sample_ditto.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "from certa.explain import CertaExplainer\n",
    "from certa.utils import merge_sources\n",
    "from certa.models.ditto.ditto import DittoModel"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "datadir = 'datasets/Beer'\n",
    "lsource = pd.read_csv(datadir + '/tableA.csv')\n",
    "rsource = pd.read_csv(datadir + '/tableB.csv')\n",
    "gt = pd.read_csv(datadir + '/train.csv')\n",
    "valid = pd.read_csv(datadir + '/valid.csv')\n",
    "test = pd.read_csv(datadir + '/test.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "\n",
    "pt_model_dict = torch.load('ditto/Structured/Beer/model.pt')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Some weights of the model checkpoint at distilbert-base-uncased were not used when initializing DistilBertModel: ['vocab_layer_norm.bias', 'vocab_transform.bias', 'vocab_layer_norm.weight', 'vocab_projector.weight', 'vocab_projector.bias', 'vocab_transform.weight']\n",
      "- This IS expected if you are initializing DistilBertModel from the checkpoint of a model trained on another task or with another architecture (e.g. initializing a BertForSequenceClassification model from a BertForPreTraining model).\n",
      "- This IS NOT expected if you are initializing DistilBertModel from the checkpoint of a model that you expect to be exactly identical (initializing a BertForSequenceClassification model from a BertForSequenceClassification model).\n"
     ]
    }
   ],
   "source": [
    "ditto_model = DittoModel(lm='distilbert', device='cpu')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": "<All keys matched successfully>"
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ditto_model.load_state_dict(pt_model_dict['model'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "from certa.models.ditto.summarize import Summarizer\n",
    "\n",
    "summarizer = Summarizer(lsource, rsource, 'distilbert')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "certa_explainer = CertaExplainer(lsource, rsource)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Some weights of the model checkpoint at distilbert-base-uncased were not used when initializing DistilBertModel: ['vocab_layer_norm.bias', 'vocab_transform.bias', 'vocab_layer_norm.weight', 'vocab_projector.weight', 'vocab_projector.bias', 'vocab_transform.weight']\n",
      "- This IS expected if you are initializing DistilBertModel from the checkpoint of a model trained on another task or with another architecture (e.g. initializing a BertForSequenceClassification model from a BertForPreTraining model).\n",
      "- This IS NOT expected if you are initializing DistilBertModel from the checkpoint of a model that you expect to be exactly identical (initializing a BertForSequenceClassification model from a BertForSequenceClassification model).\n"
     ]
    }
   ],
   "source": [
    "from certa.models.bert import EMTERModel\n",
    "\n",
    "model = EMTERModel(ditto=True, summarizer=summarizer, dk='general')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.model = ditto_model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "def predict_fn(x):\n",
    "    return model.predict(x, len=64)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "test_df = merge_sources(test, 'ltable_', 'rtable_', lsource, rsource, ['label'], [])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": "label                                                1\nltable_ABV                                      7.40 %\nltable_Beer_Name                     Honey Basil Amber\nltable_Brew_Factory_Name    Rude Hippo Brewing Company\nltable_Style                  American Amber / Red Ale\nName: 2, dtype: object"
