Spaces:

Michelangiolo
/

startup-finder

Runtime error

App Files Files Community

Michelangiolo commited on Apr 1, 2023

Commit

9fa587a

•

1 Parent(s): 942976d

v2

Browse files

Files changed (3) hide show

app.py +59 -17
data_manipulation.ipynb +416 -0
df_encoded.parquet +2 -2

app.py CHANGED Viewed

@@ -8,35 +8,77 @@ from sentence_transformers import SentenceTransformer
 model = SentenceTransformer('all-mpnet-base-v2') #all-MiniLM-L6-v2 #all-mpnet-base-v2
 df = pd.read_parquet('df_encoded.parquet')
-df.columns = [['name', 'description', 'year', 'target', 'size', 'stage', 'raised', 'tags', 'text_vector_']]
-#if parsing from a parquet, I have a list of array that does not want to get changed
-df_knn = [x[0].tolist() for x in df['text_vector_'].values.tolist()]
 df = df.reset_index(drop=True)
 from sklearn.neighbors import NearestNeighbors
-import numpy as np
 import pandas as pd
 from sentence_transformers import SentenceTransformer
-#prepare model
-nbrs = NearestNeighbors(n_neighbors=20, algorithm='ball_tree').fit(df_knn)
-def search(query):
     product = model.encode(query).tolist()
     # product = df.iloc[0]['text_vector_'] #use one of the products as sample
     distances, indices = nbrs.kneighbors([product]) #input the vector of the reference object
     #print out the description of every recommended product
-    return df.iloc[list(indices)[0]][['name', 'description', 'year', 'target', 'size', 'stage', 'raised', 'tags']]
 #the first module becomes text1, the second module file1
-def greet(text1):
-    return search(text1)
-iface = gr.Interface(
-    fn=greet,
-    inputs=['text'],
-    outputs=["dataframe"]
-)
-iface.launch(share=False)

 model = SentenceTransformer('all-mpnet-base-v2') #all-MiniLM-L6-v2 #all-mpnet-base-v2
 df = pd.read_parquet('df_encoded.parquet')
+df['tags'] = df['tags'].apply(lambda x : str(x))
+def parse_raised(x):
+    if x == 'Undisclosed':
+        return 0
+    else:
+        quantifier = x[-1]
+        x = float(x[1:-1])
+        if quantifier == 'K':
+            return x/1000
+        elif quantifier == 'M':
+            return x
+df['raised'] = df['raised'].apply(lambda x : parse_raised(x))
 df = df.reset_index(drop=True)
 from sklearn.neighbors import NearestNeighbors
 import pandas as pd
 from sentence_transformers import SentenceTransformer
+def filter_df(df, column_name, filter_type, filter_value):
+    if filter_type == '==':
+        df_filtered = df[df[column_name]==filter_value]
+    elif filter_type == '>=':
+        df_filtered = df[df[column_name]>=filter_value]
+    elif filter_type == '<=':
+        df_filtered = df[df[column_name]<=filter_value]
+    elif filter_type == 'contains':
+        df_filtered = df[df['target'].str.contains(filter_value)]
+    return df_filtered
+def search(df, query):
     product = model.encode(query).tolist()
     # product = df.iloc[0]['text_vector_'] #use one of the products as sample
+    #prepare model
+    nbrs = NearestNeighbors(n_neighbors=20, algorithm='ball_tree').fit(df['text_vector_'].values.tolist())
     distances, indices = nbrs.kneighbors([product]) #input the vector of the reference object
     #print out the description of every recommended product
+    return df.iloc[list(indices)[0]][['name', 'description', 'raised', 'year', 'target', 'size', 'stage', 'tags']]
 #the first module becomes text1, the second module file1
+def greet(size, target, raised, query):
+    df_size = filter_df(df, 'size', '==', size)
+    df_target = filter_df(df_size, 'target', 'contains', target)
+    def raised_zero(x):
+        if x == 0:
+            return 'Undisclosed'
+        else:
+            return x
+    print('a')
+    df_raised = df_target[(df_target['raised'] >= raised) | (df_target['raised'] == 0)]
+    df_knn = search(df_raised, query)
+    #we live the sorting for last
+    df_knn = df_knn.sort_values('raised', ascending=False)
+    df_knn['raised'] = df_knn['raised'].apply(lambda x : raised_zero(x))
+    return df_knn
+with gr.Blocks(theme=gr.themes.Soft(primary_hue='amber', secondary_hue='gray', neutral_hue='amber')) as demo:
+    gr.Markdown(
+    """
+    # Startup Search Engine
+    """
+    )
+    size = gr.Radio(['1-10', '11-50', '51-200', '201-500', '500+'], multiselect=False, value='11-50', label='size')
+    target = gr.Radio(['B2B', 'B2C', 'B2G', 'B2B2C'], value='B2B', multiselect=False, label='target')
+    raised = gr.Slider(0, 20, value=5, step_size=1, label="Minimum raising (in Millions)")
+    query = gr.Textbox(label='Describe the Startup you are searching for', value='age reversing')
+    btn = gr.Button(value="Search for a Startup")
+    output1 = gr.DataFrame(label='value')
+    # btn.click(greet, inputs='text', outputs=['dataframe'])
+    btn.click(greet, [size, target, raised, query], [output1])
+demo.launch(share=False)

