From 92d2d169be617119e50a3921944c2a01d036f319 Mon Sep 17 00:00:00 2001
From: Yuval Uner <yuvaluner@gmail.com>
Date: Mon, 10 Feb 2025 16:46:59 +0200
Subject: [PATCH 1/4] Migrated outlier explainer from fedex to this repository

---
 README.md                                     |  10 +-
 .../notebooks/Outlier explainer demo.ipynb    | 380 +++++++++++++++++
 requirements.txt                              |   3 +
 setup.py                                      |  42 ++
 src/external_explainers/__init__.py           |   1 +
 src/external_explainers/commons/__init__.py   |   1 +
 src/external_explainers/commons/utils.py      |  25 ++
 .../outlier_explainer/__init__.py             |   0
 .../outlier_explainer/outlier_explainer.py    | 394 ++++++++++++++++++
 9 files changed, 855 insertions(+), 1 deletion(-)
 create mode 100644 examples/notebooks/Outlier explainer demo.ipynb
 create mode 100644 requirements.txt
 create mode 100644 setup.py
 create mode 100644 src/external_explainers/__init__.py
 create mode 100644 src/external_explainers/commons/__init__.py
 create mode 100644 src/external_explainers/commons/utils.py
 create mode 100644 src/external_explainers/outlier_explainer/__init__.py
 create mode 100644 src/external_explainers/outlier_explainer/outlier_explainer.py

