prune                 78.818                1.796               78.824                 78.574\n",
       "l1_structured_prune_one_percent       77.248                1.398                 77.2                    0.5\n",
       "l1_structured_prune_five_percent      68.268                1.106                68.15                    0.5\n",
       "random_unstructured_prune             20.208                1.156               18.104                  25.48"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "model: vgg16 base model accuracy % (top5) 90.394\n"
     ]
    },
    {
     "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>no_quantize</th>\n",
       "      <th>quanto_int4_quantize</th>\n",
       "      <th>quanto_int8_quantize</th>\n",
       "      <th>quanto_float8_quantize</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>no_prune</th>\n",
       "      <td>90.394</td>\n",
       "      <td>12.57</td>\n",
       "      <td>90.4</td>\n",
       "      <td>90.348</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_unstructured_prune</th>\n",
       "      <td>90.3</td>\n",
       "      <td>19.274</td>\n",
       "      <td>90.312</td>\n",
       "      <td>90.196</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_structured_prune_one_percent</th>\n",
       "      <td>89.156</td>\n",
       "      <td>13.426</td>\n",
       "      <td>89.152</td>\n",
       "      <td>0.5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_structured_prune_five_percent</th>\n",
       "      <td>81.95</td>\n",
       "      <td>6.544</td>\n",
       "      <td>81.878</td>\n",
       "      <td>0.5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>random_unstructured_prune</th>\n",
       "      <td>20.822</td>\n",
       "      <td>1.636</td>\n",
       "      <td>28.662</td>\n",
       "      <td>28.12</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 no_quantize quanto_int4_quantize quanto_int8_quantize quanto_float8_quantize\n",
       "no_prune                              90.394                12.57                 90.4                 90.348\n",
       "l1_unstructured_prune                   90.3               19.274               90.312                 90.196\n",
       "l1_structured_prune_one_percent       89.156               13.426               89.152                    0.5\n",
       "l1_structured_prune_five_percent       81.95                6.544               81.878                    0.5\n",
       "random_unstructured_prune             20.822                1.636               28.662                  28.12"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "display_tables('accuracy % (top5)')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "5eb20192-e7d6-4453-a484-c912a0957fcd",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "model: lenet5 base model size:  0.18107\n"
     ]
    },
    {
     "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>no_quantize</th>\n",
       "      <th>quanto_int4_quantize</th>\n",
       "      <th>quanto_int8_quantize</th>\n",
       "      <th>quanto_float8_quantize</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>no_prune</th>\n",
       "      <td>1.0</td>\n",
       "      <td>5.19749</td>\n",
       "      <td>3.32141</td>\n",
       "      <td>3.31752</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_unstructured_prune</th>\n",
       "      <td>1.0</td>\n",
       "      <td>5.19749</td>\n",
       "      <td>3.32141</td>\n",
       "      <td>3.31752</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_structured_prune_one_percent</th>\n",
       "      <td>1.0</td>\n",
       "      <td>5.19749</td>\n",
       "      <td>3.32141</td>\n",
       "      <td>3.31752</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_structured_prune_five_percent</th>\n",
       "      <td>1.0</td>\n",
       "      <td>5.19749</td>\n",
       "      <td>3.32141</td>\n",
       "      <td>3.31752</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>random_unstructured_prune</th>\n",
       "      <td>1.0</td>\n",
       "      <td>5.19749</td>\n",
       "      <td>3.32141</td>\n",
       "      <td>3.31752</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 no_quantize quanto_int4_quantize quanto_int8_quantize quanto_float8_quantize\n",
       "no_prune                                 1.0              5.19749              3.32141                3.31752\n",
       "l1_unstructured_prune                    1.0              5.19749              3.32141                3.31752\n",
       "l1_structured_prune_one_percent          1.0              5.19749              3.32141                3.31752\n",
       "l1_structured_prune_five_percent         1.0              5.19749              3.32141                3.31752\n",
       "random_unstructured_prune                1.0              5.19749              3.32141                3.31752"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "model: lenet300 base model size:  1.068842\n"
     ]
    },
    {
     "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>no_quantize</th>\n",
       "      <th>quanto_int4_quantize</th>\n",
       "      <th>quanto_int8_quantize</th>\n",
       "      <th>quanto_float8_quantize</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>no_prune</th>\n",
       "      <td>1.0</td>\n",
       "      <td>6.95146</td>\n",
       "      <td>3.89043</td>\n",
       "      <td>3.88862</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_unstructured_prune</th>\n",
       "      <td>1.0</td>\n",
       "      <td>6.95146</td>\n",
       "      <td>3.89043</td>\n",
       "      <td>3.88862</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_structured_prune_one_percent</th>\n",
       "      <td>1.0</td>\n",
       "      <td>6.95146</td>\n",
       "      <td>3.89043</td>\n",
       "      <td>3.88862</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_structured_prune_five_percent</th>\n",
       "      <td>1.0</td>\n",
       "      <td>6.95146</td>\n",
       "      <td>3.89043</td>\n",
       "      <td>3.88862</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>random_unstructured_prune</th>\n",
       "      <td>1.0</td>\n",
