notebooks/storage.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "3739aa997be34ad3",
   "metadata": {
    "is_executing": true
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import pandas as pd\n",
    "from typing import List"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "60ce4c8b773ea64b",
   "metadata": {
    "collapsed": false,
    "is_executing": true,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [],
   "source": [
    "def compute_storage(train_dataset_sizes: List[int], model_size: int, prediction_x_size: int, prediction_y_size: int = 1,\n",
    "                    n_computer_train=3, n_computer_inference=3,\n",
    "                    KZG_BYTES=48, PED_BYTES=32, ECDSA_SIG_BYTES=64, ECDSA_PUBLICKEY_SIZE=32, SHA256_BYTES=32):\n",
    "\n",
    "    n_elements_per_sample = prediction_x_size + prediction_y_size\n",
    "    \n",
    "    commitments = {\n",
    "        \"pc\": {\n",
    "            \"commitment_train\": KZG_BYTES * len(train_dataset_sizes),\n",
    "            \"commitment_model\": KZG_BYTES,\n",
    "            \"commitment_prediction_x\": KZG_BYTES, # we need to keep randomness but then we can not store the commitment, so essentially the same\n",
    "            \"commitment_prediction_y\": KZG_BYTES,\n",
    "            \"commitment_randomness\": KZG_BYTES,\n",
    "            \"trusted_setup\": max(prediction_x_size, prediction_y_size) * KZG_BYTES,\n",
    "        },\n",
    "        \"pc_ipa\": {\n",
    "            \"commitment_train\": PED_BYTES * len(train_dataset_sizes),\n",
    "            \"commitment_model\": PED_BYTES,\n",
    "            \"commitment_prediction_x\": PED_BYTES,\n",
    "            \"commitment_prediction_y\": PED_BYTES,\n",
    "            \"commitment_randomness\": PED_BYTES,\n",
    "            \"trusted_setup\": 0,\n",
    "        },\n",
    "        \"cerebro\": {\n",
    "            \"commitment_train\": sum([PED_BYTES * n_elements_per_sample * size for size in train_dataset_sizes]),\n",
    "            \"commitment_model\": PED_BYTES * model_size,\n",
    "            \"commitment_prediction_x\": PED_BYTES * prediction_x_size, # we could technically recompute these?\n",
    "            \"commitment_prediction_y\": PED_BYTES * prediction_y_size,\n",
    "            \"commitment_randomness\": PED_BYTES,\n",
    "            \"trusted_setup\": 0,\n",
    "        },\n",
    "        \"sha3\": {\n",
    "            \"commitment_train\": SHA256_BYTES * len(train_dataset_sizes),\n",
    "            \"commitment_model\": SHA256_BYTES,\n",
    "            \"commitment_prediction_x\": SHA256_BYTES,\n",
    "            \"commitment_prediction_y\": SHA256_BYTES,\n",
    "            \"commitment_randomness\": SHA256_BYTES,\n",
    "            \"trusted_setup\": 0,\n",
    "        }\n",
    "    }\n",
    "\n",
    "    common = {\n",
    "        \"signature_train\": ECDSA_SIG_BYTES * len(train_dataset_sizes),\n",
    "        \"signature_modelowner\": ECDSA_SIG_BYTES,\n",
    "        \n",
    "        \"signature_computer_train\": ECDSA_SIG_BYTES, # joint sig\n",
    "        \"signature_computer_inference\": ECDSA_SIG_BYTES,\n",
    "        \n",
    "        \"identity_train\": ECDSA_PUBLICKEY_SIZE * len(train_dataset_sizes),\n",
    "        \"identity_modelowner\": ECDSA_PUBLICKEY_SIZE,\n",
    "        \"identity_computer_train\": ECDSA_PUBLICKEY_SIZE * n_computer_train,\n",
    "        \"identity_computer_inference\": ECDSA_PUBLICKEY_SIZE * n_computer_inference\n",
    "    }\n",
    "    \n",
    "    # to each entry in commitments add common items\n",
    "    for commitment in commitments.values():\n",
    "        commitment.update(common)\n",
    "        \n",
    "    return commitments\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "fda2404697007d6f",
   "metadata": {
    "collapsed": false,
    "is_executing": true,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "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>approach</th>\n",
       "      <th>dataset</th>\n",
       "      <th>commitment_train</th>\n",
       "      <th>commitment_model</th>\n",
       "      <th>commitment_prediction_x</th>\n",
       "      <th>commitment_prediction_y</th>\n",
       "      <th>commitment_randomness</th>\n",
       "      <th>trusted_setup</th>\n",
       "      <th>signature_train</th>\n",
       "      <th>signature_modelowner</th>\n",
       "      <th>signature_computer_train</th>\n",
       "      <th>signature_computer_inference</th>\n",
       "      <th>identity_train</th>\n",
       "      <th>identity_modelowner</th>\n",
       "      <th>identity_computer_train</th>\n",
       "      <th>identity_computer_inference</th>\n",
       "      <th>storage_train</th>\n",
       "      <th>storage_inference</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>pc</td>\n",
       "      <td>adult</td>\n",
       "      <td>144</td>\n",
       "      <td>48</td>\n",