data_manipulation.ipynb ADDED Viewed

	@@ -0,0 +1,416 @@

+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 78,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<div>\n",
+       "<style scoped>\n",
+       "    .dataframe tbody tr th:only-of-type {\n",
+       "        vertical-align: middle;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe tbody tr th {\n",
+       "        vertical-align: top;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe thead th {\n",
+       "        text-align: right;\n",
+       "    }\n",
+       "</style>\n",
+       "<table border=\"1\" class=\"dataframe\">\n",
+       "  <thead>\n",
+       "    <tr style=\"text-align: right;\">\n",
+       "      <th></th>\n",
+       "      <th>name</th>\n",
+       "      <th>description</th>\n",
+       "      <th>year</th>\n",
+       "      <th>target</th>\n",
+       "      <th>size</th>\n",
+       "      <th>stage</th>\n",
+       "      <th>raised</th>\n",
+       "      <th>tags</th>\n",
+       "      <th>text_vector_</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>0</th>\n",
+       "      <td>0.10 of a Second</td>\n",
+       "      <td>Smart Indicators for Connected Vehicles</td>\n",
+       "      <td>2019.0</td>\n",
+       "      <td>B2B</td>\n",
+       "      <td>1-10</td>\n",
+       "      <td>Pre-Funding</td>\n",
+       "      <td>Undisclosed</td>\n",
+       "      <td>[connected-vehicles, adas, autonomous-vehicles...</td>\n",
+       "      <td>[-0.031224824488162994, -0.06342269480228424, ...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>1</th>\n",
+       "      <td>12trix</td>\n",
+       "      <td>Math Lessons for Young Kids</td>\n",
+       "      <td>2012.0</td>\n",
+       "      <td>B2B, B2C</td>\n",
+       "      <td>1-10</td>\n",
+       "      <td>Pre-Funding</td>\n",
+       "      <td>Undisclosed</td>\n",
+       "      <td>[sdg, schools, pre-k, serious-games, games, mo...</td>\n",
+       "      <td>[-0.038649097084999084, 0.028091922402381897, ...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>2</th>\n",
+       "      <td>1E Therapeutics</td>\n",
+       "      <td>Novel RNA-targeting Drugs</td>\n",
+       "      <td>2021.0</td>\n",
+       "      <td>B2B</td>\n",
+       "      <td>51-200</td>\n",
+       "      <td>Seed</td>\n",
+       "      <td>$120M</td>\n",
+       "      <td>[pharmaceuticals, chronic-disease, immunology,...</td>\n",
+       "      <td>[0.04561534896492958, -0.017776092514395714, 0...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>3</th>\n",
+       "      <td>1MRobotics</td>\n",
+       "      <td>Retail Automation Solutions with Nano Fulfillment</td>\n",
+       "      <td>2021.0</td>\n",
+       "      <td>B2B</td>\n",
+       "      <td>11-50</td>\n",
+       "      <td>A</td>\n",
+       "      <td>$25M</td>\n",
+       "      <td>[omni-channel, ecommerce, climate-tech, artifi...</td>\n",
+       "      <td>[0.0024080690927803516, -0.03042100928723812, ...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>4</th>\n",
+       "      <td>1touch.io</td>\n",
+       "      <td>Personal Data Flow Tracking and Data Cataloging</td>\n",
+       "      <td>2017.0</td>\n",
+       "      <td>B2B</td>\n",
+       "      <td>51-200</td>\n",
+       "      <td>A</td>\n",
+       "      <td>$16.1M</td>\n",
+       "      <td>[enterprise-solutions, data-protection, cyber-...</td>\n",
+       "      <td>[-0.01007091999053955, 0.10431888699531555, -0...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>...</th>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>4981</th>\n",
+       "      <td>YOW HR</td>\n",
+       "      <td>Human Resources Engagement and Optimization Pl...</td>\n",
+       "      <td>2020.0</td>\n",
+       "      <td>B2B, B2B2C</td>\n",
+       "      <td>1-10</td>\n",
+       "      <td>Pre-Funding</td>\n",
+       "      <td>Undisclosed</td>\n",
+       "      <td>[content-creators, e-learning, software-applic...</td>\n",
+       "      <td>[0.026961881667375565, 0.002459645736962557, -...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>4982</th>\n",