       "      <td>6.95146</td>\n",
       "      <td>3.89043</td>\n",
       "      <td>3.88862</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 no_quantize quanto_int4_quantize quanto_int8_quantize quanto_float8_quantize\n",
       "no_prune                                 1.0              6.95146              3.89043                3.88862\n",
       "l1_unstructured_prune                    1.0              6.95146              3.89043                3.88862\n",
       "l1_structured_prune_one_percent          1.0              6.95146              3.89043                3.88862\n",
       "l1_structured_prune_five_percent         1.0              6.95146              3.89043                3.88862\n",
       "random_unstructured_prune                1.0              6.95146              3.89043                3.88862"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "model: alexnet base model size:  244.408362\n"
     ]
    },
    {
     "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>no_quantize</th>\n",
       "      <th>quanto_int4_quantize</th>\n",
       "      <th>quanto_int8_quantize</th>\n",
       "      <th>quanto_float8_quantize</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>no_prune</th>\n",
       "      <td>1.0</td>\n",
       "      <td>6.98622</td>\n",
       "      <td>3.99444</td>\n",
       "      <td>3.99443</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_unstructured_prune</th>\n",
       "      <td>1.0</td>\n",
       "      <td>6.98622</td>\n",
       "      <td>3.99444</td>\n",
       "      <td>3.99443</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_structured_prune_one_percent</th>\n",
       "      <td>1.0</td>\n",
       "      <td>6.98622</td>\n",
       "      <td>3.99444</td>\n",
       "      <td>3.99443</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_structured_prune_five_percent</th>\n",
       "      <td>1.0</td>\n",
       "      <td>6.98622</td>\n",
       "      <td>3.99444</td>\n",
       "      <td>3.99443</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>random_unstructured_prune</th>\n",
       "      <td>1.0</td>\n",
       "      <td>6.98622</td>\n",
       "      <td>3.99444</td>\n",
       "      <td>3.99443</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 no_quantize quanto_int4_quantize quanto_int8_quantize quanto_float8_quantize\n",
       "no_prune                                 1.0              6.98622              3.99444                3.99443\n",
       "l1_unstructured_prune                    1.0              6.98622              3.99444                3.99443\n",
       "l1_structured_prune_one_percent          1.0              6.98622              3.99444                3.99443\n",
       "l1_structured_prune_five_percent         1.0              6.98622              3.99444                3.99443\n",
       "random_unstructured_prune                1.0              6.98622              3.99444                3.99443"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "model: vgg16 base model size:  553.439626\n"
     ]
    },
    {
     "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>no_quantize</th>\n",
       "      <th>quanto_int4_quantize</th>\n",
       "      <th>quanto_int8_quantize</th>\n",
       "      <th>quanto_float8_quantize</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>no_prune</th>\n",
       "      <td>1.0</td>\n",
       "      <td>7.17329</td>\n",
       "      <td>3.99656</td>\n",
       "      <td>3.99656</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_unstructured_prune</th>\n",
       "      <td>1.0</td>\n",
       "      <td>7.17329</td>\n",
       "      <td>3.99656</td>\n",
       "      <td>3.99656</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_structured_prune_one_percent</th>\n",
       "      <td>1.0</td>\n",
       "      <td>7.17329</td>\n",
       "      <td>3.99656</td>\n",
       "      <td>3.99656</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>l1_structured_prune_five_percent</th>\n",
       "      <td>1.0</td>\n",
       "      <td>7.17329</td>\n",
       "      <td>3.99656</td>\n",
       "      <td>3.99656</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>random_unstructured_prune</th>\n",
       "      <td>1.0</td>\n",
       "      <td>7.17329</td>\n",
       "      <td>3.99656</td>\n",
       "      <td>3.99656</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 no_quantize quanto_int4_quantize quanto_int8_quantize quanto_float8_quantize\n",
       "no_prune                                 1.0              7.17329              3.99656                3.99656\n",
       "l1_unstructured_prune                    1.0              7.17329              3.99656                3.99656\n",
       "l1_structured_prune_one_percent          1.0              7.17329              3.99656                3.99656\n",
       "l1_structured_prune_five_percent         1.0              7.17329              3.99656                3.99656\n",
       "random_unstructured_prune                1.0              7.17329              3.99656                3.99656"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "for model in model_types:\n",
    "    model_df = df[df['model'] == model]\n",
    "    output_table = pd.DataFrame(index=prune_types, columns=quantization_types)\n",
    "    \n",
    "    base_model = model_df[(model_df['pruning'] == no_prune_type) & (model_df['quantization'] == no_quantize_type)]\n",
    "    base_model_size = base_model['size (MB)'].iloc[0].astype(float)\n",
    "    print('model:', model, 'base model size: ', base_model_size)\n",
    "    for prune_type in prune_types:\n",
    "        for quantization_type in quantization_types:\n",
    "            tuned_model = model_df[(model_df['pruning'] == prune_type) & (model_df['quantization'] == quantization_type)]\n",
    "            tuned_model_size = base_model_size / tuned_model['size (MB)'].iloc[0].astype(float)\n",
    "            output_table.at[prune_type, quantization_type] = tuned_model_size\n",
    "    display(output_table)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.19"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}