       "      <td>48</td>\n",
       "      <td>48</td>\n",
       "      <td>48</td>\n",
       "      <td>4368</td>\n",
       "      <td>192</td>\n",
       "      <td>64</td>\n",
       "      <td>64</td>\n",
       "      <td>64</td>\n",
       "      <td>96</td>\n",
       "      <td>32</td>\n",
       "      <td>96</td>\n",
       "      <td>96</td>\n",
       "      <td>496</td>\n",
       "      <td>720</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>pc</td>\n",
       "      <td>mnist_full</td>\n",
       "      <td>144</td>\n",
       "      <td>48</td>\n",
       "      <td>48</td>\n",
       "      <td>48</td>\n",
       "      <td>48</td>\n",
       "      <td>37632</td>\n",
       "      <td>192</td>\n",
       "      <td>64</td>\n",
       "      <td>64</td>\n",
       "      <td>64</td>\n",
       "      <td>96</td>\n",
       "      <td>32</td>\n",
       "      <td>96</td>\n",
       "      <td>96</td>\n",
       "      <td>496</td>\n",
       "      <td>720</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>pc</td>\n",
       "      <td>cifar_alexnet</td>\n",
       "      <td>144</td>\n",
       "      <td>48</td>\n",
       "      <td>48</td>\n",
       "      <td>48</td>\n",
       "      <td>48</td>\n",
       "      <td>147456</td>\n",
       "      <td>192</td>\n",
       "      <td>64</td>\n",
       "      <td>64</td>\n",
       "      <td>64</td>\n",
       "      <td>96</td>\n",
       "      <td>32</td>\n",
       "      <td>96</td>\n",
       "      <td>96</td>\n",
       "      <td>496</td>\n",
       "      <td>720</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>pc</td>\n",
       "      <td>glue_qnli_bert</td>\n",
       "      <td>144</td>\n",
       "      <td>48</td>\n",
       "      <td>48</td>\n",
       "      <td>48</td>\n",
       "      <td>48</td>\n",
       "      <td>4718592</td>\n",
       "      <td>192</td>\n",
       "      <td>64</td>\n",
       "      <td>64</td>\n",
       "      <td>64</td>\n",
       "      <td>96</td>\n",
       "      <td>32</td>\n",
       "      <td>96</td>\n",
       "      <td>96</td>\n",
       "      <td>496</td>\n",
       "      <td>720</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>pc_ipa</td>\n",
       "      <td>adult</td>\n",
       "      <td>96</td>\n",
       "      <td>32</td>\n",
       "      <td>32</td>\n",
       "      <td>32</td>\n",
       "      <td>32</td>\n",
       "      <td>0</td>\n",
       "      <td>192</td>\n",
       "      <td>64</td>\n",
       "      <td>64</td>\n",
       "      <td>64</td>\n",
       "      <td>96</td>\n",
       "      <td>32</td>\n",
       "      <td>96</td>\n",
       "      <td>96</td>\n",
       "      <td>416</td>\n",
       "      <td>608</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>pc_ipa</td>\n",
       "      <td>mnist_full</td>\n",
       "      <td>96</td>\n",
       "      <td>32</td>\n",
       "      <td>32</td>\n",
       "      <td>32</td>\n",
       "      <td>32</td>\n",
       "      <td>0</td>\n",
       "      <td>192</td>\n",
       "      <td>64</td>\n",
       "      <td>64</td>\n",
       "      <td>64</td>\n",
       "      <td>96</td>\n",
       "      <td>32</td>\n",
       "      <td>96</td>\n",
       "      <td>96</td>\n",
       "      <td>416</td>\n",
       "      <td>608</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>pc_ipa</td>\n",
       "      <td>cifar_alexnet</td>\n",
       "      <td>96</td>\n",
       "      <td>32</td>\n",
       "      <td>32</td>\n",
       "      <td>32</td>\n",
       "      <td>32</td>\n",
       "      <td>0</td>\n",
       "      <td>192</td>\n",
       "      <td>64</td>\n",
       "      <td>64</td>\n",
       "      <td>64</td>\n",
       "      <td>96</td>\n",
       "      <td>32</td>\n",
       "      <td>96</td>\n",
       "      <td>96</td>\n",
       "      <td>416</td>\n",
       "      <td>608</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>pc_ipa</td>\n",
       "      <td>glue_qnli_bert</td>\n",
       "      <td>96</td>\n",
       "      <td>32</td>\n",
       "      <td>32</td>\n",
       "      <td>32</td>\n",
       "      <td>32</td>\n",
       "      <td>0</td>\n",
       "      <td>192</td>\n",
       "      <td>64</td>\n",
       "      <td>64</td>\n",
       "      <td>64</td>\n",
       "      <td>96</td>\n",
       "      <td>32</td>\n",
       "      <td>96</td>\n",
       "      <td>96</td>\n",
       "      <td>416</td>\n",
       "      <td>608</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>cerebro</td>\n",
       "      <td>adult</td>\n",
       "      <td>76685312</td>\n",
       "      <td>95264</td>\n",
       "      <td>2912</td>\n",
       "      <td>32</td>\n",
       "      <td>32</td>\n",
       "      <td>0</td>\n",
       "      <td>192</td>\n",
       "      <td>64</td>\n",
       "      <td>64</td>\n",
       "      <td>64</td>\n",
       "      <td>96</td>\n",
       "      <td>32</td>\n",