diff --git a/README.md b/README.md
index 6ac82de..e325410 100644
--- a/README.md
+++ b/README.md
@@ -1,2 +1,10 @@
 # ExternalExplainers
-External explainer implementations for PD-EXPLAIN
+External explainer implementations for [PD-EXPLAIN](https://github.com/analysis-bots/pd-explain).\
+While you can use the explainer implementations in this repository directly, it is recommended to use them through the PD-EXPLAIN library,
+for a much better and more user-friendly experience.
+## Included Explainers
+### Outlier explainer
+This explainer is based on the [SCORPION](https://sirrice.github.io/files/papers/scorpion-vldb13.pdf) paper.\
+Its goal is to provide explanations for outliers in the data, explaining why a certain data point is an outlier.\
+This explainer is meant to work on series created as a result of groupby + aggregation operations.\
+Explainer author: [@Itay Elyashiv](https://github.com/ItayELY)
diff --git a/examples/notebooks/Outlier explainer demo.ipynb b/examples/notebooks/Outlier explainer demo.ipynb
new file mode 100644
index 0000000..04d3260
--- /dev/null
+++ b/examples/notebooks/Outlier explainer demo.ipynb	
@@ -0,0 +1,380 @@
+{
+ "cells": [
+  {
+   "metadata": {},
+   "cell_type": "markdown",
+   "source": "# Outlier explainer demo",
+   "id": "e604881e784a9f9a"
+  },
+  {
+   "metadata": {},
+   "cell_type": "markdown",
+   "source": [
+    "This notebook demonstrates the usage of the outlier explainer.\\\n",
+    "Note that it is recommended to use our other package, pd-explain, for better ease of use and more features."
+   ],
+   "id": "ca1fceec8df719a6"
+  },
+  {
+   "metadata": {},
+   "cell_type": "markdown",
+   "source": "## Imports and loading data",
+   "id": "2ac929f2c20f6248"
+  },
+  {
+   "cell_type": "code",
+   "id": "initial_id",
+   "metadata": {
+    "collapsed": true,
+    "ExecuteTime": {
+     "end_time": "2025-02-10T14:28:34.539967Z",
+     "start_time": "2025-02-10T14:28:33.726878Z"
+    }
+   },
+   "source": [
+    "import pandas as pd\n",
+    "from external_explainers import OutlierExplainer"
+   ],
+   "outputs": [],
+   "execution_count": 1
+  },
+  {
+   "metadata": {
+    "ExecuteTime": {
+     "end_time": "2025-02-10T14:28:37.085917Z",
+     "start_time": "2025-02-10T14:28:34.548929Z"
+    }
+   },
+   "cell_type": "code",
+   "source": [
+    "spotify_data = pd.read_csv('https://raw.githubusercontent.com/analysis-bots/pd-explain/refs/heads/main/Examples/Datasets/spotify_all.csv')\n",
+    "spotify_data.head()"
+   ],
+   "id": "4a877eb58c908b7d",
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "   acousticness                    artists  danceability  energy  explicit  \\\n",
+       "0      0.991000            ['Mamie Smith']         0.598   0.224         0   \n",
+       "1      0.643000  [\"Screamin' Jay Hawkins\"]         0.852   0.517         0   \n",
+       "2      0.993000            ['Mamie Smith']         0.647   0.186         0   \n",
+       "3      0.000173        ['Oscar Velazquez']         0.730   0.798         0   \n",
+       "4      0.295000                   ['Mixe']         0.704   0.707         1   \n",
+       "\n",
+       "                       id  instrumentalness  key  liveness  loudness  ...  \\\n",
+       "0  0cS0A1fUEUd1EW3FcF8AEI          0.000522    5    0.3790   -12.628  ...   \n",
+       "1  0hbkKFIJm7Z05H8Zl9w30f          0.026400    5    0.0809    -7.261  ...   \n",
+       "2  11m7laMUgmOKqI3oYzuhne          0.000018    0    0.5190   -12.098  ...   \n",
+       "3  19Lc5SfJJ5O1oaxY0fpwfh          0.801000    2    0.1280    -7.311  ...   \n",
+       "4  2hJjbsLCytGsnAHfdsLejp          0.000246   10    0.4020    -6.036  ...   \n",
+       "\n",
+       "                                                name popularity  speechiness  \\\n",
+       "0                           Keep A Song In Your Soul         12       0.0936   \n",
+       "1                               I Put A Spell On You          7       0.0534   \n",
+       "2                                       Golfing Papa          4       0.1740   \n",
+       "3  True House Music - Xavier Santos & Carlos Gomi...         17       0.0425   \n",
+       "4                                          Xuniverxe          2       0.0768   \n",
+       "\n",
+       "     tempo  valence  year  decade  popularity_score            main_artist  \\\n",
+       "0  149.976   0.6340  1920    1920                10            Mamie Smith   \n",
+       "1   86.889   0.9500  1920    1920                 0  Screamin' Jay Hawkins   \n",
+       "2   97.600   0.6890  1920    1920                 0            Mamie Smith   \n",
+       "3  127.997   0.0422  1920    1920                10        Oscar Velazquez   \n",
+       "4  122.076   0.2990  1920    1920                 0                   Mixe   \n",
+       "\n",
+       "  duration_minutes  \n",
+       "0         2.805550  \n",
+       "1         2.503333  \n",
+       "2         2.730450  \n",
+       "3         7.034783  \n",
+       "4         2.753733  \n",
+       "\n",
+       "[5 rows x 21 columns]"
+      ],
+      "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>acousticness</th>\n",
+       "      <th>artists</th>\n",
+       "      <th>danceability</th>\n",
+       "      <th>energy</th>\n",
+       "      <th>explicit</th>\n",
+       "      <th>id</th>\n",
+       "      <th>instrumentalness</th>\n",
+       "      <th>key</th>\n",
+       "      <th>liveness</th>\n",
+       "      <th>loudness</th>\n",
+       "      <th>...</th>\n",
+       "      <th>name</th>\n",
+       "      <th>popularity</th>\n",
+       "      <th>speechiness</th>\n",
+       "      <th>tempo</th>\n",
+       "      <th>valence</th>\n",
+       "      <th>year</th>\n",
+       "      <th>decade</th>\n",
+       "      <th>popularity_score</th>\n",
+       "      <th>main_artist</th>\n",
+       "      <th>duration_minutes</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>0</th>\n",
+       "      <td>0.991000</td>\n",
+       "      <td>['Mamie Smith']</td>\n",
+       "      <td>0.598</td>\n",
+       "      <td>0.224</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0cS0A1fUEUd1EW3FcF8AEI</td>\n",
+       "      <td>0.000522</td>\n",
+       "      <td>5</td>\n",
+       "      <td>0.3790</td>\n",
+       "      <td>-12.628</td>\n",
+       "      <td>...</td>\n",
+       "      <td>Keep A Song In Your Soul</td>\n",
+       "      <td>12</td>\n",
+       "      <td>0.0936</td>\n",
+       "      <td>149.976</td>\n",
+       "      <td>0.6340</td>\n",
+       "      <td>1920</td>\n",
+       "      <td>1920</td>\n",
+       "      <td>10</td>\n",
+       "      <td>Mamie Smith</td>\n",
+       "      <td>2.805550</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>1</th>\n",
+       "      <td>0.643000</td>\n",
+       "      <td>[\"Screamin' Jay Hawkins\"]</td>\n",
+       "      <td>0.852</td>\n",
+       "      <td>0.517</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0hbkKFIJm7Z05H8Zl9w30f</td>\n",
+       "      <td>0.026400</td>\n",
+       "      <td>5</td>\n",
+       "      <td>0.0809</td>\n",
+       "      <td>-7.261</td>\n",
+       "      <td>...</td>\n",
+       "      <td>I Put A Spell On You</td>\n",
+       "      <td>7</td>\n",
+       "      <td>0.0534</td>\n",
+       "      <td>86.889</td>\n",
+       "      <td>0.9500</td>\n",
+       "      <td>1920</td>\n",
+       "      <td>1920</td>\n",
+       "      <td>0</td>\n",
+       "      <td>Screamin' Jay Hawkins</td>\n",
+       "      <td>2.503333</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>2</th>\n",
+       "      <td>0.993000</td>\n",
+       "      <td>['Mamie Smith']</td>\n",
+       "      <td>0.647</td>\n",
+       "      <td>0.186</td>\n",
+       "      <td>0</td>\n",
+       "      <td>11m7laMUgmOKqI3oYzuhne</td>\n",
+       "      <td>0.000018</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0.5190</td>\n",
+       "      <td>-12.098</td>\n",
+       "      <td>...</td>\n",
+       "      <td>Golfing Papa</td>\n",
+       "      <td>4</td>\n",
+       "      <td>0.1740</td>\n",
+       "      <td>97.600</td>\n",
+       "      <td>0.6890</td>\n",
+       "      <td>1920</td>\n",
+       "      <td>1920</td>\n",
+       "      <td>0</td>\n",
+       "      <td>Mamie Smith</td>\n",
+       "      <td>2.730450</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>3</th>\n",
+       "      <td>0.000173</td>\n",
+       "      <td>['Oscar Velazquez']</td>\n",
+       "      <td>0.730</td>\n",
+       "      <td>0.798</td>\n",
+       "      <td>0</td>\n",
+       "      <td>19Lc5SfJJ5O1oaxY0fpwfh</td>\n",
+       "      <td>0.801000</td>\n",
+       "      <td>2</td>\n",
+       "      <td>0.1280</td>\n",
+       "      <td>-7.311</td>\n",
+       "      <td>...</td>\n",
+       "      <td>True House Music - Xavier Santos &amp; Carlos Gomi...</td>\n",
+       "      <td>17</td>\n",
+       "      <td>0.0425</td>\n",
+       "      <td>127.997</td>\n",
+       "      <td>0.0422</td>\n",
+       "      <td>1920</td>\n",
+       "      <td>1920</td>\n",
+       "      <td>10</td>\n",
+       "      <td>Oscar Velazquez</td>\n",
+       "      <td>7.034783</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>4</th>\n",
+       "      <td>0.295000</td>\n",
+       "      <td>['Mixe']</td>\n",
+       "      <td>0.704</td>\n",
+       "      <td>0.707</td>\n",
+       "      <td>1</td>\n",
+       "      <td>2hJjbsLCytGsnAHfdsLejp</td>\n",
+       "      <td>0.000246</td>\n",
+       "      <td>10</td>\n",
+       "      <td>0.4020</td>\n",
+       "      <td>-6.036</td>\n",
+       "      <td>...</td>\n",
+       "      <td>Xuniverxe</td>\n",
+       "      <td>2</td>\n",
+       "      <td>0.0768</td>\n",
+       "      <td>122.076</td>\n",
+       "      <td>0.2990</td>\n",
+       "      <td>1920</td>\n",
+       "      <td>1920</td>\n",
+       "      <td>0</td>\n",
+       "      <td>Mixe</td>\n",
+       "      <td>2.753733</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "<p>5 rows × 21 columns</p>\n",
+       "</div>"
+      ]
+     },
+     "execution_count": 2,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "execution_count": 2
+  },
+  {
+   "metadata": {},
+   "cell_type": "markdown",
+   "source": "## Outlier detection",
+   "id": "84056d9e106e6990"
+  },
+  {
+   "metadata": {},
+   "cell_type": "markdown",
+   "source": [
+    "The outlier explainer is meant for usage on series that are a result of aggregation operations.\\\n",
+    "First, we perform a groupby and aggregation on the data."
+   ],
+   "id": "2c10b272c0aaa9ce"
+  },
+  {
+   "metadata": {
+    "ExecuteTime": {
+     "end_time": "2025-02-10T14:28:37.372620Z",
+     "start_time": "2025-02-10T14:28:37.345307Z"
+    }
+   },
+   "cell_type": "code",
+   "source": [
+    "new_songs_df = spotify_data[spotify_data['year'] >= 1990]\n",
+    "gb_decade = new_songs_df.groupby('decade')['popularity'].mean()\n",
+    "gb_decade"
+   ],
+   "id": "c38c258934cdca14",
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "decade\n",
+       "1990    43.120769\n",
+       "2000    43.167320\n",
+       "2010    29.579203\n",
+       "2020    19.171014\n",
+       "Name: popularity, dtype: float64"
+      ]
+     },
+     "execution_count": 3,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "execution_count": 3
+  },
+  {
+   "metadata": {},
+   "cell_type": "markdown",
+   "source": "Now we can the outlier to explain suspected outliers.",
+   "id": "36ce2fef356a95ba"
+  },
+  {
+   "metadata": {
+    "ExecuteTime": {
+     "end_time": "2025-02-10T14:29:22.957533Z",
+     "start_time": "2025-02-10T14:29:21.682728Z"
+    }
+   },
+   "cell_type": "code",
+   "source": [
+    "outlier_explainer = OutlierExplainer()\n",
+    "outlier_explainer.explain(df_agg=gb_decade, df_in=new_songs_df, g_att='decade', g_agg='popularity', agg_method='mean', target=2020, dir=-1)"
+   ],
+   "id": "545e243e9f81631a",
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "<Figure size 500x500 with 1 Axes>"
+      ],
+      "image/png": "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"
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
+   "execution_count": 5
+  },
+  {
+   "metadata": {},
+   "cell_type": "markdown",
+   "source": "Once again, we recommend doing this via pd-explain for a more streamlined experience, where most of these parameters are automatically inferred via the user's previous actions.",
+   "id": "4e3338ab3267da02"
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 2
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython2",
+   "version": "2.7.6"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}
diff --git a/requirements.txt b/requirements.txt
new file mode 100644
index 0000000..62768ef
--- /dev/null
+++ b/requirements.txt
@@ -0,0 +1,3 @@
+numpy~=2.1.3
+pandas~=2.2.3
+matplotlib~=3.9.2
\ No newline at end of file
diff --git a/setup.py b/setup.py
new file mode 100644
index 0000000..7b7f978
--- /dev/null
+++ b/setup.py
@@ -0,0 +1,42 @@
+import os
+
+from setuptools import setup, find_packages
+
+
+def read(rel_path):
+    here = os.path.abspath(os.path.dirname(__file__))
+    with open(os.path.join(here, rel_path), 'r') as fp:
+        return fp.read()
+
+
+def get_version():
+    for line in read('src/fedex_generator/__init__.py').splitlines():
+        if line.startswith('__version__'):
+            delim = '"' if '"' in line else "'"
+            return line.split(delim)[1]
+    else:
+        raise RuntimeError("Unable to find version string.")
+
+
+def get_long_description():
+    with open('README.md', 'r') as fh:
+        return fh.read()
+
+
+setup(
+    name='external_explainers',
+    version='1.0.0',#get_version(),
+    package_dir={'': 'src'},
+    packages=find_packages(where='src'),
+    long_description_content_type="text/markdown",
+    long_description=get_long_description(),  # Long description read from the readme file
+    project_urls={
+        'Git': 'https://github.com/analysis-bots/ExternalExplainers',
+    },
+    install_requires=[
+        'wheel',
+        'pandas',
+        'numpy',
+        'matplotlib',
+    ]
+)
diff --git a/src/external_explainers/__init__.py b/src/external_explainers/__init__.py
new file mode 100644
index 0000000..562c41a
--- /dev/null
+++ b/src/external_explainers/__init__.py
@@ -0,0 +1 @@
+from external_explainers.outlier_explainer.outlier_explainer import OutlierExplainer
\ No newline at end of file
diff --git a/src/external_explainers/commons/__init__.py b/src/external_explainers/commons/__init__.py
new file mode 100644
index 0000000..718fc6a
--- /dev/null
+++ b/src/external_explainers/commons/__init__.py
@@ -0,0 +1 @@
+from external_explainers.commons import utils
\ No newline at end of file
diff --git a/src/external_explainers/commons/utils.py b/src/external_explainers/commons/utils.py
new file mode 100644
index 0000000..50fd8fe
--- /dev/null
+++ b/src/external_explainers/commons/utils.py
@@ -0,0 +1,25 @@
+def to_valid_latex(string, is_bold: bool = False) -> str:
+    """
+    Convert a string to a valid latex string.
+    :param string: The input string.
+    :param is_bold: Whether the string should be bold.
+    :return: The latex string.
+    """
+    latex_string = str(string)  # unicode_to_latex(string)
+    # Choose the space character based on whether the string should be bold.
+    space = r'\ ' if is_bold else ' '
+    # Escape special characters and replace spaces with the chosen space character.
+    final_str = latex_string.replace("&", "\\&").replace("#", "\\#").replace(' ', space).replace("_", space)
+    return final_str
+
+
+def to_valid_latex_with_escaped_dollar_char(string, is_bold: bool = False) -> str:
+    """
+    Convert a string to a valid latex string.
+    This function adds the $ character to the list of characters that are escaped, while the function to_valid_latex
+    does not escape the $ character. Other than that, the two functions are identical.
+    :param string: The input string.
+    :param is_bold: Whether the string should be bold.
+    :return: The latex string.
+    """
+    return to_valid_latex(string, is_bold).replace("$", "\\$")
\ No newline at end of file
diff --git a/src/external_explainers/outlier_explainer/__init__.py b/src/external_explainers/outlier_explainer/__init__.py
new file mode 100644
index 0000000..e69de29
diff --git a/src/external_explainers/outlier_explainer/outlier_explainer.py b/src/external_explainers/outlier_explainer/outlier_explainer.py
new file mode 100644
index 0000000..7e0f896
--- /dev/null
+++ b/src/external_explainers/outlier_explainer/outlier_explainer.py
@@ -0,0 +1,394 @@
+from matplotlib import pyplot as plt
+import numpy as np
+import pandas as pd
+from pandas.core.interchange.dataframe_protocol import DataFrame
+from typing import List, Tuple, Any
+from concurrent.futures import ThreadPoolExecutor, as_completed
+
+from external_explainers.commons import utils
+
+ALPHA = 0.9
+START_BOLD = r'$\bf{'
+END_BOLD = '}$'
+
+HIGH = 1
+LOW = -1
+
+
+# BETA = 0.1
+def proportion(series):
+    c = series.count()
+    return (series.count()) / (series.count().sum())
+
+
+class OutlierExplainer:
+    """
+    A class for computing the outlier measure and explaining outliers.\n
+    This is based on the article "Scorpion: Explaining Away Outliers in Aggregate Queries" by Eugene Wu, Samuel Madden.\n
+    The influence of a predicate :math:`p` on an output result :math:`o` is depends on:\n
+    .. math:: \\Delta_{agg} (o,p) = agg(g_0) - agg(g_0 - p(g_0))\n
+    Where: \n
+    - :math:`agg` is the aggregation function.\n
+    - :math:`g_0` is a set of input tuples.\n
+    In other words, it is the difference between the original result and the result after removing the tuples that satisfy the predicate.\n
+    From which we define the influence as:\n
+    .. math:: inf_{agg}(o,p) = \\frac{\\Delta_{agg}(o,p)}{\\Delta g_0} = \\frac{\\Delta_{agg}(o,p)}{|p(g_0)|}\n
+    We then include the error direction, :math:`d`, to the influence calculation:\n
+    .. math:: inf_{agg}(o,p, d) = d \\ * \\ inf_{agg}(o,p) \n
+    Next, the user may select a hold-out result, :math:`h`, that the returned predicate should not influence.\n
+    Intuitively, :math:`p` should be penalized if it influences the hold-out results in any way.\n