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "idx = 2\n",
    "rand_row = test_df.iloc[idx]\n",
    "l_id = int(rand_row['ltable_id'])\n",
    "l_tuple = lsource.iloc[l_id]\n",
    "r_id = int(rand_row['rtable_id'])\n",
    "r_tuple = rsource.iloc[r_id]\n",
    "rand_row.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": "id                                         3917\nBeer_Name                     Honey Basil Amber\nBrew_Factory_Name    Rude Hippo Brewing Company\nStyle                  American Amber / Red Ale\nABV                                      7.40 %\nName: 3917, dtype: object"
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "l_tuple"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": "id                                           2224\nBeer_Name            Rude Hippo Honey Basil Amber\nBrew_Factory_Name             18th Street Brewery\nStyle                                   Amber Ale\nABV                                        7.40 %\nName: 2224, dtype: object"
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "r_tuple"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/tteofili/.local/lib/python3.6/site-packages/spacy/pipeline/lemmatizer.py:211: UserWarning: [W108] The rule-based lemmatizer did not find POS annotation for one or more tokens. Check that your pipeline includes components that assign token.pos, typically 'tagger'+'attribute_ruler' or 'morphologizer'.\n",
      "  warnings.warn(Warnings.W108)\n",
      "/home/tteofili/.local/lib/python3.6/site-packages/spacy/pipeline/lemmatizer.py:211: UserWarning: [W108] The rule-based lemmatizer did not find POS annotation for one or more tokens. Check that your pipeline includes components that assign token.pos, typically 'tagger'+'attribute_ruler' or 'morphologizer'.\n",
      "  warnings.warn(Warnings.W108)\n",
      "/home/tteofili/.local/lib/python3.6/site-packages/spacy/pipeline/lemmatizer.py:211: UserWarning: [W108] The rule-based lemmatizer did not find POS annotation for one or more tokens. Check that your pipeline includes components that assign token.pos, typically 'tagger'+'attribute_ruler' or 'morphologizer'.\n",
      "  warnings.warn(Warnings.W108)\n",
      "/home/tteofili/.local/lib/python3.6/site-packages/spacy/pipeline/lemmatizer.py:211: UserWarning: [W108] The rule-based lemmatizer did not find POS annotation for one or more tokens. Check that your pipeline includes components that assign token.pos, typically 'tagger'+'attribute_ruler' or 'morphologizer'.\n",
      "  warnings.warn(Warnings.W108)\n",
      "/home/tteofili/.local/lib/python3.6/site-packages/spacy/pipeline/lemmatizer.py:211: UserWarning: [W108] The rule-based lemmatizer did not find POS annotation for one or more tokens. Check that your pipeline includes components that assign token.pos, typically 'tagger'+'attribute_ruler' or 'morphologizer'.\n",
      "  warnings.warn(Warnings.W108)\n"
     ]
    }
   ],
   "source": [
    "saliency_df, cf_summary, counterfactual_examples, triangles, lattices = certa_explainer.explain(l_tuple, r_tuple, predict_fn, num_triangles=100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": "{'ltable_Beer_Name': {0: 0.5113636363636364},\n 'ltable_Brew_Factory_Name': {0: 0.375},\n 'ltable_Style': {0: 0.3181818181818182},\n 'ltable_ABV': {0: 0.2954545454545454},\n 'rtable_Beer_Name': {0: 0.5113636363636364},\n 'rtable_Brew_Factory_Name': {0: 0.28409090909090906},\n 'rtable_Style': {0: 0.28409090909090906},\n 'rtable_ABV': {0: 0.28409090909090906}}"