       "      <td>96</td>\n",
       "      <td>96</td>\n",
       "      <td>76780864</td>\n",
       "      <td>76783936</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>cerebro</td>\n",
       "      <td>mnist_full</td>\n",
       "      <td>1507200000</td>\n",
       "      <td>13794560</td>\n",
       "      <td>25088</td>\n",
       "      <td>32</td>\n",
       "      <td>32</td>\n",
       "      <td>0</td>\n",
       "      <td>192</td>\n",
       "      <td>64</td>\n",
       "      <td>64</td>\n",
       "      <td>64</td>\n",
       "      <td>96</td>\n",
       "      <td>32</td>\n",
       "      <td>96</td>\n",
       "      <td>96</td>\n",
       "      <td>1520994848</td>\n",
       "      <td>1521020096</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>cerebro</td>\n",
       "      <td>cifar_alexnet</td>\n",
       "      <td>4916800000</td>\n",
       "      <td>124225856</td>\n",
       "      <td>98304</td>\n",
       "      <td>32</td>\n",
       "      <td>32</td>\n",
       "      <td>0</td>\n",
       "      <td>192</td>\n",
       "      <td>64</td>\n",
       "      <td>64</td>\n",
       "      <td>64</td>\n",
       "      <td>96</td>\n",
       "      <td>32</td>\n",
       "      <td>96</td>\n",
       "      <td>96</td>\n",
       "      <td>5041026144</td>\n",
       "      <td>5041124608</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>cerebro</td>\n",
       "      <td>glue_qnli_bert</td>\n",
       "      <td>7861254240</td>\n",
       "      <td>2740691040</td>\n",
       "      <td>3145728</td>\n",
       "      <td>32</td>\n",
       "      <td>32</td>\n",
       "      <td>0</td>\n",
       "      <td>192</td>\n",
       "      <td>64</td>\n",
       "      <td>64</td>\n",
       "      <td>64</td>\n",
       "      <td>96</td>\n",
       "      <td>32</td>\n",
       "      <td>96</td>\n",
       "      <td>96</td>\n",
       "      <td>10601945568</td>\n",
       "      <td>10605091456</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>sha3</td>\n",
       "      <td>adult</td>\n",
       "      <td>96</td>\n",
       "      <td>32</td>\n",
       "      <td>32</td>\n",
       "      <td>32</td>\n",
       "      <td>32</td>\n",
       "      <td>0</td>\n",
       "      <td>192</td>\n",
       "      <td>64</td>\n",
       "      <td>64</td>\n",
       "      <td>64</td>\n",
       "      <td>96</td>\n",
       "      <td>32</td>\n",
       "      <td>96</td>\n",
       "      <td>96</td>\n",
       "      <td>416</td>\n",
       "      <td>608</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>sha3</td>\n",
       "      <td>mnist_full</td>\n",
       "      <td>96</td>\n",
       "      <td>32</td>\n",
       "      <td>32</td>\n",
       "      <td>32</td>\n",
       "      <td>32</td>\n",
       "      <td>0</td>\n",
       "      <td>192</td>\n",
       "      <td>64</td>\n",
       "      <td>64</td>\n",
       "      <td>64</td>\n",
       "      <td>96</td>\n",
       "      <td>32</td>\n",
       "      <td>96</td>\n",
       "      <td>96</td>\n",
       "      <td>416</td>\n",
       "      <td>608</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>sha3</td>\n",
       "      <td>cifar_alexnet</td>\n",
       "      <td>96</td>\n",
       "      <td>32</td>\n",
       "      <td>32</td>\n",
       "      <td>32</td>\n",
       "      <td>32</td>\n",
       "      <td>0</td>\n",
       "      <td>192</td>\n",
       "      <td>64</td>\n",
       "      <td>64</td>\n",
       "      <td>64</td>\n",
       "      <td>96</td>\n",
       "      <td>32</td>\n",
       "      <td>96</td>\n",
       "      <td>96</td>\n",
       "      <td>416</td>\n",
       "      <td>608</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>sha3</td>\n",
       "      <td>glue_qnli_bert</td>\n",
       "      <td>96</td>\n",
       "      <td>32</td>\n",
       "      <td>32</td>\n",
       "      <td>32</td>\n",
       "      <td>32</td>\n",
       "      <td>0</td>\n",
       "      <td>192</td>\n",
       "      <td>64</td>\n",
       "      <td>64</td>\n",
       "      <td>64</td>\n",
       "      <td>96</td>\n",
       "      <td>32</td>\n",
       "      <td>96</td>\n",
       "      <td>96</td>\n",
       "      <td>416</td>\n",
       "      <td>608</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   approach         dataset  commitment_train  commitment_model  \\\n",
       "0        pc           adult               144                48   \n",
       "1        pc      mnist_full               144                48   \n",
       "2        pc   cifar_alexnet               144                48   \n",
       "3        pc  glue_qnli_bert               144                48   \n",
       "4    pc_ipa           adult                96                32   \n",
       "5    pc_ipa      mnist_full                96                32   \n",
       "6    pc_ipa   cifar_alexnet                96                32   \n",