+    The influence with the hold-out set is defined as:\n
+    .. math:: inf_{agg}(o,p,d,h) = \\lambda \\ * \\ inf_{agg}(o,p,d) - (1 - \\lambda) \\ * \\ inf_{agg}(h,p)\n
+    Where:\n
+    - :math:`\\lambda` is a parameter that represents the importance of not changing the value of the hold-out set.
+    In this project we use :math:`\\lambda = 0.9`.\n
+    """
+
+    def __init__(self):
+        super().__init__()
+
+    def calc_influence_pred(self, df_before: DataFrame, df_after: DataFrame, target: str, dir: int) -> float:
+        """
+        Calculate the influence of a predicate on a given target attribute.
+
+        This function computes the influence of a target attribute by comparing its values
+        before and after some transformation, adjusted by a direction factor. It also
+        considers the influence of other attributes in the DataFrame.
+
+        :param df_before: DataFrame containing the data before the transformation.
+        :param df_after: DataFrame containing the data after the transformation.
+        :param target (str)) The target attribute for which the influence is being calculated.
+        :param dir: Direction factor, either 1 or -1, indicating the direction of the outlier. 1 for high, -1 for low.
+
+        :return: The calculated influence score. Returns -1 if an error occurs during calculation.
+        """
+        try:
+            # Compute target influence - the ratio between the change in the output and the number of
+            # tuples that satisfy the predicate, multiplied by the direction factor.
+            target_inf = ((df_before[target] - df_after[target]) * dir) / (df_before[target] + df_after[target])
+        except:
+            return -1
+
+        # Compute the holdout influence - the sum of the square root of the absolute difference between the
+        # values of the target attribute for each tuple in the DataFrame before and after the transformation.
+        holdout_inf = 0
+        for i in df_before.index:
+            if i != target:
+                try:
+                    holdout_inf += np.sqrt(abs(df_after[i] - df_before[i]))
+                except:
+                    return -1
+        # Return the final influence score, calculated as a weighted sum of the target and holdout influences.
+        return ALPHA * (target_inf) - (1 - ALPHA) * (holdout_inf / (len(df_before.index)))
+
+    def merge_preds(self, df_agg: DataFrame, df_in: DataFrame, df_in_consider: DataFrame,
+                    preds: List[Tuple[str, Tuple[float, float], float, str, int | None]], g_att: str, g_agg: str,
+                    agg_method: str, target, dir: int) -> tuple[
+        list[tuple[str, tuple[float, float], int | None]], float, Any | None]:
+        """
+        Merge predicates to find the most influential attributes.
+
+        This function iterates over a list of predicates, applying filters to the input DataFrame to exclude
+        certain rows based on the predicates. It calculates the influence of each predicate and keeps track
+        of the most influential ones. The final influence score and the filtered DataFrame are returned.
+
+        :param df_agg: DataFrame containing the aggregated data.
+        :param df_in: DataFrame containing the input data.
+        :param df_in_consider: DataFrame containing the data to be considered for filtering.
+        :param preds: A list of tuples containing predicates with attribute name, bin or value, score, kind, and rank.
+        :param g_att: The grouping attribute.
+        :param g_agg: The aggregation attribute.
+        :param agg_method: The aggregation method to be used.
+        :param target: The target attribute for which the influence is being calculated.
+        :param dir: Direction factor, either 1 or -1, indicating the direction of the outlier. 1 for high, -1 for low.
+
+        :return: A tuple containing a list of final predicates, the final influence score, and the final aggregated DataFrame.
+        """
+        # Initialize variables to store the final predicates, influence score, and aggregated DataFrame.
+        final_pred = []
+        final_inf = 0.001
+        final_agg_df = None
+        final_filter = False
+        final_filter_df = False
+
+        prev_attrs = []
+
+        for p in preds:
+            attr, i, score, kind, rank = p
+
+            # Avoid going over previously seen attributes
+            if attr in prev_attrs:
+                continue
+            prev_attrs.append(attr)
+
+            # If the kind is 'bin', we use the bin values to filter the DataFrame.
+            if kind == 'bin':
+                bin = i
+                final_filter_test = final_filter | ((df_in_consider[attr] < bin[0]) | (df_in_consider[attr] >= bin[1]))
+                final_filter_df_test = final_filter_df | ((df_in[attr] < bin[0]) | (df_in[attr] >= bin[1]))
+            # Otherwise, we use the attribute value to filter the DataFrame.
+            else:
+                final_filter_test = final_filter | (df_in_consider[attr] != i)
+                final_filter_df_test = final_filter_df | (df_in[attr] != i)
+
+            # Apply the filter to the input DataFrame and the aggregated DataFrame.
+            df_exc_consider = df_in_consider[final_filter_test]
+            df_exc_final = df_in[final_filter_df_test]
+
+            # Perform the aggregation operation on the filtered DataFrames.
+            agged_val_consider = df_exc_consider.groupby(g_att)[g_agg].agg(agg_method)
+            agged_val = df_exc_final.groupby(g_att)[g_agg].agg(agg_method)
+
+            # Normalize the aggregated values if the aggregation method is 'count'.
+            if agg_method == 'count':
+                agged_val_consider = agged_val_consider / agged_val_consider.sum()
+                agged_val = agged_val / agged_val.sum()
+
+            # Compute the influence score for the predicate.
+            inf = self.calc_influence_pred(df_agg, agged_val_consider, target, dir) / pow(
+                (df_in_consider.shape[0] / (df_exc_consider.shape[0] + 1)), 2)
+
+            # If the influence score is greater by a factor of at least 1.3, update the final predicates.
+            if inf / final_inf > 1.3:
+                final_pred.append((attr, i, rank))
+                final_inf = inf
+
+                final_agg_df = agged_val
+                final_filter = final_filter_test
+                final_filter_df = final_filter_df_test
+            # If the influence score is not much higher, break the loop.
+            else:
+                break
+        return final_pred, final_inf, final_agg_df
+
+
+    def compute_predicates_per_attribute(self, attr: str, df_in: DataFrame, g_att: str,
+                                         g_agg: str, agg_method: str, target: str, dir: int,
+                                         df_in_consider: DataFrame, df_agg_consider: DataFrame) -> List[Tuple[str, Any, float, str, int]]:
+        """
+            Compute predicates for a given attribute.
+
+            This function calculates the influence of various predicates on a target attribute
+            by iterating over the values or bins of the given attribute. It generates a list of
+            predicates with their corresponding influence scores.
+
+            :param attr: The attribute for which predicates are being computed.
+            :param df_in: DataFrame containing the input data.
+            :param g_att: The grouping attribute.
+            :param g_agg: The aggregation attribute.
+            :param agg_method: The aggregation method to be used.
+            :param target: The target attribute for which the influence is being calculated.
+            :param dir: Direction factor, either 1 or -1, indicating the direction of the outlier.
+            :param df_in_consider: DataFrame containing the data to be considered for filtering.
+            :param df_agg_consider: DataFrame containing the aggregated data to be considered.
+
+            :return: A list of predicates with their influence scores.
+        """
+        dtype = df_in[attr].dtype.name
+        predicates = []
+        exps = {}
+
+        # Ignore attributes with high correlation to the target attribute.
+        if dtype in ['int64', 'float64']:
+            if (df_in[g_att].dtype.name in ['int64', 'float64'] and df_in[g_att].corr(df_in[attr]) > 0.7) or (
+                    df_in[g_agg].dtype.name in ['int64', 'float64'] and df_in[g_agg].corr(df_in[attr]) > 0.7):
+                return []
+
+        # Get the series for the attribute and its data type.
+        series = df_in[attr]
+        dtype = df_in[attr].dtype.name
+        flag = False
+        df_in_consider_attr = df_in_consider[[g_att, g_agg, attr]]
+
+        # If the data type is not 'float64', calculate the influence score for each value of the attribute.
+        if dtype not in ['float64']:
+            vals = series.value_counts()
+            if dtype != 'int64' or len(vals) < 20:
+                # Skip attributes with more than 50 unique values, as they are too computationally expensive.
+                if len(vals) > 50:
+                    return []
+                flag = True
+
+                for i in vals.index:
+
+                    # Exclude rows with the current value of the attribute.
+                    df_in_target_exc = df_in_consider_attr[(df_in_consider_attr[attr] != i)]
+                    # Aggregate the values for the excluded rows.
+                    agged_val = df_in_target_exc.groupby(g_att)[g_agg].agg(agg_method)
+                    if agg_method == 'count':
+                        agged_val = agged_val / agged_val.sum()
+
+                    # Calculate the influence score for the predicate.
+                    inf = self.calc_influence_pred(df_agg_consider, agged_val, target, dir) / (
+                        (df_in_consider.shape[0] / (df_in_target_exc.shape[0] + 0.01)))
+
+                    exps[(attr, i)] = inf
+                    predicates.append((attr, i, inf, 'cat', None))
+
+        n_bins = 20
+        if not flag:
+            _, bins = pd.cut(series, n_bins, retbins=True, duplicates='drop')
+            df_bins_in = pd.cut(df_in_consider_attr[attr], bins=bins).value_counts(
+                normalize=True).sort_index()  # .rename('idx')
+            i = 1
+            for bin in df_bins_in.keys():
+                new_bin = (float("{:.2f}".format(bin.left)), float("{:.2f}".format(bin.right)))
+                df_in_exc = df_in_consider_attr[
+                    ((df_in_consider_attr[attr] < new_bin[0]) | (df_in_consider_attr[attr] >= new_bin[1]))]
+                agged_val = df_in_exc.groupby(g_att)[g_agg].agg(agg_method)
+                if agg_method == 'count':
+                    agged_val = agged_val / agged_val.sum()
+
+                # Calculate the influence score for the predicate.
+                inf = self.calc_influence_pred(df_agg_consider, agged_val, target, dir) / (
+                        (df_in_consider_attr.shape[0] / df_in_exc.shape[0]) + 1)
+
+                # Store the influence score for the predicate.
+                exps[(attr, (new_bin[0], new_bin[1]))] = inf
+
+                # Add the predicate to the list of predicates.
+                predicates.append((attr, new_bin, inf, 'bin', i))
+                i += 1
+
+        return predicates
+
+    def draw_bar_plot(self, df_agg: DataFrame, final_df: DataFrame, g_att: str, g_agg: str, final_pred_by_attr: dict,
+                      target: str, agg_title: str) -> None:
+        """
+        Draw a bar plot to visualize the influence of predicates on the target attribute.
+
+        This function generates a bar plot comparing the aggregated values of the target attribute
+        before and after applying the most influential predicates. It highlights the differences
+        and provides an explanation for the outlier.
+
+        :param df_agg: DataFrame containing the aggregated data.
+        :param final_df: DataFrame containing the final aggregated data after applying predicates.
+        :param g_att: The grouping attribute.
+        :param g_agg: The aggregation attribute.
+        :param final_pred_by_attr: Dictionary containing the final predicates grouped by attribute.
+        :param target: The target attribute for which the influence is being visualized.
+        :param agg_title: Title for the aggregation method used in the plot.
+
+        :return: None. Displays the bar plot.
+        """
+        fig, ax = plt.subplots(layout='constrained', figsize=(5, 5))
+        x1 = list(df_agg.index)
+        ind1 = np.arange(len(x1))
+        y1 = df_agg.values
+
+        x2 = list(final_df.index)
+        ind2 = np.arange(len(x2))
+        y2 = final_df.values
+
+        explanation = f'This outlier is not as significant when excluding rows with:\n'
+        for_wizard = ''
+        for a, bins in final_pred_by_attr.items():
+            for b in bins:
+                if type(b[0]) is tuple:
+                    pred = f"{b[0][0]} < {a} < {b[0][1]}"
+                    inter_exp = r'$\bf{{{}}}$'.format(utils.to_valid_latex(pred))
+                else:
+                    pred = f"{a}={b[0]}"
+                    inter_exp = r'$\bf{{{}}}$'.format(utils.to_valid_latex(pred))
+                if b[1] is not None:
+                    if b[1] <= 5:
+                        inter_exp = inter_exp + '-' + r'$\bf{low}$'
+                    elif b[1] >= 25:
+                        inter_exp = inter_exp + '-' + r'$\bf{high}$'
+            inter_exp += '\n'
+            for_wizard += inter_exp
+            explanation += inter_exp
+
+        bar1 = ax.bar(ind1 - 0.2, y1, 0.4, alpha=1., label='All')
+        bar2 = ax.bar(ind2 + 0.2, y2, 0.4, alpha=1., label=f'without\n{for_wizard}')
+        ax.set_ylabel(f'{g_agg} {agg_title}')
+        ax.set_xlabel(f'{g_att}')
+        ax.set_xticks(ind1)
+        ax.set_xticklabels(tuple([str(i) for i in x1]), rotation=45)
+        ax.legend(loc='best')
+        ax.set_title(explanation)
+        bar1[x1.index(target)].set_edgecolor('tab:green')
+        bar1[x1.index(target)].set_linewidth(2)
+        bar2[x2.index(target)].set_edgecolor('tab:green')
+        bar2[x2.index(target)].set_linewidth(2)
+        ax.get_xticklabels()[x1.index(target)].set_color('tab:green')
+
+        plt.show()
+
+
+    def explain(self, df_agg: DataFrame, df_in: DataFrame, g_att: str, g_agg: str, agg_method: str, target: str,
+                dir: int, control=None, hold_out: List = [], k: int = 1) -> str | None:
+        """
+        Explain the outlier in the given DataFrame.
+
+        This function identifies and explains outliers in the given DataFrame by calculating the influence of
+        various attributes on the target attribute. It iterates over the attributes, applies filters, and
+        computes the influence score for each attribute. The most influential attributes are then used to
+        generate an explanation for the outlier.
+
+        :param df_agg: DataFrame containing the aggregated data.
+        :param df_in: DataFrame containing the input data.
+        :param g_att: The grouping attribute.
+        :param g_agg: The aggregation attribute.
+        :param agg_method: The aggregation method to be used.
+        :param target: The target attribute for which the influence is being calculated.
+        :param dir: Direction factor, either 1 or -1, indicating the direction of the outlier. 1 for high, -1 for low.
+        :param control: List of control values for the grouping attribute.
+        :param hold_out: List of attributes to be held out from the analysis.
+        :param k: Number of top attributes to consider for the explanation.
+
+        :return: None. Will generate a plot with the explanation for the outlier.
+        """
+        # Get the attributes from the input DataFrame and remove the hold-out attributes.
+        attrs = df_in.columns
+        attrs = [a for a in attrs if a not in hold_out + [g_att, g_agg]]
+
+        agg_title = agg_method
+        if agg_method == 'count':
+            df_agg = df_agg / df_agg.sum()
+
+        predicates = []
+
+        # If no control values are provided, use all values for the grouping attribute.
+        if control == None:
+            control = list(df_agg.index)
+        df_in_consider = df_in
+        df_agg_consider = df_agg  # [control]
+
+        # Iterate over the attributes, calculate the influence score, and generate predicates.
+        with ThreadPoolExecutor() as executor:
+            futures = [
+                executor.submit(self.compute_predicates_per_attribute, attr, df_in, g_att, g_agg, agg_method, target,
+                                dir, df_in_consider, df_agg_consider) for attr in attrs
+            ]
+            for future in as_completed(futures):
+                preds = future.result()
+                predicates += preds
+
+
+        # Sort the predicates by influence score in descending order.
+        predicates.sort(key=lambda x: -x[2])
+
+        # Merge the predicates to find the most influential attributes.
+        final_pred, final_inf, final_df = self.merge_preds(df_agg_consider, df_in, df_in_consider, predicates, g_att,
+                                                           g_agg, agg_method, target, dir)
+
+        # If the final DataFrame is empty, return an error message. Otherwise, generate the explanation plot.
+        if final_df is None:
+            return "There was no explanation."
+
+        # Create a new DataFrame with the aggregated values and the control values.
+        new_df_agg = df_agg.copy()
+        new_df_agg[control] = final_df[control]
+        new_df_agg[target] = final_df[target]
+        final_pred_by_attr = {}
+
+        # Group the final predicates by attribute.
+        for a, i, rank in final_pred:
+            if a not in final_pred_by_attr.keys():
+                final_pred_by_attr[a] = []
+            final_pred_by_attr[a].append((i, rank))
+
+        # Create a plot to display the explanation for the outlier.
+        self.draw_bar_plot(df_agg, final_df, g_att, g_agg, final_pred_by_attr, target, agg_title)
+        return None