+       "      <td>Yummi Home Food</td>\n",
+       "      <td>Marketplace for Homemade Food</td>\n",
+       "      <td>2012.0</td>\n",
+       "      <td>B2C</td>\n",
+       "      <td>11-50</td>\n",
+       "      <td>Pre-Funding</td>\n",
+       "      <td>Undisclosed</td>\n",
+       "      <td>[ecommerce, p2p, delivery, online-shopping, ma...</td>\n",
+       "      <td>[0.0036857957020401955, 0.03582162782549858, -...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>4983</th>\n",
+       "      <td>Yung-Etgar</td>\n",
+       "      <td>Custom Mechanized Harvesting Systems</td>\n",
+       "      <td>1982.0</td>\n",
+       "      <td>B2B</td>\n",
+       "      <td>51-200</td>\n",
+       "      <td>Mature</td>\n",
+       "      <td>Undisclosed</td>\n",
+       "      <td>[crops, agtech, harvesting, machinery, sdg, cl...</td>\n",
+       "      <td>[0.027293115854263306, 0.010461761616170406, 0...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>4984</th>\n",
+       "      <td>YuviTal</td>\n",
+       "      <td>Digital Health and Fitness Solutions for Organ...</td>\n",
+       "      <td>2017.0</td>\n",
+       "      <td>B2B, B2C, B2G</td>\n",
+       "      <td>11-50</td>\n",
+       "      <td>Pre-Funding</td>\n",
+       "      <td>Undisclosed</td>\n",
+       "      <td>[fitness, digital-wallet, discount, mobile-app...</td>\n",
+       "      <td>[0.02851911261677742, 0.05474231392145157, -0....</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>4985</th>\n",
+       "      <td>Z-square</td>\n",
+       "      <td>Microendoscope for Minimally Invasive Imaging ...</td>\n",
+       "      <td>2013.0</td>\n",
+       "      <td>B2B</td>\n",
+       "      <td>11-50</td>\n",
+       "      <td>Seed</td>\n",
+       "      <td>$10M</td>\n",
+       "      <td>[endoscopy, medical-devices, minimally-invasiv...</td>\n",
+       "      <td>[0.012587728910148144, -0.07959864288568497, -...</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "<p>4986 rows × 9 columns</p>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "                  name                                        description  \\\n",
+       "0     0.10 of a Second           Smart Indicators for Connected Vehicles    \n",
+       "1               12trix                        Math Lessons for Young Kids   \n",
+       "2      1E Therapeutics                          Novel RNA-targeting Drugs   \n",
+       "3           1MRobotics  Retail Automation Solutions with Nano Fulfillment   \n",
+       "4            1touch.io    Personal Data Flow Tracking and Data Cataloging   \n",
+       "...                ...                                                ...   \n",
+       "4981            YOW HR  Human Resources Engagement and Optimization Pl...   \n",
+       "4982   Yummi Home Food                      Marketplace for Homemade Food   \n",
+       "4983        Yung-Etgar               Custom Mechanized Harvesting Systems   \n",
+       "4984           YuviTal  Digital Health and Fitness Solutions for Organ...   \n",
+       "4985          Z-square  Microendoscope for Minimally Invasive Imaging ...   \n",
+       "\n",
+       "        year         target    size        stage       raised  \\\n",
+       "0     2019.0            B2B    1-10  Pre-Funding  Undisclosed   \n",
+       "1     2012.0       B2B, B2C    1-10  Pre-Funding  Undisclosed   \n",
+       "2     2021.0            B2B  51-200         Seed        $120M   \n",
+       "3     2021.0            B2B   11-50            A         $25M   \n",
+       "4     2017.0            B2B  51-200            A       $16.1M   \n",
+       "...      ...            ...     ...          ...          ...   \n",
+       "4981  2020.0     B2B, B2B2C    1-10  Pre-Funding  Undisclosed   \n",
+       "4982  2012.0            B2C   11-50  Pre-Funding  Undisclosed   \n",
+       "4983  1982.0            B2B  51-200       Mature  Undisclosed   \n",
+       "4984  2017.0  B2B, B2C, B2G   11-50  Pre-Funding  Undisclosed   \n",