       "7    pc_ipa  glue_qnli_bert                96                32   \n",
       "8   cerebro           adult          76685312             95264   \n",
       "9   cerebro      mnist_full        1507200000          13794560   \n",
       "10  cerebro   cifar_alexnet        4916800000         124225856   \n",
       "11  cerebro  glue_qnli_bert        7861254240        2740691040   \n",
       "12     sha3           adult                96                32   \n",
       "13     sha3      mnist_full                96                32   \n",
       "14     sha3   cifar_alexnet                96                32   \n",
       "15     sha3  glue_qnli_bert                96                32   \n",
       "\n",
       "    commitment_prediction_x  commitment_prediction_y  commitment_randomness  \\\n",
       "0                        48                       48                     48   \n",
       "1                        48                       48                     48   \n",
       "2                        48                       48                     48   \n",
       "3                        48                       48                     48   \n",
       "4                        32                       32                     32   \n",
       "5                        32                       32                     32   \n",
       "6                        32                       32                     32   \n",
       "7                        32                       32                     32   \n",
       "8                      2912                       32                     32   \n",
       "9                     25088                       32                     32   \n",
       "10                    98304                       32                     32   \n",
       "11                  3145728                       32                     32   \n",
       "12                       32                       32                     32   \n",
       "13                       32                       32                     32   \n",
       "14                       32                       32                     32   \n",
       "15                       32                       32                     32   \n",
       "\n",
       "    trusted_setup  signature_train  signature_modelowner  \\\n",
       "0            4368              192                    64   \n",
       "1           37632              192                    64   \n",
       "2          147456              192                    64   \n",
       "3         4718592              192                    64   \n",
       "4               0              192                    64   \n",
       "5               0              192                    64   \n",
       "6               0              192                    64   \n",
       "7               0              192                    64   \n",
       "8               0              192                    64   \n",
       "9               0              192                    64   \n",
       "10              0              192                    64   \n",
       "11              0              192                    64   \n",
       "12              0              192                    64   \n",
       "13              0              192                    64   \n",
       "14              0              192                    64   \n",
       "15              0              192                    64   \n",
       "\n",
       "    signature_computer_train  signature_computer_inference  identity_train  \\\n",
       "0                         64                            64              96   \n",
       "1                         64                            64              96   \n",
       "2                         64                            64              96   \n",
       "3                         64                            64              96   \n",
       "4                         64                            64              96   \n",
       "5                         64                            64              96   \n",
       "6                         64                            64              96   \n",
       "7                         64                            64              96   \n",
       "8                         64                            64              96   \n",
       "9                         64                            64              96   \n",
       "10                        64                            64              96   \n",
       "11                        64                            64              96   \n",
       "12                        64                            64              96   \n",
       "13                        64                            64              96   \n",
       "14                        64                            64              96   \n",