From 1f68d59a17020e234b238da4cc90803fe52c940b Mon Sep 17 00:00:00 2001
From: Yuval Uner <80361935+YuvalUner@users.noreply.github.com>
Date: Mon, 10 Feb 2025 16:50:39 +0200
Subject: [PATCH 2/4] Create LICENSE

---
 LICENSE | 201 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 201 insertions(+)
 create mode 100644 LICENSE

diff --git a/LICENSE b/LICENSE
new file mode 100644
index 0000000..261eeb9
--- /dev/null
+++ b/LICENSE
@@ -0,0 +1,201 @@
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright [yyyy] [name of copyright owner]
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

From ee448211a5abe610ce368013d06e58af668d2b02 Mon Sep 17 00:00:00 2001
From: Yuval Uner <yuvaluner@gmail.com>
Date: Mon, 10 Feb 2025 16:55:05 +0200
Subject: [PATCH 3/4] Added empty __init__.py for src dir

---
 src/__init__.py | 0
 1 file changed, 0 insertions(+), 0 deletions(-)
 create mode 100644 src/__init__.py

diff --git a/src/__init__.py b/src/__init__.py
new file mode 100644
index 0000000..e69de29

From 8c6c0068f85a0f39e8ee3321f7e255a8f0aa5658 Mon Sep 17 00:00:00 2001
From: Yuval Uner <yuvaluner@gmail.com>
Date: Mon, 10 Feb 2025 17:00:07 +0200
Subject: [PATCH 4/4] Added pyproject.toml and v1.0.0 distribution files.

---
 dist/external_explainers-1.0.0-py3-none-any.whl | Bin 0 -> 12524 bytes
 dist/external_explainers-1.0.0.tar.gz           | Bin 0 -> 12264 bytes
 pyproject.toml                                  |   3 +++
 3 files changed, 3 insertions(+)
 create mode 100644 dist/external_explainers-1.0.0-py3-none-any.whl
 create mode 100644 dist/external_explainers-1.0.0.tar.gz
 create mode 100644 pyproject.toml