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "saliency_df.to_dict()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "outputs": [
    {
     "data": {
      "text/plain": "ltable_Beer_Name                                          0.5\nltable_Brew_Factory_Name                                  0.5\nltable_Style                                              0.0\nltable_ABV                                                0.0\nrtable_Beer_Name                                          0.5\nrtable_Brew_Factory_Name                                  0.0\nrtable_Style                                              0.0\nrtable_ABV                                                0.0\nltable_Beer_Name/ltable_Brew_Factory_Name                 0.5\nltable_Beer_Name/ltable_Style                             0.5\nltable_Beer_Name/ltable_ABV                               0.5\nltable_Brew_Factory_Name/ltable_Style                     0.2\nltable_Brew_Factory_Name/ltable_ABV                       0.0\nltable_Style/ltable_ABV                                   0.0\nrtable_Beer_Name/rtable_Brew_Factory_Name                 0.5\nrtable_Beer_Name/rtable_Style                             0.5\nrtable_Beer_Name/rtable_ABV                               0.5\nrtable_Brew_Factory_Name/rtable_Style                     0.0\nrtable_Brew_Factory_Name/rtable_ABV                       0.0\nrtable_Style/rtable_ABV                                   0.0\nltable_Beer_Name/ltable_Brew_Factory_Name/ltable_Style    0.5\nltable_Beer_Name/ltable_Brew_Factory_Name/ltable_ABV      0.5\nltable_Beer_Name/ltable_Style/ltable_ABV                  0.5\nltable_Brew_Factory_Name/ltable_Style/ltable_ABV          0.1\nrtable_Beer_Name/rtable_Brew_Factory_Name/rtable_Style    0.5\nrtable_Beer_Name/rtable_Brew_Factory_Name/rtable_ABV      0.5\nrtable_Beer_Name/rtable_Style/rtable_ABV                  0.5\nrtable_Brew_Factory_Name/rtable_Style/rtable_ABV          0.0\ndtype: float64"
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cf_summary"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    },
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": "   label                                   ltable_Beer_Name  \\\n0         T_ebonickà © Geronimo Polotmavà 1/2 Rà 1/2 Å 3...   \n8         T_ebonickà © Geronimo Polotmavà 1/2 Rà 1/2 Å 3...   \n10                                        Honey Basil Amber   \n12                                        Honey Basil Amber   \n11                                        Honey Basil Amber   \n14                                        Honey Basil Amber   \n13                                        Honey Basil Amber   \n\n      ltable_Brew_Factory_Name              ltable_Style ltable_ABV  \\\n0   Rude Hippo Brewing Company  American Amber / Red Ale     7.40 %   \n8   Rude Hippo Brewing Company  American Amber / Red Ale     7.40 %   \n10  Rude Hippo Brewing Company  American Amber / Red Ale     7.40 %   \n12  Rude Hippo Brewing Company  American Amber / Red Ale     7.40 %   \n11  Rude Hippo Brewing Company  American Amber / Red Ale     7.40 %   \n14  Rude Hippo Brewing Company  American Amber / Red Ale     7.40 %   \n13  Rude Hippo Brewing Company  American Amber / Red Ale     7.40 %   \n\n                        rtable_Beer_Name rtable_Brew_Factory_Name  \\\n0           Rude Hippo Honey Basil Amber      18th Street Brewery   \n8           Rude Hippo Honey Basil Amber      18th Street Brewery   \n10       te de Jade Brigantine Ambrà © e      18th Street Brewery   \n12                 _ i_lavar Barley Wine      18th Street Brewery   \n11    Cà ´ te de Jade Brigantine Ambrà ©      18th Street Brewery   \n14  Cà ´ te de Jade Brigantine Ambrà © e      18th Street Brewery   \n13      Cà ´ te de Jade Brigantine Ambrà      18th Street Brewery   \n\n   rtable_Style rtable_ABV          match_score       nomatch_score  \\\n0     Amber Ale     7.40 %  0.19255195558071136  0.8074480444192886   \n8     Amber Ale     7.40 %  0.17068031430244446  0.8293196856975555   \n10    Amber Ale     7.40 %  0.10978817939758301   0.890211820602417   \n12    Amber Ale     7.40 %  0.09088551998138428  0.9091144800186157   \n11    Amber Ale     7.40 %  0.10560467094182968  0.8943953290581703   \n14    Amber Ale     7.40 %  0.08469007909297943  0.9153099209070206   \n13    Amber Ale     7.40 %  0.10777609050273895   0.892223909497261   \n\n        alteredAttributes                     droppedValues  \\\n0   ('ltable_Beer_Name',)             ['Honey Basil Amber']   \n8   ('ltable_Beer_Name',)             ['Honey Basil Amber']   \n10  ('rtable_Beer_Name',)  ['Rude Hippo Honey Basil Amber']   \n12  ('rtable_Beer_Name',)  ['Rude Hippo Honey Basil Amber']   \n11  ('rtable_Beer_Name',)  ['Rude Hippo Honey Basil Amber']   \n14  ('rtable_Beer_Name',)  ['Rude Hippo Honey Basil Amber']   \n13  ('rtable_Beer_Name',)  ['Rude Hippo Honey Basil Amber']   \n\n                                         copiedValues               triangle  \\\n0   ['T_ebonickà © Geronimo Polotmavà 1/2 Rà 1/2 Å...  0@3917 1@2224 0@72990   \n8   ['T_ebonickà © Geronimo Polotmavà 1/2 Rà 1/2 Å...  0@3917 1@2224 0@72978   \n10                ['te de Jade Brigantine Ambrà © e']  1@2224 0@3917 1@25064   \n12                          ['_ i_lavar Barley Wine']  1@2224 0@3917 1@42959   \n11             ['Cà ´ te de Jade Brigantine Ambrà ©']  1@2224 0@3917 1@25077   \n14           ['Cà ´ te de Jade Brigantine Ambrà © e']  1@2224 0@3917 1@25100   \n13               ['Cà ´ te de Jade Brigantine AmbrÃ']  1@2224 0@3917 1@25075   \n\n   attr_count  \n0           2  \n8           2  \n10          2  \n12          2  \n11          2  \n14          2  \n13          2  ",
      "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>label</th>\n      <th>ltable_Beer_Name</th>\n      <th>ltable_Brew_Factory_Name</th>\n      <th>ltable_Style</th>\n      <th>ltable_ABV</th>\n      <th>rtable_Beer_Name</th>\n      <th>rtable_Brew_Factory_Name</th>\n      <th>rtable_Style</th>\n      <th>rtable_ABV</th>\n      <th>match_score</th>\n      <th>nomatch_score</th>\n      <th>alteredAttributes</th>\n      <th>droppedValues</th>\n      <th>copiedValues</th>\n      <th>triangle</th>\n      <th>attr_count</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td></td>\n      <td>T_ebonickà © Geronimo Polotmavà 1/2 Rà 1/2 Å 3...</td>\n      <td>Rude Hippo Brewing Company</td>\n      <td>American Amber / Red Ale</td>\n      <td>7.40 %</td>\n      <td>Rude Hippo Honey Basil Amber</td>\n      <td>18th Street Brewery</td>\n      <td>Amber Ale</td>\n      <td>7.40 %</td>\n      <td>0.19255195558071136</td>\n      <td>0.8074480444192886</td>\n      <td>('ltable_Beer_Name',)</td>\n      <td>['Honey Basil Amber']</td>\n      <td>['T_ebonickà © Geronimo Polotmavà 1/2 Rà 1/2 Å...</td>\n      <td>0@3917 1@2224 0@72990</td>\n      <td>2</td>\n    </tr>\n    <tr>\n      <th>8</th>\n      <td></td>\n      <td>T_ebonickà © Geronimo Polotmavà 1/2 Rà 1/2 Å 3...