+       "4985  2013.0            B2B   11-50         Seed         $10M   \n",
+       "\n",
+       "                                                   tags  \\\n",
+       "0     [connected-vehicles, adas, autonomous-vehicles...   \n",
+       "1     [sdg, schools, pre-k, serious-games, games, mo...   \n",
+       "2     [pharmaceuticals, chronic-disease, immunology,...   \n",
+       "3     [omni-channel, ecommerce, climate-tech, artifi...   \n",
+       "4     [enterprise-solutions, data-protection, cyber-...   \n",
+       "...                                                 ...   \n",
+       "4981  [content-creators, e-learning, software-applic...   \n",
+       "4982  [ecommerce, p2p, delivery, online-shopping, ma...   \n",
+       "4983  [crops, agtech, harvesting, machinery, sdg, cl...   \n",
+       "4984  [fitness, digital-wallet, discount, mobile-app...   \n",
+       "4985  [endoscopy, medical-devices, minimally-invasiv...   \n",
+       "\n",
+       "                                           text_vector_  \n",
+       "0     [-0.031224824488162994, -0.06342269480228424, ...  \n",
+       "1     [-0.038649097084999084, 0.028091922402381897, ...  \n",
+       "2     [0.04561534896492958, -0.017776092514395714, 0...  \n",
+       "3     [0.0024080690927803516, -0.03042100928723812, ...  \n",
+       "4     [-0.01007091999053955, 0.10431888699531555, -0...  \n",
+       "...                                                 ...  \n",
+       "4981  [0.026961881667375565, 0.002459645736962557, -...  \n",
+       "4982  [0.0036857957020401955, 0.03582162782549858, -...  \n",
+       "4983  [0.027293115854263306, 0.010461761616170406, 0...  \n",
+       "4984  [0.02851911261677742, 0.05474231392145157, -0....  \n",
+       "4985  [0.012587728910148144, -0.07959864288568497, -...  \n",
+       "\n",
+       "[4986 rows x 9 columns]"
+      ]
+     },
+     "execution_count": 78,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "import pandas as pd\n",
+    "\n",
+    "df = pd.read_parquet('df_encoded.parquet')\n",
+    "df"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 89,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import os\n",
+    "os.system('pip install openpyxl')\n",
+    "os.system('pip install sentence-transformers')\n",
+    "import pandas as pd\n",
+    "import gradio as gr\n",
+    "from sentence_transformers import SentenceTransformer\n",
+    "\n",
+    "model = SentenceTransformer('all-mpnet-base-v2') #all-MiniLM-L6-v2 #all-mpnet-base-v2\n",
+    "\n",
+    "df = pd.read_parquet('df_encoded.parquet')\n",
+    "df['tags'] = df['tags'].apply(lambda x : str(x))\n",
+    "def parse_raised(x):\n",
+    "    if x == 'Undisclosed':\n",
+    "        return 0\n",
+    "    else: \n",
+    "        quantifier = x[-1]\n",
+    "        x = float(x[1:-1])\n",
+    "        if quantifier == 'K':\n",
+    "            return x/1000\n",
+    "        elif quantifier == 'M':\n",
+    "            return x\n",
+    "df['raised'] = df['raised'].apply(lambda x : parse_raised(x))\n",
+    "df = df.reset_index(drop=True)\n",
+    "\n",
+    "from sklearn.neighbors import NearestNeighbors\n",
+    "import numpy as np\n",
+    "import pandas as pd\n",
+    "from sentence_transformers import SentenceTransformer"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 92,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "c:\\Users\\ardit\\AppData\\Local\\Programs\\Python\\Python39\\lib\\site-packages\\gradio\\deprecation.py:43: UserWarning: You have unused kwarg parameters in Radio, please remove them: {'multiselect': False}\n",
+      "  warnings.warn(\n",
+      "c:\\Users\\ardit\\AppData\\Local\\Programs\\Python\\Python39\\lib\\site-packages\\gradio\\deprecation.py:43: UserWarning: You have unused kwarg parameters in Slider, please remove them: {'step_size': 1}\n",
+      "  warnings.warn(\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Running on local URL:  http://127.0.0.1:7896\n",
+      "\n",
+      "To create a public link, set `share=True` in `launch()`.\n"
+     ]