diff --git a/dist/external_explainers-1.0.0-py3-none-any.whl b/dist/external_explainers-1.0.0-py3-none-any.whl
new file mode 100644
index 0000000000000000000000000000000000000000..2b0aba144b272310800e6aeef65da876149efa10
GIT binary patch
literal 12524
zcma)@b9iLyw(cvo*|E(|haKCtt%_}RY^P(}ww-i3w$-sZ_RT)$p1amwd!6U5IiFGZ
zWB$hXRlRe3bB=mTUJ48x9RL781K=GcHAim&TPz>%ls|&-BTU?!OdM<stn^IW?5qqd
zY)l*+8TIrmY%HAg^cd{irD(?{Rm3Z)B<RPaeuT+YQjJLb_#s74voOv`KRPN+F9>ZA
zPG$THa}O(t>OqK1+h7ME|NruFJ4k9e$k-h+eB`132+IE@pOLM#wXMy+<%Q@Cg8>jh
z1A0_4a*_o5_%DSJ@VeX6tN3RVgX=5=W25R8>{XgU7cQ;bCKL((NC~u&)a(SiP$>Wb
z0E)o@0J8s2N@phvE60y{PAJO&cbSkov7QJa$-l6=U4wxhk)P^%QlF9&p<8IfD36Le
z%w~7zBT;VHXn4>x_J$s$-fws<!9s~=b;APDn@P4;y`_6w4hW?8C~fBvWjC~z`WX#K
zT`qS)s0>{_!JL?fj<>hgX@kK*ExF=mTBihatsk0#FGNuWr21Vh3s53)i*F3(21jh8
zB#ZzaKRzowkJ4}u@5OR5>Zb*icMz87X{We$D4aRUmv@1w_wGgBz;fQ2EvHufs@lrw
zj8R{Y6eJOs@i8F@p+r&eceQ&f*q5vcsp0H{{K15UATC2`%o6bysTiq&T!?Z<^GSdM
z0z@o53|1$wUZPE0N=A0wGhN1$Np@)sVn)x!cy;M~8UMr|KgPA9-LO2o5)*s*w0wvp
zy8ctSr8Aq_^lG|Sx$j-9*)^V2YdEcKB{a9zdpv~q3`HLkXDYmm0u_c7;bO!mG=boX
zFqoH`4B?r8Ku|%)%~>35E%yDsN7qIb6?dGp>r(Qrl(jx1EsE`MaXpfmWj%AFX;xUB
z&x%VIx2=uS@9)0#>Y2_uwX7#^e-hTeg#7Wv0zm)u@Uby}#NP?o*4fF*!o=YZZU0%?
zV1EKee*hC~u@l8UPLw?v0D$Ws1OFq%hnUW|fwrrnO`BkQd_TA#gsrg`?yS!ZqiFn^
zB~}6J4X#Exgy=%K1GxjNB6VKs3tVp(_ZcimKjA~_GM!xg2Vq5MCU!Ge%zk>BW-RO{
zTkkfFjk8TA)2*tStk+q_Akf*UCpF7uk1XV^s(M<{(X6Wuih$B{e0F~#6S2#q)=XlJ
zsmt5HwM4@L5|G>D7CU$BN0c$FI;i}DT?ZaM5_pmYy6Bk<YTmvjvRaQYw9y<8e`UOL
zm@1=F)zP`mi<+rAv)m|w!AegX9%VF1?QY#JucALehSshaqxUoCoT3)E#LcM+y@u#6
z{mnt^7G-SPY?5o>Y@N)`vHeo4=fiEr;)3`3c$9FDb>Q2s{<43Ntt%wO^LX6<xigd7
z`}ePP;5<j`vt_`nkJod5%;ojoQ9`!Q8%9?LNAE(hTvly&M=SDW4*51USP{`s1zkY!
z;wRf8Gr4T0hYI~fbC8|3JcOdW&R;dh0T3>miQZ>yA~MZEi2iO#!!rFo=lbpLucDP&
zd0bpvsCqwBu~U*+g}Oj+^@Uat%nHAcHgVhd9l^7uOWbWEFRu#|!h#$bCV;TuHbvrQ
z)zlFYF;_SyIl9~s6chxIXV<dxL|*H3sG;|*1!Pt)vg!@fo4qeik3}yzOl~*<b?xqe
zSatVTO}m`WzznhuCV9v2nZNKERr%e<4D~<D)nN#7ft7A%cvtX}$k_V__S5*6%A_=x
zHNS)wP<HHy%=GlIKyZa)M{yVt@^<`k$7QtHy?u&?Fj1E==-v{m(Z!QnhP%b+Qknar
z!sN4lnFa=UybcnCMT4eimaM{4ADF39-%c@S0_&NH#NI0y3^IkNDildNa^0b)L)Blf
zQd^KftE*5^kEfII$K0EVZ~ppokRi#r=_fxAx{_Uat<I~ofalh5Q-^9FF;m}SiwT1K
zdV#0qqSfMnSIkrMs&P5ODhLQS1a(2~Yw%!H$odq$dL_r}HrP_)d}SlToNMt4cFn>0
z<%paikazv4lh_cJzj>@(SbpdPr-69XQg#)Y=A4pHXdxN#s+>d5nbUgByBu0008YY>
z|4!^>SR+5}4e^JHxOp2S26AG-HYYOJa~JjXo=EAw+H8}+iXHnvS8gwW{<JY?IZ-F8
z1~js>XdU@`)A-71+QIYG6c(o$%<7l`SDi0lfR5Z=ktk4|Q#g)hcreS~Bj)gs0my!*
zjBc(dfF8RC@)?C<TXaN(z&1tV^uY3ajwK}JWeg6-ZR8r@-P#%dFe{A#qb7VhCf5*o
z6^Phj(<25G4PA^X<rt_<xmhxIzy-;xA2S{x4z!E-g|0ogX}{@82FpO+C*eIuN0HCa
zm6xqS*E7}>0l^W*t-=e*C!=<S(-B0##9Exo&v+wCFCPdEDsJ1R?qG+mKi)uEtoj*Q
z)d+VHxRF&`1d#%t7U;RyC);6Yp*vJ55P2v3@F+`@03^mshz`!E6wrkW3mvr9G8;!c
zjn)izkr+#pyr&RZ@IMO>X#Tdlo3?|tq_R?16hhA-P^7ptT(kzd?5zD|sx~TUKQy7+
z6+xH;n(<r)R%#bRm=?toj^n!Mv{$fZ))=O(A0(--2#6{%ZWn9X-gw7?Du0nh9Ix7@
z{~LBHcR^@qT~m1P+tGOoR>9V|5Rv2^6cGdRVooWx6*MGOU(`}$Z|FBtO_Nafom_14
z%f|lK*sDl&4^Twp9ag(U2$=7>fmuYXDnr~TKlmNkmOaq=grDM4zNWwTL1#}}daS^Y
z;lgGtZ<=d=v+HRQ^dni5YYoW3IyF#BkEm}oJCfD9KxpinWVy>kb|0J|xQE-OM9#s0
z35C{`NQO^R<^WjqqAsZvF{F9s`^Q@MG8$vp#VGgws#1o;EhGAbfbzYSXv%>b3qPd>
zoIIq=0#+42M^T>}@7;@|Yk0&~wZjtsZL~#8_%}h3<Z4vW>OLL#j-+|oW*atoIm5Ur
z>)?3N50pSY0+u+K`g{tH#(?1x^cD!P2&M)_s78e<(a*0f7yMs72`*X%1u|~4Bla_K
z==*-BKCyLuGMNi`dW1)vEP1T70ozi}X{Jv|XJBV*&yWr2>G*oIJe@Tt2olCZEz%nu
z9$fTf2n&CX{e{r?(`rQ%BnJ<2_|R4@nSgPsudx8*0hfF(FZ!>ZJy>YxpAkLF{zr2T
z8(N4{BWCQKv%%2Aw@Vt#RvJ{BMlO$!;?O^-46>EFcqzpgyW&HTg7Ui)-;qSf00MyL
zfsG`H?!c^DP;iYQXrKVO&wYO2ImCcg`sX+n@ZF5VfI0mQp^>=0rsA4g1#>0xuam?+
zNl`}AV|sDRopC@WP5=z*kb@SRCg=)$L7D3slwl(LUIu^C;^9e{Pht%L%mBJd(#14&
zfyRbnL5>6Jmxa!AAr>kyQ&v&8UZB&gew2%B(S}h@iEu;*B{K21SurE3-)b3Z2sp{}
zDlyWQLw;yIO8gGYMYb!Op|>DmuW2LkW2dHMYqf>z2-3<~2=dILSCH(<4lMxi6r>-w
zrE%=smeaS#en`-Qw6qKGyTMb1Gnv5LSer#Z>k$$Y_0D`M@yV2$Tfe{z$&!jyj8H3a
zX{c^H)&bgi^n0LA3ckwe4>wN^JlCy0YD>KD24ziOWg4V(1rYaO&XudKvN71Qq*827
zZEO((H@=H*``sVU!?YtnFrh(J`+Gu$nFrIrmGVK$WDbsNe{U^P8SRTjX9T&=A*b4F
zCc6Bt*M|R@C@eZgZjAvw3*2*Bxjj=bJ?GkUq)yah(26sM5VS!%f9sn!2Kw@}e9(04
zl=!ONHu{!-zD|h-p!jnyG-9@1ibmf8`MmP<A?Z%VX&PQR@0xtZP*(UTsFMU2S!GrW
z6a*D5m(Bf+63HcHhdAg9GycrI>If49Sdu9yg)wSb0Qg;wOq>GEW4CrtYL|C%AX>w+
zH2?KfB$F8(u8TS@6rF{^_s02b;VS*_+nfxi@!k>q2g9*sUWRWd)uvze25?omP<Q*I
z0t%*M3_FpQS?*XgQ*;*WA|^T|&<68EuF}3Kdp<lPvw66K_j=!ntTDS2`6GM*-NhvZ
zU`N;{>a-6xY(dosQWao*6H)hc1`F2q0D(55G@JA;@4~A(z|u9UW2L%I!KmJEHYD4y
z$Z+m)%A(^3WE6gnh!~Mzya#zaJIZ*uG6^h3>&(xL?m6q7-XVY;*ZKWbnqd^kflG!U
zl8o3wjL(yX%{7q*Unv?BZxw?5E6#>7g*!^m6K4=*1>`#g6<R<Vhh{)+9}>!89%-$d
zOa^PhYCku_Zxs`f8e9)m0~0ZIq|z}RChf%hU4{BETEqoOOnJJ}8SlyEQ74?bS$n;0
z2DCf$J5EuRqR%Cv`6ZVtbs59@w6b~cFdAy)Wj80H7(xcf)b@k~9fH{TMTo@IVp!%_
zQM&GHOK|7?-RFYO#9R+kT6RlpZEPs?L!Apw=Hfiph|GjBgiZK?J=pGfc60{pRpw5*
z$@Bf%kQ1L0nLW;fc@_mDzV3X%IhZ;MkPi{*ZiHj{26wh@JP+zyKKo6ev9AR80SK`X
zqhNqdOTlq~5#!IRD-g$7S{);aX@j!gv4j*(k&@2}{X&IiWbA9MF^3ZNZUxTw^KryK
zk?@;usDuo<6ZGycEjF(y7lo;*-b&sVOD6XKXz>&_9nD&3|KYbhyga3B0d(MgJAkU+
z6W)0uy)3W+9?8}!$Q)8zUV|!ZEeON%b~)c428jIG6+Kdpgt2F`j12plFwJ#Ui6G50
zo%a35%PQbXZlqU6_N~kP;lbwhtq~WuKNP>ESe&(o4Sl=IlBohat!!HX5~BYztweiu
z3!)RXc1#I9&EZ|L7|2;i=BDzJlc!mBW`_=z^7aW5a~Ow{w4n9irxe&np5k^`%s@bZ
zxH!fET20*hGs0dW+UJbRTL-r<#EwXcw2rJs$W*;GzlefNAz6Y-b~&DA(IKYLbWup>
zy&}^IBhU^aw{Kd^j}p!d2U+gGu$l!2beQp-?w-Opu0pj&aN^C^Fpv}IB-}c_7r1v2
zs2W%$<zwrAPleeS2LnaH*5-msjEkYr8{|%_9D@02jwdT(eWE)gP>P0Sp^r2Sj>ymY
z>dgz<krP>HKMRL|_S=9Ol}g@i=^KC1sg{e!>GHYR>8pw=f|CU;jiY_9E<t_Oh|fvl
z_~3Y4OC3*ZO%q)q7->?3qj;MDkn{3&#Th;RM^=Fv;N|ku>Wjk?{oq#Kc?)}yYH?(!
z5Fyi3Ew_>Fz;ZCh*wlU4p3L;K>&y#uTmq$i>@9pjtg$1(15T7;u-Rb1+?!@T$~zTL
zqS+T1dgDNTO1|dM*9<Q`Z-O}GXK{C7k~J2i)9A^5MMIfKAw?rn!i7Bq#a)#CsBF*Q
zMxMBl;Ex<!yv&HE>qt*Stz&}4DKw)u2&|vC1nwkL7r?oCy9};YMb+u3+Oqu5YJOnh
zFw6;=MPLyp?3R;hI#+gR;7M{2m$CY!=Kcn~Aksr|nxK%-QAHE$xlF_nWh{xNC62a?
z;pgf1Eov&n`&=lWL{4Ii!)g*WysJ+WKrN{TzJgRm`Q3OlN=ljH+$>$DbVP7&9#X);
zY4BdZkVjJfa$Rt)2(0e8clu1jDQyE^WQR)xkg4!<7L%52m(;R@m+4l806b<$CRR^!
zExy?zwM#|#8MygtN{xjF^JIp=k*z^UPv4Kg!mF$2@`XV3VAH_P+)=uG_F9`&wv^yN
z@5a+x{GJnztx|bvT;6<q#4QMZO*ei?U0I#)hXVJ5m`wISHGB?7?3mX|;<gIO$z3wM
ztjCh6HBG6JAKoII7aEI?w<vmZyi)~*iAvfN=(1s}UMLP&4nZ}&MiLczDI3sHw!2h2
zFxjU9`vwQI#}HCV|E2<J$vcx(M=z||lFNnOjBNebe|3<g{!9MVExYNf$$OlZ6;E`p
zGpqOFpl94bYkh!*adgqV_o;qH2FYO=L$yFBPMcL&wF2zpxuB(p9_8V|tIfR>_snh|
zHr^a=vjwYAN85FETkUqj@!}^5QSFP@6Xt=a(D;<R;W#f1840+q1Kp5ZRjh|@UP=Ph
zu9{zk=PG%z&$hrERneXR$N~$-0Xd67^)T`Yrp(UX2P)fMS9*Ipv>vb6?qzUqey!c~
zO5*@i<8}Qhv<~=XsoA;p3;>{^cm`}#%U^Uvjf|t9ghbt=_Lo!G<*$LjT0WF>yy9};
zIb^U>kU&@!((n=TaB0x4GNBQ`j~G)4$zUt6BFZCx;@W*gH{D9dm^u`t!0oG7Gie>=
znq4MTmNyjzWcs*A?f4tYqFcSOp=*u0Xs5<D$`KXt(FGeyaU_(*u82~#w_axC*Ee!^
zGcUKG#@DzU(G53=7#n+z7gs2dRcq;Y#qrVo^*CRe&XxhFsMqm<)z$X)_G${6xdO^4
z1uaALcyc&+UmMVY!%p@SzPsA6$O}{9h3-%a{e4u9oV!hX@2<GbpG)IlqvaBM@2{Nc
z9p8Jp<99mNrqg8!0a>)t$aL90%rrOZVA@0GvFcLauZ*B?9KjkKL(NJOc*g9S?4!a7
zH-#_hiN0QeDlX0<!wW*|b%9(bs-?TRpa#*!q;O};{JdWKQKr9B3bB(|rw4A&1xk{c
z6CGgcOJ(z1xRxppq75c-n?(VBw3m!Ghu=Q$vdaE4gtyVOL2TXBW@%nZnhBl{BW?o5
z9RpWqp_kI;AB**QTT%Ik>1Ibm?w(hafu(MhX#%;eCK80OxBhAL#E*o>iLjcg-?|Ss
z@kLk06KW>93%vzVDP~&*Zp`jWb9me%3{opHcE?t0FTZYB?4x9EcLNHm@!w>{LT|TO
z5pk#FY(AFT#>vF}^1WegH$u2YBm#KP8E0ojX<|8@Un~s@#GL?jzsLyM-%ejp>u7Pp
zS)pO~K=L~Dc`<*Er_7j2myw{@YaaVP?yphFF{#V4z`EYT*_>&mqlDf7WO=w|ppw1s
zrt+1<`bF_Q=rVhV*1NWCRLu6*UAPL6)mD$r&0{S0AhMMP*e`YPx<VO`x3ER!6BltV
zE|q<*ZAVeM_7I<ZZGrEzw!p7$g~w7lj7sxcm`{Xt_|e-`N=lj#TI}3%vEmeO0g&ea
z;i_kfYq9`+hmBG(Po_?XoTGHRSB=2r8zJ|f)f!p2I3)4D-r$bvIvvSXd<9>i=T5G5
zUJANJr@DLd?`Id_a3M5xwka~hdwQK#^FFyu8o8TM&>G_=Q_eGIuV9ASR9uTclPQK!
z<_S_a4wQBA6>L$ThRBWzFC#hD3wS%)J)LKSV;0gzm|q${lENI>*k|y`%Gc|5nX%}J
zS9u~86}sdqoyE-CE+~20^o;#N4Qkil)_95D-vX&|%gcdS(b6QoB6#af)epCzPU&7#
zR(Na0S;EuvR{7C8=bIkpVdS)V+alk^NMtk0RK$n7*jEVrx%(Jq>ZIC!@eZlBAPilh
ziyP~)<d>Nb+6FHKEWR4j+k+vK@(Q87Ndvu>#z771Cf{z40EY7Hv3;pTf)1kV80I@N