</td>\n      <td>Rude Hippo Brewing Company</td>\n      <td>American Amber / Red Ale</td>\n      <td>7.40 %</td>\n      <td>Rude Hippo Honey Basil Amber</td>\n      <td>18th Street Brewery</td>\n      <td>Amber Ale</td>\n      <td>7.40 %</td>\n      <td>0.17068031430244446</td>\n      <td>0.8293196856975555</td>\n      <td>('ltable_Beer_Name',)</td>\n      <td>['Honey Basil Amber']</td>\n      <td>['T_ebonickà © Geronimo Polotmavà 1/2 Rà 1/2 Å...</td>\n      <td>0@3917 1@2224 0@72978</td>\n      <td>2</td>\n    </tr>\n    <tr>\n      <th>10</th>\n      <td></td>\n      <td>Honey Basil Amber</td>\n      <td>Rude Hippo Brewing Company</td>\n      <td>American Amber / Red Ale</td>\n      <td>7.40 %</td>\n      <td>te de Jade Brigantine Ambrà © e</td>\n      <td>18th Street Brewery</td>\n      <td>Amber Ale</td>\n      <td>7.40 %</td>\n      <td>0.10978817939758301</td>\n      <td>0.890211820602417</td>\n      <td>('rtable_Beer_Name',)</td>\n      <td>['Rude Hippo Honey Basil Amber']</td>\n      <td>['te de Jade Brigantine Ambrà © e']</td>\n      <td>1@2224 0@3917 1@25064</td>\n      <td>2</td>\n    </tr>\n    <tr>\n      <th>12</th>\n      <td></td>\n      <td>Honey Basil Amber</td>\n      <td>Rude Hippo Brewing Company</td>\n      <td>American Amber / Red Ale</td>\n      <td>7.40 %</td>\n      <td>_ i_lavar Barley Wine</td>\n      <td>18th Street Brewery</td>\n      <td>Amber Ale</td>\n      <td>7.40 %</td>\n      <td>0.09088551998138428</td>\n      <td>0.9091144800186157</td>\n      <td>('rtable_Beer_Name',)</td>\n      <td>['Rude Hippo Honey Basil Amber']</td>\n      <td>['_ i_lavar Barley Wine']</td>\n      <td>1@2224 0@3917 1@42959</td>\n      <td>2</td>\n    </tr>\n    <tr>\n      <th>11</th>\n      <td></td>\n      <td>Honey Basil Amber</td>\n      <td>Rude Hippo Brewing Company</td>\n      <td>American Amber / Red Ale</td>\n      <td>7.40 %</td>\n      <td>Cà ´ te de Jade Brigantine Ambrà ©</td>\n      <td>18th Street Brewery</td>\n      <td>Amber Ale</td>\n      <td>7.40 %</td>\n      <td>0.10560467094182968</td>\n      <td>0.8943953290581703</td>\n      <td>('rtable_Beer_Name',)</td>\n      <td>['Rude Hippo Honey Basil Amber']</td>\n      <td>['Cà ´ te de Jade Brigantine Ambrà ©']</td>\n      <td>1@2224 0@3917 1@25077</td>\n      <td>2</td>\n    </tr>\n    <tr>\n      <th>14</th>\n      <td></td>\n      <td>Honey Basil Amber</td>\n      <td>Rude Hippo Brewing Company</td>\n      <td>American Amber / Red Ale</td>\n      <td>7.40 %</td>\n      <td>Cà ´ te de Jade Brigantine Ambrà © e</td>\n      <td>18th Street Brewery</td>\n      <td>Amber Ale</td>\n      <td>7.40 %</td>\n      <td>0.08469007909297943</td>\n      <td>0.9153099209070206</td>\n      <td>('rtable_Beer_Name',)</td>\n      <td>['Rude Hippo Honey Basil Amber']</td>\n      <td>['Cà ´ te de Jade Brigantine Ambrà © e']</td>\n      <td>1@2224 0@3917 1@25100</td>\n      <td>2</td>\n    </tr>\n    <tr>\n      <th>13</th>\n      <td></td>\n      <td>Honey Basil Amber</td>\n      <td>Rude Hippo Brewing Company</td>\n      <td>American Amber / Red Ale</td>\n      <td>7.40 %</td>\n      <td>Cà ´ te de Jade Brigantine AmbrÃ</td>\n      <td>18th Street Brewery</td>\n      <td>Amber Ale</td>\n      <td>7.40 %</td>\n      <td>0.10777609050273895</td>\n      <td>0.892223909497261</td>\n      <td>('rtable_Beer_Name',)</td>\n      <td>['Rude Hippo Honey Basil Amber']</td>\n      <td>['Cà ´ te de Jade Brigantine AmbrÃ']</td>\n      <td>1@2224 0@3917 1@25075</td>\n      <td>2</td>\n    </tr>\n  </tbody>\n</table>\n</div>"
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "counterfactual_examples"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}