       "15                        64                            64              96   \n",
       "\n",
       "    identity_modelowner  identity_computer_train  identity_computer_inference  \\\n",
       "0                    32                       96                           96   \n",
       "1                    32                       96                           96   \n",
       "2                    32                       96                           96   \n",
       "3                    32                       96                           96   \n",
       "4                    32                       96                           96   \n",
       "5                    32                       96                           96   \n",
       "6                    32                       96                           96   \n",
       "7                    32                       96                           96   \n",
       "8                    32                       96                           96   \n",
       "9                    32                       96                           96   \n",
       "10                   32                       96                           96   \n",
       "11                   32                       96                           96   \n",
       "12                   32                       96                           96   \n",
       "13                   32                       96                           96   \n",
       "14                   32                       96                           96   \n",
       "15                   32                       96                           96   \n",
       "\n",
       "    storage_train  storage_inference  \n",
       "0             496                720  \n",
       "1             496                720  \n",
       "2             496                720  \n",
       "3             496                720  \n",
       "4             416                608  \n",
       "5             416                608  \n",
       "6             416                608  \n",
       "7             416                608  \n",
       "8        76780864           76783936  \n",
       "9      1520994848         1521020096  \n",
       "10     5041026144         5041124608  \n",
       "11    10601945568        10605091456  \n",
       "12            416                608  \n",
       "13            416                608  \n",
       "14            416                608  \n",
       "15            416                608  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "s_adult = compute_storage([8684, 8746, 8618], 2977, 91, 1)\n",
    "s_mnist = compute_storage([20012, 20057, 19931], 431080, 784, 1)\n",
    "s_cifar = compute_storage([16709, 16682, 16609], 3882058, 3072, 1)\n",
    "\n",
    "hidden_state = 768\n",
    "seq_len = 128\n",
    "s_stanford_dataset = 104743\n",
    "#approx 2500\n",
    "s_stanfordqa = compute_storage([833, 833, 833], \n",
    "                               85646595, hidden_state * seq_len, 1)\n",
    "\n",
    "# add to dataframe\n",
    "df = pd.DataFrame([s_adult, s_mnist, s_cifar, s_stanfordqa], index=[\"adult\", \"mnist_full\", \"cifar_alexnet\", \"glue_qnli_bert\"])\n",
    "df_unstacked = df.unstack().apply(pd.Series).reset_index()\n",
    "df_unstacked = df_unstacked.rename(columns= {\n",
    "    \"level_0\": \"approach\",\n",
    "    \"level_1\": \"dataset\"\n",
    "})\n",
    "\n",
    "# we ignore the identity size\n",
    "df_unstacked[\"storage_train\"] = df_unstacked[[\"commitment_train\", \"signature_train\", \"signature_computer_train\", \"commitment_model\", \"commitment_randomness\"]].sum(axis=1)\n",
    "df_unstacked[\"storage_inference\"] = df_unstacked[\"storage_train\"] + df_unstacked[[\"commitment_prediction_x\", \"commitment_prediction_y\", \"signature_computer_inference\", \"signature_modelowner\"]].sum(axis=1)\n",
    "\n",
    "display(df_unstacked)\n",
    "\n",
    "# now adapt so it looks like what we expect for the etl\n",
    "df_unstacked[\"approach\"] = df_unstacked[\"approach\"].replace(\"sha3\", \"sha3s\")\n",
    "df_unstacked[\"dataset\"] = df_unstacked[\"dataset\"].replace(\"mnist\", \"mnist_full\")\n",
    "df_unstacked[\"dataset\"] = df_unstacked[\"dataset\"].replace(\"cifar\", \"cifar_alexnet\")\n",
    "df_unstacked = df_unstacked.drop(df_unstacked[df_unstacked['approach'] == 'pc_ipa'].index)\n",
    "\n",
    "df_unstacked = df_unstacked.rename(columns = {\n",
    "    \"approach\": \"consistency_args.type\",\n",
    "    \"dataset\": \"mpc.script_args.dataset\"\n",
    "})\n",
    "\n",
    "df_unstacked.to_csv(\"storage.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "7ea6dad1c32c3b9b",
   "metadata": {
    "collapsed": false,
    "is_executing": true,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "63639471426450da",
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.9.16"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}