z;sfTdnxF5FfP#*g4i3uH*2&VLnT&uCut2DJ@g?}WZ~uzyY)Q@5m9=iz5y|;>zg-Q=
z`i{6>CVj&h__*X~_eMlMu&!~P8y%bVEV9>Qrv3Y{q2E(ZC=txl5%FiHZMh3!*>Q&t
zSogjRT6OO>wj4=Aa$D6<X`_k7BxBDOvoh-HG{jpcQd2*ltBNMU<p{<5*S7vhS^o?8
z-;JBN3+?l85UbF*!aEAVM(4fY%AQKc&&(_5p9;V9V|q}OG+zw4E$^YRsXY8%TYUo2
z3F!X*RdFC`pVfVQrK$$u!qh_Ugt|W}YY1`p=dRQN!O9_rcZu;D*!_7EmxP3{{XGgX
zB6X>ENkP?V`I()}L+qSxu9KSqbV#DX)!o@U$RAy8d{;?LE)~30Pb2_<;-iyJ`}gfh
zdS(WukKNe9(TU!|#?+QkT0%%fR!L-C)7lny80Afu=-ognrvyC1gi>Z@=&?ZEp>H7<
zUZM&Y{IRBEm9&|6<8!-f-jd#?nW<h(feF^DDzSOx?AiHg2lw@lA23&8+2fTJi;^AF
zjmMc!JFapaPe>-2Y@H9sJ0j$76weAPA1?<Zs`<JsdEi%5CbKVG+Tli_bg<h!HXi%e
zt{HuXr7o&%ecRAdB51FUP3>8`vp4C^2TszpuFlWfP@5NRZEdSII`f7xx}90O+q*M;
z-=nhv>AO0R)yn2O35asrq2;bv;Mmh)nv-=X*(C1_d!nFB*kwn$P#TYLyJZ`O!lz3m
z)f*d&J^?+n2Rje<McaFi_r-mYXTpMIRif0-2Q$lF8T57lz@d3fFIwpcy)PRqw=D{<
zc64dAVs-PT?adyI1cA%9W3b{ug&s?AoQOc9+wRAKmg!RReoaKsgQ|*wF`q$32pwr-
zdD`VL;Kr+XbZxQLZ;ZrP)(bQUe}+q&wc)KOq5{>Q*@&Hp3r<O~s)Kr5m>qez)3dd<
zH5F{>QAfL~+cLGY{VTS6F&)U&Q6>^i`Rh0ja;i3;z3b-btXOteIqZujUp4EMngTx2
zSBMxMZoIdnenV*tyglk{iF(Hc_J&8D?n}mD92v7OoDgY}y8s7q{BSRO<Jq&#&5B2B
zYbH{I>8!|`7qnK&=INE&RGNx9C%fFQLL7M~`9*Y}EsE{!z?|fg{c0;5?4?dym3hz=
zfY?rT-ozFbPuUY3XlCj%?{)RpGfxjO=+`;vaW3_FHwi7@JOT$+;^<<%^D~X^%~Gz3
zdlxvjon1L<M5ReK*g|BR8_QG{48h)j<Kgf8Ay(4)_I#rX$#j*^ZMYi5p_0?-0!|+J
z5&-PERF|c=VoqL^Aq)LL?8pwKwc&0B5pnZd%Z3-wAG4(}MQe2L5Y-%*dUu}h@GhUe
zBBm^)A-t(qn;?xUb@T{%h*ggrYQGe~bvFsZR+Da2r|EvSaj;VhoK{ux`-I83<_uBX
z!*l88Z~{}Z-G)T4hHt(1%{?E7ez0a!_Tp`02}z8<wqGbYZ9D9RAtdm8SH*$f4I;SR
z+8E5qJ+>`joWW2KL=9t`&fSRqo7QL~2Fh&?Tt}w<<O!TDP_+sVa*Vg4h{|?t7v5Dq
z#-V<2Ssi9AWbrxa=6K(w-0qGvubid*S`kBS>=?@YS^#DyJRScD6yH%3MF7m6wmOvR
zMUkt{aI$xKiWnW<jtt4b2{S!P(5QLgwoMpX-I_m`zee*mXSrx^c1kW3H#f)3XxDa$
zbg4bEss(pnV1Mh26?oXL^e}UPvqzyPo}<;XQGx8`yev^JY6uNH4%s9k#onusXssow
zGopH9QnF64Q$PDMRe2jvK;C=Lm{y<lh=E6|!ol<d&Y`iEh9hX{7+KmEPclpk@zkb<
zH`>0(9&lOBHzp!Q_!Kb;)?h&7GFzBnyNq=I+Zy;#eA8!Lp$>RPuTxMJ6k@PK@Nf0)
zfrmEQpt7J@!XAA-Lw)sr#0(E)!4IUr<0>W|FGrOXQE>u*yxH2mr8!fWD{cXMUIe7b
z^-kp)Yd;(Vz!%jExdQW90Y~A|wTbv&=YG!gWl@hUL7_pzhKBu8QN=v6wBxi}g0AxT
zEMvk!E7$9*L6cK(Zte^P8h#e;m5iOR+Ad(d2EG!PC?+ym_zN9J_ez)FecCZicuaCa
zjl4oSBUhgD2r;9D0RI}m>(Gyl!*Sp_)I$O+9WDtya?Ur}VY4YenT|xv?TEVr^Us|i
zqp@e%zbQM&0A!O%)w;;Fk>xI0ffPz944n<0D)SRE{dT&=j?Oj}EyRIgVTU>hXWi4F
z;lMmBH1rBt+ZPZK=DJ;l7CS!NTH!z8%|zp4Fk}{-Rm0kWBGGsHU0x74(mSwzsAOxm
znxG>8$%$N@c28cxTmrd>--g2L2Od9}y;5v%B~DJ(UYK8h)w0D7Y3EUPEgXk-x{o)c
zjj@`tV`}GV@N4}7%}bF02%)shgHOhp5QK!-kR*Z>vc1;tBhpJ+eV?xkwb*Tevo>F0
zs^#Q}nkM_Bv%!xQ7Q7N>h={+;p5(L08GAyen-}w8xOmXQM7;AwIt4phewFXZRs2K}
z7ZCbWP4U3+q*@Cb1)Q+Ga{rql+jrp|&9w%46hULCd61;ncTnR<{0?u_&)}N)hoq!(
zZFr@<=`>li&8yMjDzp6$6_$PVMB3{T2!+qC<Ie&~9a5z`fkixd%bXUhP(UDvcrCIb
z*Zv?X%+EE>WAoYaSbG03BKlJJE$0xpnM|)BJO^f*t<f#z4!fb~Q<UqipzqClkgb6Z
z{EWXZjMD_FJCGg7k0#;0a_{)YTj^Nrff=4dn)&H-rew+cz9C@Z)1#TVxBSg#b7N*u
z=JerucEMmJUa&VD8gA1%CG!zc$5wgX9U2W|!JPDWS6<e~@nNciifqC%bTSSdK~C7I
zDU&$p)yDmMEHfgSi(3{qeFcVAV&tZ1x!Wgj_T^)uvk=)`anF%1AwA$$b8+v|#K$mf
zBgw=*VY8wD!msGH1TWpApFA~PpYh&NoXQ<u2Q;ZOtnw4tU4dN6idg_*4$su<%T(_B
zMf`1N+W1uA8&+8O3VS)wx_HWizIdN#8(L0BJWRsP$1my8I+(*xyz)gG%DxI{J@=t>
zt!R>n>j{$&W3x%9$d}|u#WJNYdbP5Nn}?T|Z5(T5AjE1Sc{(>BEqQ+OA?$RMvSy|S
zB~-s0?kWPB)xcAtNP<<862I}djDTP>9BO@{B;)GSd>W;j?6+JC90bCn=@$^SQB%(k
z=irAo{9<&>m_Sll%s@*c*Ml}8)$nd=Y0d6#%OWNsgj)D2V#pD!?%}Td$rges13HuR
zwQhN=)9{GKKx)}8ER9D#Q<=dasX*4DRgVAb5S%Yom&h&W3jIB2<zd2#Npn9%W_&bk
z?Wj0W#V80oWRQ`bsDMf0XIgHc#j)PG1a%j!zt4_Mf_Sz|L!b@Ilf10XKD@h^U+p}M
z%J>LeM3+Vhm8^Lg4?M2vCo00u(VBegtZAP6Q9+*D7kPWMXL(fe)7KR&J8(-%C!Oz^
zpMt9gl@5@T#Hzf*hVh`$Num_2_)<X%5=qj7->ggJX>{t>C?p9RZa4)L+|fAHjU4D1
z21}%7zLePx$&j_8O@@i@Mhu(Jk$yL#AJ&Cdw*Oqyavx+k?g;UffHP0+=dc+WB}Nhu
z#k7?h{gQ4wCqY#+&;=%ida#Hgn<D<J!62o$?=%B9bJ6c-^j+%Ntk_~2!;*bX$M-nc
zBCChOmK}tn(>kH=A{;nzGBZv)M{{MS=Q4<_XO{(_2RR-bHRJ5s<Uy(5s-0QPoUnFu
zV(3nmT#ix9=F?!!H|nv6_hd)kVx2J8YxC&urd8Dy2&Pg)bcp5PQ%qc+yUUJ2NsR>H
zCa*l*%2o@Oi#PQRAT^VQL^9lyhZY^thiDlPsJm<IY=qP7$`pKVgqamIDe(4TLTx(*
zlS-2{a|!l{mO{+84P|-C+5&xsb_>wURcD?Q77Rw?MG<`=1x1s0py}bx*RdhzdS`D8
zAw`anzHv|5&kP9LlwhT}A+gRRdgqJ^JYX0?_}(#!K}8Evri_|YsdPti{V1o}{`x&w
zkvARlM~{`cyzFoOMSTiZhEMSE5g5mygttCMRwCCy-1a2w!p^Vzg4!0tCRh_uT2};C
z7-P%Dq2e<P_g$dBjx)#VL9v%IwD0zQY9=9IP4#?S=!<Jx^5D(4?4QhGXR1GDaG8ih
z){lQ>?oJQ8V7&O{DlvF*aL>2V(T;V&Xlu)eb$@;G)!tFdht%#`Q=7pD(>I&3swb==
zTYyh-cOD~UI+9gpI$+{zfJgmPEimSD>il3~B9DK^)Zy#P+k`X3K4aDdR<-Rx1YwnL
zo9>$0PVvdA8lT|L@rwi7U-x|VjGbAVj8zvei|{G-3C$_ZOl$4zDa|`)?W@h~u8X|5
z&zvS=Xz<zw5k{<X>~64o5o2K8{veghnlfHljO%@qhxm5`@B<=@3$DzWVD@>ql-P^W
z#G9#_#geP}-q_a;<;@w_1Zw@$aPJ=elnS*Cr%}mL>sQFWvZWe+WVc`uJFdO?(a$Bw
zX!9P+O=0ycZZq;t1RjmYU(_`#f0lo7s;0=?SrdCIxX5;A5J>nY%b5CD>qdJF+yjx|
zcFLI~#*;ge8x9|ylhDnDl&jM&emtB~*GC(V(w3owXyEqAjO|vk(-L8Ma{xpig2LUA
z!sT&x8A%KE)BwS4EO!MQEJL;UT8rtj9XX>ak>E^JkG{*B-N-o+N^AO^6<pZe?dkK5
ziLhTpx~?ZNPsq9^MDIcES=Fyg9`m3Au_f+Qk8K)w6-KM@pQZgc1hSskf`SqDh6`n1
zrpJ2yM~K6w2=DH8C`t_&)1fAIm)X4J0?eVqL$fslUyZ(ac@i@KB{2)aLx-InOl0>&
z{5tf|7PCb!relAPqUe!T+^%46>g{75JMB%rY9n{3GWP-A)J2yU$MlR(a4SGi?oa(K
zXJ-!>L}Dgrn<IAM>Q4o=B3dULK}qsPeX73EDx3z^NmHuJh>XUVirpqu3nt8{mk<7Y
z41dX#*-ZMDB0Sv>9{hD+n>iOaq?-JFS)O1Coj-e3Hi$1Unk7lI?Y2bb>Y_jUcGmRu
zOPNU2Zi=56J7wDSj16`g35Hx6U7lwHo7|zG5sr9NUKXs*W_D$C`^>nvKIvfQT|rRA
zlJJ{rkQV04X9+a#M-O`ujr%z)&+Vx3X4|Womu$w-<TZQ1)Omj~`hIip4LhcpnsoK0
z<9n)9*`>;G19_ZVivYTY@^7KjXysPiDa(0QSt3`WH*YTWnLy=U8ME#eu5q^!gy>lK
z`{`P2YoSr~xgD&X)3F<0X+4HS0o>;I*Ve7@>~U=<SkPjAl~^(u43W6S@nxpL=vA;I
z!ddKOTrgd5ekfI!Xj8+tv#CItho6j|gaP9y-M0|9jx*NM3-)#G%8x8KS(6kRc$FCf
zyf)Ej9FtTr(g#)B5qPNRx+T|UAOha@=jmKLtr{fOokgYYH{smFFD5bBdV`E~UyXUN
zuitPm!F?^l9e~FYDvzw?9Qjx?DR~E*R|K`O#BP_5^fyPOw~O8-MxR&`98DO=94>n3
zT0tl&y?YMl9Y!|ZOXW;A{jM0#tFi@7YYrXVUWdgcAGo6LskxkoH}$6@^Srk68(w|)
zPU9*(&(^6~QAoIWSco`~Z5FFVTL??Q>O_4X$PKMJjgv2;tAq33!2bOn-gjoWO5ww_
z6!76pqx*;V@G>IG0>T2y0zpbrwyR9=UAULNL{sD8gD$J<UMo``LVgtpZRbHY*v6@~
zjbrf?)7GySVuiAnM1w9c@bOQ&#=In9qOl`gIY?t^s6T(#e`!VMN%o5g+<<<CBh17c
zyA)45g_65Ycp;Nn*mx6_SUxyWfgwbv&mnclr3+V8V4rJk@FvDKfu}KUER<J9vtj+#
z*+bOnFfg^;${HA4mp)w!Yxqfo@kPgoP&%z5Q5cQMv-%6W+W7VHZ>1!P#RoHa0wI@N
z(4@;L;^S{+XhkU7Iaa|NQBd=`v9rkyWg6?|!qvtw6ZJn*%2P82&?2&<TBwV9Pc-H)
zsBnv<6TWZ;r1UM%>vu!Tcg`-c#D!LXur2hTJ&}}mRjOSir|{{$y24eBQnc*8h^2X*
zamZtz0Ti*ZK|!4#kOWQK_#v9MtmKVfp%Jv=3Wg!50<0r69wojZ#I76Mz3iS%xo8EL
zjrK@*ZV!v33A2Dc3Ao6U*1B#D3fS_9B4=B77%KCKx?^{f|H@cLp^KUmDo8JO9f&IT
z#~S7E<g;)<o1c7vC=uBzUa$&U_u=Yp(`BqSRe;>*(bMA<DYL5G{=p&9OCyL<aLbU@
z(`j=&%iaoP8?HGU^h!a`Y3vZ9*kO7gI9FkTurifukdrL-L)qzeteq{D9xvu`;6UZE
zH&}%<OIoJ*nV!BOHYOCp150511a)$eH;cRrN5lhbco9JCW^FJ9?gB=CF*;A$0Xf}7
zuv#fyb9yLYN489fV3_ms3DstkK_XKANFh*mz1uC^K2Ot|Pd1_Z-=$d<{jjV3AxW)|
zp#F!_R1+5wk%s7HhV5s9hnUhSo?3$c43pdN09vR;O`NLp&;UXTp+VsA+|~T*M?>3j
zR+L)&Wd;nE(IBgp_zA_d@oL{i=Kc%f4I7+0o4d2vI$vu4S99vkAH$pxX@cQd!oL4L
zM6>k~TFj53eR%Z#H^<+9bikZ!?ewfnTuiJOoZOtG{?o_$3-nJez<;&bicc;Pet(!p
z%fSEunt%8?D2fQlDGEo$kH_cnBa7U>(wl3II5NW}g_LwC1j<0r2N4YlQ9-;s9cggP
zqM)LQo}cezjQoi32uS<fhkVN#U5CngMj5D)iUM7H%im}rT)$SG9FfMjna&=$Tz+L(
zJ-9fhKLyfiDcH13O7&1lTb_{#o!H}*gK}I-6JCZ_nDV-X+K$g5k*paJ{GAQp+ia$T
zi=RnRk(6H_!aaSJTZnzveS<q{_YCIkLSx<z6~p@gqEyr4X_q!qBPf=~Fu`<komX#_
zzRsGe00`Ab*IkGHDSyp~qZP1CzFfuF=W!&8c4!gAXTVPDR3DdsRgt@h%cp!$>Zx=$
zcPcX>^}F9y(_!UthTcL(M@j?}^`T|yWNkzF`i3mqRQf$BktUSOMGi#4STqxjDbrgn
zqtQ4%N#eG3R<&+>Om4JIB4N8^--jPfWf?R!`?0;Tt>_>zNtQ>nrESPU3CZKqn>QI_
zBmS_iITmwc$F`j${gWa7Nt?NPuL|W?o{>ZuaK(A)iOV7W*8QXURM<ztar16nvNz3S
zSGjMh;o)XCBRi(6Gvo8o7Y<cR>H_datDPeT_K_*IT{9G*thxbMyrq_*)A-xeN29Ys
z@ej4iRyme>j*h!5nCi*^np|dPsKz$)JvxbJv=E`(yu*9moe|pEII81ac6lifQ1t(j
z<KKI$KjiaYTa*94p8n3D`X{ILpJ9Jzoc?bV0MO)b_rJjY#zy^T=-*kR|A6lPf1v+k
zl>RgE?`*h#5Jx{2y1$$Ae>i6UY|8yJ{_jk=f8cRGPSAgb|Br+p7TrJR`Md7=A9-4_
z{x;8lRGR<WYyT_wuS@<PVDAq*{NGIAKfwQU0q|GyUn}+>WbBW^^-swERlom={cDx|
z1Ka)&vHw<c|BC->ef=MNgFn&7uNwbk)c;*=<)t7X|8*?v$L{qZ1<il`!2s}o0KTN-
AumAu6