+    },
+    {
+     "data": {
+      "text/html": [
+       "<div><iframe src=\"http://127.0.0.1:7896/\" width=\"100%\" height=\"500\" allow=\"autoplay; camera; microphone; clipboard-read; clipboard-write;\" frameborder=\"0\" allowfullscreen></iframe></div>"
+      ],
+      "text/plain": [
+       "<IPython.core.display.HTML object>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "text/plain": []
+     },
+     "execution_count": 92,
+     "metadata": {},
+     "output_type": "execute_result"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "a\n"
+     ]
+    }
+   ],
+   "source": [
+    "def filter_df(df, column_name, filter_type, filter_value):\n",
+    "    if filter_type == '==':\n",
+    "        df_filtered = df[df[column_name]==filter_value]\n",
+    "    elif filter_type == '>=':\n",
+    "        df_filtered = df[df[column_name]>=filter_value]\n",
+    "    elif filter_type == '<=':\n",
+    "        df_filtered = df[df[column_name]<=filter_value]\n",
+    "    elif filter_type == 'contains':\n",
+    "        df_filtered = df[df['target'].str.contains(filter_value)]\n",
+    "    return df_filtered\n",
+    "\n",
+    "def search(df, query):\n",
+    "    product = model.encode(query).tolist()\n",
+    "    # product = df.iloc[0]['text_vector_'] #use one of the products as sample\n",
+    "\n",
+    "    #prepare model\n",
+    "    nbrs = NearestNeighbors(n_neighbors=20, algorithm='ball_tree').fit(df['text_vector_'].values.tolist())\n",
+    "\n",
+    "    distances, indices = nbrs.kneighbors([product]) #input the vector of the reference object\n",
+    "\n",
+    "    #print out the description of every recommended product\n",
+    "    return df.iloc[list(indices)[0]][['name', 'description', 'raised', 'year', 'target', 'size', 'stage', 'tags']]\n",
+    "\n",
+    "#the first module becomes text1, the second module file1\n",
+    "def greet(size, target, raised, query): \n",
+    "    df_size = filter_df(df, 'size', '==', size)\n",
+    "    df_target = filter_df(df_size, 'target', 'contains', target)\n",
+    "    def raised_zero(x):\n",
+    "        if x == 0:\n",
+    "            return 'Undisclosed'\n",
+    "        else:\n",
+    "            return x\n",
+    "    print('a')\n",
+    "    df_raised = df_target[(df_target['raised'] >= raised) | (df_target['raised'] == 0)]\n",
+    "    df_knn = search(df_raised, query)\n",
+    "    #we live the sorting for last\n",
+    "    df_knn = df_knn.sort_values('raised', ascending=False)\n",
+    "    df_knn['raised'] = df_knn['raised'].apply(lambda x : raised_zero(x))\n",
+    "\n",
+    "    return df_knn\n",
+    "\n",
+    "with gr.Blocks(theme=gr.themes.Soft(primary_hue='amber', secondary_hue='gray', neutral_hue='amber')) as demo:\n",
+    "    gr.Markdown(\n",
+    "    \"\"\"\n",
+    "    # Gradio with History\n",
+    "    \"\"\"\n",
+    "    )\n",
+    "    size = gr.Radio(['1-10', '11-50', '51-200', '201-500', '500+'], multiselect=False, value='11-50', label='size')\n",
+    "    target = gr.Radio(['B2B', 'B2C', 'B2G', 'B2B2C'], value='B2B', multiselect=False, label='target')\n",
+    "    raised = gr.Slider(0, 20, value=5, step_size=1, label=\"Minimum raising (in Millions)\")\n",
+    "    query = gr.Textbox(label='Describe the Startup you are searching for', value='age reversing')\n",
+    "    btn = gr.Button(value=\"Search for a Startup\")\n",
+    "    output1 = gr.DataFrame(label='value')\n",
+    "    # btn.click(greet, inputs='text', outputs=['dataframe'])\n",
+    "    btn.click(greet, [size, target, raised, query], [output1])\n",
+    "demo.launch(share=False)"
+   ]
+  }
+ ],
+ "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": "3.9.13"
+  },
+  "orig_nbformat": 4
+ },
+ "nbformat": 4,
+ "nbformat_minor": 2
+}

df_encoded.parquet CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0f3ea0cc9455a2bb2a2ee793b9588e85c5df7ad4e9e04e363d45747092e93422
-size 24419137

 version https://git-lfs.github.com/spec/v1
+oid sha256:f7f7bb66d15c188968839a3464fb2e5352cd0f4cd6bee7306ed5d9b3ecff4fbe
+size 24191629