literal 0
HcmV?d00001

diff --git a/dist/external_explainers-1.0.0.tar.gz b/dist/external_explainers-1.0.0.tar.gz
new file mode 100644
index 0000000000000000000000000000000000000000..8139646b8545643998ccb11264d5c06d8c9e0853
GIT binary patch
literal 12264
zcmV<EFBi}siwFp|6RKwd|7Cb|WpZv|Y+q$~aBN{|Ze?<FEio=IE-)^1VR8WNeQR^u
zMzUx=t4jZ24xM{Qo`8@@y%OyzwOUI~bk~u6CFOXxWEBGvLlSEcKro;vuGjzlcK19m
zcu)@~c6PHsr5q6$%=Gm1yBjm^mb>+**WTUt-1B+*#V5J8<+bkr+uJ+4`QLcIzdh&=
zzF>D>e1>ZwGB1VUzxdC*24Ayz7R>ok|6u2v{x{$3_IKTHzV7cI><%`*Xs+M=#qTnn
zMqW7NcS-045l=<0?{2%>TfZ*<9~>O4=YROSzk9GZ*zNcC_n}>Xu(z}G1>0-*|C3+t
zzk>W9>{sRgP9y(6i~JuH<^SMo;QmJPH}Ze$<=OMo^UKp;iU0Ta_t)_MV1IXa-^l;n
z{XX#j&ff0s7i_!X|4)9ghpywq8_zg<8H{-(_~ynvP5&2&Cx|09aJL=yf4ykor7Jer
z-rimR)-21C!>z65a_M?BkQ=Adtxygowl+8D)vMEsA1>MP`3Zae>ip#F>g?6|C42Ge
zg8g`T>adH`*B7r&eteF%9cp!Qc6oJi_U9k*j_RWCvJ*ZDq96kZg{$tu)lxHSF){PP
zkj=Rlu?%Jmd@&cyi+nbYBVWG9CUMFZf;%kbNgDf$G2V7`U)b0WM3x4l1wLS&U_Oq@
zeKuOLOFovp^kINBUQB1~8x~LCM*!{O@nX)StO{D3R$v>)$tn$|vy8<{U|9xG!K*>G
zV%{Q~#cA*l8c}yt_ewVNG6v&LQxD#ZrqoE`q5uP*dLcWd-Yak{BAg{nfirJR{b)b{
z7tmJs6GICHWWa?S9q^i^ap*8F<@!fRfIB!Nyt|0NDuHM7c^v7kR7<uDvYG5i4(YNN
zaY~RRi!_OWesbiRr8jHUQr)y@HiC76u6#FM^3;K4OkuSUj3APKJ1mRY*n`!_cDft+
zjIfZhxfgj;j;n|xi^X`RKy%n~#t8#J1~jloofojP43My(n@#|*&?<>pkYHDnU;-#j
zcsj<uJA2#zr%U6-fFg-~)6*gYfd(MMO2A^JT<9*L&k>IR5#s<Bw9vf;slD`nix(}{
zfw%FmwAHm&AO1tc+y?#vdrO&3C%OlI2M`AWVFB3Zfe=LgL~@cPXmx6+?h*zZ15X0K
z&C5iWq<q5D6gr@%6GHqu9AqB*0nE=ME;B0~MB{Kl2m!8SQJk?bm<I>|EU1Vl*%B#D
z(1@@&J}iP}M(RoTBU?CHBus+oBBchfJRvX0lvks_165SP_o9`&2kWv3iCrdXJcozl
znHK@%hF^d}B7sdkO<Z&*R6i%oV-iQypHrNq?xZwVV51~J28#(iHP<N+1<V6(7iL`$
zYB1efDcl8iB$;{6{lH_{Dk;wWO`N`~2#O`#B&bO|A>HLH5kz__IYUSUs=3X<HDTVZ
z7lhs@)H1;mR}RWqq{gwQ$jZxQg_e4N3utY`p2Qcl4hXqkmZA0`G--HsA03zjzw_n^
zj16ys+y|<cZ(`HqB;hdj9dK6|FS|A>PIwyJ0yb|sL#&8anHF%EdPFD;=&mFpH1H!&
z;3`JMoIVbM%m9QWMHr4n%Z9A6oCV{VEg@jlGSCpftSP?@Xibr(0Y8dw7zaGZss06h
zsP(qFOZSF)kP9FhS_BV<i$mfOcrTa+U|H6zeN|E$@ii%k1c#MTs4$3>t(K7bR~jYd
zbFg6z!}7#SiDVErG?O_`dAI_mh~5$MMnE=5Fp)RsUA_1LNT+n-jY)wxwh}d%t$>9X
z;_<{@{pTp>l_gNS@@0lFJZukX(5M(tYg+>f_EuOLqGw-OET*eiVu)UY2J0r^SR9g}
zN{C_X&`5N#7=buZ@<f{_L}vs%0jT(bMj{bkF+R*{lRmsxmDmOiifbAlNq@uvp%cKy
zJx0#sECSXtvuf$iq)lMNHN3?`V8=8DA?@H&jJ%KtYMJ7T5t%59NG$<FhA3d3=g2}_
zWg=%BS_I+TtEO^sW{(2@<xqi?3_^T61OpVhv-Fv<n8ZqCd@c&29hCBdqjDLO3RO)c
z6`)>}HjOdLY|J~hSSb*Lg>u9l*u>+7AUlADoRbVyHs>3XftJ?dcN!7J>1x7+*@z?<
zFXDv&R-Sw5I~0a#ZbTW|jEi6zkxl@b#FZlK){-Pjuhuyr-D5W2x~)2fE?MMe>Y5!N
zwv22vqRgC^#$_{rXT*W9z`W!n-vMrWAd4df|9Js~6ym_+7_ctYIoih-FKW>=aM^cg
zfZ@o`4HC2s#x569^(wNh^@41svxPMW9m*^$F_b9)HyM0nIe;MvGX-NX;TfREP$G!K
z&|d~<k3@0QBN7lWZT#5-+bEr)&lRt{FkAH|DTiMHn1HwO80C9a+ba(eht!@LyaF7T
zAWc_<pC#s##Ry&oTmsQ1p$AlN?f}?Q4-2}ZOb^@VE4T(mLX&x28M8*)ld#e2v}3K)
zYY%1d?_-fV@E%VxWH|6tGHpl!h(dZ&U6#n{T8jt<KJ+&8ZaG<l1|;&u<H-aKLeLI8
z1gR|l1tAlsnOq+u8I)PA%tw+uW^#xIxm0>wFG)glR^td3pHPkROaV6zy#SCd+uHL3
zR8lWCij6dlfSZNz(twzCl7c|ejt~!wUbH1aM|8oDjw7zLCCEfDT8tY&@0DLLbCMpi
z(#0?XX$%&@Q)4V~J+vNl+1UiiAa|hzNH3&svou*Sl|Xn?4<D1@Q@(#E*JZgSp2kA-
z2x~Z#alAk)Oa2A|@L1?A#UjXXwjrNN?EqNQ;9E9BNmSk^^hjAq(1r5Vt&Ya|60P*i
z^s3CsriFf_y;dYiV^(QbR&kAXO>%xy>QGx=QfnZ;;u;#7@q{+(e7LKr(4ZK)Ll@K6
zqG8u%7u*gsxitFRTjkQOESNwx2HO59il}?6L0TEKFJZI=2tFbxw0hwGajvn8ez(-|
zYowc#`%i=wi^%4j%jKTLA$U^K0MO#>FxSH!uPdj!0IHiJ%m||NQDA`sm?sJ{+fL3M
zOS~$x^+<D+osU0AHEV_)S;NXWP;Ppo|Aj#S86!%OoWSJ>BBUAVc?uf>6nuulu~T&a
zr-TzOJ1P%8wgyjm1}x`jL)(f2kVgSPm1k&=ZN|)r!$F?QwWp(~-9d5bbF{D=+vFgM
z&2k1*6O^Hh8i*>gwrJ%^v8p>E@cftzU{FCgQ$(ud!8Ehfnl}2&lP#mp@1is_w4mHZ
zT=mxZtE;o;r!8Q}yNvLToT!FDE8QN-<~K_))UZwkzqD+2U)pu`U@bkLT-2Nbc`b5L
zR$+dounJ1rk-U&|p;<VO!LHp;J)&!|OJoO~gxnM8h!s)->Ybc{!GeNe4>h2khB8NZ
zj-?`fiF={=gDvO_<W}JHVnmAtlU(|sPMzi&vN}MVIyI>Fv|VV$&6Gz`gQ-cGZOF<1
z_fakx^qBfRoaf4{P=ql*@VL?X;2yZJS2O9rp-`*AwYBVIVMt%xL=M68$bDb5V@d!i
zP9rI{D&zo+O7VOj|4z|SvPskKMnkW#^cbda<Pr#2jl#6Z6~icr@AJrC%(anQpd2k}
zr1Pj3wIb#S%Q_YUsHkNuihY5zC2fUtQ6WT$ob@@9T14f(2wAz56p&UzDUxQd2=<~T
zV}p%Rz5tEff>m5&^jqP=nlzQ{$4XDdlNxv&i-#uUYp&LKFm}Mt@D+84W7`qJ9E3`W
zte~ciosOYuvb~X@iV+^;;FP?F(z@)Cm!xtI($~#xM&Ys_Bha#f)(%tO!8kw%n)<ZT
zF?pn9RW>@U=$I82TN?_q+yfl592Js=;mD3|Kgp|878wC*QI70Gn!3Kpeaa5b;|$+2
z=_1lRqgeWL$i`E04N<=kaAE;!LU5nU)D!ZAy)tSjX@JPkE1=T&Qd95;i3wJUTgaK?
zcYJJ%ev-!qDJh?NsZ3>+T^5sa*ar!ztrUUM(z3n%m;`Dj&2%dVh3Hp_JZa;YjEOhL
zSe&tYFp$R6Tg)7&p8%GkXW3X2um;+(!dS}9P5IA@K;^4Y;R{$ARQa^DU>U@7Oq3%$
zfNQW!#xO~>Ai1-GQQ%7KSM!!$KBXsWbc06Mfy+(;L2ei3jVJ64SgU}XmEi~j?Pw)E
zVDjG4J+lM@Eho7>d1TGW7fwk7kwe=-a53~y@p0|u7#J<Aau*|IpbT2am+b7a#r}MJ
zd3I?~{O0WH`&U0+u{XyT7suyUXQ!9!)rDOn@ahFSKK~p0%h~w}*hGQMLEmBERalb=
zNY?mP2sY;@iotqXcCEloCUlW2mDWgJ!06T4)yq={kbmAgJAZL;cK+Sz52xo>4*TKs
z;`#Rg=kcG<UY=e3jcDn`+12^!rL1N+)}6dQzJP`M@#XOad;R0Z>sObjQX9)WWr!IJ
zm_h=>29({PERb|6OLPMZn8ryOpgm8Mo&cR;8=}Nq5L>ZW8O0O=3@4ni5#xa%kuTzL
zU_5Op?NyqWLcMmLx9SaRD*W1oI~uY0+RMNj1tBHz&rmfpFp{E-K#{$`t&rlV048|r
zcGyPe&VZ6KJM<9oX&8W&Hs)Q&B;B1tq&SZwJxKc<X-i<{APh!i`4M<ijBDfx5<O^!
z<q(3h^!2<VWokjMV1UQ0To};MDzHFH?9IJtF;jrA>*|EOPC+CbbNqG!9JmXtV41Q&
zGe<^@F$t)8)q)$tcmO#D#8a8MM=hW0a!gy79c4o4!bqKkyc0xfNiAVph&y!dP0VY!
zaOPnwNp~8@{xS#)5y^L;v*RT3F!qR6!vbNSctN;GrK0k}$s)=PAX1QZwJDfvKsvT@
zB8TPz2nY!h4f=9yPj_Mh_MU$mP<nQv>Pdhb6e4t~h3a2(!#6HF9;32ARA_mOgCFOL
z++vV7Gqm-K+*?Uf-kajn2J3hh$1=W3A=F~ZpW?Y-;7vG57!ZF1qZf_2oJb-AtV$xT
zhywXM!g3cY#wL*)YVcV+3RPH(Y@00<L}(_<ybeqWc}KZrfhcB|z$y7YUZVdby-<TR
z!m8DCK1Hgti9$Q^VoY?EfuRVox{cB`m#qXbSwnfk#uD^-*vX;?6>&yaE0{>Bjr=d!
zpYSuu@#FIe_&D+vu%rE&;Jla4Ni=Cw++fn;?nRpB89Ei121yN0JvzEFKI>GY(xa8K
zMyxrl5DPhCjq$&<$kMVe4K$jTPtQ+K>(*85sTRktUqhp_zaHXRQp5oS?MjtW*tLfE
zjKEo1*(<CDeB>*Rs_H0)B#a#%0~e*>$!9t`;^fhQNx(y2Fwh;q|56={Fhk0LOk3AC
zEz7sT0EbeBD@|A=ZI#Pt`Qk3?oWxPvR4&*|tGoZNE+dbh+*|=B2oME$1qKr3QCNE2
z&J82)h!sfsJClZ`z<>k+L<hVX3e1Vg)+z{V<U2K%Gz27v_N{c-$R<s+veud15zotL
zC_igp5%^{c;G^I;3g;H;kz(FeRpcO`K!rhG7NF3l^S&n1lLt$@bUeeXxu%Ldk9oa<
z>xNwu>;P*i&3mI>Q1s$k?pBeK9J`v8b+Cb{6z%?j`p}LHiU+CRRGeO${z0VtSdy8B
zxN=*_iXFrw3cz`VIH0C#FUzU_p^{luw%r4;)k}}D>(?16sydGPu_ATV8Z69F*bUa2
z?Z41w*T%ZU-X-S+AT-gF1q1{Hvx}w+AU`l2Kx;(hN<<xc$*p`*of-_)eE-vY|Ji*1
z*<Alg-~YV+%Xb*Rc=f5j|Jfc4wg;u}KX>*A&G$c_<@$kVp6_K|PnkQ1VDAhzH_pL)
zI%GA(vAJOy5}9mnyiQ}x?)QGYczMXa3&7Dp3q^p5G6m(m7`b2rZF$hk3cS1CD9*&z
zsp|LC>h>hJF?-Kt7_)b^N&pHpg|{&J>VpZlxuLXj?*&*|hfJ9}n;RF>yb-+<utE+o
zF2uvi-2@cACn~q2#XMP6?t;afgmD%IBY<uN24yflv`yjqyFC?y?O|cO)g)19TBM<M
zH`PO;?S%H)GggNjeUF)iAUlYg+@|youV0_^PXGG)<?-41P3OT?Nc^7auIs+t+>l`f
zG-}6Sx#{Tm`T=#`1$>}b1XMyJ15hD`0;p2cxEa#2RDI6P*H1V=4l-C_-~~}^kDOgL
zN5^x-Gm1um8$zGEKqqN$k_ON#xPSm2$G^vDr9FGb&a^*bWA7O}c(s6ql$NBtzqvuF
z0&9L&T|EMJefj*=#cME0b5x2TO@aDch0-A2qFRlu#KX+UmcTkBB!Sz|AN6<K#Gha-
z8$3h&o`LT;jbRM~suIy@J^~=J7BQBP3hnA4cRCtQG+>vr75WJ(Wsi?p5(g1s9P94o
z`#5BEhFGskc}Q7*j+M$O#<IsLr=TvS^zta*6xJyLf8i_e52l!nkuG4u)z<+yh{3cf
z)jnj`e>%&+r8x~(o&YzkmK3qY>C3-$HybOXvHzRz|Nm9K{|C#z`ToDL|Iz-xI6Xf3
z;nbb`pW6OMYhT&_+Xp)X{QiG;d%Lm!Kg-qFn~lBM*qe?0*VuoJ{fF`2ov-&B`>(P8
z!2U~CDrDtm@jU#5_8%C4*8YDx1K|I?-JO18|9y_@db9{azb8NYzS-Q+0ZDemu3Lg<
zizJKVP<(rYeqjq5$o3<wRsa*}h*^zXdN!Q%%xg8A)#(36>;ESI``hXNCjZ;$|62V|
za@`$Irk^_gv%S4n&j0T2>^1uT^XPv*oer@^5mbBTO^2kyL9NSQ^fmesqe`-dR~r4_
z`2W9i|9|gbZ>RD98~tzef3o^@`2W88um0YS&i@Sd_9*`|*x7IN|L3^mHc2cNsN(6f
z`rMyadnm~}4x0p#KTN>!$8VRg`R0btC$xh_C*>hL%4Xd|S*K2$DIGz#F5Y#$k--0V
z^lv{%u|U5w9AZLaIP9t|i{$<?36D(6zsEt;!QSv&b~T2>+G)Fsxr0e^n18FE0+?w8
z_gn(0n}ayz*-&kt*wGNucV$5DnD(IdZIO<*Cfw(DL$$R-oNf(=xJ%%02z=1)x*`dK
z3||ta8~R>hLOr=iNtwVoblO8b;BeTs+sdzg!e9>AX}8+=F8wc(!0=FT)@rwkFQ}Qx
zK@f%xb<w@<-^hmenvDq$zs9^+L@3!$(=<*ytsf)(Jt3~6*-a8F_M@rWDq#zEW*Ayk
zZDzjt3bjpwxNR-StY$%Gg#{sM>d00|1V`-}8PryvNGULn+9Z?O&a=W&sz!<?hd@<F
z@7wJ|hD+K0P<$$m3iR8doj8uDrS7LZ>7gp+8)7NnQL9jg*jn?>Q9bq?#-6d4xZ{ZZ
zEe)b2-^&7j&E*F>3UV2L9WK&P9KFx++Ws!caF!oYXQ}b3k%M1`d0{xzVZWnmYZ$7b
zY+Jjav}OHd;TF|S7VhMAllnGLX_I(s?EfbI_iqya+u!N$f8B2w_CMSH2XpAx<Nv3f
z|FyF-7&Pa9HS)h%|3>~d*Ka5PYo2a>9REK}{vYfd>^JiNKSusH@&DgW{ukr_2fO>b
zP5i%+|M$p$D<t{^{=c{WZ?LQ5|NGlJaDM>5www6>=eQ{PShLKoD(la0RqyEhfcn<K
zL>U{dYV?2O|NR^Le+LKKyZbwh70~Ga8vQR1V||Jr_+<ZY(BH=TzsCP-{J&=XKd=7>
zjN15rjr@Ov{C~V3_;CIAp7#It27^72|2tR`*7$$FrT?dvS^Id4EC@xzE{*<g^#AW!
z|L^U7-PC_K`u{QdpESawg1`^%|G9f$>c9ItJD~pudwb3PKcD5og5NA2-g;r+4{<a6
zyN)agc32>Wqd4>r*(i=fcEnzIq2OKC`xc9aRhe(ArAn2b$I&fM@%U`{I&JLFs9Q$e
z>PlnT{1A_anyY(<wBvmcC5x=Kl>y|9I_!nE_?E-uTjXOC{hq8mmi^X@ZlI=TL)Fz0
zHd3?iV!bXNJs$gf2>7uvqiWTkvFGae4BBwq!^76i;VEWkpXDP#mK_|iwEdP*Rk)h8
ztO#$jwjT6UjJ<^GGhXr`g+34p4<wM?iG0ubF`fb$mBCbFPXc6dfDXpDG2*J;I<5b1
zIjr^e?SHqr`MqcI-ZSeSfa%CFtaihez9&c1RpENjyzsn8heHH@$Z5L%(2v8=ONR*h
zXCYA9%19p^gC}0d(+j?|)>s|k76yX$F0xj_9QUlCcktllP*RHNvp8EhsyVPszE~d8
zWv>)5c@Yib$P>(#vFTROxbg&ru^0Ypv71`@`{j*AI$uf>`?A$-6!q`f{%`U>zn%SG
zyZ^)X*Msfd#tUfd{~G(hQU<W~82*1+{$~fbglh6XP5h@>|IZu$Y3lzP`Tu13|4=6A
z!TgVr|GNHfkNCfd|1|QyS^v)~{|~mmY2<$+|DPoPtEIw4MR33T-`zjh&FjB+`<VYf
z80<FtzkHU9^8fjc4LaSQtndcz=#hlk=7zdUdp+RIC^7fsmJrm^S3AhaTkbecxr-;j
zrsJ6xO}UF($4^pk&WA}FXYn`=^>8Qf<OM!aJ!Gqd&d!xjUj}%5@zo*;xx<d}NP22C
zj^)Y6QRYq-*#bAMRWDr4a4(J5aU7oB@v%CUSq$kgR-SR+>~-_l!Uufe_~q;Gk8v|R
z_nXa)%d6vytKpwty*#0q{g-b?llSe-jnnfJeY5@H%Qn9D{n>Zl!_WTa#>-c4;E!Hk
z2E3oKKc8O7{`xZfjVJoTh`3*f+$CS{xiZGo1l<AP7E#tIM$$XQXI-{czU7L=ywg?t
zugJ4~%QgGuaQU%119(CQ#^bT-6f>W3z284>?Jw|*bxb>0rv>Qa1l+fg6-h52cHqH0
z$028}OLd09p*}$Y`#4^DD-9u?OK_}C1K{lcEof#fJjM2OfgAO(Hw%Yddh-Pj*$<xY
z^T-++<J^;Q!PTMV9w3?fc(5>Mhd}kS!=I9$)LFFvI7Y<PVFc=F{1Y8)%M%{?0vaCb
zu3eW=i$nJI?FkPvZ}{GuPCu|t>?B=SFZj`!4!66kr~ib%61vsZeZ0XfXrKf2v!@4z
z7id^Qlywe*j`6#dH=tX3!^6FraH2B0pCQZSD{&U7<I0zD>I;2n2A%m3;Q8-_r4i4T
zc#J(Qbs9`DO{0<G(Sgz2{G4DKB<1t?RuhLDQ6@kHKrJTnm~Ml11I!ELzvV2zv&A_c
zgpS)EWrYRvL=C#(u|&1T^xL-}Q^xNL%RtXQ=z9#%^WpxM|CCGf&kts82$)D8a4pX!
z!JU-Up(C&ij>f+KlcSSCIq+jI94|szxQAxpFdrx41Ay#*@SoBg_3P&#u4Kjs4<(^v
zqXiEE7+7~aJ-!zk`Zci45H?&5xo#*h@I-FI1uCI~IVS_VMKIYS$bwsrN37)hk|%+H
zExa)J2cAnlA#&1PNU|ycMXZb`$^(0^;Zc*YG_N8-6AL1N_>Z6o$9HCuE<$fU^1XVj
zIzF;yrw{+=hTWQ$Gr)h5T`Rn&UsnuBIW(ThEFI1qPp$?XB{3qEil@X8Vx4rOW}jOR
z@Ssoswo%|34yH43)Xq@IIe5rT2@<KKeljQ~p{^2)pl7tZ4oV`N*n5+SMZ#0i7N$Yh
zehmqFXjW>7Bt$&`^-hK(4vGR~nNj3)mjp!NCfHh2o*iO-&_PbZwaX}FI|*Yivw$(0
zh?2L8naUEPB+^lkvnkMPqy}PMrcNYxtrOHs$z7@56v>wQE_Z`!2itp;4GmybX~{(1
zx5NaAN+};jJO{8-FB0%z?n%kv`F{t)Qys>hixf8T#$c_u><q>XMVxmF#*PSlnOrVG
zDP_6=(H!Js&|a9kO7hmN84i*u6|7GJ((@tVUYLYi*JpXfYrB+qc}$NA{g@&Ky9Px?
zWkJwDk!qxf7_>Mz4yy&RJ~O@mF;d`33H#>M`97XE-|K^-jEKO@B+LtnQ}q!@@CJf2
z3#Ky%UX*(b#>*o9?PPfln+3Tb!<k)BJAwUxa0LTJs-JiaMlZ<2MQTs3h`j>uWa+B3
z|KKzBT#~%TW<GXL>Iv+yU<pVIUQ?Dbm9#kkZovr96;H~;JZKF#Yz}%MNC11p3PtQH
zW+?V5K}JCYVy@FMbbKuz+`tqx9qL<Mkm}GY8OGK2Usc=ZBl5d3PqM?>?HHvcUxQR7
zyHRwPF$;#cCK~bc0!ua0IL`DfZzSRnyQsIv?7^hi(J$6=1^19Ca0x0Bp(>^B<wi9d
zh4k(G0S*&j?*^^l@cI0%gfKJ?*nb^q>0f+UPos51e{}?orz?I=vkqXYW0oyYi>?+n
zxaoGQoz<>n&5GtIzEIUFShPb{aEV17n|<jZvn2=4$6wYm>!krqSBOHmmCZ7xm-#=e
zlhe5|ROlHJbH~OC5Ov6-PH6?Y-GcI-bLfq9w#bz>SU6>k9YobTL#=!4NA-B50S?Jw
zzb1eEniReKlQg^|75$C+zauRHJWcMO>^zS64Lq3+fsmDrFqKf^-}OVnV?1$Y7W?WA
ziMWz=L2O{{ZHYx&*I5o>RtI4bD)sOf>{oSjowf1<B0Cpy!fd*V^e~Twp@0l4_;SU4
zkBt&!xFIbv{4MHwQ*@RO0yagGgez1ElOU93Xxb-1VV(0n+y-Vquf{Sj)aia{yexU@
z`3Pi{7?Q6sL`dM|))<ofJI)h<f@S<JA53vZ1rvZ%802alzZrca=LaC>b0SHqyqh)V
zPLbNnhKcsGp;xp~YK@x$bCf%fSia}oIyR~`k1|xf#^3{hP|C_V=QVX!?~3V;4N)ky
zYxQl>B6(Q3qKJskN3irT2q_guOSgAGvW~1O^`dvxaTS~5deOO!IFM~`SDQ~&?V%mi
zxaSR<wac%``wCa!1NrNW$HU*5*?6o}y0yBwIFNi}3x#{MSI$oO*kS<Nai|q=c<yO{
zEDNqb5KS83*3u+zM5djEDk!x8JTR=R62@NeK(d?i17(o7+uQx}BP1RlpQ5NPKa||0
zONQ&8%C}X6ya~W^i-eyF1yRu&O3)gT99h&?_zK{l#9hc)&`)A_BQbn@8wWm{#uQUU
zp^pOx@j`?c_2Cxdh>Ed=j}#mV)*8D92`Gpby!a%4#RYxBBfo<`x>fkjBn{&<kZIZ=
zLfVcFNZ=EcDauME{TFI8vu>Rg;MCjEk?gGoX6*Bb1=R<Ig#(mi(5$f3|CGLhC5EqY
zVmIuM*y;5)I0^q`^)0?Vq9*-Xv$8}nbz>tTnc>%t`p{s{WbOAd5WIlqT%^h1(c^Gk
z+A#hpXspE%827<1!j6pt)meoj(7v$@fOYnbmTM1)7BY<|LokKRWH1O@oLvPWH?<u|
zVW-}{uAB8ubw00oiq2Lg%x{i6)+Sih-VvpF!0X({qWaQQ?8&+6aODo^?d~<{;~N*V
z6S<=8uIV@LjwbimzRFWRN9U5XMrnbRpa>=vY$!#KB8IYA?W)wsha}k{o7GTvEtYC-
zZq?kBDYSd~w7ar<&@ZkH!4%97Y3QONDP~Pdb0)Qiew<qum6gV>Q!|T#cZBhfBwlu^
zlPdAJD`sB8L0YK*N`b1(v#(gc+jZDry{;%v6V`M}=YXd&3@Ys?3cO}4<Q~Y1zPsbF
zMdItwZ`rD-5`%P!bMxR^>|?E{!=ALF>lhV{T6w*&m)M7Y%M-0JKFku_FB{7;5hcch
z(zGf_Yu>vL`Ivs7Gw;#zMB*61YP^#lHN|oAIF&)~RL}e{*mZKOIR)&<Y_!QPNOl<J
zOFINHI5gT#MJ@2R&K^{v6;|A!K2}hBSrLYH!Gu!2;winakDw@@wC)*&EXqNY<7ib0
zn%u07LC8<uq=>P+Ka?ywMN*#|smYJr<MN~N4>qYi1SMh0{#LPn;C30%FvhJ(ehGA*
zzydO%%`&VQgiz=6aFM${6)U#j??Oib)-j|h7fB3m876Ov3?V95&|WB!+*QHU%DqQx
zB#c2Q-%%pocWXQRs7S&`_}w4pd;bdl_owy2Kb;@Gezku6*i9ui^_U?_L8{7VRT9pl
z2Gxk6%^Ee#P%Bn^JPOGex@4}qbkB{bdaH&3i+(%5OYqkF55?Tc*)*aEU!KGy;7~D1
zwv{<9ZSvP-cP$e@0tv3-bsJ>l{w|(RMS>Ilys4%pl<z@oWzS|^59DAJ-a!a$52g!N
z$F_UWMFQ;9D+}{J+RaZL6W>`kW{rwgI=L9Ec*eftvMNN@iW#Fb=T96HQ{0l)5Jan1
zK)zI<gr}3qB=n}$C`^&%$Ofo-W)k-*K_Cr-o6-c8<zv9tHjhE8mBCqShf`3yP?vEp
z)yYB{Si^M$2VIH0o{u^nlE)&-?uF?S{9#2BNKlyU!A`pDk8H4Av)s?v<+~uM@X}mH
z=DdjQZR6GkKQFY(#o|bu*@{$H7RO3~5zBcYINiXt9PoZyRuua5I|I_U(E0s2w(FZ3
z!&Z{PRl4A{Hnq+&;ut2uSU<`$b}DllGL?~UfFLVHA706-@uo6y;Pq6fkZAL|Q+Lw}
zW!$U-H7{$l>>tgCDo5kfMBKH>FWWya0Q)ii{zo{~zt9o39O4IA@jjRMNv^Tw4;%0*
zu5oAmTp#d{i_2g+$lLCA-%89@Q`CvL)?R4$Mmw4760OPuu6VEHX|>@Y$Ttjr7YB@n
zWL+PD>?nvY6kN!Mr6nFZOtnvZcf80{WwE3B!BaUi{No6j7jOsuiiGQ^?Wb|l?pESu
z*m_8DT-D=R?vX<u;XmD?q^u$((vc+sr0a?}#q>CciVp4N;sW>MiKFenzbgY3$olNE
z$iZN<Ds7M;yYKj_c=D$ioiB$nHr*k)-)X%+bO)0UEmxLPbl{;I@=1mflJzZ8EFb9B
zMII!M){CHeD(T!Knf|DOu8)1&O}{-dZ80u=bAJ@ffcmil=!wGT7YLqTL+Jb#0*4mI
zmO44k_02BD<LW1UX_9mwGhtgwb#fYViZrr5LI!Hj_lu*Hx6)Y~-@8}zsqwV{lnJSz
zj++83B`Q>JDdQyX+m227sRuoK=@7q(SzkUQgIoS&T}-MFY03j+!_jIeV+=6#`q)w-
z5``p{1zE`HH-(E((yt2!oxlW`$6yMKKI;K=-v(llN8$@HonkC!jk~q3P{j_-nDO$8
z+4`{mnku$?ifT=*OKl-Zsb&Se$Ph(q`C|twbBbLjX?zRpC8*dEzeb7FcNdB)@;JdK
zs4uCZo^l_RM7>o_YY<Y)JD;m!qmHUi0-_8rUDeOWA~ym`6-BQf<d?;i4B`9r1t*HF
zuJC`kPh5?BVa!X{RU(nsS8E_wx_|-!)U=Z1r@3NXh-DI>J9~%rQJ8`58ewD6f%k#_
zj@qaLvJ}7B!gIHwu|T7zv*)lqTh!h4aWglNr%D5<szA%+i+pSnB`!=onsO>4yzAS2
ztol-&<LkQ2-5Oz_N6@5OH^!hk#-K39z@#ytgX$<YA40=6NgMqMn)2ttdlkVn3MN2C
z;NmUuyB+C0;es1K@bGQaE);7Fm%%^06e~;G#Rtq&R&Nm1khB|Qd|tH!$z;Sg@Xf<o
zvj)BS&WN-#et3nk{loS4J<hlF9-G5|AHZOH@6o?>r+@RIwax>|<hQu_mx=G+fB5j>
z%eFQ(<<6a@UD$BM+Pc~Lv`-O8@52#+^_gG-RE_`xdi_hPudd}EkJ#Ql&Z`Yy{`Xg`
z-D|_Y3gR$c(l?#8u3M-_0%Ljbi-$suRT;}_4HR1-<`O}FWp!Oj;#ureY-K51FGa@p
z?p*k*Lvo@AuIs=7f#AR&yC9yzWab_9U5AC<h=)h*<1n<nDtK{F=wd+LFLp6l)5WBX
zJPMrsHhP~=<%8WTbSJ>6UeODkN&7u1s1NMDk=u36@9gGTR@)>C#_t48b<IQSLv)B`
zJD88_1l>F?bq6Ldc*3!am>uoz6?+N!6wTBy9v`(wTx2y65cC~ub&AOBU*GkmbX9fX
z-5Yw1`%^xS!#M4<GjDVVN{UDA$GsYY23ZDvHtP&Z3wM1t_{g!=y;?sOOnq9yZ+8u_
zb>mve2Go<Yc-iR|bIR(g<fjz8uQS*0O$^t#>pD|RCQllNM<}Bl`T}4GG%iczDK+A-
zchnSP5PI)YRbal(oL65;7rw?;KV`PG{>!gYMLBH|ItjS2zu-2yRjlceK7i979hrn>
z?mgtg>C~_lhhFDSppjmza4$q>Rc$*ZDP(Co)sEU9$JKtpR|?`=<6Yffh%e*dXhx8g
zms#K8Y5(Ro@ezJW4SG|p{mHAfOXDimSG<lp1j`%cK>uTwqa--uiA}8%tur3_2qcp`
zXy6Kf<+>6<jUxym=UjiZ2d^ik-b^K2o{KP)exIN7CI~}|L!|R>QZV<k%bG%Sousq?
zH1n$~{G{Coc=7w`!WTAnJ+{eTx}abd^GH-5Uwf)%CWzEM3{UDbpI#QQZVLIxVkXj<
zCo)2*xTFw$e@wtr$`9j`$*>>#W|jD6CHr2=ibipn7E$1-@Q3eEZLt?VjZlM}zK@RT
zOmg$|UbnOu=+x^|cy7m2b8NuOuGJ9b2}{jVuUPI)0LxDl3_;h1R*1Z-8_smP=$3Nz
ztL!SO?P_3g>f)>N9d!NYfs8l050%F`>iYu1mNH}V+_)hw6LWsJf(_(f`TM_0$(l5Q
z_16Z(9>tDtsxGpeJGHlV<mo7s)FoHwYYOx;sfjE1lBlc2wa6ZP8+t7_2vz7CxAZA~
z)<)H6<>JT)3`jBNvQjKg{ep<H=m679op*e7q;_7oJ7m4P>%mR8IIR34ur%)DzY46@
ztkzUj`X&!yf9zSwKPU<*udMx{j-;L^6gjRbs;s3v%68P&X~=M3F3v%@^Ldi3937oE
z-@wng5Z;u#_9qdBOUtUJ6l)76Ilf?o8bPH-YhB@fA8d)+UBv|r*eZSZO*PD@6!_8t
z-{`$+yjy>vP!z%zkeJFSiF*(M$;zhsyp552Rw&Jqg*7keZ{Mx<>Q)`G>A+V_-PU)g
z66%AwYl65cyr!zGixIYJZI;orHDS25+1P5^YwxC1hqktzNh(=&;I*r*Gbz3d?mof9
zAto&*m2}CzUXHQ|uY#I0lrb;5m3)@`_O03fzuEuqY5V{8_YZal%`TwL{+~7bBl79A
z2WCgS^(6j(>iN$H&Hn$*{?E<&fA0OCN&YwcKR5FK5%T}_U%u;|oxgbXDE`0q{KxHq
zCI9#F{Kx+OcC-KU=eU00nTH93-v7ed(muQccd)r}4j$}b-S)_v8>S(V$>zpuxff;c
z$BUPT?7ILQl3A7|;&5wg3Lfxcgt4nF9g6Oa;!JFvs(w$cZcn%vr$K_QQ|~$Wt~|<m
zSE~e|$N-G3xtG54<7KqD@iG|mNbuf^0PA#Ko;^Q3zdYUCxZpo8f|QHiNg%RAECS(S
z<*wYBx^kQLwXWRFcf|(iR*^Rk#)lLZQef8K?Tr=iSnOdnQCoKQ7HMeRP4$rAMP_na
z=sM)+d(12Z?j#!WISk~<WB<_IyncPsJN@hHm&a%4H=PGpA@O^vyRQ3obK?yiQMig1
zZ0tp5dx`Y}25K+TlqazWvN&C-oC$6gE4LPgU-2AX@hOR>`^4aV8ZV|YI+BCT0Mn_L
zuAI#c8FHs@lSe$G`nJfIU$EfTx4lUkK(Fe<6#RRPA%<tTFOB+W!p7b+c%XKZ&gZka
zLHST?ex;4Ct}mazx_Eu|%0i_G(iEuARVWSOtqI@^FzF?@GrmxJS@&*3f7IV`6MxcW
z^bGNPmI*eEaU&iI{i$fX9Ws!MSWdtqcRKbC+she#EU$O$q8h*?#td>G)IQb}Z+Kfb
z!#T$aDHwhkr|)p4>N~D6MK#nmc4~X)c*0Z#H~Xr7g9?Fz7)&#7k<H@tkX`@jEb~_E
zG+cRN7Tng7BDOev`M2(7V{0__e`EhQ_J4EzyV?JjuYSCEetPL<ciAtp|NHP?+5SJ+
zZO;GtESIvMH!e<(PkuOc=l({rQl_e##q)4O@NAK|$!cTWc|g|bZyS&H88@CnT@N>w
z2XufLxWZb)cj9qxY2*8Gt+O~8hWwU?_ch;Ar^h@7E2gcieZHsRy+`hC{CWp9HCJQ*
zHTGX)|26hsWB+Y^l+yhe`)|8y|Lyg6_8R-|b6gFpH~PQP|Be1{^nauOpQ!(B_xYE^
zf3~-G_DlMIXLr9j|M#<8lqcPgdC?8ZgKp$`&qfY3`oGctjs9=0M*lw$5dTH`f3RED
z|NH&M|NpJlzv2GoYOdyLuI6g4=4!6yYOdyLuI6g4=4!6yYOdeO_5T4fd)T1>Kmh<;
CX=4EZ

literal 0
HcmV?d00001

diff --git a/pyproject.toml b/pyproject.toml
new file mode 100644
index 0000000..8cf3256
--- /dev/null
+++ b/pyproject.toml
@@ -0,0 +1,3 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
\ No newline at end of file